Evidence for Recursive Operations in Human Cognition

Abstract

Recursion has held a prominent place in the cognitive sciences since Noam Chomsky imported the notion from mathematical logic in the 1950s. Despite receiving widespread attention during the ensuing decades, the status of recursive operations—that is, operations that effect a self-call—in cognition remains obscure, in part owing to confusion about the relationship between recursive operations and recursive structures. Though often taken to play a large and far-reaching role in human cognition, such ubiquity cannot simply be assumed and must instead be argued for or demonstrated empirically. Most argumentation and demonstration to this end has followed a common line, which will be described and evaluated in this thesis. Following this, I will propose a novel signature of recursive operations and use behavioral paradigms to detect its presence: First, changes in working memory load during performance of a puzzle task will be shown to follow a broad pattern that would be expected if recursive operations underlie task performance. Then, a minimally-complex recursive computational model will be demonstrated to predict changes in human memory on a fine-grained, trial-by-trial basis using the same task. Finally, a non-recursive model featuring a push-down stack architecture will be shown to be inferior to the recursive model at predicting human memory dynamics. These results signify the first evidence for the existence of recursive operations in human cognition.