Abstract
Evidence indicates that faces are rapidly detected and prioritized in visual processing due to their social relevance. Crucially, research has shown that faces capture attention even when they are task-irrelevant, suggesting the existence of an automatic, domain-specific attentional mechanism. This prioritization appears to extend to face pareidolia-the illusory perception of faces in objects-indicating that face detection may rely on a broadly tuned mechanism that classifies stimuli as faces or non-faces based on relatively minimal information, such as the T-shaped configuration common to all faces. To explore whether such a configuration is sufficient to automatically capture attention, we conducted three behavioral experiments using a visual search paradigm. In Experiments 1A (online) and 1B (lab-based), participants searched for a butterfly target while either a real or an illusory face appeared among object distractors. Contrary to expectations, neither real nor illusory faces captured attention. In Experiment 2, in different blocks, participants were asked to search for a real face, an illusory face, or a butterfly presented among object distractors. The results showed that real faces were detected more efficiently than illusory faces or butterfly targets. Crucially, illusory faces showed a search disadvantage even compared to butterflies. Overall, these findings challenge the assumption that a minimal face configuration automatically captures attention. More importantly, these findings challenge the notion that attentional capture by irrelevant faces is entirely automatic, suggesting instead that such capture may be influenced by contextual factors or task demands.