Value conditioning modulates visual working memory processes

Learning allows the value of motivationally salient events to become associated with stimuli that predict those events. Here, we asked whether value associations could facilitate visual working memory (WM), and whether such effects would be valence dependent. Our experiment was specifically designed to isolate value-based effects on WM from value-based effects on selective attention that might be expected to bias encoding. In a simple associative learning task, participants learned to associate the color of tinted faces with gaining or losing money or neither. Tinted faces then served as memoranda in a face identity WM task for which previously learned color associations were irrelevant and no monetary outcomes were forthcoming. Memory was best for faces with gain-associated tints, poorest for faces with loss-associated tints, and average for faces with no-outcome-associated tints. Value associated with 1 item in the WM array did not modulate memory for other items in the array. Eye movements when studying faces did not depend on the valence of previously learned color associations, arguing against value-based biases being due to differential encoding. This valence-sensitive value-conditioning effect on WM appears to result from modulation of WM maintenance processes.

A threatening face in the crowd: effects of emotional singletons on visual working memory

Faces with threatening versus positive expressions are better remembered in visual working memory (WM) and are especially effective at capturing attention. We asked how the presence of a single threatening or happy face affects WM for concurrently viewed faces with neutral expressions. If threat captures attention and attention determines WM, then a WM performance cost for neutral faces should be evident. However, if threat boosts processing in an object-specific, noncompetitive manner, then no such costs should be produced. Participants viewed three neutral and one angry or happy face for 2 s. Face recognition was tested 1 s later. Although WM was better for singletons than nonsingletons and better for angry versus happy singletons, WM for neutral faces remained unaffected by either singleton. These results, combined with eye movement and response time analyses, argue against a selective attention account of threat-based benefits to WM and support object-specific enhancement via threat processing.