Attention bias for facial expressions of emotion: A meta-analytic review

Facial expressions of emotion are critical to survival and social interaction. Their importance is underscored by evolutionary adaptations that enable their automatic production and recognition. As a result, emotional faces may receive attentional prioritization, even when completely irrelevant to the task at hand. Although attentional bias is theoretically plausible, empirical findings have been inconsistent: Some studies have reported bias toward emotional faces, whereas many others have not. To clarify this discrepancy, we conducted a meta-analysis of attentional bias for task-irrelevant emotional expressions, including studies using the additional singleton and spatial cuing paradigms (which includes dot probe paradigms). We found an overall effect between zero and small (Hedges's g = 0.08), based on 160 cases. The only significant moderator was the data set from which the emotional face stimuli were drawn, with the Gur data set (Gur et al., 2002) producing the strongest bias. In a second meta-analysis, we examined studies where the emotional expression was task relevant because both the expression and the target were singletons. Here, the overall attentional bias was small to medium (g = 0.41), based on 25 cases. We conclude that facial expressions of emotion do not bias attention when they are task irrelevant. In the discussion, we highlight some empirical and theoretical challenges to emotion automaticity and offer explanations for why the effect was between zero and small. One potential explanation is studies often utilize static photographs of actors portraying facial expressions of emotion, which are low in salience and ecological validity because they lack context. (PsycInfo Database Record (c) 2025 APA, all rights reserved).