Abstract
Some current models of face perception propose that upright and inverted faces use different processes: an expert mechanism for upright faces and a generic object-recognition process for inverted faces. If so, we hypothesized that the transfer of adaptation effects between upright and inverted faces should be limited. We used a novel contrast-based adaptation technique to measure changes in the ability of human observers to recognize inverted or upright faces at test after adapting to a face in either orientation. We used two adapting durations, a short 100ms duration at which adaptors facilitate recognition of the adapted face (‘same-face’) but inhibit recognition of other faces (‘other-face’), and a 1600ms duration at which adaptors suppress all face recognition, but more so when the test and adaptor are from different identities (Oruc I, Barton JJS. Brief adaptation increases sensitivity of face recognition. VSS 2008). Eight subjects participated. We had subjects perform two orientation-congruent conditions, one for upright faces and one for inverted faces, and also two orientation-incongruent conditions, one in which the adaptor was inverted and the test upright (upright-adaptor/inverted-test), and one in which the adaptor was upright and the test inverted (inverted-adaptor/upright-test). We found that both upright and inverted orientation-congruent conditions generated significant aftereffects that differentiated between same-face and other-face trials at both 100 and 1600ms. In the incongruent conditions, more modest adaptation aftereffects were found with the upright-adaptor/inverted-test condition, but no aftereffects with the inverted-adaptor/upright-test condition. This asymmetric partial transfer of adaptation may suggest that upright faces can partially adapt the mechanisms used for inverted faces, consistent with effects on generic recognition systems that are relatively robust to the effects of orientation, but inverted faces cannot adapt the mechanisms primarily used for upright faces, which may represent an orientation-dependent expertise process.
Funding was provided by NSERC Discovery Grant RGPIN 355879-08, a CIHR Summer Studentship (XG), a Canada research Chair and Michael Smith Foundation for Health Research Senior Scholarship (JB).