Dissociating attentional and working memory load in memory for emotional faces

There is evidence that both general attention load and face-specific working memory load modulate the processing of emotional faces, but the contribution of each remains uncertain. Here, we investigated how these resources modulate the recognition of emotional faces in working memory. In Experiment 1, participants performed a face working memory (WM) task with either one face (low WM load) or three faces (high WM load) at encoding, showing happy, angry, and sad emotions. Participants stated whether a single (neutral) probe face matched identity to one of the faces just encoded. Concurrently, participants engaged in an auditory monitoring task (low vs. high attention load). The results revealed two contrasting effects: (1) increased WM load had a larger detrimental impact on WM for negative compared to happy faces; (2) increased general attentional demands from the auditory task impaired WM for angry and happy faces more than sad faces. In Experiment 2, an additional perceptual face matching task under high and low perceptual and attentional load indicated that the WM findings were not plausibly due to featural similarities between faces, and that the accuracy costs of attentional load found in Experiment 1 were specific to conditions requiring active WM maintenance. These results demonstrate that WM for emotional faces requires both general attention and face-WM resources inter-dependently, and that this differs depending on the emotional expression shown.