Attentional allocation to task-irrelevant fearful faces is not automatic: experimental evidence for the conditional hypothesis of emotional selection.

A growing body of research indicates that attentional biases toward emotional stimuli are not automatic, but may depend on the relevance of emotion to the top-down search goals of the observer. To determine whether and how this relevance factor modulates attentional allocation to irrelevant fearful faces, four spatial cueing tasks were designed, in which the goal-relevance of completely task-irrelevant (neutral or fearful) cue faces was systematically manipulated by changing the target defining feature. No attentional capture by cue faces (be they neutral or fearful) was observed when the cue faces were completely goal-irrelevant. When faces - but not facial expressions - were goal-relevant, fearful cue faces captured attention, but so did neutral cue faces to a similar extent. Only when facial expressions were explicitly goal-relevant did we observe a difference between attentional allocation to fearful and neutral cue faces, with larger cueing effects for neutral cue faces in the Neutral task, and for fearful cue faces in the Fearful task. Therefore, rather than automatic, attentional allocation to irrelevant fearful faces proved conditional to the explicit relevance of fearful expressions to top-down search goals.