The effect of anxiety on memory accuracy, response time, and confidence

Abstract

Anxiety is an emotional state that has been associated with negative outcomes on cognitive tasks, as well as low confidence in responses, specifically in the domain of long-term memory. The following series of experiments explored the effect of anxiety on long-term memory performance as well as on memory confidence, the accuracy-confidence relation, and response times. The purpose was to determine how realistic anxious participants are when making confidence judgments compared to low anxious peers and to attempt to improve the memory accuracy and confidence of highly anxious individuals. In Experiment 1, participants encoded words presented visually, followed by an anxiety induction. Those with higher anxiety scores had poorer memory accuracy for target words as well as less confidence in their memory overall, consistent with much of the literature. Response time (RTs) was slower in the high- relative to low-anxious group when making confidence assessments but not memory judgments, suggesting that rather than slowing cognitive processing, high anxiety individuals may be doubting their memory ability, resulting in more time spent appraising their performance. In Experiment 2, I tested my hypothesis that high-anxious participants are particularly slowed only when asked to critically examine and evaluate their memory decisions. The same procedure as in Experiment 1 was used except that I allowed participants an opportunity to switch their memory responses immediately after making an initial classification. Highly anxious participants made more switches in their memory judgments and this decreased their overall memory accuracy. Interestingly, in this experiment, highly anxious participants did not have lower memory confidence than their low-anxious peers, nor longer RTs when making confidence judgments. They did however take longer when deciding whether to switch an answer. Results are in line with the suggestion that high-anxious individuals defer worry until after a memorial decision is made, and that it is in the post-mnemonic stage that high-anxious individuals differ from low-anxious ones. In Experiment 3, I examined anxiety without induction to determine whether differences in memory accuracy and in response time would remain. I also included a manipulation of encoding duration to determine whether longer encoding time would improve highly anxious participant’s memory accuracy and confidence relative to the low-anxious group. Participants encoded words for either 750 or 4000 ms. High-anxious individuals had poorer accuracy and lower confidence compared to low-anxious individuals: specifically for correct memory responses. Longer encoding duration benefited both accuracy and confidence, and there was no differential benefit across groups. RTs to make memory classifications again did not differ between groups yet, as in Experiment 1, high-anxious participants were slower to make confidence judgments, though only for incorrect responses. Results suggest that high-anxious individuals have unrealistically low confidence in their memory, especially when correct, and that allowing additional encoding time does not alleviate the effect. Taken together, this series of experiments shows that individuals with high levels of anxiety take longer to evaluate their memorial decisions, suggesting that they engage in more post-mnemonic evaluation than their low-anxious peers.