Abstract
Human beings share an extremely efficient biological skill in recognizing faces, with the exception of other-race (OR) faces: the so-called other-race effect (ORE). As reported by Feingold (1914) nearly a century ago, this face recognition impairment is accompanied by the popular belief that “other-race faces all look alike!” Behavioural data and computational models have provided clear evidence that this popular belief is not accounted for by the paucity of physiognomic variations in OR faces, but by a genuine lack of expertise with the latter category. Interestingly, no study has yet provided an explanation for the neural mechanisms underlying this “visual illusion” in faces.
Here we exploited the neural adaptation approach to investigate this question. Visual adaptation reduces activity of a neural population when two successive inputs activate same but not different subpopulations. Electrophysiological studies have shown that only repetition of same identities lead to smaller amplitudes on the face-specific N170 component, revealing sensitivity of the N170 to individual face categorization. However, since different OR faces are perceived as similar, one would predict no OR identity modulations on the N170 in an adaptation paradigm.
We recorded electrophysiological signals of Western Caucasian (WC) and East Asian (EA) observers while presented with two sequentially presented WC or EA faces from same or different identities. The facial expression of the images was changed across repetitions to avoid pixel adaptation effects. Images had normalized colour-space and luminance. In line with previous electrophysiological findings, repetitions of same identities elicited more adaptation compared to different identities for SR faces. As predicted, OR faces elicited the same level of adaptation regardless of face identity for both groups of observers. These results provide a neural explanation for the ‘all look alike‘ effect and provide novel insights in the explanation of the ORE for faces.