Intensive longitudinal data was collected through theconcurrent use of a passive experience sampling (ES)smartphone application collecting objective measures ofexperience, and an ecological momentary assessment (EMA)app collecting self-reported affect. After a week-longretention interval, participants completed a memory testgenerated from paired ES and EMA data. Participants wereasked to select the GPS location at the time of a paired targetevent from four alternatives. Correct retrieval was notpredicted by self-reports grouped by negative valence/higharousal or negative valence/low arousal. Positivevalence/high arousal reported at encoding predicted greaterprobability of incorrect responses. Conversely, positivevalence/low arousal predicted greater probability of correctidentification of target. At retrieval, choice was predicted bydissimilarities in discrete emotions between target anddistractors, suggesting the use of affect as a contextualmechanism.

We investigated whether memories of different time scales(i.e., week, day, hour) are used independently (i.e., indepen-dence of scales). To overcome the limitations of previousstudies that have low ecological validity in selecting the teststimuli, we used experience sampling technology. Participantswore a smartphone around their neck for two weeks, whichwas equipped with an app.that automatically collected time,images, GPS, audio and accelerometry. After a one-week re-tention interval, participants were presented with an image thatwas captured during their data collection phase, and tested ontheir memory of when the event happened (i.e., week, day ofweek, and hour). We find that, in contrast to previous studies,memories of different time scales were not retrieved indepen-dently in everyday life. Additionally, we replicated previouslaboratory findings such as correlations between confidencerating and memory performance, and patterns found betweenvalence rating and memory accuracy.

In a Linear Associative Net (LAN), all input settles to a singlepattern, therefore Anderson, Silverstein, Ritz, and Jones (1977)introduced saturation to force the system to reach othersteady-states in the Brain-State-in-a-Box (BSB). Unfortunately,the BSB is limited in its ability to generalize because itsresponses are restricted to previously stored patterns. We presentsimulations showing how a Dynamic-Eigen-Net (DEN), a LANwith Short-Term Plasticity (STP), overcomes thesingle-response limitation. Critically, a DEN also accommodatesnovel patterns by aligning them with encoded structure. We traina two-slot DEN on a text corpus, and provide an account oflexical decision and judgement-of-grammaticality (JOG) tasksshowing how grammatical bi-grams yield stronger responsesrelative to ungrammatical bi-grams. Finally, we present asimulation showing how a DEN is sensitive to syntacticviolations introduced in novel bi-grams. We propose DENs asassociative nets with greater promise for generalization than theclassic alternatives.

To avoid interference among similar memory traces it is re-quired to form complex memory structures that include mul-tiple components of the event, which helps one to distinguishone event from another. In a laboratory setting, these complexbinding structures have been studied through a paradigm whereone has to form a memory structure that includes two items andthe context together (i.e., three-way binding). However, de-spite the long history of the theoretical concept, its importance,and the existence of the laboratory paradigm, three-way bind-ing structures have only been examined in recall paradigms.Moreover, not all memory models consider the ability to formthree-way binding structures as a default. Therefore, the cur-rent study examined the use and formation of three-way bind-ing structures in an associative recognition paradigm. Resultsprovide evidence that three-way binding structures are usedduring recognition, which implies that it is critical for mem-ory models to properly represent them.

Recognition memory studies have reliably demonstrated the word frequency effect (WFE), where low-frequency words are more accurately recognized than high-frequency words. The context noise account of WFE argues that pre-experimental exposure to stimuli generates interference that compromises high-frequency words more than low-frequency words. Because the representations of the contexts associated with more recent exposures are assumed to overlap more with the representation of the study context, stimuli that have been seen more recently are thought to generate the most interference. We asked participants to log their daily email for two months. Based on the participant’s email corpus, we constructed an individualized study-test recognition memory task to investigate the effect of recency. Results show that recency has a graded effect on recognition memory that extends for at least two months providing support for the context noise account.