Altered working memory capacity for social threat words in high versus low social anxiety

ABSTRACT

Background and objectives: Differences in working memory capacity (WMC) have been suggested in individuals with high levels of social anxiety (SA). Specifically, these individuals may preferentially maintain socially threatening material in working memory.

Design and methods: We adapted the digit span task to a series of word span tasks. We assessed WMC for lists of words that varied in terms of their threat-relatedness, in individuals either high or low in SA.

Results: Experiment 1 revealed reduced WMC for socially threatening words in those with high compared to low SA. Importantly, this relative reduction in WMC was driven by the low SA group showing expanded capacity for socially threatening words relative to neutral or generally threatening words. Furthermore, reductions in WMC for social threat were uniquely predicted by SA, and not by other theoretically related constructs such as state general anxiety, trait general anxiety, or depression. Experiment 2 showed that the semantic similarity of the words within each list was not responsible for the differences in WMC between list type or SA group.

Conclusions: Our findings suggest that individuals high in SA may fail to upregulate WMC for social information due to the activation of, or rumination upon, socially threatening concepts.

KEYWORDS:

• Working memory
• social anxiety
• memory bias
• threat
• semantic similarity

Data availability statement

Data that support the findings of this study are openly available in the Open Science Framework at https://doi.org/10.17605/OSF.IO/RVGSJ.

Disclosure statement

No potential conflict of interest was reported by the authors.

ORCID

Ryan C. Yeung http://orcid.org/0000-0002-6966-6834

Notes

1 A one-way between-subjects ANOVA comparing the mean association coefficients (as calculated by latent semantic analysis) between the three list types (neutral, general threat, social threat) was found to be significant, F(2, 105) = 4.38, MSE = 0.01, p = .02, ${\eta }_p^2$ = .08. Post hoc Tukey tests determined that the social threat word list (M = 0.16, SD = 0.12) was significantly more interrelated (p = .03) than the neutral or general threat lists (M_neutral = 0.10, SD_neutral = 0.10; M_general = 0.10, SD_general = 0.09), which were not significantly different from each other (p > .05).

Additional information

Funding

This work was supported by a Discovery grant from the Natural Sciences and Engineering Research Council of Canada awarded to author MF.