Abstract
Imagine vertically compressing a facial image to a quarter of its original width. Given the marked changes in facial configuration, one might expect that this grotesquely distorted face would be much less recognizable relative to the original. However, our current study demonstrates that identification performance remains almost entirely unaffected even at extreme compressions that result in sliver-like faces, no more than 1/6 the original width or height. Complementing this behavioral finding, we have recorded subjects’ neural activity using Magnetoencephalography (MEG) as they viewed faces and objects that were subjected to a range of compressions. Our data reveal a striking correlation between behaviorally observed psychometric curve on the one hand and neural activity when viewing faces, on the other. Specifically, we found a significant change in amplitude and latency of the M170 and M250 components (implicated in face perception) at the same compression threshold at which behavioral recognition occurs. This result has important implications for linking neural activity and perception, as well as for understanding the nature of configural information that subserves face recognition. The tight coupling we observe between the M170 and behavioral performance is puzzling since past electrophysiological evidence linking this component to identity judgments is weak, at best. We propose an alternative explanation and also describe how our results help constrain the vague notion of ‘facial configuration’.
Meeting abstract presented at VSS 2013