Abstract
We examined the effect of perceptual experience on visual representations in human cortex. Perirhinal cortex lies at the anterior end of the ventral visual stream and is thought to store complex representations of whole objects. One theory of perirhinal cortex function claims that it is important for discriminating highly similar objects that share simple visual features, and there is evidence to support this claim. However, it is not known how perceptual experience shapes and influences the object representations in perirhinal cortex. We examined the effect of short-term perceptual experience (i.e., seeing many objects from the same category over the course of a few minutes) on visual representations throughout the ventral visual stream and perirhinal cortex. Participants in the scanner viewed series of objects from the same category (e.g., teapots, shoes) while performing a 1-back repetition-detection task. In each series, individual object images were presented three times each, such that on the first presentation the object was novel and on the third presentation it was familiar. The difference in hemodynamic responses elicited by the first and third presentations –i.e., response adaptation– indicated whether a cortical region represented individual objects uniquely. In perirhinal cortex, individual stimuli for which the first and third presentations appeared early on in the series elicited significantly less response adaptation than stimuli for which the first and third presentations appeared in the last part of the series (i.e., after viewing a stream of objects from that category). We did not find this effect in regions of the ventral visual stream posterior to perirhinal cortex (i.e., lateral occipital complex). This result suggests that representations in perirhinal cortex become more specific to individual exemplars of an object category after experience with that category. This may be an effect of learning to attend to combinations of features that permit within-category discriminations.
Meeting abstract presented at VSS 2016