Purchase this article with an account.
Hugh W. Dennett, Elinor McKone, Mark Edwards, Tirta Susilo; Do face adaptation aftereffects predict face recognition? Evidence from individual differences. Journal of Vision 2011;11(11):581. doi: 10.1167/11.11.581.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Face distortion aftereffects are widely studied, yet there is currently no evidence as to whether these are associated with actual face recognition ability. Indeed, previous studies find developmental prosopagnosics typically show normal-sized face aftereffects. Here, we test whether this apparent contradiction could partly be resolved by the recent evidence that face aftereffects arise from multiple levels of the visual system (Susilo et al., 2010, JoV), and thus typically derive only partially from face-level coding. Our technique relies on the logic that (a) human psychophysical evidence (Susilo et al., 2010, Vision Research) shows broadband-opponent coding of facial dimensions such as eye- or mouth-height, and responses of face-selective cells in monkey inferotemporal cortex increase or decrease monotonically as values on facial dimensions get further from the average (Freiwald et al., 2009); (b) broadband-opponent coding predicts an adaptor a fixed distance from average will produce a larger aftereffect in individuals who have steeper neural response functions; and (c) these individuals should also show better discrimination along the dimension, potentially leading to better recognition of faces. We combine this with (d) a method (Susilo et al., 2010, JoV) for isolating the face-level component of a face aftereffect, as distinct from mid-level or shape generic components: this measures the size of the aftereffect to eye-height altered faces and to physically matched T-shapes varying vertical height of the bar. With 70 participants, we found the eye-height aftereffect alone did not correlate with face recognition (Cambridge Face Memory Test), but the face-specific component of the aftereffect (eye-height-minus-Ts) did correlate with CFMT, most strongly with the noise phase, which places the most demand on recognition of faces across view change, particularly not based on local features. Results imply a relationship between steepness of individuals' neural response functions in face space and their face recognition ability.
This PDF is available to Subscribers Only