How working memory capacity constrains the learning of relational concepts


We investigated the way in which working memory (WM) constrains the learning of relational concepts – categories defined by the way objects are assigned to roles in the structure of an underlying relation, and not by objects’ intrinsic features. By applying to a large sample a novel test of concept learning as well as the battery of WM tasks, we found that WM is a strong predictor of the scores on the test, but the WM-learning correlation decreases as the relational complexity of the to-be-learned concepts increases. Such results support those theoretical models of relational learning, which assume that learning of relational concepts (and relations, in general) consumes more WM resources than just the processing of relations which have already been learned.

Back to Table of Contents