Abstract
When we recall a past experience, rich sensory details often come to mind. Indeed, remembering engages the same perceptual regions of our brain as were active when we initially had the experience. Although this representational overlap between memory and perception may be advantageous to the phenomenology of recollection, it raises the question of how the brain distinguishes internally from externally generated information. Here, we hypothesized that, within the temporal lobe, overlap across memory and perception would be restricted to regions coding predominantly for visual features, such as in parahippocampal cortex, and not found in regions like the hippocampus that are strongly influenced by spatiotemporal information that discriminates past and present. Prior to an fMRI session, 24 observers were asked to form mental associations between pairs of images (each pair consisted of a unique face and scene). While being scanned, they were presented with blocks of images from one category, one image appearing at a time, and performed a simple categorization judgment (male vs. female for faces; natural vs. urban for scenes). Critically, they were cued to base their judgments either on the images being viewed ('perceive' condition) or, from memory, on the images that had been paired with those images beforehand ('retrieve' condition). Consistent with our hypothesis, scene-selective parahippocampal cortex showed greater pattern similarity across voxels when perceiving and retrieving the same vs. different scenes. As hypothesized, this was not observed in the hippocampus, but more unexpectedly, the effect was significantly reversed: there was greater pattern dissimilarity (anti-correlation) across voxels when perceiving and retrieving the same vs. different scenes. This finding suggests that the hippocampus can separate or differentiate representations of the same visual features when they are encountered at different points in time and/or arise from different sources (internal or external).
Meeting abstract presented at VSS 2016