The expectation-violation effect: Paradoxical effects of semantic relatedness

Abstract

This paper reports an experimental effect, the expectation-violation effect, in which items from weakly related pairs are better recalled than items from strongly related pairs. This effect occurs when strength of relation is defined from associative norms. The experiments presented below define important boundary conditions for the expectation-violation effect. Type of Memory Test, Type of Experimental Design, and Number of Weakly Related Pairs in the Study List all affect whether the expectation-violation effect will occur. The theoretical explanation of the expectation-violation effect presented herein claims that a failure to understand the relation between the items in a word pair can improve memory performance on that word pair. These failures, which are called blind-alley searches, occur when the items in word pairs represent unexpected or novel semantic combinations. The blind-alley search results in a memory representation, a blind-alley search cue, which can mediate the later retrieval of the word pairs on which the blind-alley search is committed. The expectation-violation effect occurs because weakly related pairs are more likely to represent unexpected or novel semantic combinations than are strongly related pairs. Subjects are thus more likely to commit blind-alley searches, with their attendant memory benefits, on weakly related pairs than on strongly related pairs.