Anxious mood and memory

Influenced by Bower (Am. Psychol. 36, 129-148, 1981) and Lang (Anxiety and the Anxiety Disorders, Erlbaum, Hillsdale, N.J., 1985), we tested three hypotheses concerning anxious mood and memory: (1) the mood state dependent hypothesis which states that memory retrieval will be greater when mood at encoding and at recall are the same than when they are different: (2) the encoding mood congruent hypothesis which states that information semantically related to mood at encoding is retrieved more readily than information unrelated to mood at encoding; and (3) the recall mood congruent hypothesis which states that information semantically related to mood at recall is retrieved more readily than information unrelated to mood at recall. We induced anxiety in speech anxious students by informing them that they would be delivering a speech during the experiment. Mood could be either anxious or nonanxious at encoding, recall, both, or neither. Hence, there were four groups: Anxiety-Anxiety, Anxiety-Nonanxiety, Nonanxiety-Anxiety, and Nonanxiety-Nonanxiety. Subjects were asked to rate the self-descriptiveness of anxiety (e.g. NERVOUS) and nonanxiety adjective (e.g. POLITE) during the encoding phase, and to recall them later. Anxious mood was measured by self-report scales and by heart rate. No support was obtained for any of the three hypotheses. However, post-hoc analyses indicated that anxiety words were recalled least often in subjects whose heart rate increased from encoding to recall. This suggests that attention to threat information may diminish in aroused nonclinical subjects.