An aim of research on spoken word recognition is to characterize the influence of various lexical characteristics(e.g., word frequency, neighborhood size) on lexical access. Dynamics can be coarsely estimated from single-point measureslike naming or more directly assessed using time course measures like fixation proportions over time in the visual worldparadigm (e.g., Tanenhaus et al., 1995). We propose that cumulative response probabilities (CRPs) over time may allow anew characterization of the activation dynamics of lexical access from single-point measures. We assume that the timing ofresponses in a naming task reflects probabilistic sampling of underlying continuous activation dynamics that can be recoveredby CRPs. We applied CRP analyses to visual word recognition data collected for 40,481 words from 472 participants (theEnglish Lexicon Project; Balota et al., 2007) and report initial efforts to validate this new approach.

Models in cognitive science often postulate that individuals maintain complex representations of their environment when simpler explanations, based on simple behaviors interacting with each other and environmental constraints, would suffice. As an example, I consider representational approaches to animal behavior (e.g., Gallistel, 1990; Myerson and Miezin, 1980), which posit that complex group behavior results from complex representations of events within the central nervous systems of individual animals. For example, ducks feeding from two food sources distribute themselves proportionately to the density of food available at each source. This phenomenon, probability matching, is typically explained by attributing representations of the density of food available at each source within the central nervous system (CNS) of each duck. Are such complex representations required to explain this phenomenon? I will compare the results of two simulations of probability matching in groups. In one, individuals maintain and update representations of food available at each source. Although probability matching emerges, the organisms exhibit various unrealistic behaviors. In the second, each individual follows simple behavioral rules but has no representation of the food density at each source. Probability matching emerges and the behavior observed is more realistic than that in the first simulation. This adds to demonstrations in other domains that complexity at one level of analysis need not result from complexity at lower levels (e.g., Resnick, 1994; Sigmund, 1993).

The Time-Invariant String Kernel (TISK) model of spokenword recognition (Hanngan et al., 2013) is an interactiveactivation model like TRACE (McClelland & Elman, 1986).However, it uses orders of magnitude fewer nodes andconnections because it replaces TRACE's time-specificduplicates of phoneme and word nodes with time-invariantnodes based on a string kernel representation (essentially aphoneme-by-phoneme matrix, where a word is encoded as byall ordered open diphones it contains; e.g., cat has /kæ/, /æt/,and /kt/). Hannagan et al. (2013) showed that TISK behavessimilarly to TRACE in the time course of phonologicalcompetition and even word-specific recognition times.However, the original implementation did not includefeedback from words to diphone nodes, precluding simulationof top-down effects. Here, we demonstrate that TISK can beeasily adapted to lexical feedback, affording simulation oftop-down effects as well as allowing the model todemonstrate graceful degradation given noisy inputs.

In visual word recognition, having more orthographicneighbors (words that differ by a single letter) generallyspeeds access to a target word. But neighbors can mismatch atany letter position. In light of evidence that informationcontent varies between letter positions, we consider howneighbor effects might vary across letter positions. Resultsfrom a word naming task indicate that response latencies arebetter predicted by the relative number of positional friendsand enemies (respectively, neighbors that match the target at agiven letter position and those that mismatch) at some letterpositions than at others. In particular, benefits from friendsare most pronounced at positions associated with low a prioriuncertainty (positional entropy). We consider how theseresults relate to previous accounts of position-specific effectsand how such effects might emerge in serial and parallelprocessing systems.

We examined how phonological competition effects inspoken word recognition change with word length. Cohorteffects (competition between words that overlap at onset) arestrong and easily replicated. Rhyme effects (competitionbetween words that mismatch at onset) are weaker, emergelater in the time course of spoken word recognition, and aremore difficult to replicate. We conducted a simple experimentto examine cohort and rhyme competition using monosyllabicvs. bisyllabic words. Degree of competition was predicted byproportion of phonological overlap. Longer rhymes, withgreater overlap in both number and proportion of sharedphonemes, compete more strongly (e.g., kettle-medal [0.8overlap] vs. cat-mat [0.67 overlap]). In contrast, long andshort cohort pairs constrained to have constant (2-phoneme)overlap vary in proportion of overlap. Longer cohort pairs(e.g., camera-candle) have lower proportion of overlap (inthis example, 0.33) than shorter cohorts (e.g., cat-can, with0.67 overlap) and compete more weakly. This finding hasmethodological implications (rhyme effects are less likely tobe observed with shorter words, while cohort effects arediminished for longer words), but also theoreticalimplications: degree of competition is not a simple functionof overlapping phonemes; degree of competition isconditioned on proportion of overlap. Simulations withTRACE help explicate how this result might emerge.

What determines degree of competition among phonologicallysimilar words? One proposal is that proportion of overlappredicts competition independently of word length. We arguethat proportion of overlap may provide descriptive adequacy,but does not provide an explanation. We show that TRACEcorrectly predicts patterns previously attributed to proportionof overlap. In additional simulations, with independentmanipulations of word length and proportion of overlap,proportion of overlap fails to predict the full pattern of results.We discuss how competition dynamics in TRACE modulatecompetition as word length and proportion of overlap change.These results have implications for theories of human spokenword recognition, and will motivate experiments to test thesenew TRACE predictions.