The importance of the eye region for familiar face recognition: Evidence from spatial low-pass filtering and contrast inversion

Abstract

What information is required for effective familiar face recognition? Accurate recognition for contrast chimeras – contrast positive eyes in contrast negative faces – suggests a salience of the eye-region for activating representations relevant to facial recognition. However, event-related
potentials show that contrast chimeras undergo effortful face detection, and inefficient representation repetition priming, compared to unfiltered faces. It was unclear if efficient access to the eye-region alone was insufficient to effectively activate these representations, or if inefficiency in earlier face detection caused knock-on disruption. Evidence for the latter is supported by blur
chimeras – unfiltered eyes in blurred faces – which are efficiently detected and elicit more typical repetition priming. However, no direct within-subjects comparison of blur and contrast chimeras’ electrophysiological correlates has been made to date. Moreover, it has yet to be reliably
established if un-filtering blurred faces’ eye-regions provides a recognition advantage comparable to
contrast chimeras. To address this, this dissertation provided direct behavioural and electrophysiological comparisons between chimeras.

Behavioural experiments showed that the recognition advantage between blur chimeras and contrast chimeras was comparable, unmanipulated eye-regions initiating holistic integration across both manipulations. Electrophysiological comparisons demonstrated that contrast chimeras were
detected less efficiently than blur chimeras, suggesting that knock-on disruption caused delayed repetition priming, although blur chimeras underwent effortful representation activation. While both chimeras elicited identity repetition priming, this was more inefficient than unfiltered faces.
Our results suggest that while detailed information within the eye-region is sufficient to activate recognition-relevant representations, indicating an important role for eye-region cues, efficiently accessed information in the remainder of the face facilitates this process in typical recognition. Despite chimeras inaccurately reflecting real-world face recognition, their comparison provided valuable means to investigate the information required to effectively and efficiently complete the serial stages of recognition, demonstrated here through support of several theoretical accounts of face detection and representation activation.