A Spiking Neuron Model of Word Associations for the Remote Associates Test

Abstract

Generating associations is important for cognitive tasks including language acquisition and creative problem solving. It remains an open question how the brain represents and processes associations. The Remote Associates Test (RAT) is a task, originally used in creativity research, that is heavily dependent on generating associations in a search for the solutions to individual RAT problems. In this work we present a model that solves the test. Compared to earlier modeling work on the RAT, our hybrid (i.e., non-developmental) model is implemented in a spiking neural network by means of the Neural Engineering Framework (NEF), demonstrating that it is possible for spiking neurons to be organized to store the employed representations and to manipulate them. In particular, the model shows that distributed representations can support sophisticated linguistic processing. The model was validated on human behavioral data including the typical length of response sequences and similarity relationships in produced responses. These data suggest two cognitive processes that are involved in solving the RAT: one process generates potential responses and a second process filters the responses.

On the Rational Boundedness of Cognitive Control: Shared Versus Separated Representations

“…Kool & Botvinick, 2018;Kurzban et al, 2013;Shenhav et al, 2013Shenhav et al, , 2017, as well as in theories of planning and decision making (Callaway et al, 2018;Kool, Gershman, & Cushman, 2017;Lieder et al, 2018). We also consider how the tradeoff between shared and separated representations may help provide a unified understanding of a wide range of psychological phenomena that, to date, have been treated largely as distinct from one another -including the role of "chunking" in skill acquisition (G. A. Miller, 1956; Servan-Schreiber & Anderson, 1990), interference in working memory (Bouchacourt & Buschman, 2019;Usher & Cohen, 1999;Wilken & Ma, 2004), attention in "binding" (Treisman, 1996(Treisman, , 1999Treisman & Gelade, 1980), facilitation in creativity (Kajić, Gosmann, Stewart, Wennekers, & Eliasmith, 2017;Schatz, Jones, & Laird, 2018), and the tradeoff between pattern separation vs. pattern completion in episodic vs. semantic memory (McClelland, McNaughton, & O'Reilly, 1995) -and discuss its relationship to similar principles that have begun to emerge from machine learning, such as the bias-variance tradeoff and regularization.…”

“…"sick"). Performance on this task has been interpreted in terms of a semantic graph, in which nodes represent individual words and the edges between nodes represent the semantic association between them (Kajić et al, 2017;Schatz et al, 2018). The ability to retrieve the solution word is assumed to depend on how effectively activity spreads from nodes representing the cue words to the node representing the solution word and, in particular, to ones that are not directly connected.…”

“…In such networks, concepts are generally represented as distributed patterns of activity rather than discretely as individual nodes. However, the methods we described for constructing a bipartite graph from a neural network (see Section "Graph-Theoretic Analyses") could, in principle, be used to construct a semantic graph from semantic neural networks (e.g., Hinton et al, 1986;Kajić et al, 2017;Schatz et al, 2018; T. T. Rogers & McClelland, 2004); and, from that, to construct an interference graph that could be used to determine functional dependence -that is, the prevalence of indirect sharing that could be used for inference. That, in turn, could used to predict scores in the RAT, providing a bridge from detailed, process models of semantic cognition to measures of associative abilities and creativity.…”

A Systematic Review of Creativity-Related Studies Applying the Remote Associates Test From 2000 to 2019

“…Out of the 172 researches, ten studies explored the remote associative process. These studies mainly focused on how memory affects remote association, such as semantic search, memory retrieval, and prior knowledge (Davelaar, 2015;Klein and Badia, 2015;Kajić et al, 2017), and on aspects of the brain mechanisms underlying the formation of remote associations, such as brain networks, brain structure, brain function, and brain waves (Wu et al, 2016;Bendetowicz et al, 2017Bendetowicz et al, , 2018Di et al, 2018;Pick and Lavidor, 2019;Zhou et al, 2019). In addition, some studies examined the effects of priming on remote association (Sassenberg et al, 2017).…”

Psychometric Evaluation of an Alternate Scoring for the Remote Associates Test

“…Most recently, the information processing underlying RAT performance has been studied. Different computational models of the RAT have been developed whose performances predict human performance on the RAT (Gupta, Jang, Mednick & Huber, 2012;Kaji c, Gosmann, Stewart, Wennekers & Eliasmith, 2017;Oltet ßeanu & Falomir, 2015). These computational models help to identify factors which determine item difficulty for human participants.…”