Gregory E. Cox

Toward a scalable holographic word-form representation

Phenomena in a variety of verbal tasks—for example, masked priming, lexical decision, and word naming—are typically explained in terms of similarity between word-forms. Despite the apparent commonalities between these sets of phenomena, the representations and similarity measures used to account for them are not often related. To show how this gap might be bridged, we build on the work of Hannagan, Dupoux, and Christophe, Cognitive Science 35:79-118, (2011) to explore several methods of representing visual word-forms using holographic reduced representations and to evaluate them on their ability to account for a wide range of effects in masked form priming, as well as data from lexical decision and word naming. A representation that assumes that word-internal letter groups are encoded relative to word-terminal letter groups is found to predict qualitative patterns in masked priming, as well as lexical decision and naming latencies. We then show how this representation can be integrated with the BEAGLE model of lexical semantics (Jones & Mewhort, Psychological Review 114:1–37, 2007) to enable the model to encompass a wider range of verbal tasks.

An Exemplar-Familiarity Model Predicts Short-Term and Long-Term Probe Recognition across Diverse Forms of Memory Search.

Experiments were conducted to test a modern exemplar-familiarity model on its ability to account for both short-term and long-term probe recognition within the same memory-search paradigm. Also, making connections to the literature on attention and visual search, the model was used to interpret differences in probe-recognition performance across diverse conditions that manipulated relations between targets and foils across trials. Subjects saw lists of from 1 to 16 items followed by a single item recognition probe. In a varied-mapping condition, targets and foils could switch roles across trials; in a consistent-mapping condition, targets and foils never switched roles; and in an all-new condition, on each trial a completely new set of items formed the memory set. In the varied-mapping and all-new conditions, mean correct response times (RTs) and error proportions were curvilinear increasing functions of memory set size, with the RT results closely resembling ones from hybrid visual-memory search experiments reported by Wolfe (2012). In the consistent-mapping condition, new-probe RTs were invariant with set size, whereas old-probe RTs increased slightly with increasing study-test lag. With appropriate choice of psychologically interpretable free parameters, the model accounted well for the complete set of results. The work provides support for the hypothesis that a common set of processes involving exemplar-based familiarity may govern long-term and short-term probe recognition across wide varieties of memory- search conditions.

OrBEAGLE: integrating orthography into a holographic model of the lexicon

Many measures of human verbal behavior deal primarily with semantics (e.g., associative priming, semantic priming). Other measures are tied more closely to orthography (e.g., lexical decision time, visual word-form priming). Semantics and orthography are thus often studied and modeled separately. However, given that concepts must be built upon a foundation of percepts, it seems desirable that models of the human lexicon should mirror this structure. Using a holographic, distributed representation of visual word-forms in BEAGLE, a corpustrained model of semantics and word order, we show that free association data is better explained with the addition of orthographic information. However, we find that orthography plays a minor role in accounting for cue-target strengths in free association data. Thus, it seems that free association is primarily conceptual, relying more on semantic context and word order than word form information.