Poster Session D, Monday, March 25, 8:00 – 10:00 am, Pacific Concourse
Investigating Pattern Separation in the Medial Temporal Lobe through the Parametric Manipulation of Item Similarity
Corey Loo1,2, Bradley Buchsbaum1,2; 1University of Toronto, 2Baycrest Hospital
Pattern separation is a computational process that reduces the overlap between similar neural input patterns. In the context of memory encoding, computational models posit that the degree to which memories become separated varies as a function of their similarity. In an fMRI task, we investigated the parametric relationship between mnemonic similarity and neural activation in in both the cortex and regions of the medial temporal lobe (MTL), including probabilistically defined hippocampal sub-regions. In a preliminary behavioral experiment, 90 pairs of similar images were separately assessed for their pairwise similarity using an aggregate measure based on discrimination accuracy, discrimination reaction time, and subjective similarity ratings. In encoding blocks of an fMRI task, participants were exposed to one of the images from each of the similar image pairs. In the following recognition memory phase, participants were presented with an image probe and were asked to indicate if the probe was old or new and then provide a confidence rating. We examined regions in the brain that were parametrically modulated by the degree of similarity between targets and similar lures. We found that in all of the MTL sub-regions examined, confidence rating was a strong and significant predictor of neural activity, whereas similarity was not significantly related to neural activity. These results suggest that MTL activity associated with pattern separation processes are more related to internal mnemonic state variables indexed by confidence ratings than to externally derived measures of stimulus similarity.
Topic Area: OTHER