Abstract

This study evaluates the contribution of conditional relationship contents to making inferences, depending on the span of the participant's working memory. This is achieved through the use of a deductive reasoning task with conditional statements and the "Reading Span Test" by Daneman and Carpenter, in its Spanish version. The results of the two experiments show that the execution of an inference task is mediated by participants' working memory span. WM is correlated with the most complex valid inferences, but not with the most automatic ones. Contents and the number of alternatives to the antecedent only had an effect on the inference of the high-WM group. The results are relevant in several ways, and can help us to refine our predictions about inferences.