The ‘Parahippocampal Place Area’ responds selectively to high spatial frequencies in humans and monkeys

Abstract

Defining the exact mechanisms by which the brain processes visual objects remains an unresolved challenge. Valuable clues to this process have emerged from the demonstration that clusters of neurons (‘modules’) in primate inferotemporal (IT) cortex apparently respond selectively to specific categories of visual stimuli. However, lower-level visual cues which might underlie these category selectivities have not been extensively tested. Here we show that at least part of the ‘category-selective’ response in one of these brain areas (Parahippocampal Place Area - PPA, which apparently responds selectively to images of places) actually results from a lower-level selectivity for high spatial frequencies, which are emphasized in object borders and details. This lower-level selectivity was demonstrated in multiple fMRI tests, both in humans and monkeys. In humans, the PPA responses were strongly correlated with the spatial frequency content of different computer-generated 3D shapes. In addition, we found a near-doubling of the (normally-small) response to naturalistic images of faces in the human PPA, when such images were filtered to pass only high spatial frequencies. The same fMRI experiment in awake monkeys revealed a PPA homologue in macaque, and a striking double-dissociation between sensitivity to stimulus category (places/faces) versus spatial frequency (higher/lower) in the monkey homologues of PPA and FFA (Fusiform Face Area), respectively. These results suggest that a neural selectivity for high spatial frequencies might arise naturally in PPA, to encode object borders and details of scenes, during place-related cortical processing (e.g. for navigation or environmental vigilance). By the same token, the lower-level spatial selectivity shown here likely accounts for at least some of the higher-order ‘category selectivity’ reported in the previous studies of PPA.