Purchase this article with an account.
Pan Liu, Johnathan Ong, Hong Xu; Top and bottom half faces influence equally and interact nonlinearly in face-identity adaptation. Journal of Vision 2012;12(9):623. doi: 10.1167/12.9.623.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We have shown that a whole face can produce a significantly larger facial-expression aftereffect than the linear summation of the aftereffects induced by face parts – the mouth alone and a mouth-less face – suggesting that face parts interact nonlinearly in facial-expression adaptation (Xu, et al., 2011). The current study investigates whether and how face parts interact nonlinearly in face-identity adaptation.
We split a real face ("Adam") into two size-equivalent face parts, the top-half and the bottom-half faces. We used these half faces and the whole face as the adaptor in separate conditions. For the test stimuli, the whole "Adam" face was morphed with another identity’s whole face ("Sam"). The proportion of "Adam" varied from 0% (original "Sam") to 100% (original "Adam"). In each trial, observers viewed the adaptor for 4 s, and after a 500 ms inter-stimulus interval viewed a test face for 200 ms. Observers then indicated the perceived face identity of the test face ("Adam" or "Sam") via a key press. They were also tested in a baseline condition without adaptation. The face-identity aftereffects for the three adaptors were measured as shifts of the psychometric curves from the baseline condition.
We found that both top-half and bottom-half faces generated significant face aftereffects of similar magnitude. Moreover, the whole-face adaptation aftereffect is significantly larger than the sum of the aftereffects produced by the top-half face and the bottom-half face. This finding suggests a nonlinear interaction among the two half faces in face-identity adaptation. Such nonlinearity indicates a holistic nature in face identity processing, and is consistent with physiological and computational studies that show high-level areas combine low-level features nonlinearly along the visual hierarchy.
Meeting abstract presented at VSS 2012
This PDF is available to Subscribers Only