The Impact of Affective Information on Working Memory: A Pair of Meta-Analytic Reviews of Behavioral and Neuroimaging Evidence

Abstract

Everyday life is defined by goal states that are continuously reprioritized based on available, often affective information. To pursue these goals, individuals need to process and maintain goal-relevant information, while ignoring potentially salient information that distracts resources from these goals. Empirically, this ability has typically been operationalized as working memory (WM) capacity. A growing body of research is investigating the impact of information’s affective salience on WM capacity. In the present review we address this question by exploring the potential differential impact of affective compared with neutral information on WM, and the underlying neural substrates. One-hundred and 65 studies ( N = 7,433) were included in the meta-analysis. Results showed negligible to small ( d̂ = −.07–.20) effects of affective information on behavioral measures of WM in healthy individuals ( n = 4,936) that varied as a function of valence and task-relevance. Heterogeneity analyses were significant, demonstrating the need to identify further study-specific factors and individual differences that moderate affective WM. At the neural level (33 studies; n = 683), processing affective versus neutral material during WM tasks was associated with more frequent recruitment of the vlPFC, the amygdala, and the temporo-occipital cortex. In contrast to healthy individuals, across behavioral studies those suffering from mental health problems ( n = 2,041) showed impaired WM accuracy ( d̂ = −0.21) in the presence of affective material. These findings highlight the importance of integrating behavioral and neural levels of analysis. Finally, these findings suggest that affective WM capacity may be a transdiagnostic mechanism associated with poor mental health.

Public Significance Statement

The behavioral and neuroimaging meta-analyses showed that in psychologically healthy individuals there was limited support for behavioral working memory (WM) performance to be affected by affective information, whereas at the neural level WM in the presence of affective relative to neutral information was associated with differential recruitment of the salience network and the fronto-parietal control network. These findings highlight the importance of combining behavioral and neuroimaging research syntheses. Second, in individuals with mental health problems WM was significantly impaired by affective material. This suggests that WM performance on tasks including affective compared with neutral information may be a sensitive and transdiagnostic cognitive marker of mental health status.