Role and interaction of featural and configural processing in face recognition

In order to reliably recognize faces in everyday life it is necessary to detect subtle featural and configural differences. In three experiments we investigated the role and interaction of featural and configural processing in face recognition. In Experiment 1a we showed that previously learnt faces could be recognized when they were scrambled into constituent parts, which were previously determined in a free listing experiment (eyes, nose, mouth, chin, forehead, eye-brows, cheeks). This result clearly indicates a role of featural information at a certain level of face processing. In Experiment 1b we determined the blur level that made the scrambled part versions impossible to recognize. This blur level was then applied to whole faces in order to create configural versions that by definition did not contain local featural information. In Experiment 1c we showed that configural versions of previously learned faces could be recognized reliably. In Experiment 2 we replicated these results for familiar face recognition. All experiments provided converging evidence in favor of the view that recognition of familiar and unfamiliar faces relies on both featural and configural information. In Experiment 3 we investigated whether featural and configural processing provide independent routes of recognition or whether they converge to the same recognition units. Repetition priming for celebrity faces was found from scrambled to blurred recognition and vice versa thus favoring the convergent processing hypothesis. All results are explained in terms of an integrative model of recognition of unfamiliar and familiar faces