Semantic Relatedness or Arousal: What Drives Enhancement of Memory for Emotional Words?

Abstract

Emotional stimuli are typically better remembered than neutral ones. Two accounts have been proposed for the advantage in memory for lists of words. The first is that it arises due to enhanced arousal, and the second is that it arises due to the inherent semantic relatedness of emotional words. We compared these accounts by examining memory performance for lists of words in which arousal and relatedness were manipulated. In three experiments (n = 418), conducted online, participants were asked to encode lists containing words that were either unrelated, related, or emotionally valenced, followed by free recall tests. List type was manipulated within-subject. Participants also rated the arousal and emotional valence they experienced for each word using a Self-Assessment Manikin (SAM). Latent semantic analysis (in Experiment 1) and free association norms (in Experiment 2) were used to match the semantic relatedness of words in the emotional (negative) and related lists. Free recall of related and negatively valenced word lists was significantly higher than for unrelated lists. In Experiment 3, we used free association norms to create emotional lists of mixed valence (negative and positive), effectively lowering semantic relatedness within the list, but keeping arousal high. Despite lowering relatedness of the emotional word lists, recall of related and emotional word lists was again of similar magnitude, with memory for both significantly higher than for unrelated lists. Importantly, using logistic regression we showed that participants’ arousal ratings predicted recall in all three experiments. Findings suggest that arousal, not semantic relatedness, explains why memory is better for lists of emotional words.