Towards a neural-level cognitive architecture: modeling behavior in working memory tasks with neurons

AbstractConstrained by results from classic behavioral experiments we provide a neural-level cognitive architecture for modeling behavior in working memory tasks. We propose a canonical microcircuit that can be used as a building block for working memory, decision making and cognitive control. The controller controls gates to route the flow of information between the working memory and the evidence accumulator and sets parameters of the circuits. We show that this type of cognitive architecture can account for results in behavioral experiments such as judgment of recency, probe recognition and delayed-match-to-sample. In addition, the neural dynamics generated by the cognitive architecture provides a good match with neurophysiological data from rodents and monkeys. For instance, it generates cells tuned to a particular amount of elapsed time (time cells), to a particular position in space (place cells) and to a particular amount of accumulated evidence.

Return to previous page