Proceedings of the Annual Meeting of the Cognitive Science Society

The main goal of this full-day workshop is to bring togetherresearchers from several distinct fields: behavioralpsychologists studying language acquisition, speechtechnology researchers, linguists, and computationalmodelers of cognitive development. These groups arebroadly interested in the same questions, i.e. what is thenature of speech and language, and how might a systemlearn to process it in supervised or unsupervised ways?Since the groups interested in these questions work ondifferent analysis levels, cross-pollination has been sparse.Recent technological innovations have made collectinglong naturalistic recordings of children’s home environmentfar simpler than in the past. However, the raw output of suchrecordings is not immediately usable for most analyses.Simultaneously, speech technology (ST) and machinelearning tools have improved immensely over the pastdecade, making it feasible to use such tools withincreasingly diverse and noise-laden data. Relatedly,cognitively viable computational models have made recentstrides in explaining learning and development, but fewsuch models can be applied to novel data-sets withoutencountering many hurdles about translatability acrossframeworks. This workshop brings together experts from allof these areas, and seeks to build bridges across them, withinsight from other similar interdisciplinary efforts in otherareas of cognitive science. Talks will discuss the matchbetween the theory-driven questions researchers would liketo ask, and the answers the current state of the art allows.The program committee is part of a newly formed groupcalled DARCLE (Daylong Audio Recordings of Children’sLanguage Environment); with the help of an NSF grant,DARCLE has created a repository called HomeBank forraw data, metadata, and analysis/processing tools for long-form recordings of child language. This workshop is anopportunity to network with related efforts in Europe, andfor a talk and demo of a related effort, the NSF-fundedSpeech Recognition Virtual Kitchen

Everyday interactions often depend on thinking about spaceand time: collaborators need to know where events takeplace – and in what order – to, e.g., communicate drivingdirections, build pieces of furniture, or carry out strategicoperations in military and sports settings (Núñez &Cooperrider, 2013). A simple set of driving directions mayrequire a listener to interpret and reason about the spatialrelations – such as next to and behind – and the temporalrelations – such as after and during – that a speakerdescribes. The speaker may also use gestures to substitute,supplement, or disambiguate linguistic descriptions (Holle& Gunter, 2007; Perzanowski, Schultz, & Williams, 1998).Such rapid, rich, and productive interactions are transientand difficult to analyze behaviorally, and so they pose achallenge for experimenters. They are grounded in thephysical world, and accordingly challenge computationalmodels that cannot digest rich perceptual and environmentalinput in real time. Robotic systems are geared towardsprocessing and acting upon the physical world – and theyincreasingly support human-robotic interaction (e.g., Fonget al., 2006; Kawamura et al., 2003; Kortenkamp et al.,1999). But they, too, are uniquely challenged in maintainingproductive interactive exchanges with human teammates,because they must be tolerant of human idiosyncrasies,preferences, limitations, and errors (Trafton et al., 2013).Because these challenges cut across broad interests incognitive science – such as linguistics, artificial intelligence,robotics, and psychology – progress is unlikely without theengagement of multiple approaches, from psychologicalexperimentation to the construction of autonomous,embodied systems. In recent years, progress towardsunderstanding interactive spatiotemporal cognition hasaccelerated along parallel paths: there exist new behavioraland imaging methodologies to study event segmentation(e.g., Radvansky & Zacks, 2014), spatial inference (e.g.,Knauff & Ragni, 2013), and gestural cognition (e.g.,Novack et al., 2016); novel computational theories ofunderstanding physical reasoning (e.g., Battaglia et al.,2013) and mental simulation (e.g., Khemlani & Johnson-Laird, 2013); cognitive architectures that support richinteractivity (Huffman & Laird, 2014; Trafton et al., 2013);and a wide variety of technological platforms on which totransform theory into embodied interaction.The goal of the workshop is to allow these parallelapproaches to converge. Discussants will share recent dataand theory, consider novel architectural approaches, anddemonstrate burgeoning technological advances thatadvance the science of spatiotemporal inference. Theworkshop will promote interdisciplinary collaboration byfocusing on three unifying themes

Meta-analysis is a powerful yet underused tool in cognitive sci-ence. It allows researchers to leverage entire bodies of litera-ture to get a broad and quantitative overview of a particularphenomenon, thereby promoting theory development, and tomake more precise estimates of effect sizes, which enablesrobust planning of prospective studies (e.g. through power-analyses). In this tutorial, we will introduce meta-analysis as atool with which to inform everyday research, and provide par-ticipants with hands-on experience conducting their own meta-analysis. We will also present an online platform we have de-veloped for conducting meta-analyses in the field of languagedevelopment: MetaLab (http://metalab.stanford.edu).

Humans have drawn sequential images as a means ofexpression throughout history, from cave paintings andfrescoes to wall-carvings and tapestries (McCloud, 1993). Incontemporary society, we find them most prevalently incomics of the world, and over the past two decades,increasing attention has turned to this communicativesystem in the cognitive sciences.Earlier work often focused on theory alone, drawing fromparadigms in linguistics or semiotics (for review, see Cohn,2012; Wildfeuer & Bateman, 2016) or from theoristsoutside academia (e.g., McCloud, 1993). However, newerstudies test theoretical predictions with empirical corpusanalyses and both behavioral and neurocognitiveexperimentation. As in language research, combining thesemethodologies provides converging evidence on thestructure of visual narratives, their diversity across theworld, and their comprehension by minds and brains.Recent research has especially focused on the overlappingcognition between the processing of the “visual languages”constituting drawn visual narratives and the linguisticsystems of verbal and signed languages (Cohn, 2013;Magliano, Larson, Higgs, & Loschky, 2015). Thesepresentations further such analyses, and explore questionsrelated to the degree to which these visual languages sharemechanisms with linguistic and other cognitive systems.

Does neuroscience have the potential to inform education? In this debate, three participants will describe emerging results in neuroscience with translatable links to learning and memory. Experts in learning sciences will discuss connections as well as limits in the ways that neuroscience can inform education at the individual and classroom levels. Although significant progress has been made in our understanding of how the brain learns and remembers, the question for this symposium is whether this progress does or can provide a direct bridge from brain science to education.

The past 15-20 years have witnessed a particularly strong interest in our ability to rapidly extract structured information from the environment. This fundamental process of human cognition is widely believed to underpin many complex behaviors – from language development and social interaction to intuitive decision making and music cognition – so this interest spans practically all branches of cognitive science. Research on this topic can be found in two related, yet traditionally distinct research strands, namely "implicit learning" (Reber, 1967) and "statistical learning" (Saffran, Aslin, & Newport, 1996). Both lines of research focus on how we acquire information from complex stimulus domains and both rely heavily on the use of artificial systems (e.g., finite-state grammars, pseudoword lexicons). In typical experiments, participants are initially exposed to stimuli generated by an artificial system and then tested to determine what they have learned. Given these and other significant similarities, Perruchet and Pacton (2006) argue that these distinct lines of research actually represent two approaches to a single phenomenon, and Conway and Christiansen (2006) propose combining the two in name: "implicit-statistical learning". Yet, despite frequent acknowledgements that researchers in implicit learning and statistical learning might essentially be looking at the same phenomenon, there is surprisingly little alignment between the two strands. This symposium seeks to remedy this situation by bringing together leading researchers from both areas in order to promote a shared understanding of research questions and methodologies, to discuss similarities and differences between the two approaches, and to work towards a joint research agenda. The symposium comprises four presentations, followed by a thematic discussion, which provide coverage of these phenomena in terms of development (children and adults), different language learning tasks (sublexical phonotactics, word acquisition, grammar learning), and their role in both production and comprehension, each integrating multidisciplinary perspectives. Gomez focuses on implicit-statistical learning in early development, identifying words and grammatical sequences and the memory systems that underlie this learning. Monaghan and Rebuschat measure word learning and grammar learning in adults, while varying the knowledge that participants have of the structure they are acquiring. Dell and Anderson demonstrate how their work on acquisition of phonotactic constraints is exhibited in speakers’ productions, and discuss the inter-relation in speech between implicit and statistical learning. Finally, Conway provides an overview of the two fields, and proposes a novel framework that unifies implicit learning and statistical learning.

Do the discoveries of cognitive science generalize beyondartificial lab experiments? Or do they have little hope ofhelping us to understand real-world events? Fretting on thisquestion, I bought a copy of the Wall Street Journal andfound that the three front page headlines each connect tomy own research on explanatory reasoning. I report tests ofthe phenomena of inferred evidence, belief digitization, andrevealed truth in real-world contexts derived from theheadlines. If my own corner of cognitive science has suchexplanatory relevance to the real world, then cognitivescience as a whole must be in far better shape yet.

This paper examines the ability of a dual-system, formal modelof categorization COVIS (Ashby, Paul & Maddox, 2011) topredict the learning performance of participants on the six cat-egory structures described in Shepard, Hovland and Jenkin’s(1961) seminal study. COVIS assumes that category learningis mediated by two dissociable neural systems that compete tocontrol responding. The verbal system explicitly tests verbal-izable rules, whereas the implicit system gradually associateseach stimulus with the appropriate response. Although COVISis highly influential, there are no published evaluations of theformal model against classic category learning data (COVIS ismost typically applied heuristically to the design of new exper-iments). In the current paper, we begin to address this gap inthe literature. Specifically, we demonstrate that COVIS is ableto accommodate the ordinal pattern found by Shepard et al.,provided that adjustments consistent with the model’s theoret-ical framework are made.

The predictive performance equation (PPE) is a mathematical model of learning and retention that attempts to capitalize on the regularities seen in human learning to predict future performance. To generate predictions, PPE’s free parameters must be calibrated to a minimum amount of historical performance data, leaving PPE unable to generate valid predictions for initial learning events. We examined the feasibility of using the data from other individuals, who performed the same task in the past, to inform PPE’s free parameters for new individuals (prior-informed predictions). This approach could enable earlier and more accurate performance predictions. To assess the predictive validity of this methodology, the accuracy of PPE’s individualized and prior-informed predictions before the point in time where PPE can be fully calibrated using an individual’s unique performance history. Our results show that the prior data can be used to inform PPE’s free parameters, allowing earlier performance predictions to be made.

We implemented the connectionist model of social-pragmaticword learning (Caza & Knott, 2012) to test the hypothesis thatreduced joint attention between infant and mother wouldincrease the difference in acquisition between nouns andverbs as observed in Autistic Spectrum Disorder (ASD). Theratio of objects to actions in the observed event stream wasmanipulated to create an original noun-verb asymmetry. Tensimulations were run for each of the combinations of threeconditions of communicative reliability and two conditions ofunfiltered random associative learning, which is regarded bysome researchers as the primary mechanism of languagelearning in ASD. The simulations indicated that the reductionin the reliability of communicative actions does not lead toincreased noun-verb asymmetry within the originally plannedtraining epochs. A trend in the predicted direction appearedtoward the end of training, suggesting that further simulationsmay help resolve the issue within the current architecture.

Monolinguals and bilinguals differ along a number ofdimensions, including way they label existing objectcategories (Pavlenko & Malt, 2011). In the present study, weask whether English monolinguals, Spanish-Englishbilinguals, and English-Spanish bilinguals also differ in theway they use language when forming novel categories.Previous research with monolinguals shows that a sharedlabel encourages children (e.g., Waxman & Markow, 1995)and adults (e.g., Lupyan, Rakison, & McClelland, 2007) toplace objects together. Our results further demonstrate thatwhen two objects shared a Licit Word label like “zeg,”monolinguals and bilinguals alike are encouraged to groupthem together. Illicit Words like “gxz,” on the other hand,only influence the categorization decisions of bilinguals.Thus, bilinguals appear to be more flexible in their use oflinguistic information in categorization. Neither group madeuse of non-linguistic cues (patterned frames), suggesting aunique role for language in category formation.

In this study we investigate the effect of modeling type and reflection on the acquisition of dart-throwing skills, self- efficacy beliefs and self-reaction scores by replicating a study by Kitsantas, Zimmerman, and Cleary (2000). Participants observing a coping model were expected to surpass participants observing a mastery model who in turn were expected to outperform participants who learned without a model. Reflection was hypothesized to have a positive effect. Ninety undergraduate students were tested three times on dart-throwing skills, self-efficacy beliefs, and self-reaction scores. Contrary to what was expected, we found no main effects of modeling type and reflection. No interaction effects were found either. There was an effect of trial, indicating that participants improved dart-throwing skills, self-efficacy beliefs, and self-reaction scores over time. Furthermore, self- efficacy beliefs and dart-throwing skill were highly correlated. Our results suggest that learners do not benefit from observing a model and reflecting, but practice makes perfect.

In social-dilemma situations (public-good games) people may pursue their local, egoistic interests and thereby lower the global, overall payoff of their group and, paradoxically, even their own resulting payoff. One may also speak of intra- individual dilemmas, where people pursue local goals at the expense of their overall utility. Our current experiments transfer this idea to the context of personnel evaluation and personnel selection. In our experiments, participants were put in the position of a Human Resources manager, who should for instance select workers who optimize the overall payoff of the company, rather than those who optimize only their specific payoffs. The results of the experiments, however, suggest that most, albeit not all, participants tended to focus on directly comparing individuals without considering the overall contribution to a group. Thus employees with the best overall effects for a company or organization may be evaluated the most negatively. This possible ‘tragedy of personnel evaluation’ may be linked to maladaptive incentive structures (personnel evaluation), advancement of employees (personnel promotion) and job offers (personnel selection), and may have a substantial negative impact on the effectiveness of companies or organizations.

Verbs and prepositions pose significant challenges in second language learning, as languages differ in how they map these relational terms onto events. Second language learners must put aside their language-specific lens to uncover how a new language operates, perhaps having to rediscover semantic distinctions typically ignored in the first language. The current study examines how the acquisition of these novel mappings are affected by characteristics of the learner and of the language to be learned. English monolinguals and Dutch- English bilinguals learned novel terms that corresponded to containment and support relations of either English, Dutch, or Japanese. Results show that English distinctions are learned best across groups, potentially reflecting predispositions in human cognition. No differences were found between monolinguals and bilinguals in any language condition. The characteristics of the language to be learned appear to play a prominent role in the acquisition of novel semantic categories.

In two experiments we manipulated the prior probabilityof occurrence for two alternatives. After a first learningsession, in a second session the cue to bias the decisionwas reversed. Our investigation shows that subjects areable to learn the reverse bias only when the bias of the firstsession is in line with their expected outcome. When, dur-ing the first session, the actual outcome of the bias is notin line with the expected outcome, there is an inhibitionfor the reversal bias learning in the second session. Weinvestigate this phenomenon with computational modelsof choice showing that the inhibition of reversal is due toan increase in the rate at which subjects accumulate evi-dence for repeated, unexpected stimuli. We discuss a pos-sible theoretical explanation that links this phenomenonto similar results found in the literature on reversal learn-ing and to the effect of novelty on learning.

There is considerable evidence showing that age ofacquisition (AoA) is an important factor influencing lexicalprocessing. Early-learned words tend to be processed morequickly compared to later-learned words. The effect could bedue to the gradual reduction in plasticity as more words arelearned. Alternatively, it could originate from differenceswithin semantic representations. We implemented the trianglemodel of reading including orthographic, phonological andsemantic processing layers, and trained it according toexperience of a language learner to explore the AoA effects inboth naming and lexical decision. Regression analyses on themodel’s performance showed that AoA was a reliablepredictor of naming and lexical decision performance, and theeffect size was larger for lexical decision than for naming.The modelling results demonstrate that AoA operatesdifferentially on concrete and abstract words, indicating thatboth the mapping and the representation accounts of AoAwere contributing to the model’s performance.

When finding a best explanation for observed symptoms a multitude of information has to be integrated and matched against explanations stored in memory. Although assumptions about ongoing memory processes can be derived from the process models, little process data exists that would allow to sufficiently test these assumptions. In order to explore memory processes in diagnostic reasoning, 29 participants were asked to solve a visual reasoning task (the Black Box paradigm) where critical information had to be retrieved from memory. This study focused on differentiating between processes that take place during the encoding and the evaluation of symptom information by comparing eye movement measures (the number of fixation and fixation duration per dwell). Results will be discussed in light of existing theories on sequential diagnostic reasoning. Further, it will be discussed to which extent eye movements can be informative about memory processes underlying sequential diagnostic reasoning.

Reasoning about others’ emotions is a crucial component in so-cial cognition. Here, we tested the ability of preschool childrento reason about an agent’s emotions following an unexpectedoutcome. Importantly, we controlled for the actual payoff ofthe outcome, while varying the prior expectation of the agents.Five-year-olds, but not four-year-olds, were able to correctlyjudge an agent’s emotions following an unexpected outcome(Experiment 1). When explicitly provided with the agent’s ex-pectations, 4-year-olds were then also able to correctly judgethe agent’s feelings (Experiment 2). Our results suggest thatthe ability to reason about emotions given outcomes and priorexpectations develops by 4 years of age, while the ability tospontaneously infer such prior expectations develops soon af-ter. We discuss our results in light of the developmental litera-ture on emotion understanding and counterfactual reasoning.

Although people are poor at naming odors, naming a smell helps to remember that odor. Previous studies show wine experts have better memory for smells, and they also name wine and wine-related smells differently than novices. This leads us to ask whether wine experts’ odor memory is verbally mediated? In addition, does the odor memory advantage that experts have over novices generalize to all odors, or is it restricted to odors in their domain of expertise? Twenty-four wine experts and 24 novices smelled wines, wine-related odors and common odors, and were asked to remember these. Critically, half of the participants were asked to name the smells in addition to memorizing them, while the other half just remembered the smells. Wine experts had better memory for wines, but not for wine-related or common odors, indicating their memory is restricted to odors from their domain of expertise. Wine experts were also found to be more consistent and accurate than novices in their descriptions. But there was no relationship between experts’ ability to name odors and their memory for odors. This suggests experts’ odor memory advantage is not linguistically mediated, but may be the result of differential perceptual learning.

The category-order effect (COE) is observed when the categorical properties of items within the first half of a given list affect recall performance in a mixed-list serial-recall task. The present study examines whether the advantage is due to other sub-categorical properties (e.g., orthographic similarity and word frequency) rather than an artifact of stimuli used in previous studies (e.g., numbers vs. nouns). Participants were presented with numeric stimuli and nouns from a variety of semantic categories while their orthography and word frequency were systematically manipulated. The results suggest that a large portion of the COE can be attributed to the sub-categorical properties of the items.

When estimating the strength of the relation between a cause(X) and effect (Y), there are two main statistical approachesthat can be used. The first is using a simple correlation. Thesecond approach, appropriate for situations in which thevariables are observed unfolding over time, is to take acorrelation of the change scores – whether the variablesreliably change in the same or opposite direction. The mainquestion of this manuscript is whether lay people use changescores for assessing causal strength in time series contexts.We found that subjects’ causal strength judgments were betterpredicted by change scores than the simple correlation, andthat use of change scores was facilitated by naturalisticstimuli. Further, people use a heuristic of simplifying themagnitudes of change scores into a binary code (increase vs.decrease). These findings help explain how people uncovertrue causal relations in complex time series contexts.

Speakers often produce definite referring expressions that areoverspecified: they tend to include more attributes than neces-sary to distinguish the target referent. The current paper inves-tigates how the occurrence of overspecification is affected byviewing time. We conducted an experiment in which speakerswere asked to refer to target objects in visual domains. Half ofthe speakers had unlimited time to inspect the domains, whileviewing time was limited (1000 ms) for the other half. The re-sults reveal that limited viewing time induces the occurrenceof overspecification. We conjecture that limited viewing timecaused speakers to rely heavily on quick heuristics during at-tribute selection, which urge them to select attributes that areperceptually salient. In the case of unlimited inspection time,speakers seem to rely on a combination of heuristic and moredeliberate selection strategies.

The credibility of politicians is crucial to their persuasivenessas election candidates. The paper applies a parameter-freeBaysian source credibility model (integrating trustworthinessand epistemic authority) in a real-life test predictingparticipants’ posterior belief in the goodness of an unnamedpolicy after a named candidate has publically supported orattacked it.Two studies test model predictions against policy supportand attack of five presidential candidates from the USA.Model predictions were measured against observed posteriorbelief in the goodness of the policy.The results strongly suggest the model captures essentialtraits of how participants update beliefs in policies givenappeals to a candidates’ support of attack. Further, individualdifferences suggest that people consider other factors thanthe ones elicited for the study. More studies into appeals tospecific candidates are warranted to construct more accuratemodels of the influence of source credibility on politicalreasoning.

A repeated finding in the literature of face recognition is thatexpressive faces are remembered better than neutral faces.However, a better facial-identity recognition may come at acost of a reduced precision with which the pictorial facial fea-tures, irrelevant for identity recognition, are represented inmemory. By means of a continuous-report task, we testedthis hypothesis by measuring the memory precision of ex-pressive and neutral faces. Commensurable face-identity andfacial-expressions variations were generated with the methodof Fechnerian scaling. The results confirm our hypothesis, butonly under conditions of high memory load. We interpret thepresent findings as due to the effects of the categorical pro-cesses required for facial-identity recognition.

This study aimed to develop a Thought Experiment External-izer (TE-ext) and to apply it in order to observe barriers toproblem solving. TE-ext enables students to visualize a prob-lem situation. Users of TE-ext can implement changes in thesituation and see the result as an animation. Experimentaluse of TE-ext identified three barriers to conducting an effec-tive thought experiment (TE). First, participants tended not tochange the situation from the original one; second, incorrect orinappropriate knowledge was applied to the situation; third, theparticipants did not apply the results of their TE to other situa-tions. These factors prevented participants from rejecting theirinitial incorrect model and finding a new one through TEs.

Here we used a hidden Markov model (HMM) based ap- proach to infer individual choices from eye movements in preference decision-making. We assumed that during a deci- sion making process, participants may switch between explo- ration and decision-making periods, and this behavior can be better captured with a Switching HMM (SHMM). Through clustering individual eye movement patterns described in SHMMs, we automatically discovered two groups of partici- pants with different decision making behavior. One group showed a strong and early bias to look more often at the to-be chosen stimulus (i.e., the gaze cascade effect; Shimojo et al., 2003) with a short final decision-making period. The other group showed a weaker cascade effect with a longer final de- cision-making period. The SHMMs also showed capable of inferring participants’ preference choice on each trial with high accuracy. Thus, our SHMM approach made it possible to reveal individual differences in decision making and discover individual preferences from eye movement data.

The roles of semantic and perceptual information in cognition are of widespread interest to many researchers. However, disentangling their contributions is complicated by their overlap in real-world categories. For instance, attempts to calibrate visual similarity based on participant judgments are undermined by the possibility that semantic knowledge contaminates these judgments. This study investigated whether inverting stimuli attenuates semantic contamination of visual similarity judgments in adults and children. Participants viewed upright and inverted triads of familiar animals, and judged which of two test items looked most like the target. One test item belonged to the same category as the target, and one belonged to a different category. Test items’ visual similarity to the target either corresponded or conflicted with category membership. Across age groups, conflicting category membership reduced accuracy and slowed reaction times to a greater extent in upright than inverted triads. Therefore, inversion attenuates semantic contamination of visual similarity judgments.

Learning sequential actions is an essential human ability, formost daily activities are sequential. We modify the serial reac-tion time (SRT) task, originally used to teach people a con-sistent sequence of button presses by cueing them with thenext target response, to record mouse movements, collectingcontinuous response trajectories. Further, we introduce a rein-forcement learning version of the paradigm in which the nexttarget is not cued. Instead, learners must explore response al-ternatives, and receive a penalty for each incorrect response,as well as a reward for a correct response. Participants arenot told that they are to learn a single deterministic sequenceof responses, nor that it will repeat (nor how often), nor howlong it is. Given the difficulty of the task, it is unsurprisingthat some learners performed poorly. However, many learn-ers performed remarkably well, and some acquired the full 10-item sequence within 10 repetitions. We compare the high- andlow-performers’ detailed results in this reinforcement learning(RL) task with a cued trajectory SRT task, finding both simi-larities and discrepancies. Finally, we note that humans in thistask outperform three standard RL models and have differentpatterns of errors that suggest future modeling directions.

This study examined neural activity associated with inductive inference using functional Near Infrared Spectroscopy (fNIRS). Induction is a powerful way of generating new knowledge by generalizing known information to novel items or contexts. Two key bases for identifying targets for induction are perceptual similarity, and rules that specify category- relevant features. Similarity- and rule-based induction have been argued to represent distinct mechanisms, such that only rule-based induction requires executive function processes associated with the prefrontal cortex (PFC), namely: active maintenance of representations and inhibition of salient but irrelevant features. Here, we address the lack of direct empirical evidence supporting this possibility by recording PFC activity using fNIRS while adult participants (n=24) performed an inductive inference task. We found that PFC activity during induction was greater when participants had been taught a category-inclusion rule versus when participants could only rely on overall similarity.

Ray Jackendoff (2007) claims that most work on consciousness deals “almost exclusively with visualexperience” and suggests to focus more on linguistic awareness. Jackendoff proposes that phonologicalability – to divide utterances into words and syllables – is at the core of linguistic consciousness. Thisaccount can be supplemented by empirical research on language acquisition. Focusing on the step-by-stepemergence of linguistic consciousness in infancy can offer new and potentially fruitful angles forinvestigating states of consciousness. In addition computational models of word segmentation andpossible implications for linguistic consciousness are discussed.

We propose simple parameter-free models that predict howpeople learn environmental cue contingencies, use this infor-mation to measure the usefulness of cues, and in turn, use thesemeasures to construct search orders. To develop the models,we consider a total of 8 previously proposed cue measures,based on cue validity and discriminability, and develop simpleBayesian and biased-Bayesian learning mechanisms for infer-ring these measures from experience. We evaluate the modelpredictions against people’s search behavior in an experimentin which people could freely search cues for information todecide between two stimuli. Our results show that people’sbehavior is best predicted by models relying on cue measuresmaximizing short-term accuracy, rather than long-term explo-ration, and using the biased learning mechanism that increasesthe certainty of inferences about cue properties, but does notnecessarily learn true environmental contingencies.

In previous work, we showed that a simple neurocomputa-tional model The Model, or TM) trained on the Ekman &Friesen Pictures of Facial Affect (POFA) dataset to catego-rize the images into the six basic expressions can account forwide array of data (albeit from a single study) on facial ex-pression processing. The model demonstrated categorical per-ception of facial expressions, as well as the so-called facialexpression circumplex, a circular configuration based on MDSresults that places the categories in the order happy, surprise,fear, sadness, anger and disgust. Somewhat ironically, the cir-cumplex in TM was generated from the similarity between thecategorical outputs of the network, i.e., the six numbers rep-resenting the probability of the category given the face. Here,we extend this work by 1) using a new dataset, NimsStims,that is much larger than POFA, and is not as tightly controlledfor the correct Facial Action Units; 2) using a completely dif-ferent neural network architecture, a Siamese Neural Network(SNN) that maps two faces through twin networks into a 2Dsimilarity space; and 3) training the network only implicitly,based on a teaching signal that pairs of faces are in either inthe same or different categories. Our results show that in thissetting, the network learns a representation that is very similarto the original circumplex. Fear and surprise overlap, whichis consistent with the inherent confusability between these twofacial expressions. Our results suggest that humans evolvedin such a way that nearby emotions are represented by similarappearances.

Previous studies have shown that people’s memories are changeable, and systematic incorrect memories (e.g., false memory) can be created. We hypothesize that people’s beliefs about the real world can be changed similarly to the way systematic incorrect memories and systematic incorrect beliefs (which we call memory-based false belief) are generated. We also predict that since memory-based false beliefs are consistent with abstract knowledge that is consisted with prototypical patterns and organization found in the real world, false beliefs work adaptively in making inferences about environmental information in the real world. We conducted behavioral and simulation studies in order to examine our hypotheses on people’s beliefs and inferences about the real world. The results showed that participants had systematic false beliefs about cities’ attributes (e.g., whether they have a professional baseball team), and that such false beliefs worked adaptively in making inferences about population size.

Human actions are more than mere body movements. In contrast to other dynamic events in the natural world, human actions involve mental processes that enable willful bodily movements. We reported two experiments to demonstrate that human observers spontaneously assign the role of cause to relative limb movements, and the role of effect to body motion (i.e., the position changes of the body center of mass) when observing actions of others. Experiment 1 showed that this causal action constraint impacts people’s impression on the naturalness of observed actions. Experiment 2a/b revealed that the causal constraint guides the integration of different motion cues within a relational schema. We developed an ideal observer model to rule out the possibility that these effects resulted from the learning of statistical regularity in action stimuli. These findings demonstrate that causal relations concerning bodily movements play an important role in perceiving and understanding actions.

The block design task, a standardized test of nonverbal reason-ing, is often used to characterize atypical patterns of cognitionin individuals with developmental or neurological conditions.Many studies suggest that, in addition to looking at quantita-tive differences in block design speed or accuracy, observingqualitative differences in individuals’ problem-solving strate-gies can provide valuable information about a person’s cogni-tion. However, it can be difficult to tie theories at the levelof problem-solving strategy to predictions at the level of ex-ternally observable behaviors such as gaze shifts and patternsof errors. We present a computational architecture that is usedto compare different models of problem-solving on the blockdesign task and to generate detailed behavioral predictions foreach different strategy. We describe the results of three differ-ent modeling experiments and discuss how these results pro-vide greater insight into the analysis of gaze behavior and errorpatterns on the block design task.

Every human society includes social hierarchies— relationships between individuals and groups of unequal rank or status. Recent research has shown that even preverbal infants represent hierarchical relationships, expecting larger agents and agents from larger groups to win dominance contests. However, to successfully navigate social hierarchies, infants must also integrate information about social rank into their own behavior, such as when deciding which individuals to approach and which to avoid. Here we demonstrate that two- year-old children (ages 21-31 months) preferred novel dominant agents to subordinates. That is, by the age of 21 months, toddlers not only use phylogenetically stable cues to predict the winner of dominance contests, they also like the dominant agents better. This finding suggests that young children use their ability to infer relative rank to selectively approach dominant individuals.

For drawing higher-level perspectives in group activi-ties, resolving conflicts among group members is cru-cial. We investigated group activities with four memberswherein one member had a different perspective fromthe other three. Four members engaged in a rule discov-ery task in which they were required to unify conflictsfor the solution. Through two experiments, we investi-gated two hypotheses: 1) Innovative high-level perspec-tives are more likely to emerge from a minority individ-ual than from the majority of group members, 2) Groupmembers on the majority side might tend to have moreegocentric perspectives than an individual on the minor-ity side. Both hypotheses were supported.

Mind wandering is a prevalent but highly subjective phenomenon that is difficult to measure. Typically studies use probes at random points throughout at study that pop in and ask participants “Are you mind wandering” where they indicate yes or no, and then resume the study. This study investigated a method of extracting eye blinks from raw eye tracking data while participants were reading texts that varied in degree of engagingness on a similar topic. Blink durations were found to increase for less engaging texts. We hypothesize that eye blink durations may increase with mind wandering and discuss implications for mind wandering research.

The present research builds on prior work on the social-contextualnature of children’s generosity by systematically examining bothobserver effects and whether the recipient is an in-group or out-group member. Although previous research has examined thesefactors independently, no study to date has examined them inconjunction. We also extend prior research by including bothmeasures of sharing behavior and children’s evaluations of sharingscenarios, and by investigating a larger sample (N=164) with abroader age range than is typical of prior research (5- to 9-year-olds). We found that, across the entire age range tested, childrenwere generous when observed and gave more to in-group membersthan out-group members, and that there was no interaction betweenthese effects. We also found that children’s own sharing behaviorpredicted their evaluations of sharing scenarios, with childrenrating in-group sharing as "nicer" than out-group sharing.

This study aims to address potential costs of using incorrectworked examples in teaching mathematics. While suchpractice has been shown to be effective in educationalresearch, previous findings in the memory literature suggestthat exposure to an incorrect solution may lead students tolater believe that it is correct due to increased familiarity. Wedesigned a two-session experiment with 1-week delay inwhich students studied correct and incorrect worked outexamples. We found only small changes in students’ ability tosuccessfully distinguish between correct and incorrectsolutions over time. Students did rate the previously studiedincorrect examples as being more correct after the 1-wkdelay, but this did not affect their correctness ratings of newcorrect and incorrect worked examples or their problemsolving accuracy. We conclude that the unique nature ofmathematical problem solving may protect students from thedangers of using incorrect worked examples.

Subgoal labeled expository instructions and worked examples have been shown to positively impact student learning and performance in computer science education. This study examined whether problem solving performance differed based on the order of expository instructions and worked examples and the presence of subgoal labels within the instructions for creating applications (Apps) for phones. Participants were 132 undergraduates. A significant interaction showed that when learners were presented with the worked example followed by the expository instructions containing subgoal labels, the learner was better at outlining the procedure for creating an application. However, the manipulations did not affect novel problem solving performance or explanations of solutions,. These results suggest that some limited benefit can be gained from presenting a worked example before expository instructions when subgoal labels are included.

Category learning is thought to be mediated—in at least some category structures—by hypothesis-testing processes. Verbal labels for the stimuli and stimulus individuation have been shown to facilitate the formation, testing, and application of rules of category membership (Fotiadis & Protopapas, 2014). We sought to replicate the phenomenon of facilitation due to verbal names for the stimuli by training participants for two consecutive days to either learn new names for abstract shapes, or learn shape-ideogram pairings; a third group was unexposed to the shapes. After training, participants were given a Type II categorization task—thought to be mediated by verbal processes of rule discovery—utilizing the trained shapes. We hypothesized that verbal labels for the shapes and shape individuation would provide facilitative effects in learning to categorize. Results revealed no effect of training on categorization performance. This study suggests that caution should be taken when generalizing findings across perceptual modalities or different experimental paradigms.

It does not do the act of reading literature any justice to de-scribe it as simply processing text to acquire information orknowledge. We enjoy reading stories, we become absorbed inthem. Our absorption into stories is related to their contextualstructure. We develop a statistical method for the analysis ofreading time distributions which allows us to assess the con-text of a story rather than merely its text. This analysis detectsstatistically distinct distributions of reading times, with eachdistribution representing a distinct process or mode of read-ing. Our experiments support the hypothesis that the temporalchange in these modes of reading are related to changes in thedegrees of absorption of the subjects and also in the contextualstructure of the stories being read.

The current study examined the effects of workingmemory training on working memory capacity andsecond language ability in adult learners ofSpanish. In order to maximize the effect of the trainingfor language learners, the stimuli for the training taskswere Spanish words and sentences. While the traininggroup did not show greater improvements on workingmemory assessments relative to controls, they did showmore native-like patterns in a Spanish self-paced readingtask. The combination of second language materials withworking memory training may be helping users learn tocope with the increased processing demands associatedwith learning a new language, even if they are notnecessarily improving their working memory.

The current orthodoxy in cognitive science, what I describe as a commitment to deep representationalism, faces intractable problems. If we take these objections seriously, and I will argue that we should, there are two possible responses: 1. We are mistaken that representation is the locus of our cognitive capacities — we manage to be the successful cognitive agents in some other, non-representational, way; or, 2. Our representational capacities do give us critical cognitive advantages, but they are not fundamental to us qua human beings. As Andy Clark has convincingly argued, anti- representationalism, option one, is explanatorily weak. Consequently, I will argue, we need to take the second option seriously. In the first half of the paper I rehearse the problems with the current representational view and in the second half of the paper I defend and give a positive sketch of a two-systems view of cognition – a non-representational perceptual system coupled with a representational language-dependent one – and look at some consequences of the view.

False feedback on choices has been documented to inducelasting preference change. Here we extend such effects to thepolitical domainand investigatethetemporal persistence ofinduced preferences, as well as, the possible role the length ofconfabulatory justifications may play. We conducted a two-day choice blindness experiment using political statements,with sessions being roughly one week apart. Changes inpolitical preferences remained one week after initialresponses, and were most prominent in participants who wereallowed to confabulate freely. These findings, being the firstto demonstrate lasting preference change using choiceblindness, are discussed in light of constructivist approachesto attitude formation through a process of self-perception.

Self-driving autonomous vehicles (AVs) have the potential tomake the world a safer and cleaner place. A challengeconfronting the development of AVs is how these vehiclesshould behave in traffic situations where harm is unavoidable.It is important that AVs behave in ethically appropriate waysto mitigate harm. Ideally, they should obey a system ofprinciples that both concur with human moral judgments andare ethically defensible. Here we compare people’s moraljudgments of AV programming with their judgments aboutthe behavior of human drivers, with the goal of beginning toidentify such principles. As many debates within ethicsremain unresolved, empirical investigations like ours mayguide the development of ethical AVs (Bonnefon et al., 2015).In addition, people’s judgments about the behavior of AVsmay serve as a window into the abstract principles peopleapply in their moral reasoning.

The creative process involves several cognitive processes,such as working memory, controlled attention and taskswitching. One other process is cognitive search oversemantic memory. These search processes can be controlled(e.g., problem solving guided by a heuristic), or uncontrolled(e.g., mind wandering). However, the nature of this search inrelation to creativity has rarely been examined from a formalperspective. To do this, we use a random walk model tosimulate uncontrolled cognitive search over semanticnetworks of low and high creative individuals with an equalnumber of nodes and edges. We show that a random walkover the semantic network of high creative individuals “finds”more unique words and moves further through the networkfor a given number of steps. Our findings are consistent withthe associative theory of creativity, which posits that thestructure of semantic memory facilitates search processes tofind creative solutions.

Successful communication regularly requires listeners to makepragmatic inferences — enrichments beyond the literal mean-ing of a speaker’s utterance. For example, when interpretinga sentence such as “Alice ate some of the cookies,” listenersroutinely infer that Alice did not eat all of them. A Griceanaccount of this phenomenon assumes the presence of alterna-tives (like “all of the cookies”) with varying degrees of infor-mativity, but it remains an open question precisely what thesealternatives are. To address this question, we collect empiricalmeasurements of speaker and listener judgments about vary-ing sets of alternatives across a range of scales and use these asinputs to a computational model of pragmatic inference. Thisapproach allows us to test hypotheses about how well differ-ent sets of alternatives predict pragmatic judgments by peo-ple. Our findings suggest that comprehenders likely considera broader set of alternatives beyond those logically entailed bythe initial message.

Literature on memory research shows that when memorizing, people may blend two situations, i.e. when memorizing one story, they add elements from another story. Most of the cognitive models assume that the superficial similarity between two episodes is the primary factor for blending. However, there is evidence that people blend dissimilar stories as well, if these stories share the same relational structure. We contrasted the two factors in a single study and performed experiments with the same design and stimuli with adults and with 4-5-year-old children. The results show that there is no qualitative difference between the performance of adults and children. Also, both adults and children blend either pictures that have surface or structural similarity depending on the abstractness of the objects in them.

Cognitive factors can mediate the tendency to create falsememory. We explored the role of the two systems ofreasoning in the production of false memories. Suchdifference can be assessed through the Cognitive ReflectionTest (CRT), a measure of the propensity to reflect rather thanproducing an intuitive response. By the use of a DRM-likeparadigm in a prose recognition memory task, we measuredCRT-related individual differences in producing falsememories. We observed that intuitive thinkers are more likelyto produce false memories.

The effect of labels on nonlinguistic representations is the focusof substantial debate in the developmental literature. A recentempirical study (Twomey & Westermann, 2016) suggested thatlabels are incorporated into object representations, such thatinfants respond differently to objects for which they know alabel relative to unlabeled objects. However, these empiricaldata cannot differentiate between two recent theories ofintegrated label-object representations, one of which assumeslabels are features of object representations, and one whichassumes labels are represented separately, but become closelyassociated with learning. We address this issue using aneurocomputational (auto-encoder) model to instantiate boththeoretical approaches. Simulation data support an account inwhich labels are features of objects, with the samerepresentational status as the objects’ visual and hapticcharacteristics.

Object recognition and categorization is a fundamental aspectof cognition in humans and animals. Models have been imple-mented around the idea that categories are sets of frequentlyco-occurring features. Out of these models a question has beenraised, namely what is the mechanism by which we learn a hi-erarchically organized set of categories, including types andsubtypes? In this paper we introduce such a model, the Domi-nant Property Assembly Network (DPAN). DPAN uses an un-supervised neural network to model an agent which developsa hierarchy of object categories based on highly correlated ob-ject features. Initially, the network generates representations ofhigh-level object types by identifying commonly co-occurringsets of features. Over time, the network will start to use aninhibition of return (IOR) operation to examine the featuresof a categorized object that make it unusual as an instance ofits identified category. The result is a network which, earlyin training, represents classes of objects using coarse-grainedcategories and recognizes objects as members of these generalclasses, but eventually is able to recognize subtle differencesbetween subtypes of objects within the broad classes, and rep-resent objects using these more fine-grained categories.

Can input in one sensory modality strengthen memory in adifferent sensory modality? To address this question, weasked participants to encode images presented in variouslocations (e.g., a depicted dog in the top left corner of thescreen) while they heard spatially uninformative sounds.Some of these sounds matched the image (e.g., the word“dog” or a barking sound) while others did not. In asubsequent memory test, participants were better atremembering the locations of images that were encoded witha matching sound, even though these sounds were spatiallyuninformative – an effect that was mediated by whether thesounds were verbal or non-verbal. Because the sounds did notprovide any relevant location information, better spatialmemory cannot be attributed to auditory memory; rather, it isattributed to visual memory being strengthened by thematching auditory input. These findings provide the firstbehavioral evidence for cross-modal interactions in memory.

Parenting practices are known to play an important role inshaping children’s outcomes. For example, children whoseparents engage them in high-quality conversations and whoare given opportunities for free play are at an advantage forlearning and later academic outcomes. However, communicat-ing the results of relevant scientific findings to parents remainsa challenge. One possible moderator of uptake of parentinginformation is the implicit theories parents hold with regardto child development and parenting. As a first step in inves-tigating this possibility, the present work establishes a newmeasure of parenting attitudes including three subscales cor-responding to attitudes about rules and respect, affection andattachment, and early learning. We then examine whether sub-scale scores predict uptake of new information about children’slearning. Scores on the Early Learning subscale, but not theRules and Respect subscale, predicted generalization from thearticle, providing initial evidence of the validity of this mea-sure.

Humans use rich intuitive theories to explain other people’sbehavior. Previous work in lay psychology of behavior havetended to treat emotion as causing primarily unintentional be-havior (e.g., being sad causes one to cry), neglecting how peo-ple incorporate emotions into explanations of rational, inten-tional actions. Here, we provide preliminary explorations intointegrating emotions into a theory of folk psychology. Specif-ically, we show that in the lay theory, people are willing toendorse emotions as causes of intentional actions. Moreover,people readily attribute beliefs and desires as explanations foremotional expressions. This work provides a first step in elabo-rating people’s rich understanding of emotions as an importantcomponent of intuitive social cognition.

Previous experimental studies on concessive connectiveshave only looked at their local facilitating or predictive ef-fect on discourse relation comprehension and have oftenviewed them as a class of discourse markers with simi-lar effects. We look into the effect of two connectives,but and although, for inferring contrastive vs. concessivediscourse relations to complement previous experimentalwork on causal inferences. An offline survey on AMTurkand an online eye-tracking-while-reading experiment areconducted to show that even between these two connec-tives, which mark the same set of relations, interpretationsare biased. The bias is consistent with the distribution ofthe connective across discourse relations. This suggeststhat an account of discourse connective meaning based onprobability distributions can better account for compre-hension data than a classic categorical approach, or an ap-proach where closely related connectives only have a coremeaning and the rest of the interpretation comes from thediscourse arguments.

Can novel 3D models help students develop a deeperunderstanding of core concepts in molecular biology? Weadapted 3D molecular models, developed by scientists, foruse in high school science classrooms. The models accuratelyrepresent the structural and functional properties of complexDNA and Virus molecules, and provide visual and hapticfeedback about biomolecular properties that are often implicitin traditional models. We investigated: 1) Can we measureconceptual growth on core concepts? 2) Do lessons with 3Dmodels improve student outcomes on these measures?, and 3)What factors mediate learning? Model use yielded measurablegains in conceptual knowledge and the greatest gains wererelated to how actively models were used during a lesson andthe facilitative role adopted by the teachers.

Previous research has found that comprehenders are willingto adopt non-literal interpretations of sentences whose literalreading is unlikely. Several studies found evidence that com-prehenders decide whether or not a given utterance should betaken at face value in accordance with principles of Bayesianrationality, by weighing the prior probability of potential inter-pretations against the degree to which they are (in)consistentwith the literal form of the utterance. While all of these re-sults are consistent with string-edit noise models, many errorprocesses are known to be sensitive to the underlying linguis-tic structure of the intended utterance. Here, we explore thecase of exchange errors and provide experimental evidencethat comprehenders’ noise model is structure-sensitive. Ourresults add further support to the noisy-channel theory of lan-guage comprehension, extend the set of known noise opera-tions to include positional exchanges, and show that compre-henders’ noise models are well-adapted to structure-sensitivesources of signal corruption during communication.

We examine data from over 6.6 million games of tour-nament chess between players rated by the internationalchess authority, FIDE. Previous research has focussed onthe low representation of women in chess. We repli-cate and extend previous analysis (Chabris and Glickman,2006) on an international level. We find no support fordifferential variability, differential drop-out between maleand female players, or social context (in the form of pro-portion of female players at a national level) as drivers ofdrivers of male-female differences. Further, we examinegames between mixed and same gender pairs for evidenceof a ‘stereotype threat’ effect. Contrary to previous re-ports, we find no evidence of stereotype threat. Thoughthis analysis contradicts one specific mechanism wherebygender stereotype may influence players, the persistentdifferences between male and female players suggeststhat systematic factors do exist and remain to be uncov-ered.

A commonly suggested solution to reduce misinterpretations of verbal probability expressions in risk communications is to use a verbal-numerical (mixed format) approach, but it is not known whether this increases understanding over and above a purely numerical format. Using the ‘which outcome’ methodology (Teigen & Filkuková, 2013), we examined the effect of using verbal, numerical and mixed communication formats, as well as investigating whether marking outcomes as salient would alter the outcomes people perceived as ‘unlikely’ or having a 20% chance of occurring. We observed no effect of saliency, but replicated previous findings, with general preference for values at the high end of a distribution (including maximum/above maximum values) present in both verbal and mixed communication formats. This demonstrates the relevance of these findings for real-world consequential risk communication. Whilst the estimates differed between the mixed and numerical formats, we found that the mixed format yielded the more accurate estimates.

A growing body of research suggests that individuals withdevelopmental dyslexia perform below typical readers onnon-linguistic cognitive tasks involving the learning andencoding of statistical-sequential patterns. However, theneural mechanisms underlying such a deficit have not beenwell examined. The aim of the present study was toinvestigate the ERP correlates of sequence processing in asample of children diagnosed with dyslexia using aprobabilistic visual serial learning paradigm. The behavioralresults revealed that whereas age-matched typicallydeveloping children (n=12) showed learning in the task asreflected by their response times, the children with dyslexia(n=8) likely showed difficulty in learning. In conjunction withthese behavioral results, the ERPs of the typically developingchildren showed a P300-like response indicative of thisparadigm (Jost et al., 2015); whereas, the children diagnosedwith a reading disorder showed no such ERP effects. Thesefindings are consistent with the idea that differences instatistical-sequential learning ability might underlie thereading deficits observed in developmental dyslexia.

The St. Petersburg Paradox (SPP), where people are willing topay only a modest amount for a lottery with infinite expectedgain, has been a famous showcase of human (ir)rationality.Since inception multiple solutions have been proposed,including the influential expected utility theory. Criticismsremain due to the lack of a priori justification for the utilityfunction. Here we report a new solution to the long-standingparadox, which focuses on the probability weightingcomponent (rather than the value/utility component) incalculating the expected value of the game. We show that anew Additional Transition Time (AT) based measure,motivated by both physics and psychology, can naturally leadto a converging expected value and therefore solve theparadox.

We test contrasting predictions of two recent models of proba-bility judgment: the quantum probability model (Busemeyeret al., 2011) and the probability theory plus noise model(Costello and Watts, 2014). Both models assume that peo-ple estimate probability using formal processes that follow orsubsume standard probability theory. The quantum probabil-ity model predicts people’s estimates should agree with oneset of probability theory identities, while the probability the-ory plus noise model predicts a specific pattern of violation ofthose identities. Experimental results show just the form of vi-olation predicted by the probability theory plus noise model.These results suggest that people’s probability judgments donot follow quantum probability: instead, they follow the rulesof standard probability theory, with the systematic biases seenin those judgments due to the effects of random noise.

In this study, we experimentally investigated the influence ofa three-dimensional (3D) graphic image and a 3D-printed ob-ject on a spatial reasoning task in which participants were re-quired to infer cross sections of a liver in a situation where liverresection surgery was presupposed. The results of the studyindicated that using a 3D-printed object produced more accu-rate task performance and faster mental model construction ofa liver structure than a 3D image. During the task, using a3D-printed object was assumed to reduce cognitive load andinformation accessing cost more than using a 3D image.

Both the degree to which the left-hemisphere is specialized for language and the relative ability of the right-hemisphere to subserve language function are underspecified. The present study sought to identify whether the right-frontal fMRI activation seen in a number of case studies in patients with left-sided brain lesions exists as a group-level trend in patients with left-frontal tumors. It also sought to examine the possible compensatory nature of this activation. Thus, a retrospective analysis of 197 brain tumor patients who had undergone pre-surgical fMRI language mapping was conducted. Patients with left-frontal tumors were found to be more likely to show right- or co-dominant fMRI activation during language mapping tasks compared to patients who had tumors elsewhere in the brain. Further, patients with left-frontal tumors who were identified as right- or co-dominant for language were found to possess more intact language function as measured by the Boston Naming Test.

Previous work has shown that the abstract use of the prepositions in and on retains spatial meaning, such as containment and support that includes the control relationship between a located object (the figure) and a reference object (the ground). We extend these ideas to the case of metaphorical descriptions of emotion in Spanish – some of them featuring the emotion as a located entity in the person ́s body, and some of them featuring emotion as the ground in which the person ́s body stands. Two rating experiments show that people judge emotions as more “controllable” when they are described as located entities (the figure) than when they are described as grounds.

Abductive reasoning assigns special status to the explanatorypower of a hypothesis. But how do people make explana-tory judgments? Our study clarifies this issue by asking: (i)How does the explanatory power of a hypothesis cohere withother cognitive factors? (ii) How does probabilistic informa-tion affect explanatory judgments? In order to answer thesequestions, we conducted an experiment with 671 participants.Their task was to make judgments about a potentially explana-tory hypothesis and its cognitive virtues. In the responses, weisolated three constructs: Explanatory Value, Rational Accept-ability, and Entailment. Explanatory judgments strongly co-hered with judgments of causal relevance and with a sense ofunderstanding. Furthermore, we found that Explanatory Valuewas sensitive to manipulations of statistical relevance relationsbetween hypothesis and evidence, but not to explicit infor-mation about the prior probability of the hypothesis. Theseresults indicate that probabilistic information about statisticalrelevance is a strong determinant of Explanatory Value. Moregenerally, our study suggests that abductive and probabilisticreasoning are two distinct modes of inference.

Self-directed learning confers a number of advantages relativeto passive observation, including the ability to test hypothe-ses rather than learn from data generated by the environment.However, it remains unclear to what extent self-directed learn-ing is constrained by basic cognitive processes and how thoselimits are related to the structure of the to-be-learned material.The present study examined how hypothesis generation af-fects the success of self-directed learning of categorical rules.Two experiments manipulated the hypothesis generation pro-cess and assessed its impact on the ability to learn 1D and 2Drules. Performance was strongly influenced by whether thestimulus representation facilitated the generation of hypothe-ses consistent with the target rule. Broadly speaking, the find-ings suggest that the opportunity to actively gather informa-tion is not enough to guarantee successful learning, and thatthe efficacy of self-directed learning closely depends on howhypothesis generation is shaped by the structure of the learn-ing environment.

Can the representation of event sequence influence how jurorsremember and reason in a legal case? We addressed this ques-tion by examining the interaction between an individual’s pre-ferred spatial construal of time (SCT) for an external (visual-spatial) representation and the SCT of a courtroom graphic.One hundred fifty three undergraduates played the role of ju-rors in a fictitious civil trial. The details of a case were re-counted in a multimedia presentation featuring timelines an-imated in one of four orientations: left-right, right-left, top-bottom, and bottom-top. Participants were assessed on mea-sures of comprehension and causal reasoning. Results indi-cated effects of timeline orientation and SCT choice behav-ior on comprehension and reasoning. We discuss these resultsin terms of the role of attention in temporal-causal reasoning,and implications for the design of multimedia materials for thecourtroom.

We describe a prediction method called “Attractivity Weighting” (AW). In the case of cue-based paired comparison tasks, AW predicts a weighted average of the cue values of the most successful cues. In many situations, AW’s prediction is based on the cue value of the most successful cue, resulting in behavior similar to Take-the-Best (TTB). Unlike TTB, AW has a desirable characteristic called “access optimality”: Its long-run success is guaranteed to be at least as great as the most successful cue. While access optimality is a desirable characteristic, concerns may be raised about the short-term performance of AW. To evaluate such concerns, we here present a study of AW’s short-term performance. The results suggest that there is little reason to worry about the short-run performance of AW. Our study also shows that, in random sequences of paired comparison tasks, the behavior of AW and TTB is nearly indiscernible.

For creating photorealistic images, Computer Graphics re-searchers introduced Bidirectional Texture Functions (BTFs),which use view- and illumination-dependent textures for ren-dering. BTFs require massive storage, and several proposalswere made on how to compress them, but very few take intoaccount human perception. We present and discuss an exper-imental study on how decreasing the texture resolution influ-ences perceived quality of the rendered images. In a visualcomparison task, observer quality judgments and gaze datawere collected and analysed to determine the optimal down-sampling of BTF data without significant loss of their per-ceived visual quality.

Recent research has shown that action knowledgeinfluences categorical decisions (Borghi, Flumini, Natraj &Wheaton, 2012; Chao & Martin, 2000; Iachini, Borghi &Senese, 2008; Kalénine, Shapiro, Flumini, Borghi &Buxbaum, 2013). Shipp, Vallée-Tourangeau, and Anthony,(2014) showed that action influences categorisation in aforced-choice triad task when combined with taxonomicinformation and presented within a functional context. Thepresent experiment examined whether participants wouldbe more likely to match items in a triad task based onshared actions following priming with the functionalactions of the objects. Participants engaged in the triad taskused in Shipp et al. after a priming phase where they eitherinteracted with a series of objects for their functionalcapacity (Action Priming), grouped them into categories(Taxonomic Priming) or moved them from one table toanother (Movement Priming). Items within the triads werepresented as an image either on a white background(context-lean condition) or as a functional scene with theobject being used by an agent (context-rich condition).Consistent with Shipp et al. the results showed that actionwas primarily used to base choices on the triad task whenthe action choice also shared a taxonomic relation, and waspresented in context. Additionally, participants were morelikely to select the action related item when they had beenprimed with the functional action of the objects. The resultsare discussed in terms of the transfer effect from the objectinteraction task that facilitates how the objects aresimulated (Barsalou, 1999, 2003; Yeh & Barsalou, 2006).

We present the results of two temporally extendedexperimental implementations of the Monty Hall dilemma inorder to examine the dynamics of belief. In the firstexperiment, we used the standard three-door version of thedilemma, but biased the probability of the winning doorpositionally. Participants capitalized on the increasedprobabilities but did not discover the optimal switch strategy.In the second experiment, we increased the number of doors,in each case removing all but two doors. As the number ofdoors increased, participants converged on the optimal switchstrategy, as well as increasing their confidence in theirstrategy. This suggests that the information relevant to theMHD is not win frequencies but how the different elements ofthe dilemma are related.

The causal-contrast approach is a new teaching method thatrecruits learners’ implicit causal discovery process to improvemath learning by juxtaposing contrasting information criticalto discovering the goal of each solution step. Students oftenblindly memorize mathematical procedures and havedifficulty transferring their knowledge to novel problems. Byenabling learners to infer the goal of each step, the causal-contrast approach substantially improved high-school algebraproblem solving compared to a traditional instructionalcontrol (Walker, Cheng & Stigler, 2014). The present studydeveloped Walker et al.’s instructional materials into acomputer-based teaching program and tested the newapproach on community-college students, a population forwhom the traditional approach is often ineffective. The studyadded two new conditions: a baseline that received noinstruction and a condition using a teaching video from KhanAcademy, a well-regarded online educational websiterepresentative of the traditional approach. A delayed post-testindicated that the causal-contrast condition produceddramatically greater success in solving transfer problems thanthe other three conditions.

This study introduces a neural network that models thesocial interactions from a video corpus. The corpusconsists of recordings of naturalistic observations ofsocial interactions among children and theirenvironment. The videos are annotated multimodallyincluding features like gestures. We explore how thisvideo corpus can be utilized for modelling by trainingour model on a portion of the annotated data extractedfrom the corpus, and then by using the model to predictnovel interaction sequences. We evaluate our model bycomparing its automatically generated sequences to anunseen portion of the corpus data. The initial resultsshow strong similarities between the generatedinteractions and those observed in the corpus.

As a first step towards applying ACT-R to problems of like-lihood judgment, we draw parallels between ACT-R and Hy-Gene. More specifically, in the spirit of theory integration, wedemonstrate the relation between ACT-R’s declarative memorysystem and the core of HyGene: Minerva2. We first start bytransforming ACT-R’s activation equations into what is in ourview a more intuitive form. This form then allows us to moretransparently see the correspondence of the effect of prior his-tory between the two theories and of the current context be-tween them. The results provide insights into the workings ofthe two theories and open an avenue for future attempts of the-ory integration, not only between the two theories, but also torelated theories of memory. Moreover, we hope these resultswill be important steps toward testing ACT-R’s capabilities ofaccounting for judgment phenomena.

Cognition may require access to past events, for example to understand undesirable outcomes or diagnose failures. When cognition is distributed between multiple participants, a particular representational challenge occurs because not all of the participants may have directly experienced the focal event. Language can transcend temporal and physical limitations on event accessibility. We suggest that people create complex linguistic constructs as tools to facilitate retrospective cognition. We illustrate this process by analyzing the use of a particular linguistic construct (narrative) in the domain of clinical reasoning. Results demonstrated that narratives support clinical cognition during practitioner-patient interactions. Narratives extended access to clinically relevant events providing information about circumstances, subjective experiences, patient functioning, and prior decisions. Whereas, the hermeneutic nature of narrative allowed collaborative hypothesis testing and creation of meaning. The use of narrative in clinical cognition challenges Bruner’s (1991) distinction between narrative and paradigmatic reasoning and enriches the understanding of medical narratives.

Overprecision (overconfidence in interval estimation) is a biaswith clear implications for economic outcomes in industriesreliant on forecasting possible ranges for future prices andunknown states of nature - such as mineral and petroleumexploration. Prior research has shown the ranges peopleprovide are too narrow given the knowledge they have – thatis, they underestimate uncertainty and are overconfident intheir knowledge. The underlying causes of this bias are,however, still unclear and individual differences research hasshed little light on traits predictive of susceptibility. Takingthis as a starting point, this paper directly contrasts the NaïveSampling Model and Informativeness-Accuracy Tradeoffaccounts of overprecision – seeing which better predictsperformance in an interval estimation task. This was achievedby identifying traits associated with these theories – ShortTerm Memory and Need for Cognitive Closure, respectively.Analyses indicate that NFCC but not STM predicts intervalwidth and thus, potentially, impacts overprecision.

Categorical speech content can often be perceived directlyfrom continuous auditory cues in the speech stream, buthuman-level performance on speech recognition tasksrequires compensation for contextual variables like speakeridentity. Regression modeling by McMurray and Jongman(2011) has suggested that for many fricative phonemes, acompensation scheme can substantially increasecategorization accuracy beyond even the information from 24un-compensated raw speech cues. Here, we simulate thesame dataset instead using a neurally rather than abstractlyimplemented model: a hybrid dynamic neural field model andconnectionist network. Our model achieved slightly loweraccuracy than McMurray and Jongman’s but similar accuracypatterns across most fricatives. Results also comparedsimilarly to more recent models that were also less neurallyinstantiated but somewhat closer fitting to humans inaccuracy. An even less abstracted model is an immediatefuture goal, as is expanding the present model to additionalsensory modalities and constancy/compensation effects.

Valentine’s face-space suggests that faces are represented in apsychological multidimensional space according to their per-ceived properties. However, the proposed framework was ini-tially designed as an account of invariant facial features only,and explanations for dynamic features representation were ne-glected. In this paper we propose, develop and evaluate a com-putational model for a twofold structure of the face-space, ableto unify both identity and expression representations in a singleimplemented model. To capture both invariant and dynamicfacial features we introduce the face-space duality hypothesisand subsequently validate it through a mathematical presen-tation using a general approach to dimensionality reduction.Two experiments with real facial images show that the pro-posed face-space: (1) supports both identity and expressionrecognition, and (2) has a twofold structure anticipated by ourformal argument.

Several studies examined cue competition in human learning by testing learners on a combination of conflicting cues rooting for different outcomes, with each cue perfectly predicting its outcome. A common result has been that learners faced with cue conflict choose the outcome associated with the rare cue (the Inverse Base Rate Effect, IBRE). Here, we investigate cue competition including IBRE with sentences containing cues to meanings in a visual world. We do not observe IBRE. Instead we find that position in the sentence strongly influences cue salience. Faced with conflict between an initial cue and a non-initial cue, learners choose the outcome associated with the initial cue, whether frequent or rare. However, a frequent configuration of non-initial cues that are not sufficiently salient on their own can overcome a competing salient initial cue rooting for a different meaning. This provides a possible explanation for certain recurring patterns in language change.

Experimental philosophy’s ‘sources project’ seeks to developpsychological explanations of philosophically relevantintuitions which help us assess their evidentiary value. Thispaper develops a psycholinguistic explanation of intuitionsprompted by brief philosophical case-descriptions. For proofof concept, we target intuitions underlying a classic paradoxabout perception (‘argument from hallucination’). We tracethem to stereotype-driven inferences automatically executedin verb comprehension. We employ a forced-choiceplausibility-ranking task to show that contextuallyinappropriate stereotypical inferences are made from lesssalient uses of the verb “to see”. This yields a debunkingexplanation which resolves the philosophical paradox.

We present a neural network model of how events are storedin and retrieved from episodic long-term memory (LTM). Themodel is novel in giving an explicit account of the workingmemory (WM) medium mediating access to episodic mem-ory: it makes a specific proposal about how representations ofevents and situations in WM interface with representations ofevents and situations in episodic memory. It also provides theframework for an account of how operations accessing tempo-rally remote situations are reported in language.

While learning in a multitext environment increases with therise of electronic environments, little is known about whatmakes learners feel that they should continue learning oralready learn enough from one text. The current study aimedat examining what cues learners use to regulate their effortamong multiple sources in a multitext environment. Bymanipulating the amount of new information and conceptualoverlap across texts within a topic, we created three types oftext environments to generate different trajectories of twocues to perceived learning, new information (measured byrating of perceived new information) and encoding fluency(measured by ratings of reading ease). Results showed thatthe dominant cue to gauge perceived learning was theperceived amount of new information. The study extendedtheories in animal foraging and metacognition, andestablished a novel paradigm to better investigate adultlearning in the wild.

Risk communication, where scientists inform policy-makersor the populace of the probability and magnitude of possibledisasters, is essential to disaster management – enablingpeople to make better decisions regarding preventative steps,evacuations, etc. Psychological research, however, hasidentified multiple biases that can affect people’sinterpretation of probabilities and thus risk. For example,availability (Tversky & Kahneman, 1973) is known toconfound probability estimates while the description-experience gap (D-E Gap) (Hertwig & Erev, 2009) shows lowprobability events being over-weighted when described andunder-weighted when learnt from laboratory tasks. This paperexamines how probability descriptions interact with realworld experience of events. Responses from 294 participantsacross 8 conditions showed that people’s responses, given thesame described probabilities and consequences, were alteredby their familiarity with the disaster (bushfire vs earthquake)and its salience to them personally. The implications of thisfor risk communication are discussed.

The present study investigates individual differences in pupildilation during standard word naming. We looked at (i) howindividual subjects’ pupil size changes over the course of timeand (ii) how well pupil size is predicted by the frequency ofthe stimuli. The time course of the pupil size was analysedwith generalized additive modeling. The results show large in-dividual variations in the pupil response pattern in this verysimple task. Although, we see a pupil response to both stimu-lus onset and articulation onset and offset, both the amplitudeof change and the direction of change differ substantially be-tween subjects. This raises the question of what makes thepupil response functions so diverse, and one factor indicatedby the frequency effect or the lack thereof might be shallowreading versus reading for content.

Understanding the nature of commonsense reasoning is one ofthe deepest questions of cognitive science. Prior work hasproposed analogy as a mechanism for commonsense reasoning,with prior simulations focusing on reasoning about continuousbehavior of physical systems. This paper examines howanalogy might be used in commonsense more broadly. The twocontributions are (1) the idea of common sense units,intermediate-sized collections of facts extracted fromexperience (including cultural experience) which improvesanalogical retrieval and simplifies inferencing, and (2)analogical chaining, where multiple rounds of analogicalretrieval and mapping are used to rapidly constructexplanations and predictions. We illustrate these ideas via animplemented computational model, tested on examples froman independently-developed test of commonsense reasoning.

We use visual world eye-tracking to test if a speaker’s eyegaze to a potential antecedent modulates the listener’sinterpretation of an ambiguous pronoun. Participants listenedto stories that included an ambiguous pronoun, such as “Thedolphin kisses the goldfish... He....” During the pre-pronominal context, an onscreen narrator gazed at one of thetwo characters. As expected, participants looked more at thesubject character overall. However, this was modulated by thenarrator’s eye gaze and the amount of time the participantspent looking at the gaze cue. For trials in which participantsattended to the narrator’s eye gaze for > 500ms, participantswere significantly more likely to interpret the pronoun asreferring to the object if the narrator had previously looked atthe object. Results suggest that eye gaze – a social cue – cantemper even strong linguistic/cognitive biases in pronounresolution, such as the subject/first-mention bias.

Sensory information is a priori incomplete andambiguous. Our perceptual system has to makepredictions about the sources of the sensoryinformation, based on concepts from perceptualmemory in order to create stable and reliablepercepts. We presented ambiguous anddisambiguated lattice stimuli (variants of the Neckercube) in order to measure a hysteresis effects invisual perception. Fifteen healthy participantsobserved two periods of ordered sequences oflattices with increasing and decreasing ambiguityand indicated their percepts, in two experimentalconditions with different starting stimuli of theordered sequence. We compared the stimulusparameters at the perceptual reversal betweenconditions and periods and found significantdifferences between conditions and periods,indicating memory contributions to perceptualoutcomes on three different time scales frommilliseconds over seconds up to lifetime memory.Our results demonstrate the fruitful application ofphysical concepts like hysteresis andcomplementarity to visual perception.

We examined the visual perception of joint actions, in whichtwo individuals coordinate their body movements in space andtime to achieve a joint goal. Animations of interacting actionpairs (partners in human interactions) and non-interacting ac-tion pairs (individual actors sampled from different interactionsequences) were shown in the experiment. Participants wereasked to rate how likely the two actors were interacting. Therating data were then analyzed using multidimensional scalingto recover a two-dimensional psychological space for repre-senting joint actions. A descriptive model based on ordinallogit regression with a sparseness constraint was developed toaccount for human judgments by identifying critical featuresthat signal joint actions. We found that identification of jointactions could be accomplished by assessing inter-actor correla-tions between motion features derived from body movementsof individual actions. These critical features may enable rapiddetection of meaningful inter-personal interactions in complexscenes.

We investigated how constraints imposed by the concurrentvisual context modulate the effects of prior gender and actioncues as well as of stereotypical knowledge during situatedlanguage comprehension. Participants saw videos of female ormale hands performing an action and then inspected a displayshowing the faces of two potential agents (one male and onefemale face) as they listened to German OVS sentences aboutstereotypically female or male actions. Unlike previousexperiments (Rodriguez et al., 2015), the display concurrentwith the sentence also showed a picture of the object of thevideotaped action and a ‘competitor object’ (with oppositestereotypical valence) that had not appeared in the video butcould be mentioned in the sentence. We measured eyemovements to the faces of the agents during comprehension.The design manipulated the match between the videotapedaction and the action described by the sentence (action-verb(phrase) match) and the match between the stereotypicalvalence of the verbally described action and the gender of theagent of the previous video (conveyed only by the hands;stereotypicality match). We replicated the results obtained inRodríguez et al. (2015): an overall target agent preference (i.e.the agent whose gender matched that of the hands seen in theprevious video), reduced by action-verb mismatches. However,unlike in their study, mismatch effects emerged earlier. Inaddition, stereotypicality effects emerged in the verb region.The earlier mismatch effects and added stereotypicality effectssuggest that the visual availability of the objects, perhapsjointly with the verbal input, facilitated the activation ofrepresentations from the recent videos (speeding up mismatcheffects) and the consideration of alternative representations,favoring stereotypical expectations.

In two experiments, we investigate how adults think aboutproportion across different symbolic and spatialrepresentations in a comparison task (Experiment 1) and atranslation task (Experiment 2). Both experiments showresponse patterns suggesting that decimal notation provides asymbolic advantage in precision when representing numericalmagnitude, whereas fraction notation does not. In addition,pie charts may show some advantages above number lineswhen translating between representations. Lastly, our findingssuggest that the translation between number lines andfractions may be particularly error-prone. We discuss whatthese performance patterns suggest in terms of how adultsrepresent proportional information across these differentformats and some potential avenues through which theseadvantages and disadvantages may arise, suggesting newquestions for future work.

Algebra I is considered a gatekeeper course for highereducation, high-paying jobs, and access to STEM careers, yetmany students find themselves struggling to learn algebra.Prior research links intrinsic motivation for learning mathwith mathematics achievement, particularly duringadolescence. The current study measured middle schoolstudents’ interest in algebra and their procedural skills acrossthe span of an algebra unit to determine whether students whoshow declines in algebraic problem-solving also show adecline in a particular type of intrinsic motivation – interest inalgebra. Pre-test and post-test scores were used to categorizeparticipants into those who showed declines in problem-solving skills and those who did not. Of the overall sample (N= 367), a group of 25 students showed declining skills overthe course of the unit. These students also showed significantdeclines in interest in mathematics from pre- to post-test incomparison to students who did not show procedural declines.Our findings support the relationship between performanceand motivation in the classroom, particularly in algebra class.Educational implications are discussed.

Studies in visual category learning show that participants usedifferent category generalization strategies. Some studies re-port a preference for a rule-based strategy, while others reporta preference for a similarity-based strategy. We conducted cat-egory learning experiments in which we varied three variables— family resemblance of a category, saliency of the definingrule and presentation of transfer stimulus after a delay. Ourresults show that these factors influence the choice of categorygeneralization strategy. Our study offers a possible explanationfor the divergent results in the literature.

In this paper we examine how a mechanism that learns wordclasses from distributional information can contribute to thesimulation of child language. Using a novel measure of nounrichness, it is shown that the ratio of nouns to verbs in youngchildren’s speech is considerably higher than in adult speech.Simulations with MOSAIC show that this effect can bepartially (but not completely) explained by an utterance-finalbias in learning. The remainder of the effect is explained bythe early emergence of a productive noun category, which canbe learned through distributional analysis.

How the properties of a first language (Mandarin, Korean)influence the comprehension of sentences in a secondlanguage (English) was investigated in a series of self-pacedreading time studies. Native Mandarin- and Korean-speakinglearners of English were compared with native Englishspeakers on how they resolved a temporary ambiguity aboutthe relationship between a verb and the noun following it in asentence (e.g., The club members understood [that] thebylaws would be applied to everyone.). Frequency biases ofverbs’ subcategorization structure (direct-object-bias vs.sentential-complement-bias) was manipulated in Experiment1. Results showed that L1-Mandarin learners of L2-Englishwere able to use both the verb bias and the complementizercue, and their usage of these cues was not modulated byproficiency. L1-Mandarin learners’ use of the verb bias cuecontrasts with previously reported findings with L1-Koreanlearners of L2-English, who showed sensitivity to verb biasonly in higher proficiency learners (Lee, Lu, & Garnsey,2013). The difference between L1-Mandarin and L1-Koreanlearners suggests that L1 word order (Mandarin & English,SVO; Korean SOV) influences how quickly L2 learners learnword-order-dependent cues about structures in the L2.Experiment 2 added plausibility manipulation (e.g., The clubmembers understood the bylaws/the pool...). Neither thenative speakers or the L2 groups (L1-Mandarin L2-English &L1-Korean L2-English) used plausibility to disambiguatesentences, challenging the claims that L2 learners rely moreheavily on plausibility than syntactic cues during sentenceprocessing.

A constant element of our modern environment is change. In decision-making research however, very little is known about how people make choices in dynamic environments. We report the results of an experiment where participants were asked to choose between two options: a dynamic and risky option that resulted in either a high or a low outcome, and a stationary and safe option that resulted in a medium outcome. The probability of the high outcome in the risky option decreased or increased linearly over the course of the task while the probability of the medium outcome stayed the same throughout. We find that adaptation to change is related to the direction of that change, and that the way people adapt to changing probabilities relates to their willingness to explore available options. A cognitive model based on Instance-Based Learning Theory reproduces the behavioral patterns.

The ability to remember events plays an important role in hu-man life. People can replay past events in their heads and makedecisions based on that information. In this paper, we describea novel extension to a cognitive architecture, ICARUS, that en-ables it to store, organize, generalize, and retrieve episodictraces that can help the agent in a variety of manners. Af-ter discussing previous work on the related topic, we reviewICARUS and explain the new extension to the architecture indetail. Then we discuss four architectural implications of thenew capability and list some future work before we conclude.

Witness credibility is important for establishing testimonial value.The story model posits that people construct narratives fromevidence but does not explain how credibility is assessed. Formalapproaches use Bayesian networks (BN) to represent legalevidence. Recent empirical work suggests people might alsoreason using qualitative causal networks. In two studies,participants read a realistic trial transcript and judge guilt andwitness credibility. Study 1 varied testimonial consistency anddefendant character. Guilt and credibility assessments wereaffected by consistency but not prior convictions. Study 2constructed a BN to represent consistency issues. Individualparameter estimates were elicited for the corresponding BN tocompute posterior predictions for guilt and credibility. The BNprovided a good model for overall and individual guilt andcredibility ratings. These results suggest people construct causalmodels of the evidence and consider witness credibility. The BNapproach is a promising direction for future research in legalreasoning.

Induction, the ability to generalize knowledge from known to novel instances, is essential for human learning. This study investigates how attention allocation during category learning and induction affects what information is represented and encoded to memory. In Experiment 1 5-year-olds and adults learned rule-based categories. They were then presented with an Induction-then-Recognition task. Similar to previous results with familiar categories, children exhibited better memory for items than adults. In Experiment 2, adults learned similarity-based categories and then were presented with an Induction-then-Recognition task. In this condition, adults’ memory was as good as children’s memory in Experiment 1. These results indicate that the way categories are represented affects the way induction is performed.

The importance of human-agent negotiation, and the role ofemotion in such negotiations, have been emphasized in human-agent interaction research. Thus far, studies have focused onbehavioral effects, rather than examining the neural underpin-nings of different behaviors shown in human-agent interac-tions. Here, we used a multi-round negotiation platform, in-stead of the more common single-shot negotiation, and wereable to find distinct brain patterns in emotion-related regionsof the brain during different types of offers. Using multi-voxelpattern analysis to analyze brain imaging data acquired duringfunctional MRI scanning, we show that it is possible to pre-dict whether the negotiator concedes, does not change, or asksfor more during the negotiation. Most importantly, we demon-strate that left dorsal anterior insula, which is known to be anemotion-related brain region, shows a different pattern of ac-tivity for each of the three offer types.

One way people deal with uncertainty is by asking questions.A showcase of this ability is the classic 20 questions gamewhere a player asks questions in search of a secret object. Pre-vious studies using variants of this task have found that peopleare effective question-askers according to normative Bayesianmetrics such as expected information gain. However, so far,the studies amenable to mathematical modeling have used onlysmall sets of possible hypotheses that were provided explic-itly to participants, far from the unbounded hypothesis spacespeople often grapple with. Here, we study how people eval-uate the quality of questions in an unrestricted 20 Questionstask. We present a Bayesian model that utilizes a large data setof object-question pairs and expected information gain to se-lect questions. This model provides good predictions regardingpeople’s preferences and outperforms simpler alternatives.

This article presents an experiment in which participant’sworking memory, tasks-switching and focusing skills aretrained in a game called Wollie on a smartphone. Before andafter the training period they performed three task (a recall,Stroop and task-switching). The goal of this research was tosee how the participants, from the test group, learn withinthe game and how this affects the three tasks. Only in theStroop results a clear difference between the two groups wasfound. However, we found that participants who had the mosttrouble in playing Wollie, improved the most on Stroop andtask-switching, indicating that these participants still lackedthe relevant skills for all these tasks.

People report solving division problems by mentally recastingdivision problems as multiplication (e.g., 72 ÷ 8 à 8 × [?] =72). Mediation of division by multiplication occurs mainly onlarger problems. Eye tracking data was used to determinewhether patterns of gaze durations on division problemsprovided support for mediation. Adults solved divisionproblems in two formats: traditional (e.g., 72 ÷ 8 = [ ]) andrecasted (e.g. 72 = 8 × [ ]). Processing of individual problemelements was compared across formats. Results providesupport for mediation. Processing patterns for traditionally-formatted problems were more similar to those for traditionaldivision in earlier work (72 ÷ 8) whereas problems in recastedformat (72 = 8 × [ ]) were more similar to patterns foundwhen participants solved multiplication problems (e.g., 8 × 9).These findings provide a novel source of support fordifferential processing of problems across presentationformats.

A growing body of research on early word learning suggeststhat learners gather word-object co-occurrence statistics acrosslearning situations. Here we test a new mechanism wherebylearners are also sensitive to word-word co-occurrence statis-tics. Indeed, we find that participants can infer the likely ref-erent of a novel word based on its co-occurrence with otherwords, in a way that mimics a machine learning algorithmdubbed ‘zero-shot learning’. We suggest that the interactionbetween referential and distributional regularities can bring ro-bustness to the process of word acquisition.

In complex task domains, such as games, students may ex-ceed their teachers. Such tasks afford diverse means to trade-off one type of performance for another, combining task ele-ments in novel ways to yield method variations and strategydiscoveries that, if mastered, might produce large or smallleaps in performance. For the researcher interested in the de-velopment of extreme expertise in the wild, the problem posedby such tasks is “where to look” to capture the explorations,trials, errors, and successes that eventually lead to the inven-tion of superior performance. In this paper, we present severalsuccessful discoveries of methods for superior performance.For these discoveries we used Symbolic Aggregate Approx-imation as our method of identifying changepoints withinscore progressions in the venerable game of Space Fortress.By decomposing performance at these changepoints, we findpreviously unknown strategies that even the designers of thetask had not anticipated.

In this paper we evaluate a mechanism for the learning of wordcategories from distributional information against criteria ofpsychological plausibility. We elaborate on the ideasdeveloped by Redington et al. (1998) by embedding themechanism in an existing model of language acquisition(MOSAIC) and gradually expanding the contexts it has accessto in a developmentally plausible way. In line with child data,the mechanism shows early development of a noun category,and later development of a verb category. It is furthermoreshown that the mechanism can maintain high performance atlower computational overhead by disregarding tokenfrequency information, thus improving the plausibility of themechanism as something that is used by language-learningchildren.

Despite substantial evidence for a bidirectional relationshipbetween language and representation, the roots of this relationshipin infancy are not known. The current study explores thepossibility that labels may affect object representations at theearliest stages of language acquisition. We asked parents to playwith their 10-month-old infants with two novel toys for threeminutes, every day for a week, teaching infants a novel word forone toy but not the other. After a week infants participated in afamiliarization task in which they saw each object for 8 trials insilence, followed by a test trial consisting of both objectsaccompanied by the trained word. Infants exhibited a faster declinein looking times to the previously unlabeled object. These dataspeak to the current debate over the status of labels in humancognition, supporting accounts in which labels are an integral partof representation.

Explanation reconstruction performs a crucial role not only inthe progress of science but also in educational practices anddaily activities, including the comprehension of phenomena.In this study, we conducted experiments to examine the factorsthat facilitate shifts in explanations. We focused on the tran-sition of attention on a key fact that contradicts an initial ex-planation and has a central role in its reconstruction. We useda short story as an experimental material in which participantsfirst constructed an initial explanation and then reconstructedit. In the experiment, we controlled the time of presentationof the key fact (bottom-up condition), reflective thinking (top-down condition), and the two together (bidirectional condition)to facilitate understanding of the explanatory shift. The experi-mental results are summarized as follows. First, when the priorexplanation was rejected, attention to the key fact was inhibitedalthough a new explanation was required. Second, the success-ful group increased their attention on the key fact just beforethe explanatory shift. Third, protection of the preceding expla-nation with unobserved facts was inhibited by guiding the par-ticipants ’attention toward the key fact. Finally, although theinitial explanation was not completely shifted, a explanatorypre-shift was achieved by activating reflective thinking withattention to the key fact.

Previous work on concept learning has focused on how con-cepts are acquired without addressing metacognitive aspectsof this process. An important part of concept learning froma learner’s perspective is subjectively knowing when a newconcept has been effectively learned. Here, we investigatelearners’ certainty in a classic Boolean concept-learning task.We collected certainty judgements during the concept-learningtask from 552 participants on Amazon Mechanical Turk. Wecompare different models of certainty in order to determineexactly what learners’ subjective certainty judgments encode.Our results suggest that learners’ certainty is best explained bylocal accuracy rather than plausible alternatives such as totalentropy or the maximum a posteriori hypothesis of an idealizedBayesian learner. This result suggests that certainty predomi-nately reflects learners’ performance and feedback, rather thanany metacognition about the inferential task they are solving.

While a considerable amount of research is done in the fieldof moral psychology, to our best knowledge, no systematicstudy of moral judgments made by professional groups whomake moral decisions as part of their occupational dutiesexists (e.g. firefighters, medical doctors, midwives, policeofficers). By their training and practice, such professionals areexpected to exhibit differences in moral judgment comparedto the general population. Here we report data about moraljudgments of firefighters and midwives using moral dilemmasin which one person must be sacrificed in order to save morepeople. The study reveals that midwives and firefighters areconsiderably less utilitarian compared to a control group ofstudents. Midwives almost never find the utilitarian action tobe permissible. This striking result demonstrates that furtherunderstanding of the specific mechanisms involved in specialprofessional groups’ moral judgment is needed.

Humans have an intuitive sense of how to help and informothers even in the absence of a specific request. How do weachieve this? Here we propose that even young children canreason about others’ expected costs and rewards to flexiblydecide what is best for others. We asked children to chooseone of two toys to teach to another agent while systematicallyvarying the relative costs and rewards of discovering each toy’sfunctions. Children’s choices were consistent with the predic-tions of a computational model that maximizes others’ utilitiesby minimizing their expected costs and maximizing their ex-pected rewards. These results suggest that even early in life,children draw rational inferences about others’ costs and ben-efits, and choose to communicate information that maximizestheir utilities.

The cognitive architecture and function of co-speech gesture has been the subject of a large body of research. We investigate two main questions in this field, namely, whether language and gesture are the same or two inter-related systems, and whether gestures help resolve speech problems, by examining the relationship between gesture and disfluency in neurotypical speakers. Our results support the view of separate, but inter- related systems by showing that speech problems do not necessarily cause gesture problems, and on many occasions, gestures signal an upcoming speech problem even before it surfaces in overt speech. We also show that while gestures are more common on fluent trials, speakers use both iconic and beat gestures on disfluent trials to facilitate communication, although the two gesture types support communication in different ways.

A key function of categories is to help predictions about unob-served features of objects. At the same time, humans often findthemselves in situations where the categories of the objectsthey perceive are uncertain. How do people make predictionsabout unobserved features in such situations? We propose arational model that solves this problem. Our model comple-ments existing models in that it is applicable in settings wherethe conditional independence assumption does not hold (fea-tures are correlated within categories) and where the featuresare continuous as opposed to discrete. The qualitative predic-tions of our model are borne out in two experiments.

Two experiments were conducted to examine the role of threefacets of numeracy (objective (ONS), subjective (SNS), andsymbolic number mapping (SMap)) in three anchoring tasks(experimenter-given, self-generate, and valuation). We foundthat the three numeric competencies were associated withdifferent anchoring tasks. SMap was associated with none ofthe three anchor tasks, while ONS consistently predictedstronger susceptibility to self-generated anchoring. The roleof ONS and SNS in experimenter-given and valuation taskswere inconsistent. In Experiment 1, where the direction ofadjustment from an anchor is specified, ONS and SNS werepositively associated with anchor susceptibility in a valuationtask, while they were not in an experimenter-given anchortask. On the other hand, in Experiment 2 where the directionof adjustment from an anchor is uncertain, ONS and SNSwere positively associated with anchor susceptibility in anexperimenter-given anchor task, while they were not in avaluation task.

Distributed cognition is a perspective that primarily has been applied to complex socio-technical systems such as flight decks of commercial airliners, or operating rooms where professionals perform cognitive tasks in environments specifically designed for this. For some scholars distributed cognition is exactly this kind of specialized cognitive system. On the other hand it has been claimed by some workers in the field that distributed cognition is not a kind of cognition but a perspective on all cognition. We have therefore studied an environment very different from the systems previously studied, namely single people’s homes. We find that there are many similarities between the home and the specialized socio-technical environments. To us this suggests that the specially designed complex environments can be seen as specialized cases of the general principles of distributed cognition which are not reflections of “particular work practices” but of general features of human cognition.

Sequential learning (SL) refers to the ability to learn thetemporal and ordinal patterns of one’s environment. Whereasresearch on the learning of ordinal patterns is common, thelearning of temporal patterns within sequential events hasbeen far less studied. The current study examines the effectsof synchronous and asynchronous temporal patterns on visualsequential learning. We hypothesize that entrainment (i.e.exposure to a regular rhythmic pattern) allows for betterprocessing of the ordinal structure of sequential events.Twenty healthy adult participants (11 females, 18–34 yearsold) performed two versions of a visual sequential learningparadigm while event-related potentials (ERPs) wererecorded. The SL task involved the visual presentation ofcolored circles, wherein a target circle was embedded thatwas partially predictable based on preceding predictorstimuli. One version of the task incorporated synchronoustemporal presentation of the stimuli whereas the other versioninvolved asynchronous presentation of stimuli using arandomized ISI on every trial. Reaction time datademonstrated that learning occurred in both temporalconditions. On the other hand, the mean ERP amplitudesbetween 350 and 750ms post-predictor onset in the posteriorregions of interest revealed that learning of the statisticalcontingencies between stimuli was disrupted for theasynchronous temporal condition but intact for thesynchronous condition. These neurophysiological datasuggest that the brain processes regular and irregular timingevents differently, with statistical learning of ordinal visualpatterns being improved by a synchronous temporal structure,possibly a result of heightened attention to the stimuli due toentrainment.

How much do individuals, compared to the population, knowabout the distribution of values in the world? Participantsreported the prices of consumer goods such as watches andbelts and we compared how accurately individuals vs. theoverall population knew the mean and dispersion of prices.Although individuals and the population both knew objects’average prices and relative standard deviations, the populationwas more sensitive to the absolute standard deviation ofprices. In a second experiment, we examined whetherindividuals’ impoverished distribution knowledge impairstheir ability to interpret advertisements. Consistent withpeople using Bayesian inference, the higher an object’s actualprice dispersion, the more participants relied onadvertisements; however, this effect is considerably smallerthan a simple proportional offset, suggesting again thatindividuals underestimate dispersion. Thus, despite having asense of the distribution of real world quantities, individualstend to know only a fraction of the world distribution.

Representations of numerical value have been assessed usingbounded (e.g., 0-1000) and unbounded (e.g., 0-?) number-linetasks, with considerable debate regarding whether one or bothtasks elicit unique cognitive strategies (e.g., addition orsubtraction) and require unique cognitive models. To test this,we examined 86 5- to 9-year-olds' addition, subtraction, andestimation skill (bounded and unbounded). Against themeasurement-skills hypothesis, estimates were even morelogarithmic on unbounded than bounded number lines andwere better described by conventional log-linear models thanby alternative cognitive models. Moreover, logarithmic indexvalues reliably predicted arithmetic scores, whereas modelparameters of alternative models failed to do so. Resultssuggest that the logarithmic-to-linear shift theory provides aunified framework for numerical estimation with highdescriptive adequacy and yields uniquely accurate predictionsfor children’s early math proficiency.

When learning the relation between a cause and effect, howdo people control for all the other factors that influence thesame effect? Two experiments tested a hypothesis that peoplefocus on events in which the target cause changes and allother factors remain stable. In both four-cause (Experiment 1)and eight-cause (Experiment 2) scenarios, participants learnedcausal relations more accurately when they viewed datasets inwhich only one cause changed at a time. However,participants in the comparison condition, in which multiplecauses changed simultaneously, performed fairly well; inaddition to focusing on events when a single cause changed,they also used events in which multiple causes changed forupdating their beliefs about causal strength. These findingshelp explain how people are able to learn causal relations insituations when there are many alternative factors.

Text cohesion is an important element of discourseprocessing. This paper presents a new approach to modeling,quantifying, and visualizing text cohesion using automatedcohesion flow indices that capture semantic links amongparagraphs. Cohesion flow is calculated by applyingCohesion Network Analysis, a combination of semanticdistances, Latent Semantic Analysis, and Latent DirichletAllocation, as well as Social Network Analysis. Experimentsperformed on 315 timed essays indicated that cohesion flowindices are significantly correlated with human ratings of textcoherence and essay quality. Visualizations of the globalcohesion indices are also included to support a more facileunderstanding of how cohesion flow impacts coherence interms of semantic dependencies between paragraphs.

This study explores the relationship between beliefs aboutself-control and the ability to exercise self-control in 4- to 6-year-old children. Sixty-eight children were asked a series ofquestions to gauge whether they believed that they couldfreely choose to act against their desires or inhibit themselvesfrom performing desired actions. Children were also asked toprovide qualitative explanations for why they could or couldnot exercise free will, and to complete two inhibitory controltasks: forbidden toy and day/night. Choice responses werenegatively correlated with performance on the forbidden toytask, when children performed that task first. There was also anegative correlation between a belief in an internal locus ofcontrol, and success on the forbidden toy measure. Refrainingfrom touching a forbidden toy appears to be correlated to lessbelief in free will. Though this may appear counter-intuitive,it is consistent with cross-cultural research.

Stereotypes are important simplifying assumptions we usefor navigating the social world, associating traits withsocial categories. These beliefs can be used to infer anindividual’s likely social category from observed traits (adiagnostic inference) or to make inferences about anindividual’s unknown traits based on their putative socialcategory (a predictive inference). We argue that theseinferences rely on the same explanatory logic as other sortsof diagnostic and predictive reasoning tasks, such as causalexplanation. Supporting this conclusion, we demonstratethat stereotype use involves four of the same biases knownto be used in causal explanation: A bias against categoriesmaking unverified predictions (Exp. 1), a bias towardsimple categories (Exp. 2), an asymmetry betweenconfirmed and disconfirmed predictions of potentialcategories (Exp. 3), and a tendency to treat uncertaincategorizations as certainly true or false (Exp. 4).

Do children take into account their addressees’ needs in spontaneous production? Developmental evidence for speaker adjustments is mixed. Some studies show that children are often under-informative when communicating with ignorant addressees but other studies demonstrate successes in children’s ability to integrate another person’s perspective. We asked whether children adapt their event descriptions depending on (a) the typicality of event components, and (b) the listener’s visual access to the events. We found that children’s ability to use information about the listener’s visual perspective to make specific adjustments to event descriptions emerged only in highly interactive contexts, in which participants collaborated towards mutual goals.

Humans turn abstract referents and discourse structuresinto gesture using metaphors. The semantic relation be-tween abstract communicative intentions and their phys-ical realization in gesture is a question that has not beenfully addressed. Our hypothesis is that a limited set ofprimary metaphors and image schemas underlies a widerange of gestures. Our analysis of a video corpus sup-ports this view: over 90% of the gestures in the corpus arestructured by image schemas via a limited set of primarymetaphors. This analysis informs the extension of a com-putational model that grounds various communicative in-tentions to a physical, embodied context, using those pri-mary metaphors and image schemas. This model is usedto generate gesture performances for virtual characters.

To reveal the fundamental skills involved in slacklining,this study examined a hypothesis regarding single-legstanding on a slackline. In the field of practice,instructors teach learners how to maintain balance on aswinging flat belt (slackline), such as by moving theirhands in parallel. We hypothesized that bimanualcoordination in the horizontal direction mightcontribute to dynamic balancing on a slackline. In ourpilot study, two participants at different skill levelswere asked to maintain their balance on a slackline aslong as possible. The dynamic stability of bimanualcoordination was assessed by a nonlinear time seriesanalysis (cross recurrence quantification analysis), thencompared among the participants. Bimanualcoordination stability was higher in the experiencedplayer than in the novice player. The results suggestthat the single-leg standing skill might be correlatedwith bimanual coordination stability. Furtherinvestigations are expected to clarify this notion in thefuture.

We offer an account of the role of normality—both statisti-cal and prescriptive—in judgments of actual causation. Us-ing only standard tools from the literature on causal cognition,we argue that the phenomenon can be explained simply on theassumption that people stochastically sample (counterfactual)scenarios in a way that reflects normality. We show that a for-malization of this idea, giving rise to a novel measure of causalstrength, can account for some of the most puzzling qualitativepatterns uncovered in recent experimental work

When A causes B and B causes C, under what conditions isA a good explanation for the occurrence of C? We proposethat distal causes are only perceived to be explanatory if thecausal mechanism is insensitive to inessential variations ofboundary conditions. In two experiments, subjects first ob-served deterministic A → B → C relationships in a single ex-emplar of an unknown kind. They judged A to be crucial forC by default. However, when they subsequently learned thatthe causal mechanism fails to generate the A → C dependencyin other exemplars of the same kind, subjects devalued A asa crucial explanation for C even within the first exemplar. Werelate these findings to the idea that good explanations pick outportable dependency relations, and that sensitive causes fail tomeet this requirement.

Collaborative problem solving based on different perspectivesis an effective strategy for constructing new knowledge anddiscoveries. It remains unclear what kind of interaction pro-cess underlies development of an abstract or integrated per-spective upon experiencing conflict with different perspectivesin a group. The present study investigates two factors in anexperimental setting: (1) groups with a single opposing per-spective (maverick) would hold an advantage over groups and(2) groups with positive moods would hold an advantage overgroups with negativity. We investigate the factors influencingperspective taking in problem-solving groups using conversa-tional agents. Results showed that (1) a single different per-spective in the group can be accepted for perspective takingcompared to several members with an opposing perspective,and (2) positive mood generated by group members facilitat-ing perspective taking compared to negative mood.

The current study investigated how interactivity of simulation controls affects data collection in science inquiry. A chemistry simulation was designed to allow either low or high interactivity in setting experimental variables. Adult participants were randomly assigned to one of the interactivity conditions and solved a series of assessment items. The results from the first item indicated that the highly interactive controls posed challenges in conducting a thorough investigation. Performance in the last item which is a repetition of the first item suggested that the participants were able to overcome the initial challenges over the course of their investigations. The results provide implications for designing educational simulations for learning and assessment.

Causal learning is a fundamental ability that enables human reasoners to learn about the complex interactions in the world around them. The available evidence with children and adults, however, suggests that the mechanism or set of mechanisms that underpins causal perception and causal reasoning are not well understood; that is, it is unclear whether causal perception and causal reasoning are underpinned by a Bayesian mechanism, associative mechanism, or both. It has been suggested that a Bayesian mechanism, rather than an associative mechanism, underpins causal reasoning because such a mechanism can better explain the putative backward- blocking finding in children and adults (e.g., Sobel, Tenenbaum, & Gopnik, 2004). In this paper, we report two experiments to examine to what extent infants and adults exhibit backward blocking and whether humans’ ability to reason about causal events is underpinned by an associative mechanism, a Bayesian mechanism, or both.

Qualitative Process theory provides a formal representation for human-like models of continuous processes. Prior research mapped qualitative process elements onto English language constructions, but did not connect the representations to existing frame semantic resources. Here we identify and classify QP language constituents through their instantiation in FrameNet frames to provide a unified semantics for linguistic and non-linguistic representations of processes. We demonstrate that all core QP relations can map to FN, though larger QP evoking phrasal constructions do exist outside of this mapping. We conclude with a corpus analysis showing that these frames occur in natural text involving a variety of continuous processes.

Generic sentences (e.g., “birds lay eggs”) express generaliza-tions about kinds, in contrast to non-generic sentences thatexpress facts about specific individuals or sets of individuals(e.g., “all birds lay eggs”). Although generics are pervasive innatural language, there is no unique linguistic marker of gener-icity, making the identification of generics a challenge. We in-vestigate the morphosyntactic cues that listeners use to identifywhether a sentence should receive a generic interpretation ornot. We find that two factors – the definiteness of a sentence’ssubject NP and the tense of the sentence – are extremely im-portant in guiding intuitions about whether a sentence shouldreceive a generic interpretation. We argue that the importanceof these factors can be explained by taking generic interpreta-tions to arise due to a failure to ground expressions as referringto specific entities or events.

The ability to inhibit the processing of irrelevant informationdeclines as adults age (Hasher & Zacks, 1988; Lustig, Hasherand Tonev, 2006; Mayr, 2001). However, previous researchinvestigating inhibitory control in older adults has notevaluated the extent to which irrelevant information isprocessed and later recognized. Using a dual task paradigmwith young adults, Dewald, Sinnett, and Doumas (2011)demonstrated inhibited recognition for previously ignoredwords, provided they had appeared infrequently with targetsin the primary task, compared to words that did not appearwith targets. The current study adapted this paradigm toexamine inhibitory mechanisms in a sample of older adults.Here, older adults exhibited inhibited recognition for allwords while young adults continued to show greaterinhibition for words that had appeared with targets comparedto words that had not. This finding suggests that older adultsmay experience a decline in the selective inhibition ofirrelevant information.

Recent research has revealed two different eye movement pat-terns during face recognition: holistic and analytic. The pre-sent study investigated the neural correlates of these two pat-terns through functional magnetic resonance imaging (fMRI).A more holistic pattern was associated with more activation inthe face-selective perceptual areas, including the occipitalface area and fusiform face area. In contrast, participants us-ing a more analytic pattern demonstrated more activation inareas important for top-down control of visual attention, in-cluding the frontal eye field and intraparietal sulcus. In addi-tion, participants using the analytic patterns had better recog-nition performance than those showing holistic patterns. The-se results suggest that analytic eye movement patterns are as-sociated with more engagement of top-down control of visualattention, which may consequently enhance recognition per-formance.

In recent years, there has been an increasing interest in design-ing social robots to interact with people to provide therapy andcompanionship. Most social robots currently being used arelight-weight and much smaller in size compared to people. Inthis work, we investigate designing interactions for larger andmore physically capable robots as they have more potential toassist people physically. A modified version of Baxter robotwas used, by sitting Baxter on top of an electronic wheelchair.Two experiments were designed for studying the role of facialexpressions and body movements in establishing trust with theuser and for expressing attitudes. Our results suggest that therobot is capable of expressing fine and distinguishable attitudes(proud vs. relaxed) using its body language, and the couplingbetween body movements and speech is essential for the robotto be viewed as a person.

Popular approaches to modeling analogical reasoning havecaptured a wide range of developmental and cognitivephenomena, but the use of structured symbolicrepresentations makes it difficult to account for the dynamicand context sensitive nature of similarity judgments. Here, theresults of a novel behavioral task are offered as an additionalchallenge for these approaches. Participants were presentedwith a familiar analogy problem (A:B::C:?), but with a twist.Each of the possible completions (D1, D2, D3), could beconsidered valid: There was no unambiguously “correct”answer, but an array of equally good candidates. We find thatparticipants’ recent experience categorizing objects (i.e.,manipulating the salience of the features), systematicallyaffected performance in the ambiguous analogy task. Theresults are consistent with a dynamic, context sensitiveapproach to modeling analogy that continuously updatesfeature weights over the course of experience

Curiosity has a tumultuous past. Originally curiosity wasconsidered a vice of excess leading to misconduct anddisaster. Recently, curiosity has transformed into a virtue ofself-expression resulting in success and better performance. Inclassrooms, educators try to find ways of eliciting curiosityfrom their students: allowing them to pick their own researchtopics and books, including pop culture references in lecture,and many more strategies. Recent adult studies have revealedbetter memory for trivia facts that elicit more curiosity. Thecurrent study modifies the methods used in previous adultstudies in order to make them more appropriate for children.Results from a sample of 24 7- and 8-year-olds reveal that byage eight curiosity significantly affects memory for triviafacts. This research may shed light on the cognitiveadvantages of curiosity and legitimatize the encouragement ofcuriosity in classrooms for school age children.

We report a self-paced reading study that investigated garden-path sentences like While the boy washed {a/the} dog barkedloudly and While the man hunted {a/the} deer ran into thewoods. In such sentences, the critical noun phrase (dog, deer)tends to be misparsed as an object of the preceding verb, andhas to be re-analyzed as a subject of the following clause whenthe disambiguating verb (e.g. barked, ran) is encountered. Tobetter understand how discourse level information guides real-time processing, we build on earlier corpus work in linguisticswhich found a relationship between syntactic function andinformation status: Entities in subject position tend to bealready-mentioned (old/given) information and definite, whileentities in object position are typically new information andindefinite. We investigated whether the information status ofthe ambiguous noun influences the extent of processingdifficulty, and whether this effect also depends on the argumentstructure of the first verb. Results from self-paced readingshowed that information status matters when processing theambiguous NP after optionally transitive verbs (e.g. hunt) butnot after reflexive absolute verbs (e.g. wash). These resultssuggest that access to discourse-level representations during re-analysis of the noun phrase is modulated by verb argumentstructure

Rendaku, or sequential voicing, is a morphophonemic process in Japanese in which the voiceless word-initial consonant of the second element (=E2) of a compound word becomes voiced (e.g., ori + kami →origami, ‘folding’ + ‘paper’ → ‘paper folding’, /k/ becomes /g/). In adult grammar, rendaku is subject to two conditions: It applies if and only if (a) E2 is a Yamato word (native vocabulary) in the lexicon and (b) it contains no voiced consonant (e.g., b, d, & g). Recent psycholinguistic studies have revealed that Japanese-speaking preschoolers do not follow adult’s grammar; they develop their original prosodically-based rendaku processing strategy (preschooler-specific rendaku strategy). Their strategies qualitatively change in the early middle childhood to be adult-like rendaku, creating a discontinuity in children’s word-processing strategies. This study investigated factors responsible for this developmental discontinuity. We conducted an experiment using cross- modal linguistic stimuli (prosody & orthography) to see whether children’s orthographic knowledge affects their rendaku strategy or not. Our results showed that orthographic cues affected literate children’s rendaku processing. They were aware the correspondence between types of orthography and word categories in Japanese.

Experimental research suggests that at birth infants can dis-criminate two languages if they belong to different rhythmicclasses, and by 4 months of age they can discriminate two lan-guages within the same class provided they have been previ-ously exposed to at least one of them. In this paper, we presenta novel application of speech technology tools to model lan-guage discrimination, which may help to understand how in-fants achieve high performance on this task. By combininga Gaussian Mixture Model of the acoustic space and low-dimensional representations of novel utterances with a modelof a habituation paradigm, we show that brief exposure toFrench does not allow to discriminate between two previouslyunheard languages with similar phonological properties, butfacilitates discrimination of two phonologically distant lan-guages. The implications of these findings are discussed.

We test the effects of advice and trust on risk-taking in three online experiments designed to elucidate under what conditions financial advice may increase risk-taking, irrespective of advisor performance. In our study, investors made 100 decisions, selecting between one of two alternatives: risky or conservative. We manipulate the suggestion of an advisor (risky vs. non-risky investments), the fee of the advice, as well as the trustworthiness of the advisor (by increasing the transparency of the advice presented) to test the effect of the advice on risk-taking. The results show that individuals asymmetrically follow the advice they received, with a bias towards following more risky than conservative advice. Moreover, trusted advice was more persuasive irrespective of what the advisor suggested and even the fee is higher.

What capacities enable linguistic interaction? While severalproposals have been advanced, little progress has been made incomparing and articulating them within an integrative frame-work. In this paper, we take initial steps towards a connec-tionist framework designed to systematically compare differ-ent cognitive models of social interactions. The frameworkwe propose couples two simple-recurrent network systems(Chang, 2002) to explore the computational underpinnings ofinteraction, and apply this modeling framework to predict thesemantic structure derived from transcripts of an experimen-tal joint decision task (Bahrami et al., 2010; Fusaroli et al.,2012). In an exploratory application of this framework, wefind (i) that the coupled network approach is capable of learn-ing from noisy naturalistic input but (ii) that integration of pro-duction and comprehension does not increase the network per-formance. We end by discussing the value of looking to tra-ditional parallel distributed processing as flexible models forexploring computational mechanisms of conversation.

How does the mental lexicon, the network of learned words inour semantic memory, change in old age? To address thisquestion, we employ a new network inference method to infernetworks from verbal fluency data of a group of younger andolder adults. We find that older adults produce more uniquewords in verbal fluency tasks than younger adults. In line withrecent theorizing, this suggests a larger mental lexicon forolder than for younger adults. Moreover, we find that relativeto the mental lexicon of younger adults, the mental lexicon ofolder adults is less small-world-like. Based on severalfindings linking network clustering to processing speed, thisfinding suggests that not only the size, but also the structureof the mental lexicon may contribute to apparent cognitivedecline in old age.

Preschool-aged children develop awareness of the words theydo and do not know. Awareness of one’s lexicon mayencourage word learning if children pay more attention to thedefinition of unknown words. Here, we tested 3-4-year-oldchildren (N = 91) on a word learning task embedded in an e-book. When a novel word was read, children were eitherasked if they knew the word, asked a question about thestoryline, or asked no question. Then they were given adescription without visual input and asked to identify thereferent’s picture from three choices. Participants who wereasked if they knew a word before being provided with thedefinition identified more referents than children in the otherconditions. Children’s word learning was predicted by short-term memory.

In groups and organizations, agents use both individual and so-cial learning to solve problems. The balance between these twoactivities can lead collectives to very different levels of perfor-mance. We model collective search as a combination of simplelearning strategies to conduct the first large-scale comparativestudy, across fifteen challenging environments and two differ-ent network structures. In line with previous findings in thesocial learning literature, collectives using a hybrid of individ-ual and social learning perform much better than specialistsusing only one or the other. Importantly, we find that collec-tive performance varies considerably across different task en-vironments, and that different types of network structures canbe superior, depending on the environment. These results sug-gest that recent contradictions in the social learning literaturemay be due to methodological differences between two sepa-rate research traditions, studying disjoint sets of environmentsthat lead to divergent findings.

To successfully navigate the complex social world, peopleoften need to solve the problem of perspective selection:Between two conflicting viewpoints of the self and the other,whose perspective should one take? In two experiments, weshow that four-year-olds use others’ knowledge and goals todecide when to engage in visual perspective taking. Childrenwere more likely to take a social partner’s perspective todescribe an ambiguous symbol when she did not knownumbers and wanted to learn than when she knew numbersand wanted to teach. These results were shown in children’sown responses (Experiment 1) and in their evaluations ofothers’ responses (Experiment 2). By preschool years,children understand when perspective taking is appropriateand necessary and selectively take others’ perspectives insocial interactions. These results provide novel insights intothe nature and the development of perspective taking.

The purpose of this study is to build a monaural auditorydisplay to convey four pieces of directional information(upward, downward, rightward, and leftward) to userseffectively and intuitively without the need for wearingheadphones or preparing more than one speaker. We preparedfive types of monaural auditory displays consisting of trianglewave sounds and conducted an experiment to investigatewhich kinds of displays succeeded in conveying the fourpieces of information to participants. As a result, we couldconfirm that one of the prepared monaural auditory displays,designed as a “progress bar” on the basis of the mental-number line and spatial-number association of the responsecode effect, succeeded in conveying the four pieces ofinformation more effectively compared with the othercandidate sets (its average correct rates were about 0.88). Thisresult thus strongly shows that this monaural auditory displaywas quite useful for conveying primitive spatial informationto users

Social stimuli are a highly salient source of information, and seem to possess unique qualities that set them apart from other well-known categories. One characteristic is their ability to elicit spatial orienting, whereby directional stimuli like eye- gaze and pointing gestures act as exogenous cues that trigger automatic shifts of attention that are difficult to inhibit. This effect has been extended to non-social stimuli, like arrows, leading to some uncertainty regarding whether spatial orienting is specialized for social cues. Using a standard spatial cueing paradigm, we found evidence that both a pointing hand and arrow are effective cues, but that the hand is encoded more quickly, leading to overall faster responses. We then extended the paradigm to include multiple cues in order to evaluate congruent vs. incongruent cues. Our results indicate that faster encoding of the social cue leads to downstream effects on the allocation of attention resulting in faster orienting.

Differences in working memory capacity (WM C) relate to performance on a variety of problem solving tasks. High WM C is beneficial for solving analytical problems, but can hinder performance on insight problems (DeCaro & Beilock, 2010). One suggested reason for WM C-related differences in problem solving performance is differences in strategy selection, in which high WM C individuals tend toward complex algorithmic strategies (Engle, 2002). High WM C might increase the likelihood of non- optimal performance on Luchins’ (1942) water jar task because high WM C solvers tend toward longer solutions, not noticing when shorter solutions become available. We present empirical data showing this effect, and a computational model that replicates the findings by choosing among problem solving strategies with different WM demands. The high WM C model used a memory- intensive strategy, which led to long solutions when shorter ones were available. The low WM C model was unable to use that strategy, and switched to shorter solutions.

An argument is developed to show that explanations ofbiological and physical systems can be unified via the secondlaw of thermodynamics (SLT). The SLT’s influence on theevolutionary history of life at the scale of the global Earthsystem justifies reunifying phenomena—i.e., mind andmatter—whose separation dates back to Modern Westernphilosophy and still influences contemporary scientificinvestigations. From this perspective it appears that thenecessity of ever-increasing entropy in nature may constrainthe organization and behavior of living organisms andcognitive processes. Via an example of explaining memory atthe scale of the brain-body-environment system, werecommend understanding cognition with respect to its role inincreasing entropy in nature. This framework may lead to afruitful understanding of cognition by appealing to thenecessity of physical laws.

Do students activate conceptual and procedural knowledgesimultaneously when learning fraction addition? In groundedfeedback, student actions on a target, to-be-learnedrepresentation are reflected in a more familiar feedbackrepresentation to promote conceptual learning withinprocedural practice. An experiment with 163 4th and 5thgraders shows improved learning with a grounded feedbacktutor over a symbols-only control with step-level right/wrongfeedback. Learning with grounding also transferred tosymbols-only assessment items, providing some support forthe simultaneous activation view.

Recent research has suggested that using a foreign languageto present hypothetical moral dilemmas increases the rate ofutilitarian judgments about those dilemmas (e.g., Greene et al,2001) and decreases incoherency between judgments inframing effect tasks (e.g., Tversky & Kahneman, 1981; seeCosta, Foucart, Arnon, Aparici, & Apesteguia, 2014; Costa,Foucart, Hayakawa, Aparici, Apesteguia, Heafner, & Keysar,2014; Keysar, Hayakawa, & An, 2012). However, existingresearch has mainly investigated this effect using between-participants designs (i.e., different participants in the foreignand native language conditions). Such designs are unable toexclude non-equivalent conditions as a confounding variable.In contrast, this study examined the foreign language effectusing a within-subjects design (i.e., all participants respondedto moral dilemmas (Greene et al, 2001) and framing effecttasks (Tversky & Kahneman, 1981) in both their native andforeign languages. The “foreign language effect” wasreplicated, excluding semantic non-equivalence betweenlanguage conditions as a potential confound. This resultsupports the hypothesis that the foreign language effect isindependent of meaning.

We often make decisions on behalf of others, such as pickingout gifts or making restaurant recommendations. Yet, withoutdirect access to others’ preferences, our choices on behalf ofothers depend on what we think they like. Across twoexperiments, we examined whether and how accuratelypeople are able to infer others’ preferences by observing theirchoices. Our results suggest that people are capable of makingreasonably accurate predictions about what others will choosenext, given what they have chosen before. These results laythe groundwork to systematically study how people makenovel predictions about others’ preferences, and whendifferent strategies might be appropriate.

Attention is a factor that affects the performance of various in-telligent activities in humans. Up until now, the methods formeasuring the level of attention have been mostly based onsubjective reports or employing large and costly devices. Inthis paper, a new method of estimating the level of attentionis proposed, based on posture and EEG measurements. Thesedata can be recorded using easily available and less burden-some devices. From the obtained data, the time evolution ofattention was explored. Experiments showed that there is neg-ative correlation between posture variance and attention, andalso between EEG and attention.

Though communicative goals are an important element in lan-guage production, few studies investigate the extent to whichthese goals might affect the form and content of referring ex-pressions. In this study, we directly contrast two tasks withdifferent goals: identification and instruction giving. Speak-ers had to refer to a target building nearby or further away, sothat their addressee would distinguish it between other build-ings (identification) or give route directions and use the samebuilding as a landmark (instructions). Our results showed thatirrespective of goals, the referring expressions consisted of thesame types of attributes, yet the attribute frequency and for-mulation differed. In the identification task, references werelonger, contained more locative and more post-nominal mod-ifiers. In addition, referential choices were influenced by thevisual distance between the speaker and the target: when thespeaker observed the target from far, references were longerand contained more often locative modifiers.

A remarkable ability of the cognitive system is the creation of new knowledge based on prior experiences. What cognitive mechanisms support such knowledge creation? We propose that statistical learning not only extracts existing relationships between objects, but also generates new associations between objects that have never been directly associated. Participants viewed a continuous color sequence consisting of base pairs (e.g., A-B, B-C), and learned these pairs. Importantly, they also successfully learned a novel pair (A-C) that could only be associated through transitive relations between the base pairs (Exp1). This learning, however, was not successful with three base pairs (e.g., learning A-D from A-B, B-C, C-D), revealing a limit in this transitive process (Exp2). Beyond temporal associations, novel transitive associations can also be formed across categorical hierarchies (Exp3), but with limits (Exp4&5). The current findings suggest that statistical learning provides an efficient scaffold through which new object associations are transitively created.

Predictions about information search behavior have been informed by extensive research in food foraging behavior. However, information foraging environments may differ in key ways from food foraging environments, and these differences may impact search behavior. We investigated the effect of patch distribution (depleting or non-depleting) and ability to return to previously searched patches on participants’ decision to switch from one patch to another while searching. Whether or not a participant could return after leaving a patch led to fewer samples and fewer relevant items found. Whether or not the patches depleted and whether it was possible to return to a patch influenced stopping rules, indicating that these factors may alter the size of the increment applied through the Incremental Rule.

Recent research on creative thinking has implicatedconceptual expansion as potential cognitive underpinnings.These theories were examined within the context of alaboratory study using two divergent thinking prompts.Participants generated alternative/creative uses for a brick andfor a glass bottle (separately) for two minutes and responseswere time-stamped using a Matlab GUI. Semantic distancesbetween responses and conceptual representations of the DTprompts were computed using latent semantic analysis.Results showed that semantic distance increased asresponding progressed, with significant differences betweenthe two tasks, and intraparticipant variation. Results haveimplications for theories of creative thinking and representmethodological and analytic advances in the study ofdivergent thinking.

Although the causal graphical model framework has achievedconsiderable success accounting for causal learning data, appli-cation of that formalism to multi-cause situations assumes thatpeople are insensitive to the statistical properties of the causesthemselves. The present experiment tests this assumption byfirst instructing subjects on a causal model consisting of twoindependent and generative causes and then requesting them tomake data likelihood judgments, that is, to estimate the proba-bility of some data given the model. The correlation betweenthe causes in the data was either positive, zero, or negative. Thedata was judged as most likely in the positive condition andleast likely in the negative condition, a finding that obtainedeven though all other statistical properties of the data (e.g.,causal strengths, outcome density) were controlled. These re-sults pose a problem for current models of causal learning.

A great deal of research has been conducted on how humansreason about probability, yet it remains unknown what mentalcomputations support this ability. Research on thedevelopment of the Approximate Number Sense (ANS) hasshown that performance in a magnitude (i.e., estimations ofintegers) discrimination task is well fit by a psychophysicalmodel (Halberda & Feigenson, 2008). Whether or notestimations of integers plays a role in probability judgmentshas yet to be investigated. In the present study we use datafrom two adult experiments as well as results fromcomparisons of two computational models to investigate thepotential relationship between the ANS and probabilityjudgments.

False-belief task have mainly been associated with the ex-planatory notion of the theory of mind and the theory-theory.However, it has often been pointed out that this kind of high-level reasoning is computational and time expensive. Dur-ing the last decades, the idea of embodied intelligence, i.e.complex behavior caused by sensorimotor contingencies, hasemerged in both the fields of neuroscience, psychology andartificial intelligence. Viewed from this perspective, the fail-ing in a false-belief test can be the result of the impairment torecognize and track others’ sensorimotor contingencies and af-fordances. Thus, social cognition is explained in terms of low-level signals instead of high-level reasoning. In this work, wepresent a generative model for optimal action selection whichsimultaneously can be employed to make predictions of others’actions. As we base the decision making on a hidden state rep-resentation of sensorimotor signals, this model is in line withthe ideas of embodied intelligence. We demonstrate how thetracking of others’ hidden states can give rise to correct false-belief inferences, while a lack thereof leads to failing. Withthis work, we want to emphasize the importance of sensorimo-tor contingencies in social cognition, which might be a key toartificial, socially intelligent systems.

Humans fail to understand the world around them and alsofail to recognize this lack of understanding. The illusion ofexplanatory depth (IOED) exemplifies these failures: peoplebelieve they understand the world more deeply than theyactually do and only realize that this belief is an illusion whenthey attempt to explain elements of the world. An unexploredfactor of the IOED is how people may become overconfidentby confusing their own understanding with others’understanding. In two experiments, we examine the IOED inmental disorders, a domain where society has a limitedunderstanding. In Experiment 1, we demonstrate that peopledisplay an IOED for mental disorders as well as devices, butthat it is smaller for mental disorders. In Experiment 2, weshow that exposing the IOED is specifically linked togenerating an explanation, rather than more generallythinking about a phenomenon.

We routinely encounter speakers with different accents andspeaking styles. The speech perception literature offersexamples of disruption of comprehension for unfamiliarspeech and also of listeners’ rapid accommodation tounfamiliar accents. Much of this research uses a singlemeasure and/or focuses on isolated word perception. Weinvestigated listeners’ abilities to comprehend and shadowconnected speech spoken in a familiar or unfamiliar accent.We found increases in shadowing latencies andcomprehension errors in the Dissimilar Speech relative toSimilar Speech conditions—especially for relatively informalrather than more academic style speech. Additionally, therewas less accommodation over time to Dissimilar than SimilarSpeech. These results suggest that there are costs both in theimmediate timescale of processing speech (necessary forshadowing) and in the longer time scale of listeningcomprehension when accent and other speech quality is verydifferent from one’s own speech.

Hidden Markov models (HMM) can describe the spatial andtemporal characteristics of eye-tracking recordings incognitive tasks. Here, we introduce a new HMM approach.We developed HMMs based on fixation locations and we alsoused image information as an input feature. We demonstratethe benefits of the newly proposed model in a facerecognition study wherein an HMM was developed for everysubject. Discovery of regions of interest on facial stimuli isimproved as compared with earlier approaches. Moreover,clustering of the newly developed HMMs lead to very distinctgroups. The newly developed approach also allowsreconstructing image information at each fixation.

Social interaction involves cues such as gaze direction, headorientation, and pointing gestures that serve to automaticallyorient attention to a specific referent or spatial location. In thispaper we demonstrate that an observed reaching actionsimilarly results in a reflexive shift in attention as evidencedby faster responses that are congruent with the direction of thereach, than responses that are incongruent. This facilitation isevident quickly after the onset of the reach action and is dueto the rapid prediction of the reach-goal. When the taskinvolves a saccadic response (Experiment 1) this prediction isinhibited and results in a reverse-congruence, faster responsesto incongruent than congruent cues, when the cue occurs afterthe reach is completed. This reverse-congruence is not presentwhen the task involves a key press (Experiment 2) or a mousemovement (Experiment 3). We propose that the inhibition ofthe predictive saccade is overcome when the eye movementstoward the goal are activated to guide the mouse movement.The three experiments together demonstrate that automaticattention distribution and its effects on behavior depend onthe response.

We investigate the accuracy development of the Englisharticle by learners of English as a second language. The studyfocuses on individual learners, tracking their learningtrajectories through their writings in the EF-Cambridge OpenLanguage Database (EFCAMDAT), an open access learnercorpus. We draw from 17,859 writings by 1,280 learners andask whether article accuracy in individual learners fluctuatesrandomly or whether learners can be clustered according totheir developmental trajectories. In particular, we apply k-means clustering to automatically cluster in a bottom upfashion learners with similar learning curves. We followlearners for a period covering one CEFR level. Given therelatively short learning window, the majority of learnersfollow a horizontal line. Nevertheless, we also identify groupsof learners showing a power-function and U-shaped curve.Crucially, these groups are ‘hidden’ when the aggregate oflearners is considered, a finding highlighting the importanceof individual level analysis.

People implicitly associate different emotions with differentlocations in left-right space. Which dimensions of emotion dothey spatialize? Across many studies people spatializeemotional valence, mapping positive emotions onto theirdominant side of space and negative emotions onto their non-dominant side. Yet, other results suggest a contradictorymapping of emotional intensity (a.k.a., emotional magnitude),according to which people associate more intense emotionswith the right and less intense emotions with the left, regardlessof valence. To resolve this apparent contradiction, we firsttested whether people implicitly spatialize whicheverdimension of emotion they attend to. Results showed thepredicted valence mapping, but no intensity mapping. We thentested an alternative explanation of findings previouslyinterpreted as showing an intensity mapping; these data mayreflect a left-right mapping of spatial magnitude, not emotion.People implicitly spatialize emotional valence, but there is noclear evidence for an implicit lateral mapping of emotionalintensity.

Psychological applications of human language technologycombined with multidisciplinary approaches to similarity cal-culations and data visualization offer avenues to broaden theuse of students’ own words in program assessment. Wecompared multiple analysis approaches on both simple to-ken counts (word roots and character trigrams) and top-downlanguage indicators from 85 student essays about service-learning events. Bioinformatic distance calculations on wordroot counts provided useable assessment information on at-titude change, showing patterns of word use that match theholistic goals of the assignment. Although these patterns werenot found in a subsequent batch of 81 essays, the tools we areproviding may facilitate other efforts to detect attitude changein student writing about service-learning events.

Two studies examined how lexical information contained in words affects people’s category representations. Some words are lexically suggestive regarding the taxonomic position of their referent (e.g., bumblebee, starfish). However, this information differs from language to language (e.g., in Dutch the equivalent words hold no taxonomic information: hommel, vlinder). Three language groups, Dutch, English, and Indonesian speakers, were tested in similarity and typicality judgment tasks. The results show that the lexical information affects only the users of the language (e.g., Dutch speakers rated Dutch-informative items, both in similarity and typicality tasks, higher than English and Indonesian speakers). Results are discussed in light of theories of concept representation and the language relativity hypothesis.

Previous research has shown a link between sequentiallearning (SL) and language as well as links between executivefunction (EF) and both language and SL. However, littleresearch has focused on both the development of therelationship between these factors and their neurologicalunderpinnings. Here we report a study of the event-relatedpotential (ERP) correlates of SL and behavioral measures oflanguage and EF in a sample of 7-12-year-old children.Results revealed that both SL and EF had independentassociations with language development but that thecontribution that both made toward language developmentshifted dramatically between the ages of 7 to 11-12 years. Theresults furthermore suggest that this developmental shift maybe due in part to the maturation of EF abilities and changesdue to neural entrenchment and commitment as aconsequence of language acquisition.

Embodied cognition is sometimes presented as an alternative to computational approaches, the argument being that cognition is strongly influenced by an agent's body movement. However, the exact nature of this influence is still uncertain. In the current paper, we add to the conversation by analyzing adults’ predictions in a high-ambiguity task: Adults had to decide which of two objects would sink faster (or slower) in water. Ambiguity was achieved by pitting object volume and object mass against buoyancy: The winning object of a pair was sometimes the bigger and heavier one, and sometimes it was the smaller and lighter one. The crucial manipulation was whether the stimuli were real-life objects or 2D pictures. All participants were presented with pictures of the objects during a training phase (when they received feedback on their predictions). Real-life objects were either present during the phase prior to the training (jars-first condition), or during the phase after the training (jars-last condition). Findings showed a clear influence of hands-on experiences: When allowed to hold the objects, adults were more likely to demonstrate a simplistic focus on object heaviness. These results call for a more nuanced understanding of the effect of embodied experiences on the stability of representations. While embodiment sometimes can help distinguish relevant from irrelevant information, we show that it can also destabilize representations acquired through visual information.

Questioning is a core component of formal pedagogy. Parents commonly question children, but do they use questions to teach? Research has shown that informal pedagogical situations elicit stronger inferences than the same evidence observed in non- pedagogical situations. Certain questions (“pedagogical questions”) have similar features. We investigate the frequency and distribution of pedagogical questions from mother-child conversations documented in the CHILDES database. We show that pedagogical questions are commonplace, are more frequent for middle-class mothers compared to working-class mothers, are more frequent during free play than during daily routines, and are more frequent in mothers who ask more questions. The results serve as a first step towards understanding the role of questions in informal pedagogy.

Distributional Semantic Models have been successful atpredicting many semantic behaviors. The aim of this paper isto compare two major classes of these models – co-occurrence-based models, and prediction error-driven models– in learning semantic categories from child-directed speech.Co-occurrence models have gained more attention incognitive research, while research from computationallinguistics on big datasets has found more success withprediction-based models. We explore differences betweenthese types of lexical semantic models (as representatives ofHebbian vs. reinforcement learning mechanisms,respectively) within a more cognitively relevant context: theacquisition of semantic categories (e.g., apple and orange asfruit vs. soap and shampoo as bathroom items) from linguisticdata available to children. We found that models that performsome form of abstraction outperform those that do not, andthat co-occurrence-based abstraction models performed thebest. However, different models excel at different categories,providing evidence for complementary learning systems.

Traditionally, models of the decision-making process have fo-cused on the case where a decision-maker must choose be-tween two alternatives. The most successful of these, sequen-tial sampling models, have been extended from the binary caseto account for choices and response times between multiplealternatives. In this paper, I present a geometric representa-tion of diffusion and accumulator models of multiple-choicedecisions, and show how these can be analyzed as Markovprocesses on lattices. I then introduce psychological relation-ships between choice alternatives and show how this impactsthe sequential sampling process. I conclude with two examplesshowing how one can predict distributions of responses on acontinuum as well as response times by incorporating psycho-logical representations into a multi-dimensional random walkdiffusion process.

The centering inference - p & q, therefore if p then q - is important in reasoning research because it is logically valid for some accounts of conditionals (e. g. the material and the probability conditionals), but not for others (e. g. the inferential conditional, according to which a conditional is true if and only if there is an inferential connection between p and q). We tested participants' acceptance of centering compared to valid and invalid inferences not containing conditionals, varying the presence of an inferential connection and of a common topic of discourse between p and q. Participants' acceptance of centering was more similar to valid inferences than to invalid inferences, and there was no reliable effect of a connection between p and q. Acceptance rates were higher when there was a common topic of discourse, independently of the type of inference. The findings support the probability conditional account.

Certain content of a linguistic construction can project whenthe construction is embedded in entailment-canceling environ-ments. For example, the conclusion that John smoked in thepast from the utterance John stopped smoking still holds forJohn didn’t stop smoking, in which the original utterance isembedded under negation. There are two main approaches toaccount for projection phenomena. The semantic approach addsrestrictions of the common ground to the conventional meaning.The pragmatic approach tries to derive projection from generalconversational principles. In this paper we build a probabilisticmodel of language understanding in which the listener jointlyinfers the world state and what common ground the speakerhas assumed. We take change-of-state verbs as an exampleand model its projective content under negation. Under certainassumptions, the model predicts the projective behavior and itsinteraction with the question under discussion (QUD), withoutany special semantic treatment of projective content.

In this paper, we explore the cultural differences in theproduction of speech and speech+gesture combinations byinfants at the age of 17-18 months in Mozambique and theNetherlands. We found that Dutch infants produce morespeech and gestures compared to Mozambican infants. Infantsin both communities make most use of content words. Theresults further show that Dutch infants make more use ofproximal pointing than Mozambicans, whereas Mozambicansmake more use of the offering gesture. Finally, the amount ofsemantically coherent speech+gesture combinations of theMozambican infants is higher than of the Dutch infants

In this paper, we present a spiking neural model of contextdependent decision making. Prefrontal cortex (PFC) plays afundamental role in context dependent behaviour. We modelthe PFC at the level of single spiking neurons, to explore theunderlying computations which determine its contextual re-sponses. The model is built using the Neural EngineeringFramework and performs input selection and integration as anonlinear recurrent dynamical process. The results obtainedfrom the model closely match behavioural and neural experi-mental data obtained from macaque monkeys that are trainedto perform a context sensitive perceptual decision task. Theclose match suggests that the low-dimensional, nonlinear dy-namical model we suggest captures central aspects of contextdependent decision making in primates.

Cognitive development researchers have drawn conclusionsabout young children’s developing knowledge of number bystudying their behavior, while at the same timeacknowledging that behavior is an imperfect index ofknowledge, e.g., it may be disputed whether a givenbehavioral task accurately measures, overestimates, orunderestimates children’s knowledge. The texts of publishedresearch articles from these investigations are the focus of adiscourse analysis described in the present article. The resultsof the discourse analysis suggest that claims about what aperson knows are actually generalized descriptions ofbehavior. Therefore, in studying behavior on tasks to drawconclusions about participants’ conceptual knowledge,researchers are merely making behavioral generalizations, notinvestigating hidden cognitive or epistemic content.

In the paper we discuss the relation between fuzzy sets and thegraded membership and typicality effects found in the studyof concepts. After a short overview of the topic, we presentthree experiments, carried out using the same method but withdifferent situational contexts, which examine whether gradedmembership and typicality could be considered as independentfactors capable of influencing the performance of human par-ticipants involved in sentence verification tasks, or they aresomehow interrelated. The paper concludes with a generaldiscussion of the experimental findings and the problems theypose for models of concepts based on the theory fuzzy sets.

Exploration-exploitation of functions, that is learningand optimizing a mapping between inputs and expectedoutputs, is ubiquitous to many real world situations.These situations sometimes require us to avoid certainoutcomes at all cost, for example because they arepoisonous, harmful, or otherwise dangerous. We testparticipants’ behavior in scenarios in which they haveto find the optimum of a function while at the sametime avoid outputs below a certain threshold. Intwo experiments, we find that Safe-Optimization, aGaussian Process-based exploration-exploitation algo-rithm, describes participants’ behavior well and thatparticipants seem to care first about whether a point issafe and then try to pick the optimal point from all suchsafe points. This means that their trade-off betweenexploration and exploitation indicates intelligent,approximate, and homeostasis-driven behavior

Language has been shown to influence the ability to formcategories. Here we investigate whether linguistic labels areprivileged compared to other types of cues (e.g., numbers orsymbols), and whether labels exert their effects regardless ofwhether they are introduced intentionally. In a categorizationtask, we found that adults were more likely to use labels to de-termine category boundaries compared to numbers or symbols,and that these effects persisted in all intentionality manipula-tions. These findings suggest that labels have a powerful effecton categorization compared to other cues; most strikingly, la-bels (but not other cues) are used during categorization evenwhen people are specifically asked to ignore them. These re-sults provide novel support for the position that labels indicatecategory membership.

We test the predictions of different computational models ofcross-situational word learning that have been proposed in theliterature by comparing their behavior to that of young childrenand adults in the word learning task conducted by Ramscar,Dye, and Klein (2013). Our experimental results show that aHebbian learner and a model that relies on hypothesis testingfail to account for the behavioral data obtained from both pop-ulations. Ruling out such accounts might help reducing thesearch space and better focus on the most relevant aspects ofthe problem, in order to disentangle the mechanisms used dur-ing language acquisition to map words and referents in a highlynoisy environment.

People often use norms to coordinate behavior andaccomplish shared goals. But how do people learn andrepresent norms? Here, we formalize the process by whichcollaborating individuals (1) reason about group plans duringinteraction, and (2) use task features to abstractly representnorms. In Experiment 1, we test the assumptions of our modelin a gridworld that requires coordination and contrast it with a“best response” model. In Experiment 2, we use our model totest whether group members’ joint planning relies more onstate features independent of other agents (landmark-basedfeatures) or state features determined by the configuration ofagents (agent-relative features).

Do metaphors shape people’s emotional states and mindsetsfor dealing with hardship? Natural language metaphors mayact as frames that encourage people to reappraise anemotional situation, changing the way they respond to it.Recovery from cancer is one type of adversity that manypeople face, and it can be mediated by the mindset peopleadopt. We investigate whether two common metaphors fordescribing a cancer experience – the battle and the journey –encourage people to make different inferences about thepatient’s emotional state. After being exposed to the battlemetaphor participants inferred that the patient would feelmore guilt if he didn’t recover, while after being exposed tothe journey metaphor participants felt that he had a betterchance of making peace with his situation. We discussimplications of this work for investigations of metaphor andemotion, mindsets, and recovery.

Inequality in wealth is a pressing concern in many contempo-rary societies, where it has been show to co-occur with polit-ical polarization and policy volatility, however its causes areunclear. Here we demonstrate in a simple model where socialbehavior spreads through learning that inequality can covaryreliably with other cooperative behavior, despite a lack of ex-ogenous cause or deliberate coordination. In the context ofsimulated cultural evolution selecting for trust and cooperativeexchange, we find both cooperation and inequality to be moreprevalent in contexts where the same agents play both the rolesof the trusting investor and the trusted investee, in contrast tothe condition where these roles are divided between disjointpopulations. Cooperation is more likely in contexts of hightransparency about potential partners and with a high amountof partner choice; while inequality is more likely with highinformation but no choice in partners for those that want to in-vest. While not yet a full model of contemporary society, ourapproach holds promise for examining the causality and socialcontexts underlying shifts in income inequality.

Decades of research have demonstrated that students facecritical conceptual challenges in learning mathematics. Asnew adaptive learning technologies become ubiquitous ineducation, they bring opportunities both to facilitateconceptual development in more focused ways and to gatherdata that may yield new insights into students’ learningprocesses. The present study analyzes data archives from aperceptual learning intervention designed to help studentsmaster key concepts related to linear measurement andfractions. Using algorithmic data coding on a database of78,034 errors from a sample of sixth graders, both conceptualerrors and other errors were captured and analyzed for changeover time. Results indicate that conceptual errors decreasedsignificantly. This approach suggests additional ways thatsuch datasets can be exploited to better understand how thesoftware impacts different students and how next generationsof adaptive software may be designed to code and respond tocommon error patterns in real time.

There is a long-standing assumption that gestural forms aregeared by a set of modes of representation (acting,representing, drawing, moulding) with each techniqueexpressing speakers’ focus of attention on specific aspects ofreferents (Müller, 2013). Beyond different taxonomiesdescribing the modes of representation, it remains unclearwhat factors motivate certain depicting techniques overothers. Results from a pantomime generation task show thatpantomimes are not entirely idiosyncratic but rather followgeneralisable patterns constrained by their semantic category.We show that a) specific modes of representations arepreferred for certain objects (acting for manipulable objectsand drawing for non-manipulable objects); and b) that use andordering of deictics and modes of representation operate intandem to distinguish between semantically related concepts(e.g., “to drink” vs “mug”). This study provides yet moreevidence that our ability to communicate through silentgesture reveals systematic ways to describe events and objectsaround us.

Plural nouns do not strictly refer to more than one object, whichsuggests that they are not semantically marked to mean “more thanone” and that plurality inferences are made via a scalarimplicature. Consistent with that hypothesis, recent evidence usinga picture-matching paradigm supports founds that participantswere equally fast to respond to a picture of a single object as apicture of multiple objects after reading a sentence containing aplural. This suggests that comprehenders activate both a semantic(i.e., singular) and a pragmatic interpretation (i.e., plural). Thecurrent study found that even after a 1500 ms delay,comprehenders still maintain activation of both meanings afterreading a sentence containing a plural. This suggests that theactivation of the singular meaning may not be due to theprocessing of a scalar implicature, but rather may be due to thenature of plural conceptual representations

Younger adults use both semantic and phonological cues to quickly and efficiently localize the referent during sentence comprehension. While some behavioral studies suggest that older adults use contextual information even more strongly than younger adults, ERP studies have shown that this population, as a group, is less apt at using contextual semantic cues to predict upcoming words. The current study extends the investigation of contextual cue processing in auditory sentence comprehension beyond semantic cue processing, by comparing younger and older adults in their ability to use phonological cues in indefinite articles (a/an) to localize the referent in an eye-tracking visual world paradigm. Our results suggest that both age groups use such phonological information for referent localization, but with different timelines: younger adults use the cues to anticipate an upcoming word, whereas older adults show delayed cue processing after the target word has been spoken. Together with findings from semantic context processing, these results support a model of sentence comprehension in which the use of contextual cues continues with aging, but is no longer as efficient as in the young system for anticipatory word retrieval.

Resource scarcity poses challenging demands on the human cognitive system. Budgeting with limited resources induces an attentional focus on the problem at hand. This focus enhances processing of relevant information, but it also comes with a cost. Specifically, scarcity may cause a failure to notice beneficial information that helps alleviate the condition of scarcity. In three experiments, participants were randomly assigned with a small budget (“the poor”) or a large budget (“the rich”) to order a meal from a restaurant menu. The poor participants looked longer at the prices of the items and recalled the prices more accurately, compared to the rich participants. Importantly, the poor neglected a useful discount that would save them money. This neglect may arise as a result of attentional narrowing, and help explain a range of counter- productive behaviors of low-income individuals. The current findings have important implications for public policy and services for low-income individuals.

Previous research has indicated that memories can be modified in conjunction with one’s attitudes, in particular, political beliefs. The current study extended this finding by focusing on the relationship between differing religious beliefs and false memories for news events. We predicted that religious people would be more inclined to remember fabricated news events positively depicting religion and less likely to remember events negatively depicting religion compared to non-religious people. Opposite effects were predicted for events depicting atheism. In contrast, we found that religious people were more likely to falsely remember both events depicting religion positively and negatively compared to non-religious individuals. However, the extent to which individuals felt positively about the events interacted with religious beliefs to predict reported false memories. Religious individuals were more likely to remember events if they felt positively about them whereas atheists were more likely to remember events if they felt negatively about them

Language is widely assumed to be a well designed tool for re-liably communicating propositional information between peo-ple. This suggests that its users should be sensitive to failuresof communication, such as utterances that are blatantly inco-herent with respect to an ongoing conversation. We presentexperimental work suggesting that, in fact, people are surpris-ingly tolerant of conversational incoherence. In two previousstudies, participants engaged in instant-messaging conversa-tions that were either repeatedly crossed with other conversa-tions or had lines inserted into them that deliberately contra-dicted available information. In both cases, a substantial pro-portion of participants failed to notice. In a new study, confed-erates inserted unexpected, nonsensical lines into face-to-faceconversations. The majority of participants failed to notice.We argue these findings suggest that we should be wary ofmodeling spontaneous communication in terms of faithful in-formation transmission, or language as a well designed tool forthat purpose.

Individual differences in reflectiveness have been found topredict belief in God. We hypothesize that this associationmay be due to a broader inclination for intuitive thinkers toendorse teleological explanations. In support of ourhypothesis, we find that scientifically unfounded teleologicalexplanations are more likely to be endorsed by intuitive compared to analytical thinkers, and that those who endorse teleological explanations are more likely to have religious beliefs.

Decisions over the timing of actions are critical in severalsafety, security and healthcare scenarios. These decisions, sim-ilar to discrete decisions, can be influenced by biases and in-dividual traits. In this paper, a bias of impatience is studiedin an experiment with 626 participants, with a focus on gen-der differences. Impatience was moderated with a manipula-tion of a variable-speed countdown. Men and women differedin how they expressed impatience. While men systematicallyand irrationally act earlier when become impatient followingthe slower countdowns, women react by irrationally request-ing earlier information about the outcome of each trial, andimpulsively pressing an inactive key.

What is the role of fraction magnitude knowledge in learningfraction addition? An experiment with 71 6th and 7 th gradestudents compared fraction addition instruction and practicewith a magnitude representation to a tightly controlled non-magnitude condition. In the magnitude condition, studentswith better fraction magnitude estimation skills benefittedmore from the conceptual instruction and this relationshipwas moderated by students’ knowledge of how magnituderelates to fraction addition and equivalence. However,students with better fraction magnitude estimation skillsbenefitted less from the practice problems with magnitude. Inthe non-magnitude condition, fraction magnitude estimationwas not predictive of learning. This study indicates thatstudents with magnitude knowledge can leverage it to learnfraction addition concepts from magnitude representations,but, for those students, magnitude representations may be adistraction from practicing the procedure.

Memory models typically assume that recall is a two-stage process with learning affecting both processes to the same degree. This equal learning assumption is difficult to reconcile with studies of the 'testing effect', which reveal different forgetting rates following learning from test practice versus learning from restudy. Here we present a new memory model, termed Primary and Convergent Retrieval (PCR) that assumes successful recall leads to a selective enhancement for the second stage of recall (Convergent Retrieval). We applied this model to existing testing effect data. In two new experiments, we confirmed novel predictions of the PCR model for transfer between retrieval cues and for recall latencies. This is the first formally specified model of the testing effect and it has broad implications for the nature of learning and retrieval.

Word learning involves mapping observable words to unob-servable speaker intentions. The ability to infer referential in-tentions in turn has been shown to depend in part on accessto language. Thus, word learning and intention-reading co-develop. To explore this interaction, we present an agent-basedmodel in which an individual simultaneously learns a lexiconand learns about the speaker’s perspective, given a shared con-text and the speaker’s utterances, by performing Bayesian in-ference. Simulations with this model show that (i) lexicon-learning and perspective-learning are strongly interdependent:learning one is impossible without some knowledge of theother, (ii) lexicon- and perspective-learning can bootstrap eachother, resulting in successful inference of both even when thelearner starts with no knowledge of the lexicon and unhelpfulassumptions about the minds of others, and (iii) receiving ini-tial input from a ‘helpful’ speaker (who adopts the learner’sperspective on the world) paves the way for later learning fromspeakers with perspectives which diverge from the learner’s.This approach represents a first attempt to model the hypoth-esis that language and mindreading co-develop, and a first ex-ploration of the implications for theories of word learning andmindreading development.

Categories are essential for thinking, learning, and communi-cating. Research has shown that young children and adultstreat categories very differently, with young children favor-ing whole objects while adults focus on the key informationin most cases. If so, then how can young children learn cat-egories requiring focused attention to key features? Studieshave shown that drawing attention to rules had facilitative ef-fects. We sought to identify whether the effect was driven byinstruction about rules or by stimulus-driven factors. Our re-sults suggest that even with instruction, 4-year-olds were notable to attend to key information. Simply making importantinformation more salient, however, allowed them to learn thecategory and transfer to situations when the key feature wasno longer salient.

In order to creatively interact with robots we need to understand how creative thinkers work with objects to explore new ideas physically. Our approach involves comparing the model-making strategies of architects with students to expose the creative extras architects bring to working with physical models. To study this we coded students and architects performing a design task. Architects differed from students along three dimensions. First, architects were more selective; they used fewer blocks overall and fewer variations. Second, architects appear to think more about spatial relationships and material constraints. Lastly, architects more often experiment with re-orientations: they position a block one way to see its relations to its neighbors; they reposition it another way to see how that changes how things look and feel. These findings suggest that designers interact with the material more effectively than students. This embodied know-how is something next generation robots can support and possibly enhance.

How do people rapidly learn rich, structured concepts fromsparse input? Recent approaches to concept learning havefound success by integrating rules and statistics. We describe ahierarchical model in this spirit in which the rules are stochas-tic, generative processes, and the rules themselves arise froma higher-level stochastic, generative process. We evaluate thisprobabilistic language-of-thought model with data from an ab-stract rule learning experiment carried out with adults. In thisexperiment, we find novel generalization effects, and we showthat the model gives a qualitatively good account of the exper-imental data. We then discuss the role of this kind of model inthe larger context of concept learning.

Conventional theories of visual object recognition treat objectseffectively as abstract, arbitrary patterns of image features.They do not explicitly represent objects as physical entities inthe world, with physical properties such as three-dimensionalshape, mass, stiffness, elasticity, surface friction, and so on.However, for many purposes, an object’s physical existence iscentral to our ability to recognize it and think about it. Thisis certainly true for recognition via haptic perception, i.e., per-ceiving objects by touch, but even in the visual domain an ob-ject’s physical properties may directly determine how it looksand thereby how we recognize it. Here we show how a physi-cal object representation can allow the solution of visual prob-lems, like perceiving an object under a cloth, that are other-wise difficult to accomplish without extensive experience, andwe provide behavioral and computational evidence that peoplecan use such a representation.

We investigated what happens when the spontaneous encoding of a problem is incongruent with its solving strategy. We created word problems from which two distinct semantic representations could be abstracted. Only one of these representations was consistent with the solving strategy. We tested whether participants could recode a semantically incongruent representation in order to access another, less salient, solving strategy. In experiment 1, participants had to solve arithmetic problems and to indicate which problems were unsolvable. In experiment 2, participants received solved problems and had to decide whether the solution was appropriate or not. In both experiments, participants had more difficulties acknowledging that problems inducing an incongruent representation could be solved than they had for problems inducing a congruent representation. This was confirmed by response times. These results highlight how semantic aspects can lead even adults to fail or succeed in the solving of arithmetic problems requiring basic mathematical knowledge.

The creativity of natural language poses a significant theoretical problem. One example of this is denominal verbs (those derived from nouns) such as spoon in “She spooned me some sugar”. Traditional generative approaches typically posit a unique entry in the lexicon for this usage, though this approach has difficulty scaling. Construction Grammar has evolved as a competing theory which instead allows the syntactic form of the sentence itself to contribute semantic meaning. However, how people learn syntactic constructions remains an open question. One suggestion has been that they are learned through analogical generalization. We evaluate this hypothesis using a computational model of analogical generalization to simulate Kaschak and Glenberg’s (2000) study regarding interpretation of denominal verbs.

Mindfulness meditation is a form of secular meditation thatemphasizes non-judgmental awareness of the presentmoment. Research into mindfulness has greatly expanded inrecent years (Davidson & Kasniak, 2015) and a growingliterature has documented effects of mindfulness training oncognition. However, the specific aspects of mindfulnessmeditation training for novice practitioners that mightinfluence cognition remain unexplored. The present studyused a rigorous, dual-blind design to investigate whether theattention-monitoring component of mindfulness meditationreduces mind-wandering and improves performance duringreading comprehension and sustained attention tasks. Whencompared with relaxation meditation, mindfulness trainingimproved recall of specific details from a text but did notreduce mind-wandering or affect sustained attention. Theresults are discussed with respect to design considerationswhen studying a meditation intervention.

We propose a novel framework for the analysis of learning al-gorithms that allows us to say when such algorithms can andcannot generalize certain patterns from training data to testdata. In particular we focus on situations where the rule thatmust be learned concerns two components of a stimulus beingidentical. We call such a basis for discrimination an identity-based rule. Identity-based rules have proven to be difficult orimpossible for certain types of learning algorithms to acquirefrom limited datasets. This is in contrast to human behaviouron similar tasks. Here we provide a framework for rigorouslyestablishing which learning algorithms will fail at generalizingidentity-based rules to novel stimuli. We use this frameworkto show that such algorithms are unable to generalize identity-based rules to novel inputs unless trained on virtually all possi-ble inputs. We demonstrate these results computationally witha multilayer feedforward neural network.

Speakers need to use a variety of referring expressions (REs) (e.g. full noun phrases, pronouns or null forms) in pragmatically appropriate ways to produce coherent narratives. Languages, however, differ from each other in terms of a) whether REs as arguments can be dropped or not and b) whether personal pronouns encode gender or not. Here we examine two languages that differ from each other in these two aspects and ask whether the co-reference context (i.e., referents are maintained or re-introduced) and the gender encoding options affect the use of REs differentially. We elicited narratives from Dutch and Turkish speakers about two types of three-person events, one including people of the same and the other of mixed-gender. Speakers of both languages followed a general principle of using full forms such as noun phrases (NPs) while re-introducing a previously mentioned referent into the discourse and reduced forms (overt or null pronoun) while maintaining the same referent; a language independent strategy in discourse production. Turkish speakers, unlike Dutch speakers, used pronouns mainly to mark emphasis. Furthermore, Dutch but not Turkish speakers used pronouns differentially across the two videos. Thus, we argue that linguistic possibilities available in typologically different languages might tune speakers into taking different principles into account to establish coherence in narratives in pragmatically coherent ways.

What does it mean for analogy to be rational? John StuartMill described a probabilistic underpinning for analogicalinference based on the the odds of observing systemicpairwise correspondence across otherwise independentsystems by mere chance. Although proponents and criticshave debated its validity, Mill’s approach has yet to beimplemented computationally or studied psychologically. Inthis paper we examine Mill’s approach and show how it canbe instantiated using Bayes theorem. Then we describe twoexperiments that present subjects with partially-revealed,aligned binary strings with varying degrees of intra- and inter-string regularity. Experimental results are compared to aformal rational analysis of the stimuli revealing conditionswhereby participants exhibit confidence patterns consistentand inconsistent with Mill’s rational basis of analogy.

Inductive generalizations about the properties of kinds arebased on evidence. But evidence can come either from ourobservations, or from the testimony of knowledgeableinformants. The current study explores how we combineinformation from these two sources to make inductiveinferences. Participants learned about a novel object category,and observed the property occur with some frequency in asample of category members. Different groups of participantsalso heard an informant making either Generic, Quantified, orSpecific claims about the prevalence of the property.Participants who heard generic claims were more resistant toa straightforward use of statistical evidence in theirgeneralizations. Moreover, participants who rated theinformant as more knowledgeable (across conditions) gavehigher prevalence estimates. The results suggest twopathways through which testimony translates into evidencefor category learning, and raise questions on how to bestcombine evidence from these different sources into acommon representational form.

Recent empirical and modeling research has focused on thesemantic fluency task because it is informative about seman-tic memory. An interesting interplay arises between the rich-ness of representations in semantic memory and the complex-ity of algorithms required to process it. It has remained anopen question whether representations of words and their re-lations learned from language use can enable a simple searchalgorithm to mimic the observed behavior in the fluency task.Here we show that it is plausible to learn rich representationsfrom naturalistic data for which a very simple search algorithm(a random walk) can replicate the human patterns. We sug-gest that explicitly structuring knowledge about words into asemantic network plays a crucial role in modeling human be-havior in memory search and retrieval; moreover, this is thecase across a range of semantic information sources.

Psychological essentialism is a cognitive bias that leads people to view members of a category as sharing a deep, underlying, inherent nature that causes them to be fundamentally similar to one another in non-obvious ways. Although essentialist beliefs can be beneficial, allowing people to view the social world as stable and predictable, essentialist beliefs about social categories such as race or ethnicity are also thought to underlie the development of stereotyping and prejudice. Whereas recent studies in adults have found that racial essentialism is associated with increased prejudice, the development of this relationship has rarely been examined. The present research examined the implications of essentialism for prejudice in a population of white five- and six-year old children in the United States, and revealed that essentialist beliefs about race are associated with increased implicit and explicit prejudice towards members of a minority racial group.

Recent research in the artificial grammar literature has found that a simple exemplar model of memory can account for a wide variety of artificial grammar results (Jamieson & Mewhort, 2009, 2010, 2011). This classic type of model has also been extended to account for natural language sentence processing effects (Johns & Jones, 2015). The current article extends this work to account for sentence production, and demonstrates that the structure of language itself provides sufficient power to generate syntactically correct sentences, even with no higher-level information about language provided to the model.

How does a naive language learner deal with speaker variationirrelevant to distinguishing word meanings? Experimental datais contradictory, and incompatible models have been proposed.Here, we examine basic assumptions regarding the acousticsignal the learner deals with: Is speaker variability a hurdle indiscriminating sounds or can it easily be ignored? To this end,we summarize existing infant data. We then present machine-based discriminability scores of sound pairs obtained withoutany language knowledge. Our results show that speaker vari-ability decreases sound contrast discriminability, and that somecontrasts are affected more than others. However, chance per-formance is rare; most contrasts remain discriminable in theface of speaker variation. We take our results to mean thatspeaker variation is not a uniform hurdle to discriminatingsound contrasts, and careful examination is necessary whenplanning and interpreting studies testing whether and to whatextent infants (and adults) are sensitive to speaker differences.

Fluent event processing appears to critically involveselectively attending to information-rich junctures withincontinuously unfolding sensory streams (e.g., Newtson, 1973).What counts as information-rich would seem to depend on avariety of factors, however, including the novelty/familiarityof such events, as well as local opportunity for repeatedviewings. Using Hard, Recchia, & Tversky’s “Dwell-timeParadigm,” we investigated the extent to which viewers’attention to unfolding activity streams is affected bynovelty/familiarity and a second viewing. Viewers’ dwelltimes were recorded as they advanced twice each through threeslideshows varying in familiarity but equated on otherdimensions. Dwell time patterns revealed reorganization on anumber of fronts: a) familiarity elicited decreased dwellingoverall, b) dwell-time patterns changed systematically onsecond viewing, and c) familiarity modulated the specificnature of change associated with repeated viewing. Thesefindings illuminate reorganization in attention as actioninformation is first encountered and then quickly incorporatedto guide event processing.

Syntactic recursion is argued to be a key property of naturallanguages, allowing us to create an infinite number ofutterances from a finite number of words and rules. Somehave argued that recursion is uniquely human. There are atleast two possibilities for the origins of recursion: 1)Recursion is a property of the language faculty. 2) Recursionis an historical accomplishment and is culturally constructedover millennia. Here we ask whether an emerging signlanguage, Nicaraguan Sign Language (NSL), exhibitssyntactic recursion by comparing the language of the firstthree age cohorts of signers. Signers (n=27) watched anddescribed vignettes designed to elicit relative clauses. Resultssuggest that signers from all three cohorts have strategies tofulfill the discourse function of relative clauses, picking outan individual from a set. The grammatical form of theutterances differs across cohorts, with signers from latercohorts clearly producing embedded structures.

We used eye-tracking to study the time course of analogical reasoning in adults. We considered proportions of looking times and saccades. The main question was whether or not adults would follow the same search strategies for different types of analogical problems (Scene Analogies vs. Classical A:B:C:D vs a Scene version of A:B::C:D). We then compared these results to the predictions of various models of analogical reasoning. Results revealed a picture of common search patterns with local adaptations to the specifics of each paradigm in both looking-time duration and the number and types of saccades. These results are discussed in terms of conceptions of analogical reasoning.

Contrary to a widely held belief, experts recall randommaterial better than non-experts. This phenomenon, predictedby the CHREST computational model, was first establishedwith chess players. Recently, it has been shown through ameta-analysis that it generalises to nearly all domains wherethe effect has been tested. In this paper, we carry outcomputer simulations to test whether the mechanismpostulated with chess experts – the acquisition and use of alarge number of chunks – also applies to computerprogramming experts. The results show that a simplifiedversion of CHREST (without the learning and use of high-level schemata known as templates) broadly captures the skilleffect with scrambled programs. However, it fails to accountfor the differences found in humans between different types ofrandomisation. To account for these differences, additionalmechanisms are necessary that use semantic processing.

This study determined if features of an action plan held in working memory are activated to the same extent (consistent with serial memory theories) or in a gradient (consistent with theories that assume serial order is imposed prior to response selection). Two visual events (A and B) occurred in a sequence. Participants planned an action (3-finger, key sequence) to the first event (Action A) and maintained this action in working memory while executing a speeded response (1-finger key-press) to the second event (Action B). Afterwards, participants executed Action A. We manipulated whether Action B overlapped with the first, second or final feature of Action A, and examined the pattern of correct, Action B RTs at the different overlap locations by finger (index, middle, ring), as well as the error rates of both Action A and Action B. Results indicate that 3-finger sequences were not activated equally or in a gradient. Instead, feature activation reflected a serial position curve or a reverse serial position curve dependent on finger.

Listeners quickly adapt to novel accents. There are three mainhypotheses for how they do so. Some suggest that listenersexpand their phonetic categories, allowing more variability inhow a sound is pronounced. Others argue that listeners shifttheir categories instead, only accepting deviations consistentwith the accent. A third hypothesis is that listeners both shiftand expand their categories. Most work has supported thecategory expansion hypotheses, with the key exception of Mayeet al. (2008) who argued for a shifting strategy. Here, we applythe ideal adaptor model from Kleinschmidt & Jaeger (2015)to reexamine what conclusions can be drawn from their data.We compare adaptation models in which categories are shifted,expanded, or both shifted and expanded. We show that modelsinvolving expansion can explain the data as well as, if not betterthan, the shift model, in contrast to what has been previouslyconcluded from these data.

Representational views of the mind traditionally face askeptical challenge on perceptual knowledge: if ourexperience of the world is mediated by representations builtupon perceptual inputs, how can we be certain that ourrepresentations are accurate and our perceptual apparatusreliable? J. J. Gibson's ecological approach provides analternative framework, according to which direct perceptionof affordances does away with the need to posit internalmental representations as intermediary steps betweenperceptual input and behavioral output. Gibson accordinglyspoke of his framework as providing “reasons for realism.” Inthis paper I suggest that, granting Gibson his reasons forperceptual realism, the Gibsonian framework motivates anti-realism when it comes to scientific theorizing and modeling.If scientists are Gibsonian perceivers, then it makes sense totake their use of models in indirect investigations of real-world phenomena not as representations of the phenomena,but rather as autonomous tools with their own affordances.

The nature of the dynamics of nonverbal interactions is of con-siderable interest to the study of human communication andfuture human-computer interaction. Facial expressions consti-tute an important source of nonverbal social signals. Whereasmost studies have focused on the facial expressions of iso-lated individuals, the aim of this study is to explore the cou-pling dynamics of facial expressions in social dyadic interac-tions. Using a special experimental set-up, the frontal facialdynamics of pairs of socially interacting persons were mea-sured and analyzed simultaneously. We introduce the use ofconvergent cross mapping, a method originating from dynam-ical systems theory, to assess the causal coupling of the dyadicfacial-expression dynamics. The results reveal the presence ofbidirectional causal couplings of the facial dynamics. We con-clude that convergent cross mapping yields encouraging resultsin establishing evidence for causal behavioral interactions.

What accounts for the vast diversity in the world’s languages?We explore one possibility: languages adapt to their linguis-tic environment (Linguistic Niche Hypothesis; Lupyan & Dale,2010). Recent studies have found support for this hypothesisthrough correlations between aspects of the environment andlinguistic structure. We synthesize this previous work and findthat languages spoken in cold, small regions tend to be morecomplex across a range of linguistic features. We also testa novel prediction of the Linguistic Niche Hypothesis by ex-amining the learnability of languages for first-language, childlearners.

Given the global challenges of security, both in physical and cyber worlds, security agencies must optimize the use of their limited resources. To that end, many security agencies have begun to use "security game" algorithms, which optimally plan defender allocations, using models of adversary behavior that have originated in behavioral game theory. To advance our understanding of adversary behavior, this paper presents results from a study involving an opportunistic crime security game (OSG), where human participants play as opportunistic adversaries against an algorithm that optimizes defender allocations. In contrast with previous work which often assumes homogeneous adversarial behavior, our work demonstrates that participants are naturally grouped into multiple distinct categories that share similar behaviors. We capture the observed adversarial behaviors in a set of diverse models from different research traditions, behavioral game theory, and Cognitive Science, illustrating the need for heterogeneity in adversarial models.

Semantic fluency tasks have increasingly been used to probethe structure of human memory, adopting methodologies fromthe ecological foraging literature to describe memory as a tra-jectory through semantic space. Clusters of semantically re-lated items are often produced together, and the transitions be-tween these clusters of semantically related items are consis-tent with theories of optimal foraging, where the search pro-cess exhibits a balance between exploration and exploitationbehaviors (Hills, Jones, & Todd, 2012). Here, we use a seman-tic fluency memory task in which subjects recall geographiclocations. For each pairwise transition, we measure tempo-ral, geographic, semantic, lexical, and phonetic distances. Ingeneral, the dimensions are loosely but reliably correlated witheach other. Segmentation of the retrieval sequence into patchessupports the notion that subjects strategically leave patches aswithin-patch resources diminish, but also suggests that sub-jects may shift their attention between different sources of in-formation, perhaps reflecting dynamically changing patch def-initions.

The complexity of situations makes individuals use emotionsto make sense of their environment and interdependent oth-ers. In this paper, we build on the idea that physiological re-actions give emotional information about the subject and wefocus on Electrodermal Activity (EDA), an index of arousal,to inspect deep processes of a dyadic interaction in a mixed-motive game. Our interest lies on how conflict episodes un-fold, to design intelligent agents that are more socially awareand thus able to express and recognise dyadic forms of con-flict. A qualitative analysis of the data allowed us to identifymoments where players made choices to cope with ongoingconflict or prospects of it in the future.

It is commonly believed that the central visual field (fovea andparafovea) is important for recognizing objects and faces, andthe peripheral region is useful for scene recognition. However,the relative importance of central versus peripheral informa-tion for object, scene, and face recognition is unclear. Larsonand Loschky (2009) investigated this question in the context ofscene processing using experimental conditions where a cir-cular region only reveals the central visual field and blocksperipheral information (”Window”), and in a ”Scotoma” con-dition, where only the peripheral region is available. Theymeasured the scene recognition accuracy as a function of vi-sual angle, and demonstrated that peripheral vision was indeedmore useful in recognizing scenes than central vision in termsof achieving maximum recognition accuracy. In this work,we modeled and replicated the result of Larson and Loschky(2009), using deep convolutional neural networks (CNNs).Having fit the data for scenes, we used the model to predictfuture data for large-scale scene recognition as well as for ob-jects and faces. Our results suggest that the relative order ofimportance of using central visual field information is facerecognition>object recognition>scene recognition, and vice-versa for peripheral information. Furthermore, our results pre-dict that central information is more efficient than peripheralinformation on a per-pixel basis across all categories, which isconsistent with Larson and Loschky’s data.

We present a model of moral judgment, Charon, which addsto previous models several factors that have been shown toinfluence moral judgment: 1) a more sophisticated account ofprior mental state, 2) imagination, 3) empathy, 4) thefeedback process between emotion and reason, 5) self-interest, and 6) self-control. We discuss previous classes ofmodels and demonstrate Charon’s extended explanatorypower with a focus on psychopathy and autism.

There is extensive evidence that listeners use general knowledge to predict upcoming sentence endings; however, less is known about how novel information is integrated when there is disagreement between general knowledge and novel information. The present studies use the visual world paradigm to study the semantic competition between new information and general knowledge. Experiment 1 demonstrates that listeners learn to use limited exposure to new information and their general knowledge to anticipate sentence endings that align with the action of the sentence. Experiment 2 demonstrates participants learn to use combinatorial information from stories to elicit anticipatory eye movements to the target over the general knowledge distractor. Evidence from these experiments indicates even in the presence of semantic conflict with general knowledge, listeners rapidly increase the weight of novel information rather than general knowledge.

Two experimental protocols, pairwise rating and triplet rank-ing, have been commonly used for eliciting perceptual similar-ity judgments for faces and other objects. However, there hasbeen little systematic comparison of the two methods. Pairwiserating has the advantage of greater precision, but triplet rank-ing is potentially a cognitive less taxing task, thus resulting inless noisy responses. Here, we introduce several information-theoretic measures of how useful responses from the two pro-tocols are for the purpose of response prediction and parame-ter estimation. Using face similarity data collected on AmazonMechanical Turk, we demonstrate that triplet ranking is signif-icantly better for extracting subject-specific preferences, whilethe two are comparable when pooling across subjects. Whilethe specific conclusions should be interpreted cautiously, dueto the particularly simple Bayesian model for response gener-ation utilized here, the work provides a information-theoreticframework for quantifying how repetitions within and acrosssubjects can help to combat noise in human responses, as wellas giving some insight into the nature of similarity representa-tion and response noise in humans. More generally, this workdemonstrates that substantial noise and inconsistency corruptsimilarity judgments, both within- and across-subjects, withconsequent implications for experimental design and data in-terpretation.

Current judgments are systematically biased by priorjudgments. Such biases occur in ways that seem to reflect thecognitive system’s ability to adapt to the statisticalregularities within the environment. These cognitivesequential dependencies have been shown to occur undercarefully controlled laboratory settings as well as more recentstudies designed to determine if such effects occur in realworld scenarios. In this study we use these well-knownfindings to guide our analysis of over 2.2 million businessreview ratings. We explore how both within-reviewer andwithin-business (between reviewer) ratings are influenced byprevious ratings. Our findings, albeit exploratory, suggestthat current ratings are influenced in systematic ways by priorratings. This work is couched within a broader program thataims to determine the validity of laboratory findings usinglarge naturally occurring behavioral data.

Children and adults can use cross-situational information toidentify words’ referents. What do learners retain about thepotential referents that occur with a word: do they encodemultiple referents or a single guess? We tested this questionusing novel mouse tracking and finger tracking paradigms.Adults were exposed to novel words in a series of ambiguoustraining trials and then tested on the words’ referents. In sometest trials, participants saw the target and three referents thathad never occurred with the word; other test trials included ahigh-probability competitor that had repeatedly occurred withthe word. Participants’ mouse movements were slower, lessaccurate, and took a more complex path to the selectedreferent when the competitor was present, indicating thatparticipants were aware that both the target and competitorhad previously occurred with the word. This suggests thatlearners can accrue information about multiple potentialreferents for a word, and that mouse tracking provides apromising way of assessing this knowledge. However, thisknowledge was not evident in participants’ finger movements,suggesting that the dynamics of finger movements might notcapture real-time competition between referents.

We use a vector space model (VSM) to simulate semantic relat-edness effects in sentence processing, and use this connectionto predict N400 amplitude in an ERP study by Federmeierand Kutas (1999). We find that the VSM-based model is ableto capture key elements of the authors’ manipulations and re-sults, accounting for aspects of the results that are unexplainedby cloze probability. This demonstration provides a proof ofconcept for use of VSMs in modeling the particular contextrepresentations and corresponding facilitation processes thatseem to influence non-cloze-like behavior in the N400.

Language interacts with olfaction in exceptional ways. Olfaction is believed to be weakly linked with language, as demonstrated by our poor odor naming ability, yet olfaction seems to be particularly susceptible to linguistic descriptions. We tested the boundaries of the influence of language on olfaction by focusing on a non-lexical aspect of language (grammatical gender). We manipulated the grammatical gender of fragrance descriptions to test whether the congruence with fragrance gender would affect the way fragrances were perceived and remembered. Native French and German speakers read descriptions of fragrances containing ingredients with feminine or masculine grammatical gender, and then smelled masculine or feminine fragrances and rated them on a number of dimensions (e.g., pleasantness). Participants then completed an odor recognition test. Fragrances were remembered better when presented with descriptions whose grammatical gender matched the gender of the fragrance. Overall, results suggest grammatical manipulations of odor descriptions can affect odor cognition.

It has been argued that people use multiple frames of reference (FORs) for representing and updating spatial relationships between objects in a complex environment. When there are conflicts among representations of multiple FORs, they compete to determine behavior. “Frame of Reference-based Map of Salience” theory (FORMS) suggests that FORs with high salience may be processed in priority. Here, we report a computational neural network model for a two-cannon task, which naturally involves multiple FORs with different levels of salience: intrinsic frame of reference (IFOR) and egocentric frame of reference (EFOR). The goal is to investigate the computational neural mechanisms underlying human spatial performance. Our simulation results fit earlier behavioral results well. The model suggests although multiple FORs may be initially represented independently, they interfere with each other by the inhibitory competition of neurons in the later process (in hidden layer) for conflict resolution. Moreover, salience may modulate the competition by prioritizing FORs with high salience levels. These results represent a connectionist support for the FORMS theory.

The psycholinguistic theory of Communication Accommoda-tion proposes that people modify communication dynamics(e.g. vocal patterns, gesture, word choice, syntax, etc.) tominimize (or maximize) their social differences. Research oncommunication accommodation has shown that people whowant social approval will modify their linguistic style to matchthat of their interactant; however, most studies have been con-ducted on small-scale datasets and in laboratory situations. Inthis work, we investigate the relationship between linguisticsyntactic usage and conversation participation in a more nat-uralistic conversational setting: social media conversations onReddit.com. We introduce a novel approach for calculatingdocument-level syntactic similarity by relying on natural lan-guage processing methods (parse tree generators) and graphtheory techniques (minimum weight perfect matching on com-plete bipartite graphs). Using the proposed method, we presentthe results of two experiments which demonstrate that userswho comment on a post tend to use syntax similar to that ofthe original post. Specifically, we provide evidence that com-ments on a post are more likely to follow the syntactic structureof the original post, compared to both random comments andalso posts by the author of the comment.

Infants, children and adults are capable of implicitly extracting regularities from their environment through statistical learning (SL). SL is present from early infancy and found across tasks and modalities, raising questions about the domain generality of SL. However, little is known about its’ developmental trajectory: Is SL fully developed capacity in infancy, or does it improve with age, like other cognitive skills? While SL is well established in infants and adults, only few studies have looked at SL across development with conflicting results: some find age-related improvements while others do not. Importantly, despite its postulated role in language learning, no study has examined the developmental trajectory of auditory SL throughout childhood. Here, we conduct a large-scale study of children's auditory SL across a wide age-range (5-12y, N=115). Results show that auditory SL does not change much across development. We discuss implications for modality-based differences in SL and for its role in language acquisition.

In a classic experiment, Saffran, Aslin, and Newport (1996)used a headturn preference procedure to show that infants candiscriminate between familiar syllable sequences (“words”)and new syllable sequences (“non-words” and “part-words”).While several computational models have simulated aspects oftheir data and proposed that the learning of transitional prob-abilities could be mediated by neural-net or chunking mech-anisms, none have simulated the absolute values of infants’listening times in the different experimental conditions. In thispaper, we used CHREST, a model based on chunking, to sim-ulate these listening times. The model simulated the fact thatinfants listened longer to novel words (non-words and part-words) than familiar words. While the times observed with themodel were longer than those observed with infants, we makea novel finding with regard to phonological store trace decay.We also propose how to modify CHREST to produce data thatfits closer to the human data.

Structured sequence processing (SSP) refers to theneurocognitive mechanisms used to learn sequential patternsin the environment. SSP ability seems to be important forlanguage (Conway, Bauernschmidt, Huang, & Pisoni, 2010);however, there are few neural studies showing an empiricalconnection between SSP and language. The purpose of thisstudy was to investigate the association between SSP andlanguage processing by comparing the underlying neuralcomponents elicited during each type of task. Healthy adultsubjects completed a visual, non-linguistic SSP taskincorporating an artificial grammar and a visual morpho-syntactic language task. Both tasks were designed to causeviolations in expectations of items occurring in a series.Event-related potentials (ERPs) were used to examine theunderlying neural mechanisms associated with theseexpectancy violations. The results indicated the P3acomponent elicited by the SSP task and the P600 componentelicited by the language task shared similarities in theirtopographic distribution. These preliminary analyses suggestthat the P3a and P600 may reflect processes involvingdetection of sequential violations in non-language andlanguage domains, which is consistent with the idea thatlanguage processing relies on general-purpose SSPmechanisms.

Language is ordered in time and an incremental processingsystem encounters temporary ambiguity in the middle of sen-tence comprehension. An optimal incremental processing sys-tem must solve two computational problems: On the one hand,it has to keep multiple possible interpretations without choos-ing one over the others. On the other hand, it must rejectinterpretations inconsistent with context. We propose a re-current neural network model of incremental processing thatdoes stochastic optimization of a set of soft, local constraintsto build a globally coherent structure successfully. Bifurcationanalysis of the model makes clear when and why the modelparses a sentence successfully and when and why it does not—the garden path and local coherence effects are discussed. Ourmodel provides neurally plausible solutions of the computa-tional problems arising in incremental processing

Models of concept learning and theory acquisition often in-voke a stochastic search process, in which learners generatehypotheses through some structured random process and thenevaluate them on some data measuring their quality or value.To be successful within a reasonable time-frame, these mod-els need ways of generating good candidate hypotheses evenbefore the data are considered. Schulz (2012a) has proposedthat studying the origins of new ideas in more everyday con-texts, such as how we think up new names for things, can pro-vide insight into the cognitive processes that generate good hy-potheses for learning. We propose a simple generative modelfor how people might draw on their experience to proposenew names in everyday domains such as pub names or actionmovies, and show that it captures surprisingly well the namesthat people actually imagine. We discuss the role for an anal-ogous hypothesis-generation mechanism in enabling and con-straining causal theory learning.

Here we investigated how music reading experience modu-lates visual spans in language reading. Participants wereasked to identify music notes, English letters, Chinese charac-ters, and novel symbols (Tibetan letters) presented at randomlocations on the screen while maintaining central fixation. Wefound that for music note reading, musicians outperformednon-musicians at some peripheral positions in both visualfields, and for English letter reading, musicians outperformednon-musicians at some peripheral positions in the RVF butnot in the LVF. In contrast, in both Chinese character andnovel symbol reading, musicians and non-musicians did notdiffer in their performance at peripheral positions. Since bothmusic and English reading involve a left-to-right reading di-rection and a RVF/LH advantage, these results suggest thatthe modulation of music reading experience on visual spans inlanguage reading depends on the similarities in the cognitiveprocesses involved.

A successful design accounts for the structure of the problemit is aimed at solving. When it is a human-directed design,this includes the expectations of its users. How do we arriveat such a design? One approach starts from first principles(e.g., simplicity, unity, symmetry, balance) to evaluate thequality of proposed designs. Here, we introduce design fromzeroth principles, a form of human-in-the-loop computationthat synthesizes a design that conforms to its users’ expecta-tions. The technique begins by constructing a transmissionchain seeded with a random design. Each user in the chain isexposed to the design and then recreates it, passing alongtheir recreation to the next user, who does the same. Throughthis iterative process, the users’ perceptual, inductive, and re-constructive biases directly transform the initial design intoone that is better fit to human cognition. Such designs are eas-ier to learn and harder to forget. We evaluated the approach inthree domains — stimulus–response mappings, vanity phonenumbers, and letter placement in typeset words — and showthat it produces a good design in each.

Moral judgments are elicited using dilemmas presentinghypothetical situations in which an agent must choosebetween letting several people die or sacrificing one person inorder to save them. The evaluation of the action or inaction ofa human agent is compared to those of two artificial agents –a humanoid robot and an automated system. Ratings ofrightness, blamefulness and moral permissibility of action orinaction in incidental and instrumental moral dilemmas areused. The results show that for the artificial cognitive agentsthe utilitarian action is rated as more morally permissible thaninaction. The humanoid robot is found to be less blameworthyfor his choices compared to the human agent or to theautomated system. Action is found to be more appropriate,morally permissible, more right, and less blameworthy thaninaction only for the incidental scenarios. The results areinterpreted and discussed from the perspective of perceivedmoral agency.

The rich literature on multiple object tracking (MOT)conclusively demonstrates that humans are able to visuallytrack a small number of objects. There is considerably lessagreement on what perceptual and cognitive processes areinvolved. While it is clear that MOT is attentionallydemanding, various accounts of MOT performance centrallyinvolve pre-attentional mechanisms as well. In this paper wepresent an account of object tracking in the ARCADIAcognitive system that treats MOT as dependent upon both pre-attentive and attention-bound processes. We show that withminimal addition this model replicates a variety of corephenomena in the MOT literature and provides an algorithmicexplanation of human performance limitations.

Many studies have shown that observers can accuratelyperceive and evaluate the statistical summary of presentedobjects’ attribute values, such as the average, withoutattending to each object. However, it remains controversialhow the visual system integrates the attribute values (e.g.,information on size) of multiple items and computes theaverage value. In this study, we tested how distributions ofitem sizes affect the precision and bias in judging averagevalues. We predicted that if observers utilize all of theavailable size information equally, the distribution wouldhave no effect, and vice versa. Our results showed that, withnovice observers, judgement precision differed among sizedistributions and that the observers overestimated the size ofthe average value compared to the actual size under allconditions. These results imply that observations of someitems in a set could be weighted more easily than others, withthe possibility that this process is easier for larger items thansmaller ones. However, this was not the case for experiencedobservers, who showed no effects of distribution type onaverage assessment performance. Our findings imply that theprocess of representing the average value may not beexplained by a single definitive mechanism and, is rathermediated by a mixture of multiple cognitive processes.

We care tremendously about what other people think of us.Motivated by two lines of prior work – children’s inferentialand communicative capacities and strategic reputation man-agement – we examine how children infer what others thinkof them given others’ observations of their performance, andhow they influence these beliefs through disclosing their per-formance. In Experiment 1, 3-5 year-olds played a luck-basedgame; one confederates watched the child win and anotherconfederate watched the child lose. We asked the child to dis-close an additional, unobserved win to one of the two confed-erates. We find that younger children overwhelmingly choosethe person who previously saw them win. However, as ageincreased, children were more likely to choose to disclose tosomeone who previously saw them lose. In Experiment 2,adults played a similar third person version and selectivelychose the person who saw the main character previously lose.

Metaphors pervade discussions of critical issues and influ-ence how people reason about these domains. For instance,when crime is a beast, people suggest enforcement-orientedapproaches to crime-reduction (e.g., by augmenting the po-lice force); when crime is a virus, on the other hand, peoplesuggest systemic reforms for the affected community. In thecurrent study, we find that extending metaphoric language intothe descriptions of policy interventions bolsters the persuasiveinfluence of metaphoric frames for an array of important is-sues. When crime is a beast, people are even more likely toendorse “attacking” the problem with harsh enforcement tac-tics; when crime is a virus people are even more likely to toendorse “treating” the problem through social reform.

Generic language (e.g., “tigers have stripes,” “girls hatemath”) is a powerful vehicle for communicating essentialistbeliefs. One way generic language likely communicates thesebeliefs is by leading children to generate kind-basedexplanations about particular properties; e.g., if a child hears“girls hate math,” he may infer that there must be an inherentcausal basis for the generalization, which in turn supportsessentialist beliefs. However, it is also possible that simplyhearing a category described with generics elicits the beliefthat the category is an appropriate kind to generalize about.On this account, even if the generic is negated (“girls don’thate math”), the generic language might nonetheless leadchildren to essentialize the category. The current studysupports the latter possibility, suggesting that even hearingnegated generics (“girls don’t hate math”) may still fostersocial essentialism.

The limited ability to simultaneously perform multiple tasksis one of the most salient features of human performance anda defining characteristic of controlled processing. Based onthe assumption that multitasking constraints arise from sharedrepresentations between individual tasks, we describe a graph-theoretic approach to analyze these constraints. Our resultsare consistent with previous numerical work (Feng, Schwem-mer, Gershman, & Cohen, 2014), showing that even modestamounts of shared representation induce dramatic constraintson the parallel processing capability of a network architecture.We further illustrate how this analysis method can be appliedto specific neural networks to efficiently characterize the fullprofile of their parallel processing capabilities. We presentsimulation results that validate theoretical predictions, and dis-cuss how these methods can be applied to empirical studiesof controlled vs. and automatic processing and multitaskingperformance in humans.

A racial priming article claimed that, relative to a controlcondition, an exotic variety of variability, called 1/ƒ noise, isaltered when stereotypes impact participants’ judgments in animplicit prejudice task (Correll, 2008). However, Madurskiand LeBel (2014) recently described two powerful, faithfullycloned, and apparently decisive studies that each failed toreturn a successful literal replication of Correll’s report.Madurski and LeBel outlined and subsequently eliminatedseveral potential extraneous reasons for their replicationfailures, such as different participant demographics,participant non-compliance, poor psychometrics, andhardware discrepancies. By contrast, this article reports asuccessful conceptual replication of the pattern reported byCorrell (cf. Schmidt, 2009). Notably, this conceptualreplication required adjustments to Correll’s original methodand statistical analyses. All the changes were dictated by asystems theory of 1/ƒ noise that was largely in place prior toCorrell’s report (Kello, Beltz, Holden, & Van Orden, 2007;Van Orden, Holden, & Turvey, 2003; 2005). Implications forthe replication debate are discussed, with emphasis oncontextualizing implicit cues.

People can estimate numerical quantities, like the number of grapes in a bunch, using the Approximate Number System (ANS). Individual differences in this ability (ANS acuity) are emerging as an important predictor in research areas ranging from math skills to judgment and decision making. One commonly used ANS acuity metric is the size of the Numerical Distance Effect (NDE): the amount of savings in RT or errors when distinguishing stimuli values as the numerical distance between them increases. However, the validity of this metric has recently been questioned. Here, we model the relationship between the NDE-size and ANS acuity. We demonstrate that the relationship between NDE- size and ANS acuity should not be linear, but rather should resemble an inverted J-shaped distribution, with the largest NDE-sizes typically being found for near average ANS acuities.

We search for various things every day – food, information onthe Internet or someone’s name in memory. Despite the dif-ferent nature of these tasks, they all have a common feature –a final goal with an unknown location in a complex environ-ment. This property of the search raises a problem of trade-offbetween exploration of new opportunities and exploitation ofthe known information. We used the data from the semanticfluency task experiment to investigate how humans switch be-tween exploration and exploitation strategies when they searchin memory and whether they do it optimally. On comparingfour different search models, the one that assumes that humansswitch search strategies according to the semantic quality ofthe current neighbourhood best fits the data. Moreover, par-ticipants who set higher thresholds for the words with betterquality of the neighbourhood tend to retrieve more words frommemory. We also used regression analysis to find out whichfactors affect efficiency of both search strategies.

Metaphors suffuse language and affect how people think. Ameta-analysis of metaphor framing studies conducted between1983 and 2000 concluded that metaphors are about 6% morepersuasive than literal language (Sopory & Dillard, 2002).However, each of these studies was conducted in English withsamples drawn from populations of native English speakers.Here, we test whether and how language proficiency moderatesthe influence of metaphor frames. Sampling from a populationof non-native, but generally proficient English speakers fromIndia, we found that metaphor frames systematically affectedpeople who reported using English primarily in informal con-texts (i.e., among friends and family and through the media)but not those who reported using English primarily in formalcontexts (i.e., for school or work). We discuss the implica-tions of this finding for countries like the US, where Englishis increasingly a non-native language for its residents, and fortheories of language processing more generally.

Many propositions are not known to be true or false, andmany phenomena are not understood. What determineswhat propositions and phenomena are perceived asknowable or unknowable? We tested whether factorsrelated to scientific methodology (a proposition’sreducibility and falsifiability), its intrinsic metaphysics (themateriality of the phenomena and its scope ofapplicability), and its relation to other knowledge (itscentrality to one’s other beliefs and values) influenceknowability. Across a wide range of naturalistic scientificand pseudoscientific phenomena (Studies 1 and 2), as wellas artificial stimuli (Study 3), we found that reducibilityand falsifiability have strong direct effects on knowability,that materiality and scope have strong indirect effects (viareducibility and falsifiability), and that belief and valuecentrality have inconsistent and weak effects onknowability. We conclude that people evaluate theknowability of propositions consistently with principlesproposed by epistemologists and practicing scientists.

Past research on action planning has shed light on the neuralmechanisms underlying the selection of simple motor actions,along with the cognitive mechanisms underlying the planningof action sequences in constrained problem solving domains.We extend this research by describing a neural model thatrapidly plans action sequences in relatively unconstrained do-mains by manipulating structured representations of objectsand the actions they typically afford. We provide an analysisthat indicates our model is able to reliably accomplish goalsthat require correctly performing a sequence of up to 5 actionsin a simulated environment. We also provide an analysis ofthe scaling properties of our model with respect to the num-ber of objects and affordances that constitute its knowledgeof the environment. Using simplified simulations we find thatour model is likely to function effectively while picking from10,000 actions related to 25,000 objects.

Using hierarchical Bayesian estimation of RT distributions, wepresent a novel application of Systems Factorial Technology(Townsend & Nozawa, 1995) to the retrieval of item and asso-ciative information from episodic memory. We find that itemand associative information are retrieved concurrently, withpositive memory evidence arising from a holistic match be-tween the test pair and the contents of memory, in which bothitem and associative matches are pooled together into a sin-gle source. This retrieval architecture is inconsistent with bothstrictly serial processing and independence of item and asso-ciative information. Pooling of item and associative matchesimplies that while item and associative information may beseparable, they are not qualitatively different, nor are quali-tatively different processes (e.g., familiarity vs. recollection)used to retrieve these kinds of information.

hildren acquire exact meanings for number words in distinctstages. First, they learn one, then two, and then three andsometimes four. Finally, children learn to apply the countingprocedure to their entire count list. Although these stages areubiquitous and well documented, the foundation of thesemeanings remains highly contested. Here we ask whetherchildren assign preliminary meanings to number words beforelearning their exact meanings by examining their responses onthe Give-a-Number task to numbers for which they do not yethave exact meanings. While several research groups haveapproached this question before, we argue that because thesedata do not usually conform to a normal distribution, typicalmethods of analysis likely underestimate their knowledge.Using non-parametric analyses, we show that children acquirenon-exact meanings for small number words like one, two,three, four and possibly for higher numbers well before theyacquire the exact meanings.

We humans spend almost a third of our lives asleep, andthere is mounting evidence that sleep not only maintains, butactually improves many of our cognitive functions. Mem-ory consolidation–the process of crystallizing and integratingmemories into knowledge and skills–is particularly benefittedby sleep. We survey the evidence that sleep aids memory con-solidation in various declarative and implicit tasks and reviewthe basic neurophysiological structure of sleep with a focus onunderstanding what neural systems are involved. Drawing onmachine learning research, we discuss why it might be usefulfor humans–and robots, perhaps–to have such an offline pe-riod for processing, even though humans are clearly capable oflearning incrementally, online. Finally, we propose and simu-late two mechanisms for use in computational memory modelsto accomplish sleep-based consolidation via either or both 1)re-encoding knowledge representations and 2) reactivating andstrengthening recent memories.

Young children share fairly and expect others to do the same.Yet little is known about the underlying cognitivemechanisms that support fairness. Across two experiments,we investigated whether children’s numerical competenciesare linked with their sharing behavior. Preschoolers (aged2.5-5.5) participated in either third-party (Experiment 1) orfirst-party (Experiment 2) resource allocation tasks.Children’s numerical competence was then assessed using theGive-N-Task (Sarnecka & Carey, 2008; Wynn, 1990).Numerical competence – specifically knowledge of thecardinal principle explained age-related changes in fairsharing in both the third- and first-party contexts. Theseresults suggest that an understanding of the cardinal principleserves as an important mechanism for fair sharing behavior.

How do people understand the pragmatics of spatial language?We propose a rational-speech act model for spatial reasoning,and apply it to the terms ‘in’ and ‘near’. We examine people’sfine-grain spatial reasoning in this domain by having them lo-cate where an event occurred, given an utterance. Our prag-matic listener model provides a quantitative and qualitative fitto people’s inferences.

In the current study, we present the methods for creating andvalidating a science curiosity scale. We find that the scalepresented here is unidimensional and highly reliable.Moreover, it predicts engagement with a science documentaryclip more accurately than do measures of science intelligenceor education. Although more steps are needed, this providesinitial evidence for the utility of our measure of sciencecuriosity.

Is language production dynamically regulated by cognitive control? If so, how domain-general is this process? In two experiments, we studied conflict adaptation, or conflict-driven adjustments of control, in two paradigms: Picture-Word Interference (PWI), which induces linguistic conflict, and Prime-Probe (PP), which induces visuospatial conflict. Exp. 1 tested within-task conflict adaptation separately in PWI and PP. Exp. 2 tested cross-task adaptation by alternating the two tasks in a task-switching paradigm. We found reliable within- task conflict adaptation in both PWI and PP, but neither an analysis of individual differences (Exp. 1), nor a direct manipulation of between-task conflict (Exp. 2) revealed cross- task adaptation. We further report a robust 2-back within-task adaptation in Exp. 2 to refute alternative accounts of null cross- task adaptation. These findings support models of dynamic, top-down control in language production that posit at least some degree of domain-specificity.

Early visual object recognition in a world full of cluttered vi-sual information is a complicated task at which toddlers areincredibly efficient. In their everyday lives, toddlers con-stantly create learning experiences by actively manipulatingobjects and thus self-selecting object views for visual learn-ing. The work in this paper is based on the hypothesis that ac-tive viewing and exploration of toddlers actually creates high-quality training data for object recognition. We tested thisidea by collecting egocentric video data of free toy play be-tween toddler-parent dyads, and used it to train state-of-the-artmachine learning models (Convolutional Neural Networks, orCNNs). Our results show that the data collected by parentsand toddlers have different visual properties and that CNNscan take advantage of these differences to learn toddler-basedobject models that outperform their parent counterparts in aseries of controlled simulations.

Iconicity is a fundamental feature of human language. However its processing consequences at the behavioral and neural level in spoken word comprehension are not well understood. The current paper presents the behavioral and electrophysiological outcome of an auditory lexical decision task in which native speakers of Dutch listened to onomatopoeic words and matched control words while their electroencephalogram was recorded. Behaviorally, onomatopoeic words were processed as quickly and accurately as words with an arbitrary mapping between form and meaning. Event-related potentials time-locked to word onset revealed a significant decrease in negative amplitude in the N2 and N400 components and a late positivity for onomatopoeic words in comparison to the control words. These findings advance our understanding of the temporal dynamics of iconic form-meaning mapping in spoken word comprehension and suggest interplay between the neural representations of real-world sounds and spoken words.

Recent work suggests that cultural transmission can lead to the emergence of linguistic structure as speakers’ weak individual biases become amplified through iterated learning. However, to date, no published study has demonstrated a similar emergence of linguistic structure in children. This gap is problematic given that languages are mainly learned by children and that adults may bring existing linguistic biases to the task. Here, we conduct a large-scale study of iterated language learning in both children and adults, using a novel, child-friendly paradigm. The results show that while children make more mistakes overall, their languages become more learnable and show learnability biases similar to those of adults. Child languages did not show a significant increase in linguistic structure over time, but consistent mappings between meanings and signals did emerge on many occasions, as found with adults. This provides the first demonstration that cultural transmission affects the languages children and adults produce similarly.

People have a common-sense notion of intelligence and use itto evaluate decisions and decision-makers. One can attributeintelligence by evaluating the strategy or the outcome of agoal-directed agent. We propose a model of intelligence at-tribution, based on inverse planning in Partially ObservableMarkov Decision Processes (POMDPs) in a probabilistic envi-ronment, inferring the most likely planning parameters givenobserved actions. The model explains the agent’s decisionsby a combination of probabilistic planning, a softmax decisionnoise, prior knowledge about the world and forgetting, estimat-ing the agent’s intelligence by a proxy measure of efficientlyoptimising costs and rewards. Behavioural evidence from twoexperiments shows that people cluster into those who attributeintelligence to the strategy and those who attribute intelligenceto the outcome of the observed actions. People in the strat-egy cluster attribute more intelligence to decisions that min-imise the agent’s overall cost, even if the outcome is unlucky.People in the outcome cluster attribute intelligence to the out-come, judging low-cost outcomes as a sign of intelligence evenif the outcome is accidental and make neutral judgements be-fore they observe the result. Our model explains human in-telligence judgements better than perceptual cues such as thenumber of revisits or moves.

Habitual sentences (e.g. Bill smokes.) generalize an event overtime, but how do you know when a habitual sentence is true?We develop a computational model and use this to guide exper-iments into the truth conditions of habitual language. In Ex-pts. 1 & 2, we measure participants’ prior expectations aboutthe frequency with which an event occurs and validate thepredictions of the model for when a habitual sentence is ac-ceptable. In Expt. 3, we show that habituals are sensitive totop-down moderators of expected frequency: It is the expec-tation of future tendency that matters for habitual language.This work provides the mathematical glue between our intu-itive theories’ of others and events and the language we useto talk about them.

Many grammatical dependencies in natural language involve elements that are not adjacent, such as between the subject and verb in the child always runs. To date, most experiments showing evidence of learning non- adjacent dependencies have used artificial languages in which the to-be-learned dependencies are presented in isolation by presenting the minimal sequences that contain the dependent elements. However, dependencies in natural language are not typically isolated in this way. In this study we exposed learners to non-adjacent dependencies in long sequences of words. We accelerated the speed of presentation and learners showed evidence for learning of non-adjacent dependencies. The previous pause-based positional mechanisms for learning of non-adjacent dependency are challenged.

Fuzzy-trace theory assumes that decision-makers processqualitative “gist” representations and quantitative “verbatim”representations in parallel. Here, we develop a formal modelof fuzzy-trace theory that explains both processes. The modelalso integrates effects of individual differences in numeracy,metacognitive monitoring and editing, and sensation seeking.Parameters of the model varied in theoretically meaningfulways with differences in numeracy, monitoring, and sensationseeking, accounting for risk preferences at multiple levels.Relations to current theories and potential extensions arediscussed.

Dual-processing accounts of reasoning havegained renewed attention in the past decade,particularly in the fields of social judgment,learning, and decision-making under uncertainty.Although the various accounts differ, thecommon thread is the distinction between twoqualitatively different types of reasoning:explicit/implicit, rational/affective, fast/slow, etc.Consequently, much research has focused oncharacterizing the two different processes. Lessextensive are the attempts to find mediators thatinfluence which process is used. In this paper, weargue that the missing perspective on these dual-processing theories is rooted in dynamicalsystems theory. By shifting the perspective to thedynamic interaction and transitions betweendifferent types of reasoning, we provide atheoretical framework for dual-processing withan emphasis on phase transitions. As a specialcase, we focus on dual-processing in decision-making and judgment under uncertainty forwhich we will propose suggestions for futureexperimental evaluation.

Successfully navigating the social world requires reasoningabout both high-level strategic goals, such as whether to co-operate or compete, as well as the low-level actions neededto achieve those goals. While previous work in experimentalgame theory has examined the former and work on multi-agentsystems has examined the later, there has been little work in-vestigating behavior in environments that require simultaneousplanning and inference across both levels. We develop a hierar-chical model of social agency that infers the intentions of otheragents, strategically decides whether to cooperate or competewith them, and then executes either a cooperative or competi-tive planning program. Learning occurs across both high-levelstrategic decisions and low-level actions leading to the emer-gence of social norms. We test predictions of this model inmulti-agent behavioral experiments using rich video-game likeenvironments. By grounding strategic behavior in a formalmodel of planning, we develop abstract notions of both co-operation and competition and shed light on the computationalnature of joint intentionality.

We investigated the role of speech-accompanying gestures inanalogical problem solving. Participants attempted to solveDuncker’s (1945) Radiation Problem after reading and retelling astory that described an analogous solution in a different domain.Participants were instructed to gesture, instructed not to gesture, orgiven no instructions regarding gesture as they retold the story.Participants who were instructed to gesture as they retold theanalogous story were more likely to mention perceptual details intheir description and less likely to apply the analogous solution tothe problem than participants who were instructed not to gesture.These results suggest that gestures can be detrimental to analogousproblem solving when the perceptual elements of a story areirrelevant to its schematic similarity with a problem.

Why do children learn some words earlier than others? Reg-ularities and differences in the age of acquisition for wordsacross languages yield insights regarding the mechanismsguiding word learning. In a large-scale corpus analysis,we estimate the ages at which 9,200 children learn 300-400words in seven languages, predicting them on the basis ofindependently-derived linguistic, environmental, and concep-tual factors. Predictors were surprisingly consistent across lan-guages, but varied across development and as a function oflexical category (e.g., concreteness predicted nouns while lin-guistic structure predicted function words). By leveraging dataat a significantly larger scale than previous work, our analyseshighlight the power that emerges from unifying previously dis-parate theories, but also reveal the amount of reliable variationthat still remains unexplained.

Although previous studies have found a link between thequantity and quality of child-directed speech learners receiveand their vocabulary development, no previous studies havefound a parallel link between overheard speech measured ata very young age and vocabulary development (Shneidman& Goldin-Meadow, 2012; Shneidman, Arroyo, Levine, &Goldin-Meadow, 2013; Weisleder & Fernald, 2013). This isdespite the fact that children are able to learn words from over-heard speech in laboratory settings (Shneidman & Woodward,2015). Drawing on the idea that children preferentially at-tend to stimuli that are at a manageable level of complexity(Kidd, Piantadosi, & Aslin, 2012, 2014), the present researchexplores the possibility that children do not initially tune intooverheard speech because it is initially too complex for theirstage of lexical development (i.e., contains too great a propor-tion of unfamiliar words). Using transcripts from CHILDESand the Santa Barbara Corpus, and estimates of vocabularyby age from the MB-CDI, we find that child-directed speechis significantly less complex than overheard speech throughat least 30 months. If attention based on complexity at leastpartially accounts for the statistical independence of overheardspeech and vocabulary development in early childhood, thenchildren might only begin learning from more complex, over-heard speech sometime after 30 months.

There is substantial variation in language experience betweenlearners, yet there is surprising similarity in the languagestructure they eventually acquire. While it is possible that thiscanalisation of language structure may be due to constraintsimposed by modulators, such as an innate language system, itmay instead derive from the broader, communicativeenvironment in which language is acquired. In this paper, thelatter perspective is tested for its adequacy in explaining therobustness of language learning to environmental variation. Acomputational model of word learning from cross-situational,multimodal information was constructed and tested. Key tothe model’s robustness was the presence of multiple,individually unreliable information sources that could supportlearning when combined. This “degeneracy” in the languagesystem had a detrimental effect on learning when compared toa noise-free environment, but was critically important foracquiring a canalised system that is resistant to environmentalnoise in communication.

Previous research showed that children’s spatial language production predicts their spatial skills, but the mechanisms underlying this relation remain a source of debate. This study examined whether 4-year-olds’ spatial skills were predicted by their attention to task-relevant information—in tasks that emphasize either memory or language—above and beyond their spatial word production. Children completed three types of tasks: (1) a memory task assessing attention to task-relevant color, size, and location cues; (2) a production task assessing adaptive use of language to describe scenes, varying in color, size, and location; and (3) spatial tasks. After controlling for age, gender, and vocabulary, children’s spatial skills were significantly predicted by their memory for task-relevant cues, above and beyond their task-related language production and adaptive use of language. These findings suggest that attending to relevant information is a process supporting spatial skill acquisition and underlies the relation between language and spatial cognition.

Novel labels provide feedback that may enhance categoricalalignment between interlocutors. However, the nature of thisfeedback may not always be linguistic. Lupyan (2008) hasdemonstrated the effects of labels on individualcategorization, and even non-word labels have seeminglyproduced greater consistency in sorting strategies (Lupyan &Casasanto, 2014). We extend this to alignment bydemonstrating that arbitrary labels can increase sortingconsistency to bring people’s categories closer together, evenwithout dialogue. Importantly, we argue that increasedalignment is not always due to labeling in a linguistic sense.Results suggest that it is not the content of the non-wordlabels driving the alignment effects, but the very presence ofthe labels acting as ‘anchors’ for category formation. Thisdemonstrates a more general cognitive effect of arbitrarylabels on categorization.

Real-world scenes contain low-level visual features (e.g., edges, colors) and high-level semantic features (e.g., objects and places). Traditional visual perception models assume that integration of low-level visual features and segmentation of the scene must occur before high-level semantics are perceived. This view implies that low-level visual features of a scene alone do not carry semantic information related to that scene. Here we present evidence that suggests otherwise. We show that high-level semantics can be preserved in low-level visual features, and that different high-level semantics can be preserved in different types of low-level visual features. Specifically, the ‘disorder’ of a scene is preserved in edge features better than color features, whereas the converse is true for ‘naturalness.’ These findings suggest that semantic processing may start earlier than thought before, and integration of low-level visual features and segmentation of the scene may occur after semantic processing has begun, or in parallel.

We investigate the age-old questions of what makes us whowe are and what features of identity, if changed, would makeus a different person. Previous approaches to identity havesuggested that there is a type of feature that is most definingof identity (e.g., autobiographical memories or moralqualities). We propose a new approach to identity thatsuggests that, like concepts in general, more causally centralfeatures are perceived as more defining of the self-concept. Inthree experiments, using both measured and manipulatedcausal centrality, we find that changes to features of identitythat are perceived as more causally central are moredisruptive to both the identity of the self and others.

Prominent theories suggest that time and number arerepresented by a common magnitude system. However,distinct patterns of temporal and numerical processingoccur in the presence of emotional stimuli, calling intoquestion theories of a common magnitude system, whilealso unveiling questions regarding the mechanismsunderlying these temporal and numerical biases. Wetested whether numerical processing, like temporalprocessing, may be impacted by increased arousal levels,yet have a higher threshold level in order to impactestimates. If so, then induced arousal may reverse thetypical pattern of numerical underestimation in thepresence of emotions. Adults (N = 85) participated ineither a stress-induction or a control version of the task.Then, participants completed a numerical bisection task inthe presence and absence of emotional content. Increasingarousal had no impact on numerical processing, except inthe presence of happy faces, providing further evidencefor distinct processing mechanisms.

Language is a powerful cognitive tool. For example, labelingobjects or features of problems can support categorization andrelational thinking. Less is known about their role in makinginferences about the structure of mathematics problems. Wetest the impact of labeling decimals such as 0.25 using formalplace value labels (“two tenths and five hundredths”)compared to informal labels (“point two five”) or no labels onchildren’s problem-solving performance. Third- and fourth-graders (N = 104) were randomly assigned to one of threeconditions (formal labels, informal labels, or no labels) andlabeled decimals while playing a magnitude comparison gameand number line estimation task. Formal labels facilitatedperformance on comparison problems that requiredunderstanding the role of zero, which highlighted place valuestructure. However, formal labels hindered performance whenexplicit understanding of place value magnitudes wasrequired. Findings highlight how the language teachers andstudents use can impact problem-solving success.

Moral judgment often involves pinning causation for harm toa particular person. Since it reveals “who one sides with”, ex-pression of moral judgment can be a costly social act that peo-ple may be motivated to conceal. Here, we demonstrate thata simple, well-studied psycholinguistic task (implicit causal-ity) can be leveraged as a novel implicit measure of morallyrelevant causal attributions. Participants decided whether tocontinue sentences like “Amy killed Bob because...” with ei-ther the pronoun he or she. We found that (1) implicit causal-ity selections predicted explicit causal judgments, (2) select-ing the object (victim) for harm/force events (e.g., kill, rape)predicted endorsement of moral values previously linked tovictim-blame, and (3) higher hostile sexism predicted select-ing the female as the cause in male-on-female harm/force. Theimplicit causality task is a new measure of morally motivatedcausal attribution that may circumvent social desirability con-cerns.

Any adaptive organism faces the choice between taking actions with known benefits (exploitation), and sampling new actions to check for other, more valuable opportunities available (exploration). The latter involves information- seeking, a drive so fundamental to learning and long-term reward that it can reasonably be considered, through evolution or development, to have acquired its own value, independent of immediate reward. Similarly, behaviors that fail to yield information may have come to be associated with aversive experiences such as boredom, demotivation, and task disengagement. In accord with these suppositions, we propose that boredom reflects an adaptive signal for managing the exploration-exploitation tradeoff, in the service of optimizing information acquisition and long-term reward. We tested participants in three experiments, manipulating the information content in their immediate task environment, and showed that increased perceptions of boredom arise in environments in which there is little useful information, and that higher boredom correlates with higher exploration. These findings are the first step toward a model formalizing the relationship between exploration, exploitation and boredom.

Infants’ speech perception adapts to the phonemic categoriesof their native language, a process assumed to be driven bythe distributional properties of speech. This study investigateswhether deep neural networks (DNNs), the current state-of-the-art in distributional feature learning, are capable oflearning phoneme-like representations of speech in anunsupervised manner. We trained DNNs with unlabeled andlabeled speech and analyzed the activations of each layer withrespect to the phones in the input segments. The analysesreveal that the emergence of phonemic invariance in DNNs isdependent on the availability of phonemic labeling of theinput during the training. No increased phonemic selectivityof the hidden layers was observed in the purely unsupervisednetworks despite successful learning of low-dimensionalrepresentations for speech. This suggests that additionallearning constraints or more sophisticated models are neededto account for the emergence of phone-like categories indistributional learning operating on natural speech.

Human languages can be seen as socially evolved systems thathave been structured to optimize information flow incommunication. Communication appears to proceed both moreefficiently and more fluently when information is distributedevenly across the linguistic signal. In previous work (Ramscaret al., 2013), we used tools from information theory to examinehow naming systems evolved to meet this requirementhistorically, and how, over the past several hundred years,social legislation and rapid population growth have disruptednaming practices in the West, making names ever harder toprocess and remember. In support of these observations, wepresent findings from three experiments investigating namefluency, recognition, and recall. These results provideconverging empirical evidence for an optimal solution to namedesign, and offer a more nuanced understanding of how socialengineering has impaired the structure of names in memory.

What makes some metaphors easier to understand than oth-ers? Theoretical accounts of metaphor processing appeal todimensions like conventionality and aptness to explain vari-ability in metaphor comprehensibility. In a typical experiment,one group of naive participants rates a set of metaphoric sen-tences along these dimensions, while another is timed readingthe same sentences. Then, the ratings are used to predict re-sponse times in order to identify the most relevant linguistic di-mension for metaphor comprehension. However, surprisinglyhigh correlations between ratings of theoretically orthogonalconstructs and the results of an experiment in which a con-text manipulation affected ratings of metaphor conventionalityand aptness suggest that these measures should be treated asdependent, rather than explanatory, variables. We discuss theimplications of this perspective for theories of language pro-cessing.

The present accent rating study investigates the interactionbetween accent strength and grammatical correctness on per-ceived accentedness. German native (L1) listeners rated Ger-man sentences produced by L1 and non-native (L2) speakers.Sentences either contained a grammatical error or were gram-matically correct. Results showed that grammatical correct-ness affected the accent rating of sentences produced by L1speakers, but not of those by L2 speakers. The inverse influ-ence of grammatical errors on sentences spoken with strongeraccents suggests that phonological information plays a moreimportant role for global perception of speech accentednessthan grammatical correctness does, revealing a hierarchical im-portance of factors that form an L2 accent. This finding is inline with recent findings from an online processing ERP study(Hanul ́ıkov ́a, van Alphen, van Goch, & Weber, 2012) in whichL1 listeners were tolerant towards grammatical errors made byL2 speakers, i.e. showed no P600 effect for grammatically in-correct sentences.

Humans demonstrate remarkable abilities to predict physicalevents in complex scenes. Two classes of models for physicalscene understanding have recently been proposed: “IntuitivePhysics Engines”, or IPEs, which posit that people make pre-dictions by running approximate probabilistic simulations incausal mental models similar in nature to video-game physicsengines, and memory-based models, which make judgmentsbased on analogies to stored experiences of previously en-countered scenes and physical outcomes. Versions of the lat-ter have recently been instantiated in convolutional neural net-work (CNN) architectures. Here we report four experimentsthat, to our knowledge, are the first rigorous comparisonsof simulation-based and CNN-based models, where both ap-proaches are concretely instantiated in algorithms that can runon raw image inputs and produce as outputs physical judg-ments such as whether a stack of blocks will fall. Both ap-proaches can achieve super-human accuracy levels and canquantitatively predict human judgments to a similar degree,but only the simulation-based models generalize to novel sit-uations in ways that people do, and are qualitatively consis-tent with systematic perceptual illusions and judgment asym-metries that people show.

The event segmentation theory (EST) postulates that humanssystematically segment the continuous sensorimotor informa-tion flow into events and event boundaries. The basis for theobserved segmentation tendencies, however, remains largelyunknown. We introduce a computational model that groundsEST in the interaction abilities of a system. The model learnsevents and event boundaries based on actively gathered senso-rimotor signals. It segments the signals based on principles ofprobabilistic predictive coding and surprise. The implementedmodel essentially simulates, anticipates, and learns event pro-gressions and event transitions online while interacting withthe environment by means of dynamic, predictive Bayesianmodels. Besides the model’s event segmentation capabilities,we show that the learned encodings can be used for higher-order planning. Moreover, the encodings systematically con-ceptualize environmental interactions and they help to identifythe factors that are critical for ensuring interaction success.

Daily life often makes us decide between two goals: maximizing immediate rewards (exploitation) and learning about the environment so as to improve our options for future rewards (exploration). An adaptive organism therefore should place value on information independent of immediate reward, and affective states may signal such value (e.g., curiosity vs. boredom: Hill & Perkins, 1985; Eastwood et al. 2012). This tradeoff has been well studied in “bandit” tasks involving choice among a fixed number of options, but is equally pertinent in situations such as foraging, hunting, or job search, where one encounters a series of new options sequentially. Here, we augment the classic serial foraging scenario to more explicitly reward the development of knowledge. We develop a formal model that quantifies the value of information in this setting and how it should impact decision making, paralleling the treatment of reward by the marginal value theorem (MVT) in the foraging literature. We then present the results of an experiment designed to provide an initial test of this model, and discuss the implications of this information-foraging framework on boredom and task disengagement.

It has been suggested that the way that number words are used mayplay an important role in the development of number concepts.However, little is currently known about the overall ways in whichnumber words are used in child-directed speech. To address this,we performed an analysis of how number words are used in theCHILDES database. We looked at four statistics: 1) lexicalfrequency, 2) contextual diversity, 3) word co-occurrence, and 4)distributional similarity, to see if these distributional statisticssuggest why some aspects of number acquisition are easy andothers are hard, and if these statistics are informative about specificdebates in number acquisition. We found that that are manyimportant differences in how small and large number words areused (such as differences in frequency, co-occurrence patterns, anddistributional similarity), differences that may play an role inshaping hypotheses about children’s acquisition of numberconcepts. Keywords: number representation, language acquisition,concept acquisition, statistical learning, corpus analyses

The physical behavior of moving fluids is highly complex, yetpeople are able to interact with them in their everyday liveswith relative ease. To investigate how humans achieve thisremarkable ability, the present study extended the classicalwater-pouring problem (Schwartz & Black, 1999) to examinehow humans take into consideration physical properties of flu-ids (e.g., viscosity) and perceptual variables (e.g., volume) ina reasoning task. We found that humans do not rely on simplequalitative heuristics to reason about fluid dynamics. Instead,they rely on the perceived viscosity and fluid volume to makequantitative judgments. Computational results from a prob-abilistic simulation model can account for human sensitivityto hidden attributes, such as viscosity, and their performanceon the water-pouring task. In contrast, non-simulation mod-els based on statistical learning fail to fit human performance.The results in the present paper provide converging evidencesupporting mental simulation in physical reasoning, in addi-tion to developing a set of experimental conditions that rectifythe dissociation between explicit prediction and tacit judgmentthrough the use of mental simulation strategies.

A common obstacle for students in the transition from arithmetic to algebra is developing a conceptual understanding of equations representing functions. Two experiments manipulated isomorphic problems in terms of their solution requirements (computation vs. interpretation) and format to test for understanding of linear functions. Experiment 1 provided problems in a story context, and found that performance on slope comparison problems was low, especially when problems were presented with equations. Experiment 2 tested whether performance on slope comparison problems improves when problem prompts include explicit mathematical terminology rather than just natural language consistent with the problem story. Results suggest that many undergraduate students fail to access the mathematical concept of slope when problem prompts are presented with natural language. Overall, the results suggest that even undergraduate students lack understanding of the slope concept and equations of linear functions, both which are foundational for advanced algebraic thinking.

Using cross-modal priming, we investigated the processing of idioms in non-native listeners in varying experimental contexts. As idiomatic processing models have presented evidence for an idiomatic mode of processing that can be activated for non-native speakers in highly figurative contexts (Bobrow & Bell, 1973), this experiment revisits those claims while also examining access to figurative meaning in addition to the literal meaning of individual words within an idiom. This experiment showed increased priming for visual targets related to the figurative meaning of an idiom when the experimental list contained a large proportion of idiomatic sentences compared to when the list contained only a small proportion of idiomatic sentences. Non-native speakers not only showed online access to figurative meaning but were also sensitive to highly idiomatic contexts; though, responses to the targets related to literal meaning of the final word of the idiom were faster in all instances than figuratively-related targets.

Studies of sound symbolism have shown that people can associate sound and meaning in consistent ways when presented with maximally contrastive stimulus pairs of nonwords such as bouba/kiki (rounded/sharp) or mil/mal (small/big). Recent work has shown the effect extends to antonymic words from natural languages and has proposed a role for shared cross-modal correspondences in biasing form- to-meaning associations. An important open question is how the associations work, and particularly what the role is of sound-symbolic matches versus mismatches. We report on a learning task designed to distinguish between three existing theories by using a spectrum of sound-symbolically matching, mismatching, and neutral (neither matching nor mismatching) stimuli. Synthesized stimuli allow us to control for prosody, and the inclusion of a neutral condition allows a direct test of competing accounts. We find evidence for a sound-symbolic match boost, but not for a mismatch difficulty compared to the neutral condition.

Previous work has found that people feel significantly moresatisfied with explanations of psychological phenomena whenthose explanations contain neuroscience information — evenwhen this information is entirely irrelevant to the logic of theexplanations. This seductive allure effect was firstdemonstrated by Weisberg, Keil, Goodstein, Rawson, & Gray(2008), and has since been replicated several times inindependent labs (e.g., Fernandez-Duque, Evans, Christian, &Hodges, 2014; Rhodes, Rodriguez, & Shah, 2014; Weisberg,Taylor, & Hopkins, 2015). However, these studies onlyexamined psychological explanations with addedneuroscience information. The current study thus investigatedthe generality of this effect and found that the seductive allureeffect occurs across several scientific disciplines wheneverthe explanations include reference to smaller components ormore fundamental processes. These data suggest that peoplehave a general preference for reductive explanations.

Child-directed speech is often temporally organized such that successive utterances refer to the same topic. This type of extended discourse on the same referent has been shown to possess several verbal signatures that could facilitate learning. Here, we reveal multiple non-verbal correlates to extended discourse that could also aid learning. Multimodal analyses of extended discourse episodes reveal that during these episodes, toddlers and parents exhibit greater sustained attention on objects, and greater coordination between their behaviors. The results indicate the interconnections between multiple aspects of the language-learning environment, and suggest that parents’ speech may both shape and be shaped by non-verbal processes. Implications for understanding how the learning environment influences development are discussed.

The approximate number system (ANS) is frequentlyconsidered to be a foundation for the acquisition of uniquelyhuman symbolic numerical capabilities. However, themechanism by which the ANS influences symbolic numberrepresentations and mathematical thought remains poorlyunderstood. Here, we tested the relation between ANS acuity,cardinal number knowledge, approximate arithmetic, andsymbolic math achievement in a one-year longitudinalinvestigation of preschoolers’ early math abilities. Our resultssuggest that cardinal number knowledge is an intermediaryfactor in the relation between ANS acuity and symbolic mathachievement. Furthermore, approximate arithmeticperformance contributes unique variance to math achievementthat is not accounted for by ANS acuity. These findingssuggest that there are multiple routes by which the ANSinfluences math achievement. Therefore, interventionstargeting both the precision and manipulability of the ANSmay prove to be more beneficial for improving mathematicalreasoning compared to interventions targeting only one ofthese factors.

The Model of Intuitive Morality and Exemplars (MIME)predicts a mutual dependency between the moral scrutiny ofmediated narratives and media exposure. This study proposesmoral judgments of media content are not only related to basalmoral domain salience and exemplars, but also to the immediateprocessing state of the individual at the moment of exposure. Anexperiment manipulating construal level prior to exposure to amediated narrative was conducted to test this proposal. The resultssuggest that evaluations of moral violations are modulated byconstrual level. High-level construal led to harsher, moreconsistent judgments of domain-violator morality, eliminating theeffect of baseline moral intuitions. Low-level construal induced anapparent trade-off in moral evaluation strategy which is sensitive toboth narrative outcome and domain salience. When domainviolators were punished, intuitive moral salience was negativelycorrelated with moral evaluations; however, when domainviolators were rewarded, the opposite trend emerged. Thesefindings indicate the need for an adjustment to the MIME model toallow for processing states to interact with moral domain salienceand moral judgments of media content. They also suggest that thestrength and quality of moral intuitions are not robust to broadercognitive processes, but interact with them.

Although many theories of causal cognition are based on causalgraphical models, a key property of such models—the inde-pendence relations stipulated by the Markov condition—isroutinely violated by human reasoners. Two accounts of whypeople violate independence are formalized and subjected toexperimental test. Subjects’ inferences were more consistentwith a dual prototype model in which people favor networkstates in which variables are all present or all absent than aleaky gate model in which information is transmitted throughnetwork nodes when it should normatively be blocked. Thearticle concludes with a call for theories of causal cognitionthat rest on foundations that are faithful to the kinds of causalinferences people actually draw.

Studies of child development of tool use precursors show suc-cessive but overlapping phases of qualitatively different typesof behaviours. We hypothesize that two mechanisms in par-ticular play a role in the structuring of these phases: the in-trinsic motivation to explore and the representation used toencode sensorimotor experience. Previous models showedhow curiosity-driven learning mechanisms could allow theemergence of developmental trajectories. We build uponthose models and present the HACOB (Hierarchical ActiveCuriosity-driven mOdel Babbling) architecture that activelychooses which sensorimotor model to train in a hierarchy ofmodels representing the environmental structure. We studythis architecture using a simulated robotic arm interacting withobjects in a 2D environment. We show that overlapping phasesof behaviours are autonomously emerging in hierarchical mod-els using active model babbling. To our knowledge, this isthe first model of curiosity-driven development of simple tooluse and of the self-organization of overlapping phases of be-haviours. In particular, our model explains why and how in-trinsically motivated exploration of non-optimal methods tosolve certain sensorimotor problems can be useful to discoverhow to solve other sensorimotor problems, in accordance withSiegler’s overlapping waves theory, by scaffolding the learningof increasingly complex affordances in the environment.

Foraging is a search process common to all mobile organisms. Spatial memory can improve foraging efficiency and efficacy, and evidence indicates that many species—including humans—actively utilize spatial memory to aid in their foraging, yet most current models of foraging do not include spatial memory. In this study, a simple online foraging game was used to attempt to replicate and extend findings from a recent study (Kerster, Rhodes, & Kello, 2016) to further investigate the role of spatial memory in foraging. The game involved searching a simple 2d space by clicking the mouse to try and find as many resources as possible in 300 clicks. Spatial information was displayed that provided complete information about search history in order test how “perfect” spatial memory improves search performance. Over 1000 participants were recruited to participate in the task using Amazon’s Mechanical Turk, which allowed this test to be performed across a wide parameter space of different resource distributions. Results replicated many of the findings of earlier studies, and demonstrated that spatial memory can have a dramatic effect on search performance.

People use spatial metaphors to talk about temporal concepts.They also gesture frequently during speech. Thecharacteristics of these gestures give information regardingthe mental timelines people form to experience time. Thepresent study investigates the expression of temporal conceptson a natural setting with Turkish speakers. We found thatTurkish speakers used more metaphoric temporal phrases(e.g., short period, time flies quickly) than words referring totime without spatial content (e.g., today, nowadays) in asession where they talked about people’s fortune.Spontaneous gestures were mainly classified as metaphoricand beat gestures and were mostly produced on the sagittalaxis, which contradicts with the previous findings. Yet, wealso found that people used vertical axis to represent currentand future events. These findings suggest that lateral axis maynot always be the most common direction for co-speechtemporal gesture use, and the pragmatic constraints of theenvironment may influence the spatial conceptualization oftime.

To investigate children’s awareness of authority to changerules, we showed children (ages 4-7) videos of one childplaying a game alone or three children playing a gametogether. In the group video, the game rule was initiated either:by one of the children, by three children collaboratively or byan adult. They then were asked whether the characters in thevideos could change the rules. Children believed that thecharacter could change the rule when playing alone. Theirresponses to the group video depended on how the rule wasinitiated. They attributed authority to change rules only to thechild who initiated the rule, unless the rule was createdcollaboratively. We also asked children whether they couldchange norms (school/moral/artifact norms) in daily life; andfound moral/artifact distinction in children’s endorsement ofnorm changing. These results suggest that children recognizeflexibility in changing rules even in preschool years.

In this paper, we develop a memory model that predicts retrieval characteristics of real-world facts. First, we show how ACT-R’s memory model can be used to predict people’s knowledge about real-world objects. The model assumes the probability of retrieving a chunk of information about an object and the time to retrieve this information depend on the pattern of prior environmental exposure to the object. Second, we use frequencies of information appearing on the Internet as a proxy for what information people would encounter in their natural environment, outside the laboratory. In two Experiments, we use this model to account for subjects’ associative knowledge about real-world objects as well as the associated retrieval latencies. Third, in a computer simulation, we explore how such model predictions can be used to understand the workings and performance of decision strategies that operate on the contents of declarative memory.

Theory of mind is a powerful cognitive ability: by the ageof six, people are capable of accurately reasoning about oth-ers’ beliefs and desires. An influential series of language un-derstanding experiments by Keysar and colleagues, however,showed that adults systematically failed to take a speaker’sbeliefs into account, revealing limitations on theory of mind.In this paper we argue that these apparent failures are in factsuccesses. Through a minimal pair of replications comparingscripted vs. unscripted speakers, we show that critical utter-ances used by Keysar and colleagues are uncooperative: theyare less informative than what a speaker would actually pro-duce in that situation. When we allow participants to naturallyinteract, we find that listener expectations are justified and er-rors are reduced. This ironically shows that apparent failuresof theory of mind are in fact attributable to sophisticated ex-pectations about speaker behavior—that is, to theory of mind.

As new concepts and discoveries accumulate over time, theamount of information available to speakers increases as well.One would expect that an utterance today would be more in-formative than an utterance 100 years ago (basing informationon surprisal; Shannon, 1948), given the increase in technol-ogy and scientific discoveries. This prediction, however, is atodds with recent theories regarding information in human lan-guage use, which suggest that speakers maintain a somewhatconstant information rate over time. Using the Google Ngramcorpus (Michel et al., 2011), we show for multiple languagesthat changes in lexical information (a unigram model) are actu-ally negatively correlated with changes in structural informa-tion (a trigram model), supporting recent proposals on infor-mation theoretic constraints.

The learnability of center-embedded recursive structures has attracted much attention (Corballis, 2007; Friederici, 2004; Rey, Perruchet, & Fagot, 2012). However, most of the previous studies adopted the artificial grammar learning paradigm (Reber, 1967) and did not apply natural language stimuli. Rather, they applied synthetic meaningless training materials, which hardly represent the richness and complexity of natural language. Accordingly, in the current study, we attempt to tighten the link between artificial language learning and natural language acquisition in the auditory modality, by enriching our learning environment with phonological cues that occur in natural, spoken information; in particular, Chinese tones. In a grammaticality judgment task, we examined the syntactical processing by participants from different language backgrounds. Through the cross-language comparison between Chinese and Dutch native speakers, we aim to test the influence of language-specific phonological cues on processing complex linguistic structures. The results showed that tones had a more beneficial learning effect for Chinese than for Dutch participants. In other words, when participants learned a new language, they were likely to bring their own language routines implicitly from the familiar native language into processing the unfamiliar one.

Semantic networks have been used extensively in psychologyto describe how humans organize facts and knowledge inmemory. Numerous methods have been proposed to constructsemantic networks using data from memory retrieval tasks,such as the semantic fluency task (listing items in a category).However these methods typically generate group-levelnetworks, and sometimes require a very large amount ofparticipant data. We present a novel computational methodfor estimating an individual’s semantic network usingsemantic fluency data that requires very little data. Weestablish its efficacy by examining the semantic relatedness ofassociations estimated by the model.

We examined the effects of individual versus joint action on aSimon task using motion tracking to explore the implicitcognitive dynamics underlying responses. In both individual andjoint conditions, participants were slower to respond, and weredifferentially attracted to the distracter response location, whenthe spatial component of the stimulus was incompatible with theresponse location. When two people completed similar twochoice tasks together, the results were not statistically differentfrom the individual condition, even though the magnitude of thestimulus-response compatibility effect was slightly larger.Neither was there an increased effect when the partner had nostimulus-response conflict to resolve. We found no evidence foran action conflict when the responses of the two partners weredifferent. These data imply that the literature regarding the JointSimon task is still in the process of determining the relevantevents that interact with and support joint action.

Listeners track distributions of speech sounds along percep-tual dimensions. We introduce a method for evaluating hy-potheses about what those dimensions are, using a cognitivemodel whose prior distribution is estimated directly from speechrecordings. We use this method to evaluate two speaker nor-malization algorithms against human data. Simulations showthat representations that are normalized across speakers predicthuman discrimination data better than unnormalized representa-tions, consistent with previous research. Results further revealdifferences across normalization methods in how well eachpredicts human data. This work provides a framework forevaluating hypothesized representations of speech and lays thegroundwork for testing models of speech perception on naturalspeech recordings from ecologically valid settings.

Two experiments examined the effects of interactive tutorial features (compared to “passive” features) on learning spatial tasks, an area seldom explored in interactivity research. Experiment 1 results indicated that for simple spatial tasks, interactive tutorials hindered learning for participants of higher spatial ability but improved learning for lower-ability participants. This interaction can be explained by “compensation,” the notion that people of higher ability can compensate for poor external support (passive tutorials) while people of lower ability need the better support. It is likely that the increased cognitive load of interactivity (Kalyuga, 2007) hindered high-spatial participants on a relatively easy task. In Experiment 2, task difficulty was increased, and the results revealed that the interactive tutorial produced better learning than the passive tutorial, regardless of spatial abilities. With the relatively difficult task, the benefits of interactivity became clearer because most people actually needed the interactive features despite the associated cognitive load.

Parsing actions entails that relations between objects are dis-covered. A pervasively neural account of this process requiresthat fundamental problems are solved: the neural pointer prob-lem, the binding problem, and the problem of generating dis-crete processing steps from time-continuous neural processes.We present a prototypical solution to these problems in a neuraldynamic model that comprises dynamic neural fields holdingrepresentations close to sensorimotor surfaces as well as dy-namic nodes holding discrete, language-like representations.Making the connection between these two types of represen-tations enables the model to parse actions as well as groundmovement phrases—all based on real visual input. We demon-strate how the dynamic neural processes autonomously gen-erate the processing steps required to parse or ground object-oriented action.

We examine how people judge the probabilities of real-world events, such as natural disasters in different countries. We find that the associations between the words and phrases that constitute these events, as assessed by vector space semantic models, strongly correlate with the probabilities assigned to these events by participants. Thus, for example, the semantic proximity of “earthquake” and “Japan” accurately predicts judgments regarding the probability of an earthquake in Japan. Our results suggest that the mechanisms and representations at play in language are also active in high- level domains, such as judgment and decision making, and that existing insights regarding these representations can be used to make precise, quantitative, a priori predictions regarding the probability estimates of individuals.

We examine the effect of variability in model parameters on the predictions of expected utility theory and cumulative prospect theory, two of the most influential choice models in decision making research. We find that zero-mean and symmetrically distributed noise in the underlying parameters of these models can systematically distort choice probabilities, leading to false conclusions. Likewise, differences in choice proportions across decision makers might be due to differences in the amount of noise affecting underlying parameters rather than to differences in actual parameter values. Our results suggest that care and caution are needed when trying to infer the underlying preferences of decision makers, or the effects of psychological, biological, economic, and demographic variables on these preferences.

Doing long sums in the absence of complementary actions or artefacts is a multi-step procedure that quickly taxes working memory; congesting the phonological loop further handicaps performance. In the experiment reported here, participants completed long sums either with hands down—the low interactivity condition—or by moving numbered tokens—the high interactivity condition—while they repeated ‘the’ continuously, loading the phonological loop, or not. As expected, articulatory suppression substantially affected performance, but more so in the low interactivity condition. Independent measures of basic arithmetic skill and mathematics anxiety moderated the impact of articulatory suppression on performance in the low but not in the high interactivity condition. These findings suggest that working memory resources are augmented with interactivity, underscoring the importance of characterizing the properties of the system as it is configured by the dynamic agent-environment coupling.

Cross-linguistic research has shown that boundaries for lexical categories differ from language to language. The aim of this study is to explore these differences between languages in relation to the categorization differences within a language. Monolingual Dutch- (N=400) and French-speaking (N=300) Belgian adults provided lexical category judgments for three lexical categories that are roughly equivalent in Dutch and French. Each category was represented by good, borderline, and bad examples. A mixture modeling approach enabled us to identify latent groups of categorizers within a language and to evaluate cross-linguistic variation in relation to within- language variation. We found complex patterns of lexical variation within as well as between language groups. Even within a seemingly homogeneous group of speakers sharing the same mother tongue, latent groups of categorizers display a variability that resembles patterns of lexical variation found at a cross-linguistic level of comparison.

Although there is a clear and intuitive mapping between lin-guistic arguments of verbs and event participants, the mappingis not perfect. We review the linguistic evidence that indicatesthat the mapping is imperfect. We also present the results of anew experimental study that provides further support for a dis-sociation between event participants and linguistic arguments.The study consists of two tasks. The first task elicited intu-itions on conceptual event participants, and the second taskelicited intuitions on linguistic arguments in instrument verbsand transaction verbs. The results suggest that while instru-ment phrases and currency/price phrases are considered neces-sary event participants, they are not linguistic arguments.

Decisions often rely on judgments about the probabilitiesof various explanations. Recent research has uncovered ahost of biases that afflict explanatory inference: Wouldthese biases also translate into decision-making? We findthat although people show biased inferences when makingexplanatory judgments in decision-relevant contexts (Exp.1A), these biases are attenuated or eliminated when thechoice context is highlighted by introducing an economicframing (price information; Exp. 1B–1D). However, biasedinferences can be “locked in” to subsequent decisions whenthe judgment and decision are separated in time (Exp. 2).Together, these results suggest that decisions can be morerational than the corresponding judgments—leading tochoices that are rational in the output of the decisionprocess, yet irrational in their incoherence with judgments.

Pupils’ poor achievement in mathematics has recently been aconcern in many Western countries. In order to address this is-sue, it has been proposed to teach chess in schools. However,in spite of optimistic claims, no convincing evidence of the ac-ademic benefits of chess instruction has ever been provided,because no study has ever controlled for possible placebo ef-fects. In this experimental study, a three-group design (i.e., ex-perimental, placebo, and control groups) was implemented tocontrol for possible placebo effects. Measures of mathematicalability and metacognitive skills were taken before and after thetreatment. We were interested in metacognitive skills becausethey have been claimed to be boosted by chess instruction, inturn positively influencing the enhancement of mathematicalability. The results show that the experimental group (partici-pants attending a chess course) achieved better scores in math-ematics than the placebo group (participants attending a Gocourse) but not than the control group (participants attendingregular school lessons). With regard to metacognition, no dif-ferences were found between the three groups. These resultssuggest that some chess-related skills generalize to the mathe-matical domain, because the chess lessons compensated for thehours of school lessons lost, whereas the Go lessons did not.However, this transfer does not seem to be mediated by meta-cognitive skills, and thus appears to be too limited to offer ed-ucational advantages.

Three experiments investigate whether and why people accept explanations for symptoms that appeal to mental disorders, such as: “She experiences delusions because she has schizophrenia.” Such explanations are potentially puzzling, as mental disorder diagnoses are made on the basis of symptoms, and the DSM implicitly rejects a commitment to some common, underlying cause. Do laypeople nonetheless conceptualize mental disorder classifications in causal terms? Or is this an instance of non-causal explanation? Experiment 1 shows that such explanations are indeed found explanatory. Experiment 2 presents participants with novel disorders that are stipulated to involve or not involve an underlying cause across symptoms and people. Disorder classifications are found more explanatory when a causal basis is stipulated, or when participants infer that one is present (even after it’s denied in the text). Finally, Experiment 3 finds that merely having a principled, but non-causal, basis for defining symptom clusters is insufficient to reach the explanatory potential of categories with a stipulated common cause. We discuss the implications for accounts of explanation and for psychiatry.

Shared preferences are a critical component of social attrac-tion. Knowing that someone likes the same things as you do isindicative of broader underlying similarities that support suc-cessful social partnerships. However, not all overlaps in prefer-ences are equally informative. Here we propose that the rarityof overlaps in preferences may be a particularly salient cue forsocial affiliation. We find evidence that people are sensitiveto the rarity of overlaps in preferences and affiliate themselves(Experiment 1) or predict others’ affiliations (Experiment 2)with potential social partners who share a relatively rare pref-erence. Because preferences provide information about bothwhat people know and what they like, we also tested the ef-fect of overlaps in knowledge (without taste) and overlaps intaste (without knowledge) to understand why we are drawn topeople who share our preferences.

According to existing accounts of causation, people rely on asingle criterion to identify the cause of an event. Thephenomenon of causal illusions raises problems for suchviews. Causal illusions arise when a particular factor isperceived to be causal despite knowledge indicatingotherwise. According to what we will call the Dual-ProcessHypothesis of Causal Identification, identifying a causeinvolves two cognitive processes: 1) an automatic, intuitiveprocess that identifies possible causes on the basis ofperceptual cues (spatial and temporal) and 2) a slow,reflective process that identifies possible causes on the basisof causal inference, in particular, a consideration of possiblemechanism. Consistent with this hypothesis, we found that inresponse to a causal illusion shown in a naturalistic setting,people’s initial judgments of causation were higher than theirultimate judgments of causation (Experiment 1). Using anonline measure of the time-course of people’s causaljudgments, we found that people initially view animations ofcausal illusions as causal before concluding that they are non-causal (Experiment 2). Finally, we obtained similar resultsusing a deadline procedure, while also finding that the lowerthe cognitive reflectiveness (as measured by the CRT), thestronger people’s impressions of causation were (Experiment3). Implications for different classes of theories of causationare discussed.

Humans, as social beings, are capable of employing variousbehavioral cues, such as gaze, speech, manual action, and bodyposture, in everyday communication. However, to extract fine-grained interaction patterns in social contexts has beenpresented with methodological challenges. Cross-RecurrencePlot Quantification Analysis (CRQA) is an analysis methodinvented in theoretical physics and recently applied tocognitive science to study interpersonal coordination. In thispaper, we extend this approach to analyzing joint activities inchild-parent interaction. We define a new representation asCross Recurrence Block based on CRQA. With thisrepresentation, we are able to capture interpersonal dynamicsfrom more than two behavioral streams in one CrossRecurrence Plot and derive a suite of measures to quantifydetailed characteristics of coordination. Using a datasetcollected from a child-parent interaction study, we show thatthese quantitative measures of joint activities revealdevelopmental changes in coordinative behavioral patternsbetween children and parents.

Human behavior is plagued by shortsightedness. When faced withtwo options, smaller rewards are often chosen over larger rewards,even when such choices are potentially costly. In threeexperiments, we use big data techniques to examine how suchchoices might be driven by people’s temporal horizons. InExperiment 1, we determine the average distance into the futurepeople talk about in their tweets in order to determine the temporalhorizon of each U.S. state. States with further future horizons hadlower rates of risk taking behavior (smoking, binge drinking) andhigher rates of investment (e.g., education, infrastructure). InExperiment 2, we used an individual’s tweets to establish theirtemporal horizon and found that those with longer temporalhorizons were more willing to wait for larger rewards. InExperiment 3, we were once again able to predict the choicebehaviors of individuals from their tweets, this time showing thatthose with longer future horizons were less likely to take risks. Thefindings help establish a powerful relationship between people’sthoughts about the future and their decisions.

The present study explored passive, active, and constructive methods of learning problem solving procedures. Using subgoal learning, which has promoted retention and transfer in procedural domains, the study compared the efficacy of different methods for learning a programming procedure. The results suggest that constructive methods produced better problem solving performance than passive or active methods. The amount of instructional support that learners received in the three different constructive interventions also affected performance. Learners performed best when they either received hints about the subgoals of the procedure or received feedback on the subgoal labels that they constructed, but not when they received both. These findings suggest that in some cases constructing subgoal labels is better than passively or actively engaging with subgoal labels. There is an optimal level of instructional support for students engaging in constructive learning and that providing too much support can be equally as detrimental as providing too little support.

Fear of a negative stereotype about one’s performance canlead to temporary underperformance on tests; e.g. womenmay underperform on a math test when prompted to thinkabout gender. The current study extends this literature toexamine whether stereotype threat not only leads tounderperformance on tests, but also may impact reasoningand learning more broadly. We focus in particular on theeffects of stereotype threat on analogical learning, a complexreasoning process that imposes a high working memory load.In this study, we examined the effects of gender stereotypeswhen females were asked to learn by comparing themathematical concepts of combinations and permutations.Overall, participants given a threat before learning gained lessfrom the instruction, as reflected by assessments administeredimmediately after the lesson and after a 1-week delay. Thiscould lead to systematic differences in the quality of abstractrepresentational knowledge for individuals from negativelystereotyped groups.

The ability to improve in speed and accuracy as a result of re-peating some task is an important hallmark of intelligent bio-logical systems. Although gradual behavioural improvementsfrom practice have been modelled in spiking neural networks,few such models have attempted to explain cognitive devel-opment of a task as complex as addition. In this work, wemodel the progression from a counting-based strategy for ad-dition to a recall-based strategy. The model consists of twonetworks working in parallel: a slower basal ganglia loop, anda faster cortical network. The slow network methodically com-putes the count from one digit given another, correspondingto the addition of two digits, while the fast network gradually“memorizes” the output from the slow network. The faster net-work eventually learns how to add the same digits that initiallydrove the behaviour of the slower network. Performance ofthis model is demonstrated by simulating a fully spiking neu-ral network that includes basal ganglia, thalamus and variouscortical areas. Consequently, the model incorporates variousneuroanatomical data, in terms of brain areas used for calcula-tion and makes psychologically testable predictions related tofrequency of rehearsal. Furthermore, the model replicates de-velopmental progression through addition strategies in termsof reaction times and accuracy, and naturally explains observedsymptoms of dyscalculia.

Young infants are prolific word learners even though they are facing the challenge of referential uncertainty (Quine, 1960). Many laboratory studies have shown that human infants are skilled at inferring the correct referent of an object from ambiguous contexts (Swingley, 2009). However, little is known regarding how children visually attend to and select the target object among many other objects in view when parents name it during free play interactions. In the current study, we explored the looking pattern of 12-month-old infants using naturalistic first person images with varying degrees of referential ambiguity. Our data suggest that infants’ attention is selective and they tend to only select a small subset of objects to attend to at each learning instance despite the complexity of the data existed in the real world. This work allows us to better understand how perceptual properties of objects in infants’ view influence their visual attention, which is also related to how they select candidate objects to build word-object mappings.

Humans can seamlessly infer what other people like, based onwhat they do. Broadly, two types of accounts have beenproposed to explain different aspects of this ability. A firstaccount focuses on inferences from spatial information:agents choose and move towards things they like. A secondaccount focuses on inferences from statistical information:uncommon choices reveal preferences more clearly comparedto common choices. Here we argue that these two kinds ofinferences can be explained by the assumption that agentsmaximize utilities. We test this idea in a task where adultparticipants infer an agent’s preferences using a combinationof spatial and statistical information. We show that our modelpredicts human answers with higher accuracy than a set ofplausible alternative models.

To act effectively in a complicated, uncertain world, peopleoften rely on task-sets (TSs) that define action policies over arange of stimuli. Effectively selecting amongst TSs requiresassessing their individual utility given the current world state.However, the world state is, in general, latent, stochastic, andtime-varying, making TS selection a difficult inference for theagent. An open question is how observable environmentalfactors influence an actor's assessment of the world state andthus the selection of TSs. In this work, we designed a noveltask in which probabilistic cues predict one of two TSs on atrial-by-trial basis. With this task, we investigate how peopleintegrate multiple sources of probabilistic information in theservice of TS selection. We show that when action feedback isunavailable, TS selection can be modeled as “biased Bayesianinference”, such that individuals participants differentiallyweight immediate cues over TS priors when inferring thelatent world state. Additionally, using the model’s trial-by-trial posteriors over TSs, we calculate a measure of decisionconfidence and show that it inversely relates to reactiontimes. This work supports the hierarchical organization ofdecision-making by demonstrating that probabilistic evidencecan be integrated in the service of higher-order decisions overTSs, subsequently simplifying lower-order action selection.

The N400 component of the event-related brain potential iswidely used in research on language and semantic memory,but the cognitive functions underlying N400 amplitudes arestill unclear and actively debated. Recent simulations with aneural network model of word meaning suggest that N400amplitudes might reflect implicit semantic prediction error.Here, we extend these simulations to sentencecomprehension, using a neural network model of sentenceprocessing to simulate a number of N400 effects obtained inempirical research. In the model, sequentially incoming wordsupdate a representation capturing probabilities of elements ofsentence meaning, not only reflecting the constituentspresented so far, but also the model’s best guess at all featuresof the sentence meaning based on the statistical regularities inthe model’s environment internalized in its connectionweights. Simulating influences of semantic congruity, clozeprobability, a word’s position in the sentence, reversalanomalies, semantic and associative priming, categoricallyrelated incongruities, lexical frequency, repetition, andinteractions between repetition and semantic congruity, wefound that the update of the predictive representation ofsentence meaning consistently patterned with N400amplitudes. These results are in line with the idea that N400amplitudes reflect semantic surprise, defined as the change inthe probability distribution over semantic features in anintegrated representation of meaning occasioned by the arrivalof each successive constituent of a sentence.

The ability to ask questions during learning is a key aspect ofhuman cognition. While recent research has suggested com-mon principles underlying human and machine “active learn-ing,” the existing literature has focused on relatively simpletypes of queries. In this paper, we study how humans constructrich and sophisticated natural language queries to search for in-formation in a large yet computationally tractable hypothesisspace. In Experiment 1, participants were allowed to ask anyquestion they liked in natural language. In Experiment 2, par-ticipants were asked to evaluate questions that they did not gen-erate themselves. While people rarely asked the most informa-tive questions in Experiment 1, they strongly preferred moreinformative questions in Experiment 2, as predicted by an idealBayesian analysis. Our results show that rigorous information-based accounts of human question asking are more widely ap-plicable than previously studied, explaining preferences acrossa diverse set of natural language questions.

Sequential images have frequently been used as experimentalstimuli in the cognitive and psychological sciences to exploretopics like theory of mind, temporal cognition, discourse,social intelligence, and event sequencing, among others. Theassumption has been that sequential images provide a fairlyuniversal and transparent stimuli that require little to nolearning to decode, and thus are ideal for non-verbal tasks indevelopmental, clinical, and non-literate populations.However, decades of cross-cultural and developmentalresearch have actually suggested something different: thatsequential image comprehension is contingent on exposureand practice with a graphic system. I here review thisliterature and advocate for more sensitivity to the “fluency”needed to understand sequential images.

Working memory is a limited-capacity form of human mem-ory that actively holds information in mind. Which memoriesought to be maintained? We approach this question by showingan equivalence between active maintenance in working mem-ory and a Markov decision process in which, at each moment,a cognitive control mechanism selects a memory as the targetof maintenance. The challenge of remembering is then findinga maintenance policy well-suited to the task at hand. We com-pute the optimal policy under various conditions and defineplausible cognitive mechanisms that can approximate these op-timal policies. Framing the problem of maintenance in thisway makes it possible to capture in a single model many of theessential behavioral phenomena of memory maintenance, in-cluding directed forgetting and self-directed remembering. Fi-nally, we consider the case of imperfect metamemory — wherethe current state of memory is only partially observable — andshow that the fidelity of metamemory determines the effective-ness of maintenance.

In practice, mathematics education is blocked (i.e., teachingone topic at a time; CCSS, 2010), but research generallypromotes interleaving (i.e., teaching multiple topics together;Rohrer & Taylor, 2007). For example, fraction arithmetic isblocked with students being taught fraction addition beforefraction multiplication. Since students often confuse fractionoperations to produce arithmetic errors, interleaved fractionarithmetic instruction might be more productive than blockedinstruction to teach students to discriminate between theoperations. Additionally, a cognitive task analysis suggeststhat fraction multiplication may be a prerequisite to fractionaddition and thus reversing the blocking order may enhancelearning. Two experiments with fraction addition and fractionmultiplication were run. Experiments 1 and 2 show thatinterleaved instruction is generally better than the currentblocked instruction. Experiment 2 provides evidence thatblocking that reverses the standard order -- providing practiceon fraction multiplication before fraction addition -- producesbetter learning.

Game elements like points and levels are a popular tool tonudge and engage students and customers. Yet, no theory cantell us which incentive structures work and how to design them.Here we connect the practice of gamification to the theory ofreward shaping in reinforcement learning. We leverage thisconnection to develop a method for designing effective incen-tive structures and delineating when gamification will succeedfrom when it will fail. We evaluate our method in two behav-ioral experiments. The results of the first experiment demon-strate that incentive structures designed by our method helppeople make better, less short-sighted decisions and avoid thepitfalls of less principled approaches. The results of the sec-ond experiment illustrate that such incentive structures can beeffectively implemented using game elements like points andbadges. These results suggest that our method provides a prin-cipled way to leverage gamification to help people make betterdecisions.

Natural languages exhibit properties that are difficult to explainfrom a purely functional perspective. One of these properties isthe systematic lack of upper-bounds in the literal meaning ofscalar expressions. This investigation addresses the develop-ment and selection of such semantics from a space of possiblealternatives. To do so we put forward a model that integratesBayesian learning into the replicator-mutator dynamics com-monly used in evolutionary game theory. We argue this syn-thesis to provide a suitable and general model to analyze thedynamics involved in the use and transmission of language.Our results shed light on the semantics-pragmatics divide andshow how a learning bias in tandem with functional pressuremay prevent the lexicalization of pragmatic inferences.

Damage to the Medial Temporal Lobe (MTL) impairs declar- ative memory and perception. The Representational-Hierar- chical (RH) Account explains such impairments by assuming that MTL stores conjunctive representations of items and events, and that individuals with MTL damage must rely upon representations of simple visual features in posterior visual cortex. A recent study revealed a surprising anti-perceptual learning effect in MTL-damaged individuals: with exposure to a set of visual stimuli, discrimination performance worsened rather than improved. We expand the RH account to explain this paradox by assuming that visual discrimination is per- formed using a familiarity heuristic. Exposure to a set of highly similar stimuli entails repeated presentation of simple visual features, eventually rendering all feature representations equally (maximally) familiar and hence inutile for solving the task. Since the unique conjunctions represented in MTL do not occur repeatedly, healthy individuals are shielded from this perceptual interference. We simulate this mechanism with a neural network previously used to simulate recognition memory, thereby providing a model that accounts for both mnemonic and perceptual deficits caused by MTL damage us- ing a unified architecture and mechanism.

Children rapidly learn a tremendous amount about languagedespite limitations imposed on them by their developing cog-nitive abilities. One possible explanation for this rapid learn-ing is that caregivers tune the language they produce tothese limitations, titrating the complexity of their speech todevelopmentally-appropriate levels. We test this hypothesis ina large-scale corpus analysis, measuring the contingency be-tween parents’ and children’s speech over the first 5 years.Our results support the linguistic tuning hypothesis, showinga high degree of mostly parent-led coordination early in de-velopment that decreases as children become more proficientlanguage learners and users.

A common intuition, often captured in fiction, is that some impossible events (e.g., levitating a stone) are “more impossible” than others (e.g., levitating a feather). We investigated the source of this intuition, hypothesizing that graded notions of impossibility arise from explanatory considerations logically precluded by the violation at hand but still taken into account. Studies 1-2 involved college undergraduates (n = 192), and Study 3 involved preschool-aged children (n = 32). In Study 1, participants saw pairs of magical events (spells) that violated one of 18 causal principles—six physical, six biological, and six psychological—and were asked to indicate which spell would be more difficult to learn. Both spells violated the same causal principle but differed in their relation to a subsidiary principle. Participants’ judgments of spell difficulty honored the subsidiary principle, even when participants were given the option of judging the two spells equally difficult. Study 2 replicated the effects of Study 1 with Likert-type ratings, and Study 3 replicated those effects in children. Taken together, these findings suggest that events that defy causal explanation are interpreted in terms of explanatory considerations that hold in the absence of such violations.

Statistical learning (SL) is increasingly invoked as a set ofgeneral-purpose mechanisms upon which language learning isbuilt during infancy and childhood. Here we investigated theextent to which SL is related to adult language processing. Inparticular, we asked whether SL proclivities towards relationsthat are more informative of English are related to efficiency inreading English sentences by native speakers of Korean. Wefound that individuals with a stronger statistical learningsensitivity showed a larger effect of conditional wordprobability on word reading times, indicating that they moreefficiently incorporated statistical regularities of the languageduring reading. In contrast, L2 English proficiency was relatedto overall reading speed but not to the use of statisticalregularities.

Real people sometimes appear in fiction, for example,Napoleon in War and Peace. Readers may also believe that aperson who never actually appears in a novel couldpotentially appear there. In two experiments, we find evidencethat readers think that a real person could appear in specificnovels and physically interact with a character. This effect ismagnified when the person and character share spatial andtemporal elements of their setting.

This paper examines the development of therepresentativeness heuristic in early childhood. Using a novelparadigm, we investigated 3- to 6-year-old children’s abilityto use base-rate and individuating information in theirpredictive inferences. In Experiment 1, we presented childrenwith base-rate and individuating information separately to testtheir ability to use each independently. In Experiment 2, wepresented children with base-rate and individuatinginformation together. Two critical trial types were used, onein which the base-rate information and individuatinginformation pointed to the same response and one in whichthe base-rate and individuating information pointed toconflicting responses. Results suggest that children progressto adult-like heuristic-based responding at 6 years of age.

While speakers tailor referring expressions to the knowledgeof their addressees, they do so imperfectly. Our goal here is toprovide an explanation for this type of pattern by extending aprobabilistic model introduced to explain perspective-takingbehavior in comprehension. Using novel production data froma type of knowledge mismatch not previously investigated inproduction, we show that production patterns can also beexplained as arising from the probabilistic combination of thespeaker’s and the addressee’s perspectives. These resultsshow the applicability of the multiple-perspectives approachto language production, and to different types of knowledgemismatch between conversational partners.

Statistical learning (SL) is believed to be a mechanism that enables successful language acquisition. Language acquisition in turn is heavily influenced by environmental factors such as socioeconomic status (SES). However, it is unknown to what extent SL abilities interact with SES in affecting language outcomes. To examine this potential interaction, we measured event-related potentials (ERPs) in 38 children aged 7-12 while performing a visual SL task consisting of a sequence of stimuli that contained covert statistical probabilities that predicted a target stimulus. Hierarchical regression results indicated that SL ability moderated the relationship between SES (average of both caregiver’s education level) and language scores (grammar, and marginally with receptive vocabulary). For children with high SL ability, SES had a weaker effect on language compared to children with low SL ability, suggesting that having good SL abilities could help ameliorate the disadvantages associated with being raised in a family with lower SES.

Group work is frequently part of idea generation, despite evidence that group interaction may reduce productivity during brainstorming sessions. Idea quantity is one aspect of creativity, but the originality of ideas generated is also important. In this paper, we examine how different aspects of group interaction, such as who makes the most contributions to an idea and the number of group members contribute to an idea, impact the originality of concepts generated by engineering students. We found that the most original concepts were produced when the concept originator was the top contributor to the design, and when the majority of group members contributed to the concept, particularly among senior students. These results are discussed in relation to previous work and suggestions are made for future research that assesses the interaction between design fixation and group processes.

Teaching a new concept with gestures – hand movements thataccompany speech – facilitates learning above-and-beyondinstruction through speech alone (e.g., Singer & Goldin-Meadow, 2005). However, the mechanisms underlying thisphenomenon are still being explored. Here, we use eyetracking to explore one mechanism – gesture’s ability todirect visual attention. We examine how children allocatetheir visual attention during a mathematical equivalencelesson that either contains gesture or does not. We show thatgesture instruction improves posttest performance, andadditionally that gesture does change how children visuallyattend to instruction: children look more to the problem beingexplained, and less to the instructor. However lookingpatterns alone cannot explain gesture’s effect, as posttestperformance is not predicted by any of our looking-timemeasures. These findings suggest that gesture does guidevisual attention, but that attention alone cannot account for itsfacilitative learning effects.

Two experiments using novel complex working memory spantasks were performed, both requiring the participants toremember a span of letters whilst being distracted by theprocessing of words. Word processing could either be self-referential (SRP) or not. In the first experiment recallperformance was compared between SRP and non-SRPconditions using the same words. In the second experiment,we compared SRP and non-SRP in two tasks equalized insemantic processing but using different words. In bothexperiments recall performance was significantly lower afterSRP compared to non-SRP, indicating that SRP has adisruptive effect on the recall task. A cognitive modelimplemented in PRIMs, using goal competition during SRP,interfering with rehearsal of letters, could account for theobserved experimental results. If SRP interferes withsubsequent tasks in this manner it should also interfere withtasks other than recall, such as SRP occurring in daily life.

The forced-choice triad task has become increasinglypopular in use over recent years. While it is seen as beinga categorisation task (Lin & Murphy, 2001) variation intask instructions often leads to different results. Shipp,Vallée-Tourangeau, and Anthony (2014) used the triadtask to show that when participants are asked to choosean option that ‘goes best with the target’, they are morelikely to select the choice that shares an action relationwhen it also shares taxonomic information. Howeverusing the instruction to select the item that “goes best” isvague and might encourage a strategy other than acategorical decision. The present experiment used thesame triads as in Shipp et al. to test whether participantswould match items based on shared actions or sharedtaxonomic relations when given specific categorisationinstructions. The task instructions were manipulated sothat participants either selected the item that “goes best”,“goes best to form a category” or is “most similar” to thetarget. The results found instances where the instructionsof “goes best to form a category” led to a higherprobability that participants would select the actionchoices over the instructions of “goes best”. Howeverwhen participants were encouraged to use similarityoverall action choices were lower. Therefore the triadtask does encourage a natural categorisation strategy anddifferences in task instructions across research are aresult of the stimuli used.

Prospection is an important cognitive achievement, and isrelated to uniquely human abilities such as planning, delay ofgratification, and goal attainment. While prospection developsrapidly during early childhood, little is known about themechanisms that support its development. Here we exploredwhether encouraging children to talk about their extendedselves (self outside the present context) boosts theirprospective abilities. Preschoolers (N = 81) participated in a5-minute interaction with an adult in which they were askedto talk about events in the near future, distant future, nearpast, or present. Compared with children discussing theirpresent and distant future, children asked to discuss events intheir near future or near past displayed better planning andprospective memory. Additionally, those two conditions weremost effective in eliciting self-projection (use of personalpronouns). Results suggest that experience communicatingabout the close-in-time, extended self contributes tochildren’s future-oriented thinking.

Psychologists have studied syllogistic inferences for morethan a century, but no extant theory gives an adequate accountof them. Reasoners appear to reason using different strategies.A complete account of syllogisms must therefore explainthese strategies and the resulting differences from oneindividual to another in the patterns of conclusions that theydraw. We propose a dual-process theory that solves these twoproblems. It is based on the manipulation of mental models,i.e., iconic simulations of possibilities. We also propose a newway in which to analyze individual differences, whichdepends on implementing a stochastic computer program. Theprogram, mReasoner, generates an initial conclusion bybuilding and scanning a mental model. It can vary fourseparate factors in the process: the size of a model, itscontents, the propensity to consider alternative models, andthe propensity to revise its heuristic conclusions. The formertwo parameters control intuitive processes and the latter twocontrol deliberative processes. The theory accounts forindividual differences in an early study on syllogisms(Johnson-Laird & Steedman, 1978). The computational modelprovides an algorithmic account of the different processes onwhich three subsets of participants relied (Simulation 1). Italso simulates the performance of each individual participantin the study (Simulation 2). The theory and itsimplementation constitute the first robust account ofindividual differences in syllogistic reasoning.

Since Harnad (1990) pointed out the symbol grounding prob-lem, cognitive science research has demonstrated that ground-ing in perceptual or sensorimotor experience is crucial to lan-guage. Recent embodied cognition theories have argued thatlanguage is more important for grounding abstract than con-crete words; abstract words are grounded via language. Dis-tributional semantics has recently addressed the embodied na-ture of language and proposed multimodal semantic models.However, these models are not cognitively plausible becausethey do not address the recent embodiment view of abstractconcepts. Therefore, we propose a novel multimodal distribu-tional semantics in which abstract words are represented indi-rectly through grounded representations of their semanticallyrelated concrete words. A simulation experiment demonstratedthat the proposed model achieved better performance in com-puting the word similarity than other multimodal or text-baseddistributional models. This finding suggests that the indirectembodiment view is plausible and contributes to the improve-ment of multimodal distributional semantics.

Overhearing can be seen as active learning, and overheardspeech provides an increasingly viable source of linguisticinput across development. This study extends previous re-sults showing learning from overhearing simplified, pedagogicspeech to a more ecologically valid context. Children learnmultiple words and facts corresponding to novel toys eitherthrough an overheard phone call or through direct instruction.Remarkably, 4.5–6-year-olds learned four new words equallywell in both conditions. Their performance on a set of six factswas even better, especially when taught directly. Analysis ofthe videos revealed that older children with high test accuracyboth looked toward the experimenter often, and tracked ob-jects as she discussed them. 3–4.5-year-olds only learned factsfrom overhearing, and exhibited greater varability in attention.These results suggest learning from overhearing is driven byattention to the indirect input, and may be a skill that under-goes substantial development during the preschool years.

The ability to associate words is an important cognitive skill.In this study we investigate different methods for representingword associations in the brain, using the Remote AssociatesTest (RAT) as a task. We explore representations derived fromfree association norms and statistical n-gram data. Althoughn-gram representations yield better performance on the test, acloser match with the human performance is obtained with rep-resentations derived from free associations. We propose thatword association strengths derived from free associations playan important role in the process of RAT solving. Furthermore,we show that this model can be implemented in spiking neu-rons, and estimate the number of biologically realistic neuronsthat would suffice for an accurate representation.

A lot of people show a decline in performance when they haveto report a second target stimulus in a stream of distractorstimuli. Curiously, this decline only happens when the secondtarget appears approximately 200-500ms after the first target.Recently, Choi, Chang, Shibata, Sasaki, and Watanabe (2012)have shown that a short, one-hour training can eliminate this“attentional blink”. Up to now, it is still unclear why thistraining works. In this paper, we have evaluated a range ofdifferent training paradigms to test several hypotheses aboutthe mechanism behind the reduction of the attentional blink.Our results show that none of these training paradigms havea large training effect when administered in isolation. Thetraining by Choi et al. (2012) outperforms them all. The mostlikely explanation for this effect are temporal expectationsrelative to the first target.

The deployment of selective attention has been studied in depth as a mechanism of visual categorization for decades. However, little work has investigated how attentional mechanisms operate for non-visual domains, and many models of categorization tacitly presume domain-general attention use. In three experiments, we investigated whether learners deploy attention to novel auditory features when learning novel words in a similar fashion to the prevailing visual categorization findings. These studies yielded evidence of non-isomorphism, as selective attention in the auditory domain shows high context specificity, in contrast to the wide generalization of attention in the visual domain.

The current study examined cardiac and behavioral responsesto changing auditory and visual information while usingmodified oddball tasks. When instructed to press the samebutton for auditory and visual oddballs, auditory dominancewas found with cross-modal presentation slowing downvisual response times and decreasing visual accuracy. Wheninstructed to make separate responses to auditory and visualoddballs, visual dominance was found with cross-modalpresentation slowing down response times and decreasingauditory accuracy. However, examination of cardiacresponses that were time-locked to stimulus onset show cross-modal facilitation effects, with discrimination of oddballs andstandards occurring earlier in the course of processing in thecross-modal condition than in the unimodal conditions. Thesefindings shed light on potential mechanisms underlyingmodality dominance effects and have implications on tasksthat require simultaneous processing of auditory and visualinformation.

The contents and structure of semantic networks have been the focus of much recent research, with major advances in the development of distributional models. In parallel, connectionist modeling has extended our knowledge of the processes engaged in semantic activation. However, these two lines of investigation have rarely brought together. Here, starting from a standard textual model of semantics, we allow activation to spread throughout its associated semantic network, as dictated by the patterns of semantic similarity between words. We find that the activation profile of the network, measured at various time points, can successfully account for response times in the lexical decision task, as well as for subjective concreteness and imageability ratings.

What forces have shaped the evolution of the lexicon? Lan-guages evolve under the pressure of having to communicatean unbounded set of ideas using a finite set of linguistic struc-tures. This suggests why the transmission of ideas should becompressed such that one word will develop multiple senses.Previous theory also suggests how a word might develop newsenses: Abstract concepts may be construed in terms of moreconcrete concepts. Here, we bring these two perspectives to-gether to examine metaphorical extensions of English wordmeanings over the past millennium, analyzing how sensesfrom a source domain are extended to new ones in a target do-main. Using empirical and computational methods, we foundthat metaphorical mappings are highly systematic and can beexplained in terms of a compact set of variables. Our workshows how metaphor can provide a cognitive device for com-pressing emerging ideas into an existing lexicon.

When people perform joint actions together, task knowledge is sometimes distributed asymmetrically such that one person has information that another person lacks. In such situations, interpersonal action coordination can be achieved if the knowledgeable person modulates basic parameters of her goal-directed actions in a way that provides relevant infor- mation to the less knowledgeable partner. We investigated whether systematic violations of predicted movement duration provide a sufficient basis for such communication. Results of a joint movement task show that knowledgeable partners spontaneously and systematically violated the pre- dictions of Fitts’ law in order to communicate if their partners could not see their movements. Unknowing partners had a benefit from these violations and more so if the violations provided a good signal-to-noise ratio. Together, our findings suggest that generating and perceiving systematic deviations from the predicted duration of a goal-directed action can enable non-conventionalized forms of communication during joint action.

This work examines the extent to which people hold independent sequential events (e.g., players making correct/incorrect guesses) responsible for overall outcomes (e.g., the team winning/losing the game). Two types of events are found to garner the majority of responsibility for overall outcomes: (1) final events and (2) events that are perceived to disrupt momentum (e.g., an incorrect guess after a sequence of correct guesses). While previous research has shown that final events tend to be perceived as more responsible for overall outcomes, the current experiments are the first to document the role of perceived momentum on responsibility judgments. Specifically, we demonstrate that the effect is mediated by perceived momentum changes after the time of the event and moderated when exogenous factors (e.g., a delay between events) disrupt perceived momentum. We discuss how these findings relate to pivotality, the counterfactual simulation model, and the role of unexpectedness in responsibility judgments.

Investigations of the Sapir-Whorf hypothesis often ask whetherthere is a difference in the non-linguistic behavior of speak-ers of two languages, generally without modeling the underly-ing process. Such an approach leaves underexplored the rela-tive contributions of language and universal aspects of cogni-tion, and how those contributions differ across languages. Weexplore the naming and non-linguistic pile-sorting of spatialscenes across speakers of five languages via a computationalmodel grounded in an influential proposal: that language willaffect cognition when non-linguistic information is uncertain.We report two findings. First, native language plays a smallbut significant role in predicting spatial similarity judgmentsacross languages, consistent with earlier findings. Second, thesize of the native-language role varies systematically, such thatfiner-grained semantic systems appear to shape similarity judg-ments more than coarser-grained systems do. These findingscapture the tradeoff between language-specific and universalforces in cognition, and how that tradeoff varies across lan-guages.

The current study examined how relevant and irrelevantauditory stimuli affect the speed of responding to structuredvisual sequences. Participants were presented with a dot thatappeared in different locations on a touch screen monitor andthey were instructed to quickly touch the dot. Response timessped up over time, suggesting that participants learned thevisual sequences. Response times in Experiment 1 were slowerwhen the dot was paired with random sounds, suggesting thatirrelevant sounds slowed down visual processing/responding.Dots in Experiment 2 were paired with correlated sounds (bothauditory and visual information provided locationinformation). While the redundant intersensory informationdid not speed up response times, it did partially attenuateauditory interference. These findings have implications ontasks that require processing of simultaneously presentedauditory and visual information and provide evidence ofauditory interference and possibly dominance on a task thattypically favors the visual modality.

Some argue that category learning is mediated by two com-peting learning systems: one explicit, one implicit (Ashby etal., 1998). These systems are hypothesised to be responsi-ble for learning rule-based and information-integration cate-gory structures respectively. However, little experimental workhas directly investigated whether people are conscious of cat-egory knowledge supposedly learned by the implicit system.Here we report one experiment that directly compared explicitrecognition memory for exemplars between these two categorystructures. Contrary to the predictions of the dual-systems ap-proach, we found preliminary evidence of superior exemplarmemory after information-integration category learning com-pared to rule-based learning. This result is consistent with thehypothesis that participants learn information-integration cate-gory structures by using complex rules.

Speakers of Chinese and English share decodable neuralsemantic representations, which can be elicited by words ineach language. We explore various, common models ofsemantic representation and their correspondences to eachother and to these neural representations. Despite very strongcross-language similarity in the neural data, we find that twoversions of a corpus-based semantic model do not show thesame strong correlation between languages. Behavior-basedmodels better approximate cross-language similarity, butthese models also fail to explain the similarities observed inthe neural data. Although none of the examined modelsexplain cross-language neural similarity, we explore how theymight provide additional information over and above cross-language neural similarity. We find that native speakers’ratings of noun-noun similarity and one of the corpus modelsdo further correlate with neural data after accounting forcross-language similarities.

We model two patterns related to the acquisition of color termsin Russian and English: children produce overextension errorsfor some colors but not others, and language-specific distinc-tions affect color discrimination in a non-linguistic task. Botheffects, as well as a reasonable convergence with adult linguis-tic behavior, are shown by a Self-Organizing Map trained onnaturalistic input. We investigate the effect of different waysof representing colors, i.e., as perceptual features or in terms ofthe cognitive biases on categorization extracted from crosslin-guistic color naming data. We also consider the influence ofcolor term frequency. Our results suggest effects of all three ofterm frequency, cognitive biases, and perceptual features.

Nominal reference is very flexible—the same object may becalled a dalmatian, a dog, or an animal when all are literallytrue. What accounts for the choices that speakers make in howthey refer to objects? The addition of modifiers (e.g. big dog)has been extensively explored in the literature, but fewer stud-ies have explored the choice of noun, including its level of ab-straction. We collected freely produced referring expressionsin a multi-player reference game experiment, where we ma-nipulated the object’s context. We find that utterance choiceis affected by the contextual informativeness of a description,its length and frequency, and the typicality of the object forthat description. Finally, we show how these factors naturallyenter into a formal model of production within the RationalSpeech-Acts framework, and that the resulting model predictsour quantitative production data.

We conceptualize objects based on sensorimotor information gleaned from real-world experience. To what extent is conceptual information structured according to higher-level linguistic features? We investigate whether classifiers, a grammatical category, shape the conceptual representations of objects. In three experiments native Mandarin speakers (a classifier language) and native Dutch speakers (a language without classifiers) judged the similarity of a target object with four objects (presented as words or pictures). One object shared a classifier with the target, the other objects did not. Overall, the target object was judged as more similar to the object with the shared classifier than distractor objects in both Dutch and Mandarin speakers, with no difference between the two languages. Thus, even speakers of a non-classifier language are sensitive to object similarities underlying classifier systems, and using a classifier system does not exaggerate these similarities. This suggests that classifier systems reflect, rather than affect, conceptual structure.

Recent evidence indicates that color categories can exert astrong influence over color matching in both perception andmemory. We explore this phenomenon by analyzing the costfunction for perceptual error. Our analysis is developed withinthe mathematical framework of rate–distortion theory. Ac-cording to our approach, the goal of perception is to minimizethe expected cost of error while subject to a strong constrainton the capacity of perceptual processing. We propose that thecost function in color perception is defined by the sum of twocomponents: a metric cost associated with the magnitude of er-ror in color space, and a cost associated with perceptual errorsthat cross color category boundaries. A computational modelembodying this assumption is shown to produce an excellent fitto empirical data. The results generally suggest that what ap-pear as ‘errors’ in working memory performance may reflectreasonable and systematic behaviors in the context of costs.

Abstract mental representation is fundamental for humancognition. Forming such representations in time, especiallyfrom dynamic and noisy perceptual input, is a challenge forany processing modality, but perhaps none so acutely as forlanguage processing. We show that LISA (Hummel &Holyaok, 1997) and DORA (Doumas, Hummel, & Sandhofer,2008), models built to process and to learn structured (i.e.,symbolic) representations of conceptual properties andrelations from unstructured inputs, show oscillatory activationduring processing that is highly similar to the cortical activityelicited by the linguistic stimuli from Ding et al. (2016). Weargue, as Ding et al. (2016), that this activation reflectsformation of hierarchical linguistic representation, andfurthermore, that the kind of computational mechanisms inLISA/DORA (e.g., temporal binding by systematicasynchrony of firing) may underlie formation of abstractlinguistic representations in the human brain. It may be thisrepurposing that allowed for the generation or emergence ofhierarchical linguistic structure, and therefore, humanlanguage, from extant cognitive and neural systems. Weconclude that models of thinking and reasoning and models oflanguage processing must be integrated—not only forincreased plausiblity, but in order to advance both fieldstowards a larger integrative model of human cognition.

Cognitive models of choice almost universally implicate se-quential evidence accumulation as a fundamental element ofthe mechanism by which preferences are formed. When to stop evidence accumulation is an important question that suchmodels do not currently try to answer. We present the first cog-nitive model that accurately predicts stopping decisions in in-dividual economic decisions-from-experience trials, using anonline learning model. Analysis of stopping decisions acrossthree different datasets reveals three useful predictors of sam-pling duration - relative evidence strength, how long it takesparticipants to see all rewards, and a novel indicator of con-vergence of an underlying learning process, which we call pre-dictive volatility. We quantify the relative strengths of thesefactors in predicting observers’ stopping points, finding thatpredictive volatility consistently dominates relative evidencestrength in stopping decisions.

To account for natural variability in cognitive processing, it is standard practice to optimize the parameters of a model to account for behavioral data. However, variability reflecting the information to which one has been exposed is usually ignored. Nevertheless, most language theories assign a large role to an individual’s experience with language. We present a new way to fit language-based behavioral data that combines simple learning and processing mechanisms using optimization of language materials. We demonstrate that benchmark fits on multiple linguistic tasks can be achieved using this method and will argue that one must account not only for the internal parameters of a model but also the external experience that people receive when theorizing about human behavior.

In this paper, we present a new information-processing modelof math problem solving in which representation change the-ory can be implemented. Specifically, we divided the problemrepresentation process into two. One is to straightforwardlytranslate problem texts into formulas in a conservative exten-sion of Zermelo-Fraenkel’s set theory, and the other is to in-terpret the translated formulas in local mathematical theories.A ZF formula has several interpretations, and representationchange is thus implementable as a choice of an appropriate in-terpretation. Adopting the theory of real closed fields as an ex-ample of local theory and its quantifier elimination algorithmsas an approximate process of searching for solutions, we de-velop a prototype system. We use more than 400 problemsfrom three sources as benchmarks: exercise books, univer-sity entrance examination, and the International MathematicalOlympiad problems. Our experimental results suggest that ourmodel can serve as a basis of a quantitative study on represen-tation change in the sense that the performance of our proto-type system reflects difficulties of the problems quite precisely.

Moral Foundations Theory (MFT) posits that people moralize at least six distinct kinds of virtues. These virtues are divided into “individualizing” and “binding” virtues. Despite widespread enthusiasm for MFT, it is unknown how plausible it is as a model of people’s conceptualizations of the moral domain. In this research, we take a bottom-up approach to characterizing people’s conceptualization of the moral domain, and derive a taxonomy of morality that does not resemble MFT. We find that this model more accurately reflects people’s theories of morality than does MFT.

People exhibit a tendency to generalize a novel noun to thebasic-level of a hierarchical taxonomy – a cognitively salientcategory such as “dog” – with the degree of generalization de-pending on the number and type of exemplars. Recently, achange in the presentation timing of exemplars has also beenshown to have an effect, surprisingly reversing the prior ob-served pattern of basic-level generalization. We explore theprecise mechanisms that could lead to such behavior by ex-tending a computational model of word learning and word gen-eralization to integrate cognitive processes of memory and at-tention. Our results show that the interaction of forgetting andattention to novelty, as well as sensitivity to both type and to-ken frequencies of exemplars, enables the model to replicatethe empirical results from different presentation timings. Ourresults reinforce the need to incorporate general cognitive pro-cesses within word learning models to better understand therange of observed behaviors in vocabulary acquisition.

Spatial metaphors for affective valence are common inEnglish, where up in space=happy/positive and down inspace=sad/negative. Past research suggests that thesemetaphors have some measure of psychological reality:people are faster to respond to valenced words and faceswhen they are presented in metaphor-congruent regions ofspace. Here we explore whether the orientation of a stimulus– rather than its position – is sufficient to elicit such spatial-valence congruency effects, and, if so, which spatial referenceframe(s) people use to represent this orientation. InExperiment 1, participants viewed images of happy and sadprofile faces in different orientations and had to identify theemotion depicted in each face. In Experiment 2, participantscompleted this task while lying down on their sides, therebydisassociating environmental and egocentric reference frames.Experiment 1 revealed a metaphor-congruent interactionbetween emotion and orientation, while Experiment 2revealed that this spatial-valence congruency effect was onlyreliable in the environmental frame of reference.

During language acquisition, children must learn when togeneralize a pattern – applying it broadly and to new words(‘add –ed’ in English) – and when to restrict generalization,storing the pattern only with specific lexical items. One effortto quantify the conditions for generalization, the TolerancePrinciple, has been shown to accurately predict children’sgeneralizations in dozens of corpus-based studies. Thisprinciple hypothesizes that a general rule will be formedwhen it is computationally more efficient than storing lexicalforms individually. It is formalized as: a rule R will generalizeif the number of exceptions does not exceed the number ofwords in the category N divided by the natural log of N(N/lnN). Here we test the principle in an artificial language of9 nonsense nouns. As predicted, children exposed to 5 regularforms and 4 exceptions generalized, applying the regular formto 100% of novel test words. Children exposed to 3 regularforms and 6 exceptions did not extend the rule, even thoughthe token frequency of the regular form was still high in thiscondition. The Tolerance Principle thus appears to capture abasic principle of generalization in rule formation.

We explore how context change and item variation duringnatural category learning influence memory and generalizationto new examples. Participants studied either images of the samebird or varied birds from each of several categories. Theseimages could be presented in a constant background color ordifferent background colors. During test, birds were presentedin only one of the studied background colors. Performance attest depended on the context overlap between study and test,with better performance when there was minimal contextchange during study. Also, contrary to previous findings, wefound that learners generalized better when items wererepeated during study and remembered old items better whenitems were varied during study. When there is a moderatedegree of context change, there is no benefit of repetition orvariation for either novel or old items. These results indicatethat context change and item variation have complementaryeffects on learning.

An important objective in higher-order cognition research isto understand how relational categories are acquired andapplied. Much of the research on relational category learninghas investigated the role of within-category comparisonopportunities in category acquisition and transfer – guided bypredictions from structure mapping theory that alignmentleads to highlighting and abstraction of shared relationalstructure (Gentner, 1983). Recent research has yielded awithin-category comparison advantage under the supervisedobservational learning mode (relative to twice as many single-item trials), but not under the supervised classification mode(Patterson & Kurtz, 2015). In the present study we investigatethe role that pressure to succeed at the training task – a criticaldifference between the two learning modes – plays in theapparent ineffectiveness of learning by comparison within theclassification mode. In a 2x2 between-subjects design wecrossed two levels of performance pressure (elevated andstandard) with two presentation formats (single-item andwithin-category pairs). The main findings are: (1) asignificant interaction showing a negative impact of increasedperformance pressure for single-item learners, but not forcomparison learners; and (2) a theoretically predicted, butempirically elusive effect of comparison over single-item inthe classification mode. We conclude that: (1) performancepressure exerts a deleterious effect on relational categorylearning (in accord with findings in the attribute categoryliterature) that opportunities to compare may compensate for;and (2) pressure to perform does not appear to underlielackluster comparison + classification performance (relativeto observational learning). Further, we offer new evidence onthe role that within-category comparison plays in relationalcategory learning.

Rational agents should update their beliefs in the light of newevidence. Equally, changes in belief should depend only onthe quality of the evidence, and not on factors such as the or-der in which the evidence is acquired, or whether intermedi-ate judgements are requested during evidence acquisition. Incontrast we show that requests for intermediate judgments caninhibit belief updating for real decision makers, which repre-sents a new type of decision making fallacy. This behaviour isparadoxical from the point of view of classical Bayesian mod-els, but we show that it is consistent with an a priori, parameterfree prediction of a cognitive model based on quantum theory.

In this paper we present a model of how events and their partic-ipants are represented in working memory (WM). The model’scentral assumption is that events are experienced through se-quentially structured sensorimotor (SM) routines—as are theindividuals that participate in them. In the light of this assump-tion, we propose that events and individuals are stored in WMas prepared SM routines. This proposal allows a new mech-anism for binding representations of individuals to semanticroles such as AGENT and PATIENT. It also enables a novelaccount of how expectations about forthcoming events can in-fluence SM processing in real time as events are perceived.Finally, it supports an account of the interface between WMrepresentations and language.

Speech perception is made much harder by variability betweentalkers. As a result, listeners need to adapt to each differenttalker’s particular acoustic cue distributions. Thinking of thisadaptation as a form of statistical inference, we explore the rolethat listeners’ prior expectations play in adapting to an unfa-miliar talker. Specifically, we test the hypothesis that listenerswill have a harder time adapting to talkers whose cue distribu-tions fall outside the range of normal variation across talkers.We also show that it is possible to infer listeners’ shared priorexpectations based on patterns of adaptation to different cuedistributions. This provides a potentially powerful tool for di-rectly probing listeners’ prior expectations about talkers thatdoes not rely on speech produced by many different talkers,which is costly to collect and annotate, and only indirectly re-lated to listeners’ subjective expectations.

Decision-making under uncertainty is pervasive. This worksought to understand the role of working memory (WM) in losssensitivity by utilizing two widely used tasks, the IowaGambling Task (IGT) and the Soochow Gambling Task (SGT),and manipulating WM with a dual-task paradigm. Wehypothesized that WM load would reduce attention to both lossvalue and frequency in the decision-making tasks. To betterdelineate the psychological processes underpinning choicebehavior, we developed an Expectancy-Frequency-Perseveration (EFP) model which parsimoniously capturesthree critical factors driving choices: expected value,frequency of gains and losses, and perseveration. Behavioraland computational modeling results indicate that WM loadcompromised performance in the IGT due to reduced attentionto loss value but enhanced performance in the SGT because ofdiminished attention to loss frequency. Our findings suggestthat WM heightens attention to losses, but that greater attentionis given to loss frequency than loss value.

Deep neural networks have become increasingly successful atsolving classic perception problems such as object recognition,semantic segmentation, and scene understanding, often reach-ing or surpassing human-level accuracy. This success is duein part to the ability of DNNs to learn useful representationsof high-dimensional inputs, a problem that humans must alsosolve. We examine the relationship between the representa-tions learned by these networks and human psychological rep-resentations recovered from similarity judgments. We find thatdeep features learned in service of object classification accountfor a significant amount of the variance in human similarityjudgments for a set of animal images. However, these fea-tures do not capture some qualitative distinctions that are a keypart of human representations. To remedy this, we develop amethod for adapting deep features to align with human sim-ilarity judgments, resulting in image representations that canpotentially be used to extend the scope of psychological exper-iments.

In this study, we manipulated gain/loss framing context duringa simulated negotiation between a human user and a virtualagent. Task instructions placed users either in a loss or gainframed context, such that those in the loss frame had tominimize expenses whereas those in the gain frame had tomaximize profits. The virtual agent displayed facial emotionsso that we could also test how interpersonal emotions interactwith framing. Results suggest that individuals are moremotivated to minimize their losses than maximizing their gains.The loss frame caused individuals to demand more during thenegotiation, hence to minimize expenses. Neurophysiologicalresults suggest that cardiovascular patterns of challenge (i.e.,positive motivations) were present in the loss frame condition,most strongly when the virtual human smiled. We discuss theseresults in regards to Prospect Theory. This work also hasimplications for designing and rigorously evaluating human-like virtual agents.

Theoretical and empirical work in the field of classificationlearning is centered on a ‘reference point’ view, where learn-ers are thought to represent categories in terms of stored pointsin psychological space (e.g., prototypes, exemplars, clusters).Reference point representations fully specify how regions ofpsychological space are associated with class labels, but theydo not contain information about how features relate to oneanother (within- class or otherwise). We present a novel exper-iment suggesting human learners acquire knowledge of within-class feature correlations and use this knowledge during gen-eralization. Our methods conform strictly to the traditional ar-tificial classification learning paradigm, and our results can-not be explained by any prominent reference point model (i.e.,GCM, ALCOVE). An alternative to the reference point frame-work (DIVA) provides a strong account of the observed perfor-mance. We additionally describe preliminary work on a noveldiscriminative clustering model that also explains our results.

Deictic time words (e.g., “tomorrow,” “yesterday”) refer totime periods relative to the present moment. While childrenproduce these words by age 2-3, they use them incorrectly forseveral more years. Here, as a case study in abstract wordlearning, we explored what children know about these wordsduring this delay. Specifically, we probed children’sknowledge of three aspects of meaning: deictic (past/future)status, sequential ordering (e.g., “tomorrow” is after“yesterday”), and remoteness from now. We asked 3- to 8-year-olds to place these words on a timeline extending fromthe past (left) to the future (right). Even 4-year-olds couldmeaningfully represent the words’ deictic status and order,and by 6, the majority displayed adult-like performance.Adult-like knowledge of remoteness, however, emergedindependently, after age 7. Thus, even while children usethese terms incorrectly, they are gradually constructing astructured semantic domain, including information about thedeictic, sequential, and metric relations among terms.

People tend to look at uninformative, blank locations inspace when retrieving information. This gaze behaviour,known as looking at nothing, is assumed to be driven by theuse of spatial indices associated with external information.We investigated whether people form spatial indices andlook at nothing when retrieving words from memory.Participants were simultaneously presented four words.During retrieval participants looked at the relevant, blanklocation, where the probe word had appeared previously,longer than the other blank locations. Additionally, wordpresentation was sometimes followed by a visual cue eitherco-located or not with the probe word. Valid cues functionedas visual reinforcement while invalid cues causedinterference. Finally, participants with better visuospatialmemory looked less at the relevant, blank location,suggesting a dynamic relationship between so-called“external” and “internal” memory. Overall findings suggestan automatic, instantaneous spatial indexing mechanism forwords and a dynamic looking at nothing behaviour.

Kehler and Rohde (2013) proposed a Bayesian theory of pro-noun interpretation where the influence of world knowledgeemerges as effects on the prior and the influence of informationstructure as effects on the likelihood: P(referent|pronoun) μP(pronoun|referent)P(referent). Here we present two experi-ments on Mandarin Chinese that allow us to test the generalityof the theory for a language with different syntactic-semanticassociations than English. Manipulations involving two dif-ferent classes of implicit-causality verbs and passive vs. activevoice confirmed key predictions of the Bayesian theory: effectsof these manipulations on the prior and likelihood in produc-tion were consistently reflected in pronoun interpretation pref-erences. Quantitative analysis shows that the Bayesian modelis the best fit for Mandarin compared to two competing anal-yses. These results lend both qualitative and quantitative sup-port to a cross linguistically general Bayesian theory of pro-noun interpretation.

Talk of levels is everywhere in cognitive science. Whether it isin terms of adjudicating longstanding debates or motivatingfoundational concepts, one cannot go far without hearingabout the need to talk at different ‘levels’. Yet in spite of itswidespread application and use, the concept of levels hasreceived little sustained attention within cognitive science.This paper provides an analysis of the various ways the notionof levels has been deployed within cognitive science. Thepaper begins by introducing and motivating discussion viafour representative accounts of levels. It then turns to outliningand relating the four accounts using two dimensions ofcomparison. The result is the creation of a conceptualframework that maps the logical space of levels talk, whichoffers an important step toward making sense of levels talkwithin cognitive science

Game Theory -based decision aids have been successfully em-ployed in real-world policing, anti-terrorism, and wildlife con-servation efforts (Tambe, Jiang, An, & Jain, 2013). Cognitivemodeling, in concert with model tracing and dynamic parame-ter fitting techniques, may be used to improve the performanceof such decision aids by predicting individual attacker behav-ior in repeated security games. We present three simulations,showing that (1) cognitive modeling can aid in greatly improv-ing decision-aid performance in the security domain; and (2)despite the fact that individual attackers will differ in initialpreferences and in how they learn, model parameters can beadjusted dynamically to make useful predictions for each at-tacker.

Individuals with Autism Spectrum Disorder (ASD) often struggle with making inductive generalizations. Yet for typically developing children, the capacity to make such generalizations is a hallmark of human learning. This ability requires some understanding of “intuitive statistics” (i.e., the understanding that there is a relationship between samples and populations), which have been previously demonstrated to emerge early on in infancy. We hypothesized that the challenges with inductive generalization among the ASD population may have its roots in weaknesses in probabilistic reasoning. In the current study, we gave children with ASD a probability prediction task adapted from the method used with infants in Teglas et al. (2007), and our results over two experiments with two groups (one from the U.S. and one from Singapore) suggest that compared with typically developing children, children with autism may have difficulties in engaging in probabilistic reason

We test the robustness of American college students’ mentaltimeline to dual tasks that have interfered with spatial andverbal reasoning in prior work. We focus on the left-right axisfor representing sequences of events. We test Americancollege students, who read from left to right. We test forautomatic space-time mappings using two established space-time association tasks. We find that their tendency toassociate earlier events with the left side of space and laterevents with the right remains under conditions of visuospatialand verbal interference. We find this both when participantsmade time judgments about linguistic and non-linguisticstimuli. We discuss the relationship between these results andthose obtained for mental timelines that result from learningnew metaphors in language (Hendricks & Boroditsky, 2015),and the effects of the same interference tasks on number tasks(mental number-line and counting; van Dijck et al., 2009;Frank et al., 2012).

Scalar implicatures—inferences from a weak description (“Iate some of the cookies”) that a stronger alternative is true(“I didn’t eat all”)—are paradigm cases of pragmatic infer-ence. Children’s trouble with scalar implicatures is thus animportant puzzle for theories of pragmatic development, giventheir communicative competence in other domains. Previousresearch has suggested that access to alternatives might be key.Here, we explore children’s reaction times in a new paradigmfor measuring scalar implicature processing. Alongside fail-ures on scalar implicatures with “some,” we replicate previ-ous reports of failures with “none,” and find evidence of aspeed-accuracy trade-off for both quantifiers. Motivated bythese findings, we explore the relationship between accuracyand reaction time with a Drift Diffusion Model. We find evi-dence consistent with the hypothesis that preschoolers lack ac-cess to the alternatives for scalar implicature computation, al-though this set of alternatives may be broader than previouslyassumed.

The spatial prepositions in and on apply to a wide range ofcontainment and support relations, making exhaustivedefinitions difficult. Theories differ in whether they endorsegeometric or functional properties and how these properties arerelated to meaning and use. This study directly examines theroles of geometric and functional information in adults’ andchildren’s use of in and on by developing a large sample ofrelations situated within a small gradable geometric andfunctional feature space. We propose that variation in featuresacross items is systematically related to the use of in and onand demonstrate that feature-language relationships changeacross development: adults’ expression use is sensitive to bothgeometric and functional features, while children’s use variesonly according to geometric features.

With the ever increasing amount of information available, the ability to prioritize the most relevant items for full processing is increasingly necessary to maintain expertise in a domain. As a result, accurate triage decisions--initial decisions about the relevance of a given article, book, or talk in order to determine whether to pursue that information further--are very important. In the present paper, we present a model of triage decision making that includes both an information search component to determine reading strategy and a decision making component to make the final decision. We apply the model to human relevance ratings as well as binary decisions of relevance for a set of emails.

Age-related declines in scores on neuropsychological tests arewidely believed to reveal that human cognitive capacitiesdecline across the lifespan. In a computational simulation, weshow how the behavioral patterns observed in PairedAssociate Learning (PAL), a particularly sensitive measure ofage-related performance change (Rabbitt & Lowe, 2000), arepredicted by the models used to formalize associative learningprocesses in other areas of behavioral and neuroscientificresearch. The simulation further predicts that manipulatinglanguage exposure will reproduce the experience-relatedperformance differences erroneously attributed to age-relateddecline in age-matched adults. Consistent with this, olderbilinguals outperformed native speakers in a German PALtest, an advantage that increased with age. These analyses andresults show that age-related PAL performance changesreflect the predictable effects of learning on the associabilityof test items, and indicate that failing to control for theseeffects is distorting our understanding of cognitive and braindevelopment in adulthood.

Throughout our lives, we are faced with a variety of causalreasoning problems. Arguably, the most successful models ofcausal reasoning, Causal Graphical Models (CGMs), performwell in some situations, but there is considerable variation inhow well they are able to account for data, both across scenar-ios and between individuals. We propose a model of causalreasoning based on quantum probability (QP) theory that ac-counts for behavior in situations where CGMs fail. WhetherQP or classical models are appropriate depends on the repre-sentation of events constructed by the reasoner. We describean experiment that suggests the representation of events canchange with experience to become more classical, and that therepresentation constructed can vary between individuals, in away that correlates with a simple measure of cognitive ability,The Cognitive Reflection Task.

A standard view in cognitive psychology is that people esti-mate probabilities using heuristics that do not follow proba-bility theory. We describe a model of probability estimationwhere people do follow probability theory in estimation, butare subject to random error or noise. This model predicts thatpeople’s conditional probability estimates will agree closelywith probability theory for certain noise-cancelling expres-sions, but deviate from probability theory for other expres-sions. We describe an experiment which strongly confirmsthese predictions, suggesting that people estimate conditionalprobabilities in a way that follows standard probability theory,but is subject to the biasing effects of random noise.

Much of what we learn comes from a mix of information thatwe select (active) and information that we receive (passive).But which type of training is better for different kinds of learn-ing problems? Here, we explore this question by comparingdifferent sequences of active/passive training in an abstractconcept learning task. First, we replicate the active learningadvantage from Markant & Gureckis (2014) (Experiments 1aand 1b). Then, we provide a test of whether experiencing ac-tive learning first or passive learning first improves the effec-tiveness of concept learning (Experiment 2). Across both ex-periments, active training led to better learning of the targetconcept, but “passive-first” learners were more accurate than“active-first” learners and more efficient than “active-only”learners. These findings broaden our understanding of whendifferent sequences of active/passive learning are more effec-tive, suggesting that for certain problems active explorationcan be enhanced with prior passive experience.

Recent research has shown that when people multitask, boththe subtask structure and the temporal constraints of thecomponent tasks strongly influence people’s task-switchingbehavior. In this paper, we propose an integrated theoreticalaccount and associated computational model that aims toquantify how people balance structural and temporalconstraints in everyday multitasking. We validate the theoryusing data from an empirical study in which drivers performeda visual-search task while navigating a driving environment.Through examination of illustrative protocols from the modeland human drivers as well as the overall fit on the aggregateglance data, we explore the implications of the theory andmodel for time-critical multitasking domains.

How does the number of features impact category learning?One view suggests that additional features creates a “curse ofdimensionality” - where having more features causes the sizeof the search space to grow so quickly that discovering goodclassification rules becomes increasingly challenging. The op-posing view suggests that additional features provide a wealthof additional information which learners should be able to useto improve their classification performance. Previous researchexploring this issue appears to have produced conflicting re-sults: some find that learning improves with additional features(Hoffman & Murphy, 2006) while others find that it does not(Minda & Smith, 2001; Edgell et al., 1996). Here we inves-tigate the possibility that category structure may explain thisapparent discrepancy – that more features are useful in cate-gories with family resemblance structure, but are not (and mayeven be harmful) in more rule-based categories. We find whilethe impact of having many features does indeed depend on cat-egory structure, the results can be explained by a single unifiedmodel: one that attends to a single feature on any given trialand uses information learned from that particular feature tomake classification judgments.

This study explores developmental changes in the ability toask informative questions. We hypothesized an intrinsic linkbetween the ability to update beliefs in light of evidence andthe ability to ask informative questions. Four- to ten-year-oldchildren played an iPad game asking them to identify a hiddenbug. Learners could either ask about individual bugs, or makea series of feature queries (e.g., “Does the hidden bug haveantenna?”) that could more efficiently narrow the hypothesisspace. Critically the task display either helped children inte-grate evidence with the hypothesis space or required them toperform this operation themselves. Although we found thathelping children update their beliefs improved some aspects oftheir active inquiry behavior, children required to update theirown beliefs asked questions that were more context-sensitiveand thus informative. The results show how making a taskmore difficult may actually improve children’s active inquiryskills, thus illustrating a type of desirable difficulty.

The relationship between working memory capacity andwriting ability was examined via a linguistic analysis ofstudent essays. Undergraduate students (n = 108) wrotetimed, prompt-based essays and completed a battery ofcognitive assessments. The surface- and discourse-levellinguistic features of students’ essays were then analyzedusing natural language processing tools. The results indicatedthat WM capacity was related to surface-level, but notdiscourse-level features of student essays. Additionally, theresults suggest that these relationships were attenuated forstudents with high inferencing skills, as opposed to those withlower inferencing skills

Recent work has proposed that prominence perception inspeech could be driven by predictability of prosodic patterns,connecting prominence perception to the concept of statisticallearning. In the present study, we tested the predictabilityhypothesis by conducting a listening test where subjects werefirst exposed to a 5-minute stream of sentences with a certainproportion of sentence-final words having either a falling orrising pitch trajectory. After the exposure stage, subjects wereasked to grade prominence in a set of novel sentences withsimilar pitch patterns. The results show that the subjects weresignificantly more likely to perceive words with low-probability pitch trajectories as prominent independently ofthe direction of the pitch change. This suggests that evenshort exposure to prosodic patterns with a certain statisticalstructure can induce changes in prominence perception,supporting the connection between prominence perceptionand attentional orientation towards low-probability events inan otherwise predictable context.

We are very familiar with certain objects; we can quickly rec-ognize our cars, friends and collaborators despite heavy occlu-sion, unusual lighting, or extreme viewing angles. We can alsodetermine if two very different views of a stranger are indeedof the same person. How can we recognize familiar objectsquickly, while performing deliberate, perceptual inference onunfamiliar objects? We describe a model combining an iden-tity classification network for familiar faces with an analysis bysynthesis approach for unfamiliar faces to make rich inferencesabout any observed face. We additionally develop an onlinenon-parametric clustering algorithm for recognition of repeat-edly experienced unfamiliar faces, and show how new facescan become familiar by being consolidated into the identityrecognition network. Finally, we show that this model predictshuman behavior in viewpoint generalization and identity clus-tering tasks, and predicts processing time differences betweenfamiliar and unfamiliar faces.

Adaptive fact learning systems have been developed to makeoptimal use of testing and spacing effects by taking intoaccount individual differences in learning efficiency.Measures derived from these systems, capturing theindividual differences, predict later performance in similarand different fact learning tasks. Additionally, there is a richbody of literature showing that individual differences ingeneral cognitive ability or working memory capacity canpredict scores on achievement tests. If these measures alsoinfluence fact learning, incorporating them might furtherenhance adaptive systems. However, here we provideevidence that performance during fact learning is neitherrelated to working memory capacity nor general cognitiveability. This means that the individual differences captured byour adaptive learning system encapsulate characteristics oflearners that are independent of their general cognitive ability.Consequently, adaptive learning methods should focusprimarily on memory-related processes.

Recent studies report a striking decline in children’s ability to notice same-different relations around age 3 (Walker et al., 2015). We propose that such a decline results from an object focus related to children’s avid noun-learning. To test this, we examine children’s performance on a classic relational task – the relational match-to-sample task (RMTS). Prior work has shown that 4-year-olds can pass this task (Christie & Gentner, 2014). However, if nominal language induces an object focus, their performance should be disrupted by a noun-labeling pretask. In two experiments, 4-year-olds either labeled objects or actions in a naming pretask. Then they completed the RMTS task. Consistent with the noun-focus explanation, the object-naming group failed the RMTS task, whereas the action-naming group and a control group both succeeded. This suggests that nominal language can lead to an object focus, and that this could explain the temporary decline in children’s relational processing.

Although syntactic priming is well studied and commonly assumed to involve implicit learning, the mechanisms behind this phenomenon are still under debate. We tested whether implicit learning of adjacent and non-adjacent sequences occurs in a non-linguistic, finger sequence task (Serial Reaction Time task), and if so, whether these implicitly-learned dependencies can cause syntactic priming in the linguistic domain. We followed the logic that exposure to statistical patterns in the SRT task may influence language users’ relative clause (RC) attachment biases, and trained participants on SRT sequences with adjacent or non-adjacent dependencies. Participants then wrote completions to relative clause fragments in a situation where they could opt for adjacent or non-adjacent linguistic structures. Participants successfully learned the adjacent and non-adjacent dependency implicitly during the SRT task, but, strikingly, their RC continuations did not exhibit priming effects. Implications for theories of syntactic priming and its relations to implicit learning are discussed.

The visual word form area (VWFA) is a region of the cortex lo-cated in the left fusiform gyrus, that appears to be a waystationin the reading pathway. The discovery of the VWFA occurredin the late twentieth century with the advancement of func-tional magnetic resonance imaging (fMRI). Since then, therehas been an increasing number of neuroimaging studies to un-derstand the VWFA, and there are disagreements as to its prop-erties. One such disagreement is regarding whether or not theVWFA is more selective for real words over pseudowords1. Arecent study using fMRI adaptation (Glezer, et al., 2009) pro-vided evidence that neurons in the VWFA are selectively tunedto real words. This contradicts the hypothesis that the VWFAis tuned to the sublexical structure of visual words, and there-fore has no preference for real words over pseudowords. Inthis paper, we develop a realistic model of the VWFA by train-ing a deep convolutional neural network to map printed wordsto their labels. The network is able to achieve an accuracy of98.5% on the test set. We then analyze this network to see ifit can account for the data Glezer et al. found for words andpseudowords, and find that it does.

When tested appropriately, infants appear to demonstrate false-belief understanding in the first year of life. Some have argued that this is inconsistent with the well-established relationship between social experience and preschoolers’ false-belief performance. We argue that these two sets of findings are not inconsistent because the ability to attribute false beliefs to others is necessary but not sufficient for false- belief performance, and we propose several ways that one social factor, hearing and using mental-state language, might relate to false-belief performance throughout the lifespan. We tested this account by examining the relationship between adults’ use of mental-state language and their false-belief understanding. Participants’ use of mental-state language was related to how quickly they could accurately predict the behavior of agents on the basis of desires and beliefs. These findings provide the first evidence that mental-state talk and false-belief performance are related into adulthood.

How can heuristic strategies emerge from smaller build-ing blocks? We propose Approximate Bayesian Com-putation (ABC) as a computational solution to thisproblem. As a first proof of concept, we demonstratehow a heuristic decision strategy such as Take The Best(TTB) can be learned from smaller, probabilisticallyupdated building blocks. Based on a self-reinforcingsampling scheme, different building blocks are com-bined and, over time, tree-like non-compensatory heuris-tics emerge. This new algorithm, coined ApproximatelyBayesian Computed Take The Best (ABC-TTB), is ableto recover data that was generated by TTB, leads tosensible inferences about cue importance and cue direc-tions, can outperform traditional TTB, and allows totrade-off performance and computational effort explic-itly.

A crucial component of event recognition is understanding the roles that people and objects take: did the boy hit the girl, or did the girl hit the boy? We often make these categorizations from visual input, but even when our attention is otherwise occupied, do we automatically analyze the world in terms of event structure? In two experiments, participants made speeded gender judgments for a continuous sequence of male-female interaction scenes. Even though gender was orthogonal to event roles (whether the Agent was male or Female, or vice- versa), a switching cost was observed when the target character’s role reversed from trial to trial, regardless of whether the actors, events, or side of the target character differed. Crucially, this effect held even when nothing in the task required attention to the relationship between actors. Our results suggest that extraction of event structure in visual scenes is a rapid and automatic process.

In the past few years, deep convolutional neural networks(CNNs) trained on large image data sets have shown impres-sive visual object recognition performances. Consequently,these models have attracted the attention of the cognitive sci-ence community. Recent studies comparing CNNs with neuraldata from cortical area IT suggest that CNNs may—in addi-tion to providing good engineering solutions—provide goodmodels of biological visual systems. Here, we report evidencethat CNNs are, in fact, not good models of human visual per-ception. We show that a 3D shape inference model explainshuman performance on an object shape similarity task betterthan CNNs. We argue that deep neural networks trained onlarge amounts of image data to maximize object recognitionperformance do not provide adequate models of human vision.

Human error while performing well-learned tasks on a com-puter is an infrequent, but pervasive problem. Such errors areoften attributed to memory deficits, such as loss of activation orinterference with other tasks (Altmann & Trafton, 2002). Weare arguing that this view neglects the role of the environment.As embodied beings, humans make extensive use of externalcues during the planning and execution of tasks. In this paper,we study how the visual interaction with a computer interfaceis linked to user errors. Gaze recordings confirm our hypoth-esis that the use of the environment increases when memorybecomes weak. An existing cognitive model of sequential ac-tion and procedural error (Halbrügge, Quade, & Engelbrecht,2015) is extended to account for the observed gaze behavior.

A novel connectionist model of sentence production is pre-sented, which employs rich situation model representationsoriginally proposed for modeling systematicity in comprehen-sion (Frank, Haselager, & van Rooij, 2009). The high overallperformance of our model demonstrates that such represen-tations are not only suitable for comprehension, but also formodeling language production. Further, the model is able toproduce novel encodings (active vs. passive) for a particularsemantics, as well as generate such encodings for previouslyunseen situations, thus demonstrating both syntactic and se-mantic systematicity. Our results provide yet further evidencethat such connectionist approaches can achieve systematicity,in production as well as comprehension.

Grammatical dependencies often involve elements that are not adjacent. However, most experiments in which non-adjacent dependencies are learned bracketed the dependent material with pauses, which is not how dependencies appear in natural language. Here we report successful learning of embedded NAD without pause bracketing. Instead, we induce learners to compute structure in an artificial language by entraining them through processing English sentences. We also found that learning becomes difficult when grammatical entrainment causes learners to compute boundaries that are misaligned with NAD structures. In sum, we demonstrated that grammatical entrainment can induce boundaries that can carry over to reveal structures in novel language materials, and this effect can be used to induce learning of non-adjacent dependencies.

In this paper, we bring together research on active learningand intuitive physics to explore how people learn about“microworlds” with continuous spatiotemporal dynamics.Participants interacted with objects in simple two-dimensionalworlds governed by a physics simulator, with the goal ofidentifying latent physical properties such as mass, and forcesof attraction or repulsion. We find an advantage for activelearners over passive and yoked controls. Active participantsspontaneously performed several kinds of “natural exper-iments” which reveal the objects’ properties with varyingsuccess. While yoked participants’ judgments were affectedby the quality of the active participant they observed, they didnot share the learning advantage, performing no better thanpassive controls overall. We discuss possible explanations forthe divergence between active and yoked learners, and outlinefurther steps to categorize and explore active learning in thewild.

Attention is thought to be a part of a larger cluster of mecha-nisms that serve to orient a cognitive system, to filter contentswith respect to their task relevance, and to devote more com-putation to certain options than to others. All these activitiesproceed under the plausible assumption that not all informationcan be or ought to be processed for a system to satisfice in anever changing world. In this paper, we describe an attention-centric cognitive system called ARCADIA that demonstratesthe orienting, filtering, and resource-skewing functions men-tioned above. The demonstration involves maintaining focuson cognitive tasks in a dynamic environment. While ARCA-DIA carries out a task, limits on its attentional capacity resultin “inattentional blindness” under circumstances analogous tothose where people fail to perceive otherwise salient stimuli.

The literature on first-order false-belief is extensive, but less isknown about the second-order case. The attainment of second-order false-belief mastery seems to mark a cognitively signifi-cant stage, but what is its status? Is it an example of complex-ity only development, or does it indicate that a more funda-mental conceptual change has taken place? In this paper weextend Bra ̈uner’s hybrid-logical analysis of first-order false-belief tasks (Bra ̈uner, 2014, 2015) to the second-order case,and argue that our analysis supports a version of the concep-tual change position.

Does syntactic information affect the production of bare nouns?Research into this issue has explored word-specific features (e.g.,gender). However, word-independent syntactic distributions mayalso play a role. For example, studies of word recognition haveuncovered strong effects of the diversity of a word's syntacticdistribution – its syntactic flexibility – on response times in thelexical decision paradigm. By contrast, studies of sentenceproduction have produced strong but conflicted effects of syntacticflexibility. We propose that syntactic flexibility also affectsproduction of individual words. We reanalyze a database ofpreviously collected timed picture naming data using two novelmeasures of syntactic flexibility, one based on the relationsstemming from the noun, and one based on the relations extendingto the noun. Our results show that nouns that project a diversearray of structures are produced faster, and those that are integratedinto a diverse array of structures are produced slower.

The spread of cultural variants, such as dress or speech pat-terns, may be promoted or inhibited by different types of bias.In model-based bias, variants are differentially adopted accord-ing to characteristics of individuals exhibiting them. A surpris-ing case of cross-group adoption comes from sociolinguisticfieldwork in which White speakers were observed exhibiting afeature of African-American Vernacular English, in spite of ex-pressing aggressively negative attitudes towards their African-American neighbors. A likely explanation for this is that thefeature in question had become dissociated for these speakersfrom the inalienable trait Blackness, but had retained associa-tions with the more alienable trait of being “street” or tough.We tested this by conducting an artificial-language experimentin which groups of four participants played a computer gamethat involved typing instant messages to each other, tradingresources, and fighting. Participants were assigned to one oftwo mutually antagonistic “alien species” (weaker Wiwos andtougher Burls) and learned an alien language with two species-specific dialects. In one condition, the Wiwos were told thatthat Burl dialect was mainly used by Burls; in the other con-dition they were told it was mainly used by “tougher aliens”.Burl variants were significantly more likely to be used by Wi-wos in the latter condition than in the former, even though theywere associated with tougher aliens in both conditions. Thissuggests that cultural variants linked to more alienable traitsare more likely to be adopted than those linked to inalienableones, even if the practical implications of the two traits are verysimilar.

We apply Bayesian data analysis to a structured cognitivemodel in order to determine the priors that support humangeneralizations in a simple concept learning task. We mod-eled 250,000 ratings in a “number game” experiment wheresubjects took examples of a numbers produced by a program(e.g. 4, 16, 32) and rated how likely other numbers (e.g. 8vs. 9) would be to be generated. This paper develops a dataanalysis technique for a family of compositional “Language ofThought” (LOT) models which permits discovery of subjects’prior probability of mental operations (e.g. addition, multi-plication, etc.) in this domain. Our results reveal high cor-relations between model mean predictions and subject gener-alizations, but with some qualitative mismatch for a stronglycompositional prior.

Recent research shows that Chinese, when they gesture about time, tend to put the past “ahead” and the future “behind”. Do they think of time in the way as suggested by their gestures? In this study we investigate whether Chinese people explicitly have such past-in-front mappings. In experiment 1 we show that when time conceptions are constructed with neutral wording (without spatial metaphors), Chinese people are more likely to have a past- in-front-mapping than Spaniards. This could be due to cultural differences in temporal focus of attention, in that Chinese people are more past-oriented than Europeans. However, additional experiments (2 & 3) show that, independent of culture, Chinese people’s past-in-front mapping is sensitive to the wording of sagittal spatial metaphors. In comparison to a neutral condition, they have more past-in-front mappings when time conceptions are constructed with past-in-front spatial metaphors (“front day”, means the day before yesterday), whereas fewer past- in-front mappings are constructed with future-in-front metaphors. There thus appear to be both long-term effects of cultural attitudes on the spatialization of time, and also immediate effects of the space-time metaphors used to probe people’s mental representations.

The emergence of language around a child’s first birthday isone of the greatest transformations in human development.Does this transition require a fundamental shift in the child’sknowledge or beliefs, or could it instead be attributable to moregradual changes in processing abilities? We present a simplemodel of cognitive performance that supports the second con-clusion. The premise of this model is that any cognitive op-eration requires multiple steps, each of which require sometime to complete and have some probability of failure. Weuse meta-analysis to estimate these parameters for two com-ponents of simple ostensive word learning: social cue use andword recognition. When combined in our model, these esti-mates suggest that learning should be very difficult for chil-dren younger than around a year, especially with gaze alone.This model takes a first step towards quantifying performancelimitations for cognitive development and may be broadly ap-plicable to other developmental changes.

Humans spatialize time. This occurs within individual mindsand also in larger, shared cultural systems like language.Understanding the origins of space-time mappings requiresanalyses at multiple levels, from initial individual biases tocultural evolution. Here we present a laboratory experimentthat simulates the cultural emergence of space-timemappings. Dyads had to communicate about temporalconcepts using only a novel, spatial signaling device. Overthe course of their interactions, participants rapidlyestablished semiotic systems that mapped systematicallybetween time and space. These semiotic systems exhibited anumber of similarities, but also striking idiosyncrasies. Byforegrounding the interaction of mechanisms that operate ondisparate timescales, laboratory experiments can shed light onthe commonalities and variety found in space-time mappingsin languages around the world.

The main question we try to answer in this paper is whetherstage-like progression in cognitive development can beexplained by transfer of cognitive skill among tasks. Wefocus on the following question: To what extent does trainingon one task improve the performance on another task? Thetasks are Piaget’s (1959) Balance Scale Task and NumberConservation Task, and a task that we will call the Une-Sentence Task, which is taken from Karmiloff-Smith's (1979)experiment on the acquisition of determiners in French. Were-implemented already existing models within theframework of the PRIMs cognitive architecture (Taatgen,2013). Each task was subdivided in certain stages related tothe complexity of the problem-solving strategies. We showthat mastery of a certain stage of a problem becomes easier ifa higher stage of another task is mastered first.

We examined whether the classical framing effect observedwith the Asian Disease problem could be reversed when peoplemake decisions from experience. Ninety-five universitystudents were randomly allocated to one of three conditions:Description, Sampling (where the participants were allowed tosample through the outcomes presented as a pack of cards) andInteractive (where the participants were invited to spread outall possible outcomes in a sample) and made three gain-framedchoices and three loss-framed choices, with two filler tasksafter the first three choices. The results revealed a significantinteraction effect between framing and choice condition. In theDescription choice condition, participants were more risk-seeking with loss-framed problems. This pattern was reversedin the Sampling choice condition where participants were morerisk-seeking with gain frames. Finally, the Interactive choicecondition resulted in a classic pattern of framing effect,whereby people were more risk averse in the domain of gains.

Not much is known about event apprehension, the earliest stage of information processing in elicited language production studies, using pictorial stimuli. A reason for our lack of knowledge on this process is that apprehension happens very rapidly (<350 ms after stimulus onset, Griffin & Bock 2000), making it difficult to measure the process directly. To broaden our understanding of apprehension, we analyzed landing positions and onset latencies of first fixations on visual stimuli (pictures of real-world events) given short stimulus presentation times, presupposing that the first fixation directly results from information processing during apprehension.

Possibility and probability expressions, like possibly or prob-ably, are frequently assumed to communicate that the proba-bility of a proposition is above a certain threshold. Most pre-vious empirical research on these expressions has focused oncases of known objective chance: if the true objective proba-bility is given, would a speaker use possibly, probably or oneof their kin? Here, we investigate the use of probability expres-sions when speakers have subjective uncertainty about objec-tive chance, i.e., higher-order uncertainty. Experimental datasuggest that speakers’ choices of a probability expression is acomplex function of their state of higher-order uncertainty. Weformulate a computational probabilistic model of pragmaticspeaker behavior that explains the experimental data.

We model the causal links between child language (CL) andchild-directed language (CDL). We take pairs of sequences oflinguistic measurements from a longitudinal study. Each child-mother pair of sequences is considered as an instance of thetrajectory of a high-dimensional dynamical system. We thenuse Multispatial Convergent Cross Mapping to ascertain thedirections of causality between the pairs of sequences, that is,whether the complexity of CL drives that of CDL, the com-plexity of CDL drives that of CL, both, or neither. We find thatchildren are responsive to the amount of speech and the diver-sity of words produced by their mothers, but not vice-versa.However, the syntactic diversities of the children’s utterancesdrive the syntactic diversity of the mothers’ utterances. This isevidence for fine-grained fine-tuning of CDL in response onlyto the syntax of CL.

Recent research has highlighted the operation of a ratioprocessing system that represents the analog magnitudes ofnonsymbolic ratios. This study investigated whether suchrepresentations would demonstrate spatial associationsparallel to the SNARC (spatial numeric association ofresponse codes) effect previously demonstrated with wholenumber magnitudes. Participants judged whethernonsymbolic ratio test stimuli were larger or smaller thanreference stimuli using response keys located alternatelyeither on the left or on the right side of space. Larger ratiomagnitudes were associated with the right side of space andsmaller magnitudes with the left. These results demonstratethat nonsymbolic ratio magnitudes – defined relationally bypairs of components – are characterized by a left-to-rightspatial mapping. The current focus on ratio magnitudesexpands our understanding of the basic human perceptualapparatus and how it might provide tools that grant intuitiveaccess to more advanced numerical concepts beyond wholenumbers.

In studies on prospective time perception, a prolonging effectof arousal on time estimates is commonly reported fordurations under 2s while the effect vanishes for longerintervals. In this study, we investigated how arousal andpleasure induced by aural stimuli varying in volume andvalence influenced reproductions in the range from 1.1s to 5s.As expected, higher arousal was associated with higherestimates for 1.1s durations. However, this effect was alsofound for 3.8s durations. An additional analysis with linearmixed models revealed an interaction between volumemanipulation and subjective ratings regarding arousal andpleasure. Based on these results we propose that subjectiveexperience of the emotional quality of stimuli might beinteresting for further research on prospective timeperception. Moreover, the results showed that not only withinsubject variation should be statistically controlled whenanalyzing such data. Instead, statistical models should alsoinclude parameters controlling for stimulus material.

People’s beliefs about everyday events are both of theoreti-cal interest in their own right and an important ingredient inmodel building—especially in Bayesian cognitive models ofphenomena such as logical reasoning, future predictions, andlanguage use. Here, we explore several recently used methodsfor measuring subjective beliefs about unidimensional contigu-ous properties, such as the likely price of a new watch. Asa first step towards a way of assessing and comparing beliefelicitation methods, we use hierarchical Bayesian modeling forinferring likely population-level beliefs as the central tendencyof participants’ individual-level beliefs. Three different depen-dent measures are considered: (i) slider ratings of (relative)likelihood of intervals of values, (ii) a give-a-number task, and(iii) choice of the more likely of two intervals of values. Ourresults suggest that using averaged normalized slider ratingsfor binned quantities is a practical and fairly good approxima-tor of inferred population-level beliefs.

This paper reports an experiment testing whether volitionalcontrol over the presentation of stimuli leads to enhancedrecognition memory in 6- to 8-year-old children. Childrenwere presented with a simple memory game on an iPad.During the study phase, for half of the materials childrencould decide the order and pacing of stimuli presentation(active condition). For the other half of the materials, childrenobserved the study choices of another child (yokedcondition). We found that recognition performance was betterfor the objects studied in the active condition as compared tothe yoked condition. Furthermore, we found that the memoryadvantages of active learning persisted over a one-week delaybetween study and test. Our results support pedagogicalapproaches that emphasize self-guided learning and show thateven young children benefit from being able to control howthey learn.

This study investigates how cohesion manifests in readers’thought processes while reading texts when they areinstructed to engage in self-explanation, a strategy associatedwith deeper, more successful comprehension. In Study 1,college students (n = 21) were instructed to either paraphraseor self-explain science texts. Paraphrasing was characterizedby greater cohesion in terms of lexical overlap whereas self-explanation included greater lexical diversity and moreconnectives to specify relations between ideas. In Study 2,adolescent students (n = 84) were provided with instructionand practice in self-explanation and reading strategies across8 sessions. Self-explanations increased in lexical diversity butbecame more causally and semantically cohesive over time.Together, these results suggest that cohesive featuresexpressed in think alouds are indicative of the depth ofstudents’ comprehension processes.

Previous research on basic-level object categories shows thereis cross-cultural variation in basic-level concepts, arguingagainst the idea that the basic level reflects an objectivereality. In this paper, I extend the investigation to the domainof events. More specifically, I present a case study of verbs ofingestion in Maniq illustrating a highly specific categorizationof ingestion events at the basic level. A detailed analysis ofthese verbs reveals they tap into culturally salient notions.Yet, cultural salience alone cannot explain specificity ofbasic-level verbs, since ingestion is a domain of universalhuman experience. Further analysis reveals, however, thatanother key factor is the language itself. Maniq’s preferencefor encoding specific meaning in basic-level verbs is not apeculiarity of one domain, but a recurrent characteristic of itsverb lexicon, pointing to the significant role of the languagesystem in the structure of event concepts.

This research introduces the switch task, a novel learning modethat fits with calls for a broader explanatory account of hu-man category learning (Kurtz, 2015; Markman & Ross, 2003;Murphy, 2002). Learning with the switch task is a processof turning each presented exemplar into a member of anotherdesignated category. This paper presents the switch task to fur-ther explore the contingencies between learning goals, learn-ing modes, outcomes, and category representations. The pro-cess of successfully transforming exemplars into members of atarget category requires generative knowledge such as within-category feature correspondences – similar to inference learn-ing. Given that the ability to switch items between categoriesnicely encapsulates category knowledge, how does this relateto more familiar tasks like inferring features and classifyingexemplars? To address this question we present an empiri-cal investigation of this new task, side-by-side with the well-established alternative of classification learning. The resultsshow that the category knowledge acquired through switchlearning shares similarities with inference learning and pro-vides insight into the processes at work. The implications ofthis research, particularly the distinctions between this learn-ing mode and well-known alternatives, are discussed.

Human learning in complex environments critically dependson the ability to perform model selection, that is to assess com-peting hypotheses about the structure of the environment. Im-portantly, information is accumulated continuously, which ne-cessitates an online process for model selection. While modelselection in human learning has been explored extensively, it isunclear how memory systems support learning in an online set-ting. We formulate a semantic learner and demonstrate that on-line learning on open model spaces results in a delicate choicebetween either tracking a possibly infinite number of compet-ing models or retaining experiences in an intact form. Sincenone of these choices is feasible for a bounded-resource mem-ory system, we propose an episodic learner that retains an op-timised subset of experiences in addition to semantic memory.On a simple model system we demonstrate that this norma-tive theory of episodic memory can effectively circumvent thechallenge of online model selection.

Inhibitory control, the ability to stop inappropriate actions, isan important cognitive function often investigated via the stop-signal task, in which an infrequent stop signal instructs the sub-ject to stop a default go response. Previously, we proposed arational decision-making model for stopping, suggesting theobserver makes a repeated Go versus Wait choice at each in-stant, so that a Stop response is realized by repeatedly choosingto Wait. We propose an alternative model here that incorpo-rates a third choice, Stop. Critically, unlike the Wait action,choosing the Stop action not only blocks a Go response at thecurrent moment but also for the remainder of the trial – thedisadvantage of losing this flexibility is balanced by the bene-fit of not having to pay attention anymore. We show that thisnew model both reproduces known behavioral effects and hasinternal dynamics resembling presumed Go neural activationsin the brain.

Complex numerals (e.g., four hundred) have a multiplicative structure (four hundred = 4 x 100). This paper investigates whether children are sensitive to the meaning of the multiplicative structure. We designed a novel word learning paradigm and taught 4- to 6-year-old children the meaning of a novel numeral phrase (e.g., ‘one gobi houses’ to mean a group of three houses). We then asked whether they could generalize it to a novel context (e.g., ‘two gobi butterflies’ to mean two groups of three). Experiment 1 showed that only English-speaking children who received multiplier syntax training were able to generalize. Experiment 2 extended findings from Experiment 1 to Cantonese-speaking children and found that they could also generalize a novel multiplier to novel contexts. These results suggest that children as young as 4 can create a mapping between the structure of complex numerals and a multiplicative meaning.

Do neural systems for planning motor actions play a func-tional role in understanding action language? Across multi-ple neuroimaging studies, processing action verbs correlateswith somatotopic activity in premotor cortex (PMC). Yet, onlyone neurostimulation study supports a functional role for PMCin action verb understanding: paradoxically, inhibiting PMCmade people respond faster to action verbs. Here we investi-gated effects of PMC excitation and inhibition on action verbunderstanding using tDCS. Right-handers received excitatoryor inhibitory stimulation to left PMC hand areas, then madelexical decisions on unimanual action verbs and abstract verbs.tDCS polarity selectively affected how accurately participantsresponded to unimanual action verbs. Inhibitory stimulationto left PMC caused a relative improvement in performance forright-hand responses, whereas excitatory left PMC stimulationcaused a relative impairment. tDCS polarity did not differ-entially affect responses to abstract verbs. Premotor areas thatsubserve planning actions also support understanding languageabout these actions.

The basic cognitive architecture of the human brain remainsunknown. However, there is evidence for the existence ofdistinct behavioral control systems shared by humans andnonhumans; and there is further evidence pointing to distincthigher-level problem solving systems shared by humans andother primates. To clarify the nature of these proposedsystems and examine how they may interact in the brain, wepresent a four-level model of the primate brain and compareits performance to three other brain models in the face of achallenging foraging problem (i.e., with transparent, and thus,invisible barriers). In all manipulations (e.g., size of problemspace, number of obstacles), our model never performed thebest outright; however, it was always among the best,appearing to be a jack-of-all-trades. Thus, the virtues of ourprimate brain lie not only in the heights of thinking it canreach, but also in its range and versatility.

Learning concepts and categories in the real world is often ac-companied by verbal labels. The existing theoretical accountsof how labels influence what we learn range from facilitationto overshadowing, with changes occurring over development.Studies investigating how labels influence what people learnhave typically been confined to a category learning framework,where participants were tasked to learn how to discriminatecategories or infer missing category properties. Here, we in-vestigate how the absence or presence of labels, both commonand unique, alter how people attend and what they remember ina more general setting. Our results suggest that unique labelsmay promote visual exploration of objects; whereas, there wasno evidence to support the claim that hearing the same labelassociated with different members of a to-be-learned categorydirected attention to common features.

Learning a new language involves significant vocabulary ac-quisition. Learners can accelerate this process by relying onwords with native-language overlap, such as cognates. Forbilingual third language learners, it is necessary to determinehow their two existing languages interact during novel lan-guage learning. A scaffolding account predicts transfer fromeither language for individual words, whereas an accumula-tion account predicts cumulative transfer from both languages.To compare these accounts, twenty English-German bilingualadults were taught an artificial language containing 48 novelwritten words that varied orthogonally in English and Germanwordlikeness (neighborhood size and orthotactic probability).Wordlikeness in each language improved word production ac-curacy, and similarity to one language provided the same bene-fit as dual-language overlap. In addition, participants’ memoryfor novel words was affected by the statistical distributions ofletters in the novel language. Results indicate that bilingualsutilize both languages during third language acquisition, sup-porting a scaffolding learning model.

Philosophers, psychologists, and neuroscientists have oftendivided the mind into fundamental component parts. Does thisintuition carry over into folk philosophy of mind? In a seriesof large-scale studies, we explore intuitive distinctions amongdifferent kinds of mental phenomena and consider how thesedistinctions might organize the conceptual space of thediverse “intelligent” and “social” entities in the modernworld. Across studies, independent exploratory factoranalyses reveal a common latent structure underlying mentalcapacity attributions, centered on three types of phenomenalexperiences: physiological experiences of biological needs(e.g., hunger, pain); social-emotional experiences of self- andother-relevant emotions (e.g., guilt, pride); and perceptual-cognitive abilities to detect and use information about theenvironment (e.g., hearing, memory). We argue for anexpanded model of folk philosophy of mind that goes beyondagency and experience (H. M. Gray, Gray, & Wegner, 2007)to make basic and important distinctions among differentvarieties of experience.

When two individuals perform a task together, they combinetheir individual skills to achieve a joint goal. Previousresearch has shown that interindividual skill differencespredict a group’s collective benefit in joint perceptualdecision-making. In the present study, we tested whether thisrelationship also holds for other task domains, using adynamic object control task in which two participants eachcontrolled either the vertical or horizontal movement directionof an object. Our findings demonstrate that the difference inindividuals’ skill levels was highly predictive of the dyad’scollective benefit. Differences in individuals’ subjectiveratings of task difficulty reflected skill differences and thusalso turned out to be a predictor of collective benefit.Generally, collective benefit was modulated by spatial taskdemands. Overall, the present study shows that previousfindings in joint decision-making can be extended to dynamicmotor tasks such as joint object control.

People conceptualize both time and numbers as unfoldingalong a horizontal line, either from left to right or from rightto left. The direction of both the mental timeline (MTL) andthe mental number line (MNL) are widely assumed to dependon the direction of reading and writing within a culture.Although experimental evidence supports this assumptionregarding the MTL, there is no clear evidence that readingdirection determines the direction of the MNL. Here wetested effects of reading experience on the direction of boththe MTL and MNL. Participants read English text eithernormally (from left to right) or mirror-reversed (from right toleft). After normal reading, participants showed the space-time associations and space-number associations typical ofWesterners. After mirror reading, participants’ space-timeassociations were significantly reduced but their space-number associations were unchanged. These results suggestthat the MTL and MNL have different experiential bases.Whereas the MTL can be shaped by reading experience, theMNL is shaped by other culture-specific practices throughwhich people experience numbers arrayed in left-right space.

The social and cognitive mechanisms of cultural evolutionhave been studied in detail for different domains: language,technology, the economy, art, etc. However, a model thatincorporates the function of a cultural tradition and that isable to compare evolutionary dynamics across culturaldomains has not been formulated. By exploring the dynamicsof comparable linguistic, technological and artisticexperimental tasks, we test the effect of domain-specificfunction on evolutionary mechanisms such as inheritance,innovation and selection. We find evidence that culturaldomain shapes both the structure of the traditions and the waythe cultural-evolutionary mechanisms operate. Thesimplifying effects of cultural transmission are noticeable inlanguage and technology, but not in art; innovation is highestin art and lowest in language; and functional pressures lead todifferent morphological adaptations across domains. Thisspeaks of a crucial role of function and domain in theevolution of culture.

Previous attempts at modeling the neuro-cognitive mecha-nisms underlying word processing have used connectionist ap-proaches, but none has modeled spoken word architectures asthe input is presented in real-time. Hence, such models rely onthe ingenuity of the modeler to establish a mapping of real-time stimulus to the model’s input which may not preserveprocessing that happens during each time step. We present aneural field model which successfully replicates the effect ofimmediate auditory repetition of monosyllabic words and fitsit to a component of a well-studied mechanism for analyzinglanguage processing, the event-related potential (ERP). Thisrepresents a new modeling approach to studying the neuro-cognitive processes, one that is based on the bottom-up inter-action of real-time sensory information with higher-level cate-gories of cognitive processing.

Conveying information in a false or indirect manner in consid-eration of listeners’ wants (i.e. being polite) seemingly contra-dicts an important goal of a cooperative speaker: informationtransfer. We propose that a cooperative speaker considers bothepistemic utility, or utility of providing the listener new and ac-curate information, and social utility, or utility of maintainingor boosting the listener’s self-image (being polite). We for-malize this tradeoff within a probabilistic model of languageunderstanding and test it with empirical data on people’s infer-ences about the relation between a speaker’s goals, utterancesand the true states of the world.

When did something almost happen? In this paper, we in-vestigate what brings counterfactual worlds close. In Exper-iments 1 and 2, we find that participants’ judgments aboutwhether something almost happened are determined by thecausal proximity of the alternative outcome. Something almosthappened, when a small perturbation to the relevant causalevent would have been sufficient to bring it about. In contrastto previous work that has argued that prior expectations areneglected when judging the closeness of counterfactual worlds(Kahneman & Varey, 1990), we show in Experiment 3 thatparticipants are more likely to say something almost happenedwhen they did not expect it. Both prior expectations and causaldistance influence judgments of “almost”. In Experiment 4, weshow how both causal proximity and beliefs about what wouldhave happened in the absence of the cause jointly explain judg-ments of “almost caused” and “almost prevented”.

Miscommunication is often seen as a detrimental aspectof human communication. However, miscommunicationcan differ in cause as well as severity. What distinguishesa miscommunication where conversation partnerscontinue to put forth the effort from miscommunicationwhere conversation partners simply give up? In this eye-­‐‑tracking study, participants heard globally ambiguousstatements that were either a result of an experimentalerror or speaker underspecification; participants eitherreceived positive or negative feedback on theseambiguous trials. We found that negative feedback,paired with the reliability of the message, will impact theamount of processing effort a comprehender putsforth—specifically, listeners were less forgiving of errorswhen they were penalized and when speakers’instructions lacked effort. This suggests that languageusers weigh conversational contexts and outcomes aswell as linguistic content during communication.

Despite their considerable communicative abilities, youngchildren often have difficulty interpreting complex linguisticstructures in context. Two examples of this phenomenon arenegation and pragmatic implicature, both of which pose some-times surprising difficulties for preschoolers. Both of thesestructures require children to resist a more salient alternativeinterpretation; since executive function abilities develop ex-tensively during childhood, perhaps failures are due to prob-lems in inhibition. To test this hypothesis, we designed tasksto measure inhibitory control, negation, and implicature com-prehension in children and adults. Using standard analyses aswell as drift diffusion models, we found different patterns ofprocessing on all three tasks, and no support for the hypothesisthat inhibitory control per se is playing a role in either adults’or children’s negation or implicature processing. Instead, ouranalyses reveal qualitatively different developmental trajecto-ries, suggesting task-specific factors driving these changes.

Queries about singular causation face two problems: It needsto be decided whether the two observed events are instanti-ations of a generic cause-effect relation. Second, causationneeds to be distinguished from coincidence. We propose acomputational model that addresses both questions. It accessesgeneric causal knowledge either on the individual or the grouplevel. Moreover, the model considers the possibility of a co-incidence by adopting Cheng and Novick’s (2005) power PCmeasure of causal responsibility. This measure delivers theconditional probability that a cause is causally responsible foran effect given that both events have occurred. To take uncer-tainty about both the causal structure and the parameters intoaccount we embedded the causal responsibility measure withinthe structure induction (SI) model developed by Meder et al.(2014). We report the results of three experiments that showthat the SI model better captures the data than the power PCmodel.

In this study we explored the impact of student versus tutorpedagogical decision-making on learning. More specifically,we examined what would happen if we let students decidehow to handle the next task: to view it as a worked exam-ple or to solve it as a problem solving. We examined this im-pact at two levels of task granularity: problem vs. step. This2 × 2 study was conducted on an existing Intelligent TutoringSystem (ITS) called Pyrenees. 279 students were randomlyassigned to four conditions and the domain content and re-quired steps were strictly controlled to be equivalent acrossfour conditions: all students used the same system, followedthe same general procedure, studied the same training materi-als, and worked through the same training problems. The onlysubstantive differences among the four conditions were deci-sion agency {Student vs. Tutor} and granularity {Problem vs.Step}. That is: who decided to present an example or to solvea problem; and was the decision made problem-by-problem orstep-by-step? Our results showed that there was a significantinteraction effect between decision agency and granularity onstudent learning and a significant main effect of granularity ontime on training. That is, step level decisions can be more ef-fective than problem level decisions but the students were morelikely to make effective pedagogical decisions at problem levelthan step level. In general, on both problem and step levels, thestudents were significantly more likely to decide to do problemsolving rather than study it as a worked example.

The purpose of our study was to determine whether activemusical engagement alleviates decline in inhibitory controldue to cognitive aging. Given that musical training in youngadults has been shown to improve attentional performance,we can expect this benefit to persist for older adults as well.With the help of the stop-signal procedure, we measuredresponse inhibition of young and older adults who provided aself-reported assessment of their musical engagement, usingthe recently validated Goldsmiths Musical SophisticationIndex. The Gold-MSI addresses a variety of musical activitiesand thus offers a more comprehensive measure than ability toplay a musical instrument used in the past. Results of theexperiment showed that older participants had longer stop-signal reaction times, independently of their musical trainingand engagement, but musical training and ensemble practicewere negatively related to the proportion of missed responsessuggesting a weak effect of certain types of musical activitieson inhibitory control.

The current study investigated the development of numerical estimation in 3- to 5-year-old children sampled monthly for six months. At each session, children completed a task that assesses verbal number knowledge (Give-N task) and a numerical estimation task that assesses approximate number knowledge (Fast Cards). Results showed that children who acquired the cardinal principle (CP) during the course of the study showed marked improvement on the estimation task. Following CP acquisition, estimation became more accurate overall but also fluctuated widely. We discuss the implications of our findings for number word learning, particularly the mapping between verbal number and the approximate number system (ANS).

usually collect information to serve specific goals andoften end up with samples that are unrepresentative of the un-derlying population. This can introduce biases on later judg-ments that generalize from these samples. Here we show thatgoals influence not only what information we collect, but alsowhen we decide to terminate search. Using an optimal stop-ping analysis, we demonstrate that even when learners have nocontrol over the content of a sample (i.e., natural sampling),the simple decision of when to stop sampling can yield sampledistributions that are non-representative and could potentiallybias future decision making. We test the prediction of thesetheoretical analyses with two behavioral experiments

Different models of categorization are often treated as compet-ing accounts, but specific models are often used to understandindividual differences, by estimating individual-level param-eters. We develop an approach to understanding categoriza-tion that allows for individual differences both between andwithin models, using two prominent categorization models thatmake different theoretical assumptions: the Generalized Con-text Model (GCM) and General Recognition Theory (GRT).We develop a latent-mixture model for inferring whether anindividual uses the GCM or GRT, while simultaneously allow-ing for the use of special-case simpler strategies. The GCMsimple strategies involve attending to a single stimulus dimen-sion, while the GRT simple strategies involve using unidimen-sional decision bounds. Our model also allows for simple con-taminant strategies. We apply the model to four previouslypublished categorization experiments, finding large and inter-pretable individual differences in the use of both models andspecific strategies, depending on the nature of the stimuli andcategory structures.

Participants in most binary-choice tasks with multiple trialstend to probability-match (Vulkan, 2000) — i.e., provide re-sponses that match the probability distribution of the presentedpopulation. Given a single trial, however, participants usuallychoose the majority option (James & Koehler, 2011). By us-ing a method that visually presents the probabilities of the twocompeting options, we examine responses when participantsare given only a single trial, and initial responses when partic-ipants are given multiple trials. While we still observe aggre-gate probability-matching in the multiple-trial condition, wefind robust sequence effects in participants’ initial responses,including robust maximizing behavior on the first response.This suggests that both maximizing in single-trial experimentsand aggregate probability-matching in multiple-trial ones canbe explained by a single, underlying mechanism; one thatseeks to provide a representative sample at each point duringsequence generation.

Networks of every kind and in numerous fields areomnipresent in today’s society (e.g. brain networks, socialnetworks) and are the intense subject of research. It wouldbe of great utility to have a computationally efficient andgenerally applicable method for assessing similarity ofnetworks. The field (going back to the 1950s) has not comeup with such a method (albeit a few moves in this directionexist, such as Jaccard coefficients, QAP--quadraticassignment procedure, and more recently Menezes & Roth,2013, and Asta & Shalizi, 2014). I present a Bayesian-basedmetric for assessing similarity of two networks, possibly ofdifferent size, that include nodes and links between nodes. Iassume the nodes are labeled so that both the nodes andlinks between two nodes that are shared between the twonetworks can be identified.The method calculates similarity as (a monotonictransformation of) the odds that the two observed networks,termed V and W, were produced by random sampling froma single master network, termed G, as opposed to generationby two different but similar networks, termed Gv and Gw.The simplest form of the method ignores strengths thatcould be assigned to nodes and links, and considers onlynodes and links that are, or are not, shared by the networks.Suppose there are n V nodes and N V links only in V, n Wnodes and N W links only in W and n c nodes and N c linksshared between the networks. Thus the number of nodes inV is n c + n V and the number in W is n c + n W . The number ofunique nodes in both V and W is n c + n V + n W = n. Thenumber of links in V is N c + N V and the number in W is N c +N W . The number of unique links in both V and W is N c + N V+ N W = N.The single master network, G, is assumed to consist of theunion of the nodes and links in the two networks, and has nnodes and N links. The probability a given shared node willbe randomly and independently sampled twice is[(n V +n c )/n][(n W +n c )/n]. The probability a given shared linkwill be randomly and independently sampled twice is[(N V +N c )/N][(N W +N c )/N].If there are two generating networks I assume they eachhave n nodes and N links. I also assume they are similar, because we would not be comparing dissimilar networks.The degree of similarity is controlled by ‘tuning’parameters 1 : Gv and Gw are assumed to share αn nodes andβN links. The probability a given shared node will besampled twice is then α[(n V +n c )/n][(n W +n c )/n], and theprobability a given shared link will be sampled twice isβ[(N V +N c )/N][(N W +N c )/N]. The likelihood ratio λ js for G vs(GV, GW) as generator of a given shared node is then 1/αand the likelihood ratio π js of a given shared link is then 1/β.For a non-shared node, say in V, similar reasoning gives alikelihood ratio λ kV of[1-(n W +n c )/n)] /[1– α(n W +n c )/n]and for a non-shared link a likelihood ratio π kV of[1-(N W +N c )/n)] /[1– α(N W +N c )/N]For a non-shared node or link in W substitute a Wsubscript for the V subscript in these likelihood ratios.Computational efficiency is a necessity if the similaritymetric is to be applied to large networks. For this reason Ido not calculate the exact probabilities for the numbers ofshared and non-shared nodes and links that are observed(the combinatoric complexity of such calculations isenormous). Instead I make the simplifying assumption thateach node and link contribute the likelihood ratios givenabove and that the total odds is obtained by multiplying allthe likelihood ratios together. This simplification canperhaps be justified if similar distortion is produced by thissimplifying assumption for both the cases of G and (G V ,G W )as generators. Under this simplifying assumption the overallodds becomes:φ(1/2) = (λ js ) nc (λ kV ) nV (λ jW ) nW (π js ) Nc (π kV ) NV (π jW ) NWTaking the log of this product converts the calculation tosums and makes calculation highly efficient.This abstract is too short to permit giving the different andmore complex results that hold for the several cases whenthe nodes and/or links have associated strengths. I give asummary of some of the results here. The results for linksand nodes are similar so consider the results for nodes. Letthere be just one set of strength values, Si for the i-th node.Norm these to sum to 1.0. For either generation by G or(Gv,Gw) assume sampling is made without replacement andproportional to strength. Let Ziv and Ziw be theprobabilities that node i will be sampled by n v +n c samples,or n w +n c samples respectively. The Z’s would be difficult to obtain analytically but could be estimated by Monte Carlosampling. Consider two possibilities for the way that Gv andGw overlap. In Case A the probability a node will be sharedis simply α, independent of strength. In Case B, theprobability a node will be shared is an increasingfunction of strength, Y i .For Case A the likelihood ratio for a shared node i is:1/α. For a node k only in V the likelihood ratio is: λ kV =(1-­‐Zkw)/{1 – α (1-­‐Zkw)}. For a node only in W exchangethe subscripts v and w. Then we have for the odds due tonodes: φ D = (1/α) nc Π k (λ kV )Π j (λ jW ).For Case B the likelihood ratio for a shared node i is1/Y i . For a node k only in V the likelihood ratio is: λ kV =(1-­‐Zkw)/{1–Y k (1-­‐Zkw)}. Again switch v and w subscriptsfor a node only in W. Then we have for the odds due tonodes: φ D = Π i (1/Y i )Π k (λ kV )Π j (λ jW ).These expressions would have analogous forms forlinks, with different Ns, Z’s and Y’s, and the overall oddswould, as before, be a product of the odds for nodes andthe odds for links.The critical difference between Cases A and B is thedegree to which evidence based on an observed sharednode or link is strength dependent: For Case B thisevidence rises as strength decreases. This should raiseconcerns: However strengths are obtained there is likelyto be measurement noise that reduces the reliability oflow strength values. This might argue in favor of usingCase A, or if one preferred Case B to restrict the Yi valuesto lie above a lower bound. The idea would be to letevidence depend most on the nodes (or links with highstrength values.It should be observed that the existence of acomputationally efficient and generally applicable metricfor network similarity would allow alignment of non-labeled networks. One would search for the alignment ofnodes that would maximize the metric.I have many relevant publications demonstrating somedegree of expertise in Bayesian modeling (e.g.: Shiffrin &Chandramouli, in press; Shiffrin, Chandramouli, &Grünwald, 2015; Chandramouli & Shiffrin, 2015; Nelson &Shiffrin, 2013; Cox & Shiffrin, 2012; Shiffrin, Lee, Kim, &Wagenmakers, 2008; Cohen, Shiffrin, Gold, Ross, & Ross,2007; Denton & Shiffrin; Huber, Shiffrin, Lyle, & Ruys,2001; Shiffrin & Steyvers, 1997). I note that the presentresults are in a vague sense an extension of the metricproposed for matching memory probes to memory tracesthat are given in Cox and Shiffrin (2012) and in theappendix of Nelsonb and Shiffrin (2013).

Every day, we navigate our environments with astonishingease. Most of our paths are familiar to us and can benavigated without (much) conscious thought; in other cases,we use various strategies to find our way (Tenbrink &Wiener, 2007). Since these processes are at the heart ofhuman spatial cognition they have been researchedextensively, often based on route directions as the mostcommon verbalizations of navigation. Our research extendsthis tradition across various wayfinding contexts, addressingstreet network scenarios (Hölscher, Tenbrink, & Wiener,2011), complex buildings (Tenbrink, Bergmann, &Konieczny, 2011), alpine environments (Egorova, Tenbrink,& Purves, 2015), and including effects of automatic systemsas producers (Tenbrink & Winter, 2009) or recipients(Moratz & Tenbrink, 2006; Tenbrink et al., 2010) of spatialdirections. In all of these studies natural language data areused to address concepts of navigation, some of which areexpressed explicitly, while others remain implicit and onlyindirectly reflected through the ways in which speakers uselanguage in spatial navigation contexts.

Some researchers argue that gestures serve an interpersonal function, such as making the intended message clear(e.g., Gallagher & Frith, 2003; cf. Kita, 2000). In this study, we tested whether gestures serve an interpersonal function,specifically predicting that the higher participants’ autism spectrum quotient, the less frequently they would gesture. Participantscompleted the Autism Spectrum Quotient questionnaire (Baron-Cohen, Wheelwright, Skinner, Martin, & Clubley, 2001). Toelicit gestures, participants did two tasks. In one, they explained spatial and social concepts. In another, they told the story ofa cartoon. The dependent variable is the gesture rate (gestures per word), to account for individual differences in volubility.Participants completed a standardized vocabulary test. The initial results show no correlation between gesture rate in either taskand ASQ scores. There is a negative correlation between ASQ and vocabulary scores. These results are inconsistent with theargument that gestures serve an important interpersonal function.

In this study, we aim to understand the origins of human intuitions about Euclidean geometry by simulating geo-metric concepts acquisition with unsupervised learning in hierarchical generative models. Specifically, we build a deep neuralnetwork that learns a hierarchical generative model of sensory inputs. The results show that hidden layer activities can supportthe categorization of different geometric objects and distinguish among various spatial relationships between geometric figures.Specifically, hidden layer activities can be decoded to compare line orientations, detect right triangles, and judge whether twotriangles are similar or not. We further analyze the response profiles of hidden layers units and find some units resembling pari-etal neurons in the brain. Using unsupervised deep learning, the current modeling work provides a possible explanation of howEuclidean geometrical intuitions might emerge from daily visual experience, which has significant implications for cognitivepsychology and computational neuroscience.

Vivid representations are often made by traumatic events with intense emotions. The emotions may be activatedautomatically from memory on the mere exposure of an affect-loaded stimulus. The aims of this study were to investigate theimplicit emotional processing of traumatic events and the moderation of priming by semantic level of the events, using primednaming task at short stimulus onset asynchrony (150ms). A 3 semantic level of traumatic primed events (general, domestic,or foreign words) by 3 target emotions (high-arousal negative, moderate positive, low-arousal negative words) repeated designwas used. When the primed words were general (e.g. terror) or domestic (e.g. Sewol ferry disaster) events, response time ofhigh-arousal negative words (e.g. fear or angry) were significantly longer than other emotion words (e.g. happy or sadness).Our findings suggest contrast effects of affective priming as a result of automatic implicit regulation.

The Cognitive Science Research group at ETS conducts research and development at the forefront of educationalassessment, using cognitive theory in the design of assessments, building cognitive models to guide interpretation of test-takers’performance, and researching cognitive issues in the context of assessment. Moving beyond traditional (e.g., multiple-choice)tests, the group explores assessments that use innovative, highly interactive digital environments such as online games, virtuallabs or other simulations, and human-agent conversation-based interactions. Researchers also investigate how to draw appropri-ate inferences about test-takers’ knowledge and skills from complex data sources such as eye-movements, interaction logs, andother sequential information. I will provide an overview of the group’s research, including the use of cognitive models to inter-pret test-takers’ actions within interactive assessment tasks and empirical studies on how test-takers externalize their knowledgewhen problem solving in domains such as science inquiry, inter-cultural competence, and mathematics argumentation.

There are two widespread strategies for testing experimentally whether a conclusion follows naively from a sequenceof premises. The free-form response strategy (FFR) presents participants with the premises and asks them “what, if anything,follows?” In the simplest case, participants’ responses are coded as to whether they made the predicted inference. On the yes/noquestion strategy (YNQ), after presenting the premises, the researcher puts forth a sentence C and asks whether C follows fromthe premises.We compare the two methodologies with respect to six types of fallacious problems involving propositional connectives fromthe mental-models literature, to address the question of whether the methodologies are equally valid. We found that the twomethodologies overwhelmingly yield identical results. Interestingly, the exceptions we found show that in some cases FFR failsto detect an attractive fallacious conclusion that can be reliably probed with YNQ.

How do we learn non-physical properties of physical objects? We explored how the statistical structure of eventscan be a source of object property learning. Twenty-five participants saw sequences of visual events surrounding two distinctobjects. Object identity determined 1) the direction of a high transition probability between two events, and 2) the frequency oftwo other events. Learning was unsupervised and unguided. However, participants spontaneously noticed these regularities. Inan explicit, verbal judgment task, they discriminated between frequent vs. rare events (t(24) = 10.7, p< 0.00001) and betweenpredictive vs. non-predictive event pairs (t(24) = 3.04, p<0.01), as appropriate to the object. These statistics gave rise to distinctconceptual interpretations: participants ascribed a causal interpretation to the predictive statistics (t(24) = 1.91, p<0.05) morethan to events frequently co-occurring with the objects (t(24) = 3.00, p<0.01). Such learning may underlie concept acquisition,particularly of functional kinds like artifacts.

Fluent event processing involves segmenting streaming sensory information into discrete units. Adults and childrenselectively attend to these meaningful moments within event streams, which predicts later memory. In natural environments,however, uninterrupted attention is unlikely. Consequently, some information is missed, including event boundary information.To what extent does missing information alter the attentional dynamics of processing, specifically viewers’ ability to targetremaining boundaries with enhanced attention? Adults advanced at their own pace through slideshows of unfolding activity.Slides were systematically deleted to enable comparison of viewers’ attentional dynamics when specific content was presentversus absent. Average dwelling per slide increased with missing content. However, the attentional dynamics of processingwere unaltered; attention to boundaries displayed comparable enhancement regardless of missing content. Attention modulationduring processing of relatively familiar events appears to be highly robust to missing information. What occurs with more novelevents is an interesting question for future research.

The plausability of the retracting information, which can affect whether the retracted information be replaced bythe retracting information, had not been studied much in previous studies on continued influence effect. An experiment wasconducted to investigate how the plausability of a retracting information affects the understanding of an event in which someof the information had been retracted. In the experiment, the plausibility of the retracting information (low versus high) andthe manner of correcting information (simple retraction versus supplying an alternative) were manipulated, and the participantswere asked to answer memory questions and inference questions. Retracted information was better remembered for memoryquestions, but was used less frequently for inference questions when the retracting information was less plausible comparedto when the retracting information was quite plausible. The results were interpreted to support the ‘memory-based’ theory ofmemory updating.

The present study investigates whether, and if so in what way, adult learners are sensitive to the properties ofthe statistical input, such as frequency and skewedness, when learning and generalizing category labels. Participants werepresented with novel objects belonging to four different categories and heard category labels in a cross-situational learningtask. The four categories were matched for the total amount of exposure but varied in category size and shape of distribution.Participants learned object-to-label mappings better for categories with a skewed distribution of fewer objects. Moreover,object-to-label mapping performance was positively related to the ability to extend category knowledge to novel items. Co-occurrence frequency or category size alone were not good predictors of label learning and generalization. The results indicatethe importance of input distribution in word and category learning processes.

Young children are remarkably prosocial, yet the mechanisms driving their prosociality are poorly understood. Intwo studies, we looked at whether a need for reciprocity drives children’s prosocial behavior. In Experiment 1, children weregiven a puzzle task to complete in which they were either missing 2 pieces (experimental group) or not (control group). Allchildren then received 2 puzzle pieces from a confederate, resulting in either necessary instrumental help (experimental group)or unnecessary help (control group). Children were more prosocial (shared a greater proportion of their resources) with theconfederate after receiving instrumental help than after receiving unnecessary help. In Experiment 2, we investigated the typesof principles children use when paying back help. We found that children employed a mix of exact reciprocation and “needs-based” help when paying back individuals. Our results suggest an important role of gratitude and reciprocity in the developmentof early prosociality.

Everyday decision-making is filled with choices about what to act on, with outcomes playing a critical role in learn-ing. Information gain is oft cited as a valuable approach to maximize potential learning, but its computation is costly. It entailsevaluating the probability of multiple outcomes given any possible action, and then considering the degree of belief-change overall possibilities. Given the computational complexity of this evaluation, it becomes important to ask whether learners employcues to information gain; are there heuristics that drive choice in active learning? Our experiments ask participants to choosebetween two options (varying in distributional characteristics) in either a “learning-condition” or “collecting-condition”. Ourresults suggest that adults are sensitive to cues (e.g. variance) that tend to correlate with information gain. These cues areonly favored in learning-goal contexts, suggesting that certain distributional qualities are not always appealing, but rather areselectively-employed heuristics towards information gain.

How language represents meaning remains a central topic of debate in linguistics. On some accounts, the nounphrases in a sentence are identified semantically by a list of independent atomic labels (thematic roles), ordered relative to oneanother depending on the position of the nouns around the verb (e.g., AGENT-THEME-GOAL). Others instead capture suchinterdependencies with complex, non-atomic event structures (e.g., [x CAUSE [y TO-COME-TO-BE-AT z]]). Here, we usestructural priming to investigate the psychological reality of these two theories of semantic representation. On the thematicrole approach, we should expect to see priming between theme-first locatives and prepositional-object datives (both VP-NP-PPsyntactically) precisely because their thematic ordering is consistent across the two constructions. The event structure approachposits no such minimal semantic structural similarity, such that we should not see priming cross-constructionally. We find onlywithin-construction priming (N=52) and not across-construction priming (total N=344), in favor of event structures.

Re-representation is a mechanism for aligning non-identical structurally corresponding predicates during compari-son. Re-representation is therefore a shift in the encoding of a stimulus from an initial set of elements to an altered set. Forexample, when comparing two analogous sentences, different verbs could be re-represented to allow a match – thereby alter-ing construal. Re-representation is theoretically and intuitively compelling, but difficult to demonstrate. We had participantscompare pairs of short text passages and judge them as potential analogies. At test, participants read an altered version of aprior passage and were asked to detect changes from the original. The alterations increased the semantic match between thepassage and its analog. Participants who had compared the original passage to the analog were less likely to detect changesthan those who compared to a non-analogous passage. As would be expected due to re-representation, analogical comparisonmade participants less sensitive to the changes.

We describe a novel model-based theory of how individuals reason deductively about temporal relations. It positsthat temporal assertions refer to mental models – iconic representations of possibilities – of events (Khemlani, Harrison, &Trafton, 2015; Schaeken, Johnson-Laird, & d’Ydewalle, 1996). In line with recent accounts of spatial reasoning (Ragni &Knauff, 2013), the theory posits that individuals tend to build a single preferred model of a temporal description. The moremodels necessary to yield a correct answer, the harder that problem is. The theory is implemented in a computer program,mReasoner, which draws temporal deductions by building models. It varies four separate factors in the process: the size of amodel, its contents, the propensity to consider alternative models, and the propensity to revise initial conclusions. Two studiescorroborated the predictions of the theory and its computational implementation. We conclude by discussing temporal andrelational inference more broadly.

While people are faster to categorize objects at an intermediate or basic level of specificity (e.g. “bird”), severalrecent studies have shown them to have much earlier access to more general category information (e.g. “animal”). Ultra-rapidsuperordinate classification has been taken as evidence that recognition processes are largely feed-forward. In simulations witha deep neural network model, we show that this conclusion does not follow: even a model that is fully recurrent and interactiveshows ultra-rapid superordinate classification patterns when tested with analogs of behavioral tasks such as rapid serial visualpresentation or deadline classification. Moreover, this recurrent model explains recently-observed similarities and differencesin the time-course of classification as estimated by electro-encephlography (EEG) versus human electro-corticography (ECoG),and also account for the well-known basic-level advantage in non-speeded classification. These results provide evidence thatultra-rapid and unconstrained visual object recognition is supported by interactive processes in the brain.

The relationship between handedness and mathematical abilities is controversial. Whilst some researchers haveclaimed that left-handers are gifted in mathematics and strong right-handers perform the worst in mathematical tasks, it hasbeen more recently proposed that mixed-handers are actually the most disadvantaged group. To disentangle these discrepancies,we conducted five experiments in several Italian schools (total participants: N = 2,308) involving students of different ages(6 to 17 years) and a range of mathematical tasks. The results showed that (a) the percentage of variance in mathematicsscores explained by handedness was moderate (about 5%) but statistically significant, and (b) the shape of the relationshipbetween handedness and mathematical ability depended on age, task, and gender. We concluded that the different outcomesreported in the literature probably reflected the dissimilarities between the studies about the above variables. Therefore, a morecomprehensive model is needed, which explains how these variables interact.

We used a non-linguistic experimental paradigm to explore the instantaneous creation of new communicative con-ventions. Participants played a computer game, in which they sent and interpreted minimal signals to obtain shared rewardswithin a virtual scene. Trials manipulated the space of possible signals that could be sent, and the meanings to be expressed (lo-cations and quantities of rewards); as such, optimal success in the task required participants to jointly construct signal-meaningmappings that functioned as part of a system, rather than in isolation.We observed different signalling strategies among participants, but with some individuals using ‘system-mapping’ conven-tions that globally reorganized in light of changing task constraints. Such behaviour reflects the principle of pre-emption inpragmatics, where the inferred meaning of an utterance depends on its relationship among a set of alternatives. Our initialfindings provide a basis for future research, investigating contexts that are conducive to this phenomenon.

The associative sequence learning (ASL) model states that error patterns in observed actions during physical imi-tation and verbal description are identical, because of the critical role played by the process of visual input compared to theprocess after visual input. Action models were presented that comprised four elements: using right or left hand, using right orleft stick, tapping right or left side of a box, and placing a stick on right or left side. In the condition in which identical elementsof video stimuli and manipulated objects placed in front of participants had the same color, the colored element was correctlyperformed compared to the condition with different colored elements. However, colored element was not correctly performedin the condition in which particular elements of video stimuli were colored, whereas manipulated objects were not colored.These results suggest the important role of the process occurring after the visual input.

The ”Anchoring Deficit” hypothesis (Ahissar et al., Nat Neurosci. 2006) proposed that Dyslexics have a difficultyin automatic extraction of simple stimulus regularities in sound sequences. JaffeDax et al. (J Neurosci. 2015) modelled thesedifficulties as yielding noisy priors.The current study was aimed to assess the impact of long-term regularities in language, which listeners had life long experi-ence with. Our assumption was that this familiarity would enhance Controls’ performance more than Dyslexics’ due to a noisierprior formation among Dyslexics. This question was addressed in a series of experiments - in each there was one condition forwhich information accumulated over the life span could be utilized.In all three experiments Dyslexics did not benefit as much as Controls from the long term statistics associated with the input.These results suggest that Dyslexics could not compensate for the deficit despite multiple exposures to lingual input with thesame statistics.

Time is an abstract concept that is better understood when it is mapped onto space. One mechanism to accomplishthis mapping is a reference frame. Previous research has shown the orientation and direction parameters of a spatial referenceframe are involved in understanding time. For example, for English speakers, time is organized horizontally and runs from left(past) to right (future). The current experiments focus on the scale parameter. Experiment 1 changes temporal scale across trials,and illustrates that the scale parameter is set, as evidenced by a cost when the parameter value changes. Experiment 2 examinesthe correspondence between the spatial scale and the temporal scale, requiring participants to map small or large temporaldistances to small or large spatial distances. The results illustrate flexibility in this mapping. Together these experimentssupport the idea that all the parameters of a spatial reference frame are used when understanding time.

We evolve simple neural network controllers in swimming robots in order to test the hypothesis that, given distinctdimensions of control for the tail structure, evolution will favor the emergence of modular neural networks as most likely toenhance fitness (successful light harvesting). Evolution does lead to improved fitness, but this does not appear to result fromincreases in modularity. However, an unexpected result highlights the importance of embodiment for the evolution of the agent.The output of the neural network controller is high frequency with many extreme excursions, but the actual movements of thetail are damped by the physics of the body as it interacts with the aquatic environment. Subsequent simulations establish therole of these physical parameters in dampening noisy network controller output. Thus, morphology can increase evolvabilityby acting as a low pass filter of high-frequency controller dynamics.

Young children acquire informal math ideas through everyday experiences. Research demonstrates that storybookscan link their informal notions to abstract concepts (Murphy, 1999). The integration of visual and written depictions of math-ematics, along with conversations arising from the story, can bolster children’s capacity to communicate and think mathemati-cally. Despite the growing literature supporting use of math storybooks, little is understood about how educators perceive andinterpret the embedded math content. Our study presents storybook pages to educators and asks questions to determine whetherthey are more likely to identify concepts in math (e.g., number) or non-math (e.g., vocabulary) domains. We also analyze theassociation between domain and the degree of abstraction in the language used to describe content in that domain. Preliminarydata suggest that educators ask questions about number concepts at higher levels than expected and character’s actions at lowerlevels than expected.

When acting in a social context, people have an automatic tendency to represent another person’s task – to theextent that another’s task constraints may influence one’s own movement performance. Task co-representation will also affectco-actors’ performance in joint action coordination; however, how exactly movement parameters are influenced is unclear. Weinvestigated this question in four experiments. Pairs of participants performed arm movements back and forth between twotargets, instructed to synchronize their landing times while external metronome tones provided timing cues. We predicted thatactors would represent their co-actors’ task constraints such that when the co-actor moved over an obstacle the actor withoutobstacle would move higher as well. Results confirmed this prediction, suggesting that joint action partners co-representedeach other’s task constraints. Moreover, this obstacle effect increased significantly when timing cues were removed, indicatinga stronger need for co-representing the partner when demands on interpersonal coordination are amplified.

In this study we attempted to predict individual participants single trial behavior (response and reaction time) on anon-symbolic number comparison task. Experimental sessions included the completion of the number comparison task alongwith concurrent EEG measures. We then used a dynamic systems model with evolutionary algorithm updating to predict be-havior for each participant independently. The computational model approximated neural coding of number by calculatingtuning curves implemented through multilayered dynamic systems architecture. Typically dynamical systems models of cogni-tion have fixed parameters tailored to the particular task being modeled and selected by the researcher. The models used weredesigned to adapt such that each participant’s model is individually customized to their particular data. Average ERP amplitudeacross occipitoparietal areas were used as model input in addition to participant’s prior responses and reaction time.

The ability to perceive and understand experience is influenced by a process that divides it into meaningful parts.This process, called event segmentation, is frequently studied by asking participants to identify meaningful units of activity infilms that depict it from a third person perspective. However, because segmentation is associated with changes in the perceptualand conceptual features of film, it could differ for films that present the same activity from a first person perspective. This studytherefore examined segmentation for simultaneously recorded films that depicted identical activities from different perspectives.In several experiments participants were asked to segment these movies into natural and meaningful units of activity. Measuresof segmentation frequency and agreement provided little evidence that segmentation reliably differed across first and thirdperson perspectives. These preliminary findings suggest that participants identify similar events when they are viewed fromdifferent perspectives.

The question of whether language affects nonlinguistic processes remains unresolved. Whereas many studies findthat effects of language on such processes are disrupted when verbalization is inhibited, others show that they persist. Weexplored individual differences in the tendency to verbalize as a potential resolution to this discrepancy. We hypothesized thatif language is spontaneously accessed during nonlinguistic tasks, individual differences in verbalization should predict taskperformance. Participants completed a visual change-detection task and the Verbalizer-Visualizer Questionnaire (VVQ), a self-report measure of cognitive styles linked to modality-specific neural systems. We found that higher scores on the “verbalizer”dimension of the VVQ predicted faster but less accurate change detection. These results suggest that some individuals aremore likely than others to use language when performing tasks that do not require it, and hence that effects of language onnonlinguistic processes are more likely to be observed in such individuals.

Subliminal information has been suggested to facilitate insight problem solving. The present study examinedwhether subliminal hints may influence not only retention of the goals but also planning of sequential steps required to solve thenine-dot problem. Using continuous flash suppression, participants were subliminally presented with either an image depictingthe entire solution of the problem, the three steps to solve the problem in a sequential order, or an image of the nine dots thatdoes not involve solution of the problem. Participants presented with the entire solution of the problem tended to show bettersolution performance and greater relaxation of constraints than those in the latter two conditions, whose performance failedto significantly differ from each other. These results indicate that subliminal information may be stored as a global and staticvisuo-spatial representation to influence solution but may not involve planning of each step to achieve insight problem solving.

When we cannot concentrate on reading a book, we have problems with voluntary attention. When we stand up andstumble on a chair leg, we have a problem with involuntary attention. A bimanual Stroop task (ST) and the Ericson’s flankertask (EFT) were used for the analysis of voluntary and involuntary attention, respectively. Electrophysiological markers ofattention were applied in adult ADHD patients and yoked control individuals. Behavioral incongruence effects were strongerin patients than in controls in the EFT. P3 latency in the incongruent condition was identical in patients and controls in ST butstrongly delayed in patients compared with controls in EFT. A significant positive lateralized readiness potential indicating theactivation of the false response channel was obtained in the incongruent condition of EFT, being significantly larger in patientsthan controls. The data indicate a disorder of automatic attentional control in ADHD adults despite nearly normal voluntaryattention.

Children’s spatial language abilities relate to their spatial skills. We propose that this relation arises from attentionto spatial dimensions influencing both spatial word and spatial skill acquisition. This study tests whether attending to spatialdimensions in a word learning task predicts spatial vocabulary. Three to 5-year-olds completed a novel word assessment testingcategorization of angles, shapes, and a test of spatial vocabulary. In the novel word assessment, children were presented withan exemplar angle with a novel label and asked to select another angle sharing the label. Foils matched the exemplar in degree,orientation, color, or size. Significant age differences occurred in children’s bias to select foils based on angle degree (butno age differences occurred in exemplar choices based on shape). Children showing an angle bias had significantly higherspatial vocabulary than those who did not. These findings show that attending to relevant spatial dimensions predicts spatialvocabulary.

Scrabble-like tasks have increased in popularity as a means of exploring cognitive phenomena, such as embodiedproblem solving, mastery, and creativity. Many of these tasks make assumptions about the key factors driving relative difficultyof word-finding problems; these factors include average frequency of words produced, number of words produced, and numberof readily apparent bi-grams in the initial presentation of the letters. This study measures the effects of each of these factors oncognitive load by systematically and empirically exploring such factors, comparing how these various attributes influence thenumber of words participants produce in different circumstances.

This project examined whether the cultural and linguistic experiences of English and Chinese speakers can result indifferent metaphorical representations of emotion in those individuals. The Brief Implicit Association Test (BIAT) techniquewas used to measure how strongly various colors are associated with anger, sadness, happiness, fear, envy, shame and shyness.The results showed that some metaphorical associations like red-anger are common in both English and Chinese speakers,whereas other associations are culturally-specific (e.g., red is also associated with happiness in Chinese, while only Englishindividuals associate blue with sadness). Some interesting gender differences were also obtained, such that Chinese femalesassociate shyness with pink, but males with red. Black was associated with fear in both genders in Chinese, but only presentin English males. This study thus demonstrates that the conceptual representations of different emotions are shaped by anindividual’s linguistic and cultural experience.

Feature distinctiveness is a measure representing the uniqueness of objects’ features. Previous research found linksbetween noun feature distinctiveness and age of acquisition (i.e. nouns referring to objects with relatively unique featuresare learned earlier). The present work investigates the links between feature distinctiveness and age of acquisition in verbs.Using high-dimensional vector space modelling, noun and verb feature distinctiveness was represented as Manhattan distancebetween word nodes. Both nouns and verbs showed negative correlations between feature distinctiveness and age of acquisition(words of more distinctive objects learned earlier), suggesting a general distinctiveness bias. This effect was stronger for nouns.An investigation of child directed speech (CDS) from the CHILDES corpus showed a correlation between child directed wordfrequency and feature distinctiveness for nouns (featurally distinctive nouns are more common in CDS), but not for verbs. Thepossible link between distinctiveness in CDS and age of acquisition effects is discussed.

We examined how grammatical aspect and visual perspective taking (first- or third-person) influence the abilityto imagine accomplishments. Our main prediction was that it would be easier to imagine completed (I had built the fence.)than ongoing events (I was building the fence.) because accomplishments include natural temporal end points. Slow corticalbrain potentials were examined as an index of the difficulty associated with imagining. Our results showed that participantshad more difficulty imagining ongoing than completed accomplishments, and that it was easier to imagine from the third-versus first-person perspective. The ability of participants to imagine from different visual perspectives was not influencedby grammatical aspect. Participants indicated that the people in their imagined events were more vivid when they imaginedongoing versus completed accomplishments, as well as when they imagined from a third- versus first-person perspective. Asexpected, grammatical aspect influenced which temporal components of events were imagined.

Learning about fractions is a critical step on the path to high school mathematics, yet many children never masterbasic knowledge such as fraction arithmetic procedures. To better understand these difficulties, the present study describes acomputational model of fraction arithmetic problem solving. The model demonstrates that the majority of empirically observederrors over all four arithmetic operations can be explained by only two error-generating mechanisms – overgeneralization andrepair. Further, by assuming probabilistic selection of solution procedures using associative strengths learned from experience,the model predicts two other empirical phenomena: (1) variation in error rates and relative frequencies of specific errorsas a function of problem features, and (2) variable strategy selection within and between problems and individuals. Beyondproviding a formal account of errors, the model was used to simulate the effects of variation of instructional parameters, leadingto novel predictions regarding potentially effective instructional designs.

When planning the most efficient route from one location to another, people tend to prefer southern routes overnorthern routes of equal length and complexity. This asymmetry has been attributed to implicit associations between cardinaldirection and relative elevation (i.e., north = higher), and holds even when regional topography conflicts with these associations.No such asymmetry has been observed between eastern and western routes. Here we provide evidence for an eastern-westernasymmetry in participants residing in an environment with east-west topography differences. Residents of Colorado Springs,CO, where topography is mountainous to the west, showed a reliable preference for eastern routes over equal-length westernroutes on a Colorado Springs map, but not an unfamiliar map. This pattern held even though the represented areas contain min-imal elevation differences. Our findings suggest that regional topography can induce a novel, physically unfounded asymmetryin otherwise unbiased representations of the spatial environment.

Much research indicates that concepts of magnitude and valence are represented spatially, with more/less and posi-tive/negative relations mapped to vertical and horizontal axes. While these mappings are sometimes manifested linguisticallythrough conventional metaphors (e.g., ”prices fell”), recent evidence suggests that they may be built into the very forms ofwords – traditionally assumed to be arbitrarily related to their meanings. Following previous research, we examined whetherthe directions of hand motions constituting words in two sign languages predicted the meanings of their English translationequivalents. Upward-moving signs were more positively valenced than downward-moving signs, as found previously, but werealso greater in magnitude, or intensity. Additionally, rightward-moving signs (from the signer’s perspective) were more posi-tively valenced than leftward-moving signs, consistent with the bodily experience of right-handers. Our findings demonstratesystematic encoding of multiple spatial-conceptual mappings in words, adding to the growing literature showing non-arbitrarylinks between linguistic form and meaning.

Abstract: Recent research has suggested informant trust is an important factor in preschoolers’ observational learn-ing. This poster will present data from an ongoing study examining if 3.5- to 6.5-year-old children (current n=24) relateperceptions of effort and trust. Children watched two informants solving problems using different solutions, exerting eitherhigh or low effort. Children’s eye gaze, trust of each informant, and learning from informants were measured. There were nosignificant differences in trust of the two informants, but children were significantly more likely to learn the solution demon-strated by the high effort informant, t(23) = 2.161, p = 0.041. High effort informant trust was also significantly related totime spent looking at the high effort informant, r = 0.675, p < 0.01. These findings indicate children are more likely to watchinformants who exert high effort and are more likely to use those solutions when faced with a novel problem.

The majority of task-switching research has focused on shifting attention between multiple tasks in the same percep-tual modality (i.e., visual) within a single task domain. However, typical environments are not unisensory, and typical responsedecisions often involve multiple task domains. This study examines multisensory task-switching costs and the interactions ofseveral variables, including perceptual modality of the cue, perceptual modality of the target task, type of task completed (i.e.,spatial or identity decisions), and availability of foreknowledge. The design is marked by no redundant multisensory infor-mation and minimal memory demands. Performance costs varied as a function of whether participants had foreknowledge ofupcoming task and/or modality presentation. Consistent with previous research, the current results also show that performancecosts between tasks were significantly smaller (and essentially, eliminated) when the sensory modality of the task switchedversus when it repeated. However, this result was contingent on manipulations of the experimental design.

Learning facts without first considering what they could be can lead to hindsight bias. How might this impact (a)memory and (b) understanding of novel topics? Foresight participants read about five psychology studies, including meanperformance of one group; then they estimated the mean performance of another group and stated causes for the difference;finally, they received the second group’s actual performance. Hindsight participants learned about both groups’ performance atthe beginning, then imagined what estimates and causes they would have indicated had they not seen actual means. A weeklater, half of each group recalled the means they had learned, and other half estimated means for a novel set of studies. Weconsidered the extent to which: 1. foresight promotes long-term memories as opposed to providing an anchor that biasesmemories; 2. foresight cultivates a habit of considering alternative possible outcomes that might help one understand noveltopics.

Building and validating models of skill acquisition that explain speedup effects has been limited by difficulty dis-tinguishing quickly executed cognitive processes (e.g. Anderson, 1982; Logan, 1988; Rickard, 1997). In this experiment,magnetoencephalography (MEG) data are collected from participants solving a repeated math problem set. We use MEG signalto test the three-phase model of skill acquisition that describes the transition from problem-solving strategies of computation,to retrieval, to an automatic stimulus-response process (Fitts & Posner, 1967). We hypothesize that the processes of familiarityand recollection are early features that distinguish the three phases of skill acquisition. Analyzing event-related fields, we testtwo predictions. First, early frontal activation (akin to the FN400 old-new effect of ERP studies) should diminish in strengthwith each successive phase transition. Second, parietal activation (corresponding to the ERP P600 old-new effect) should bepresent in the second phase, but not in the first or last phase.

From children’s parties to acts in Las Vegas, magic is one of the world’s most timeless forms of entertainment.Current psychological research on magic has started to focus on how magicians are best able to elicit the observer reactionsassociated with their craft, such as what methods are most successful, as well as what cognitive mechanisms are specificallydriving the observers’ reactions. However, while research examining the practice of magic from a psychological perspectivehas been expanding, few studies have looked at how cognitive individual differences influence an observer’s magic perceptionsand experiences. In a collaboration with award-winning magician, Joshua Jay, we examined the impacts of Need for Cognition(NFC) and Need for Cognitive Closure (NFCC) on magic perceptions. Results showed that NFC and NFCC had opposite effectson engagement (i.e., rewatching and solution generation) and that frustration levels were behavior drivers for participants withhigh NFC or low NFCC.

We examined differences between the processing of inflectional versus derivational morphology, using Greek nounsand verbs with a primed lexical decision task. Previous work suggested that both noun and verb targets were significantlyprimed by the same grammatical class. However, when preceded by different grammatical class, verb but not noun targetsshowed priming. We attributed the asymmetrical priming to the materials used: noun stimuli were derived by their verbcounterparts, suggesting an important inherent asymmetry between nouns and verbs. To further investigate this suggestion,we used materials with the opposite asymmetry (verbs derived by nouns) expecting an asymmetry in the opposite direction toemerge for derivationally related words. A clear explanation of the asymmetry would allow us conclusions about the (debated)existence of differences in representation and processing between inflectional and derivational morphological relations and thusprovide evidence for or against a fully decompositional view of processing morphologically complex words.

The ability to acquire non-linearly separable (NLS) classifications is well documented in the study of human categorylearning. In particular, one experiment (Medin & Schwanenflugel, 1981; E4) is viewed as the canonical demonstration that,when within- and between- category similarities are evenly matched, NLS classifications are not more difficult to acquire thanlinearly separable ones. The results of this study are somewhat at issue due to non-standard methodology and small samplesize. We present a replication and extension of this classic experiment. We did not find any evidence of an advantage forlinearly separable classifications. In fact, the marginal NLS advantage observed in the original study was strengthened: wefound a significant advantage for the NLS classification. These results are discussed with respect to accounts provided byformal models of human classification learning.

Individuals attribute more psychological (e.g., forget) than biological (e.g., eat) or physical (e.g., be touched) prop-erties to supernatural beings (Shtulman, 2008). It is unclear how those domains each contribute to an overall conception of asupernatural being (e.g., God). Undergraduate students (N = 341) responded to nine questions representing the three domainsor factors (psychological, biological, and physical), composing an overall measure of God’s anthropomorphic properties.A confirmatory factor analysis was performed to assess the structure of undergraduates’ anthropomorphic concept of God. Fitindices suggest acceptable model fit, χ2(24) = 73.09, p < .001, CFI = 0.952, SRMR = .051. All loadings were significant. Bi-ological (0.99; 0.01) and physical (0.90; 0.19) factors loaded more strongly onto anthropomorphism, and had smaller variances,than the psychological (0.67; 0.56) factor. These findings suggest there are varied ways of conceptualizing the psychological(versus non-psychological) properties of God; thus, non-psychological properties are more predictive in God concepts.

Studying infants in daycare or school settings enables us to ask whether infants influence each others’ development,and if so, whether peer influences are similar to influences from adult caregivers. Answering these questions will not onlyinform infant educators but can also help us understand the mechanisms underlying infant learning. We have collected audiorecordings from 21 1- to 21-month-old infants in two infant rooms in our campus early childhood education center. Recordingstook place nearly every school day over a continuous 8 month period, for an average of 231 hours of recording per child.Multiple infants within the same room were recorded simultaneously. We will present our approach to synchronizing, coding,and analyzing these recordings toward the goal of understanding peer influences on vocal communication development, presentpreliminary results, and seek input on how to further analyze this large and unique dataset.

Many event descriptions are true only when the event comes to its natural end point: e.g., a “feeding” event culmi-nates when the feed-ee has eaten, not simply when food is provided. Do non-linguistic event conceptualizations reflect attentionto natural culmination points? We tested adults and 14-month-olds to ask: provided two events with the same ACTION butdifferent ENDPOINTs - one a naturally expected result, the other only partially achieved - do adults and infants perceive themas members of the same event category or of different categories? Adults were asked to rate the similarity between the twoevents; infants were habituated to one event and tested for dishabituation when it was switched to the other. Adult data suggestthe difference between a complete and a partially-complete event is registered, and carries more psychological weight than amere perceptual difference. Infant data (ongoing) will show the developmental origin of such conceptualizations.