Abstract
Strong evidence that faces are perceived holistically/configurally comes from the composite face effect/illusion: identical top halves of a face are perceived as being slightly different if they are aligned with different bottom halves. The illusion disappears when the bottom half of the face is moved away from the top half, so that the top and bottom halves are spatially misaligned. One unresolved issue is whether the composite illusion disappears in the misaligned condition because the two halves do not form a whole object anymore (i.e., they form two segmented parts), or because the bottom part is located away from the top part. To address this issue, 12 participants had to match two top halves of composite faces presented sequentially. The experiment was characterized by 3 conditions (“same”: both halves identical; “different”: both halves different; “composite”: top half identical, bottom part different) and 7 levels of alignment between the top and bottom halves: from spatially aligned, forming a whole coherent face, to completely misaligned (100% width of the face). Twenty-three base face identities were used. In the composite trials, accuracy rates decreased significantly compared to “same” trials, only in the completely aligned condition. Correct response times were also significantly increased only in that condition. Strikingly, even a small spatial misalignment (16.7% width of the face) between top/bottom parts in the composite face paradigm disrupted completely the influence of the bottom part on the perception of the top part, so there was no difference between all other levels of (mis)alignment. This observation implies, practically, that it is not necessary to use a wide spatial misalignment in such studies. From a theoretical standpoint, these data show that it is the spatial alignment/separation of the facial halves rather than the metric distance between the face parts that explain the presence/absence of the composite face effect.