Back

Pattern Separation Contributes to Categorical Face Perception

Baker, S. W.; Youm, A.; Levy, Y.; Moscovitch, M.; Rosenbaum, R. S.

2020-05-15 neuroscience
10.1101/2020.05.13.094276 bioRxiv
Show abstract

Traditionally considered a memory structure, the hippocampus has been shown to contribute to non-memory functions, from perception to language. Recent evidence suggests that the ability to differentiate highly confusable faces could involve pattern separation, a mnemonic process mediated by the hippocampal dentate gyrus (DG). Hippocampal involvement, however, may depend on existing face memories. To investigate these possibilities, we tested BL, a rare individual with bilateral lesions selective to the DG, and healthy controls. Both were administered morphed images of famous and nonfamous faces in a categorical perception (CP) identification and discrimination experiment. All participants exhibited nonlinear identification of famous faces with a midpoint category boundary. Controls identified newly learned nonfamous faces with lesser fidelity, while BL showed a notable shift in category boundary. When discriminating face pairs, controls showed typical CP effects of better between-category than within-category discrimination -- but only for famous faces. BL showed extreme within-category "compression," reflecting his tendency to pattern complete following suboptimal pattern separation. We provide the first evidence that pattern separation contributes to CP of faces.

Matching journals

The top 4 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.