Attention Selectively Reshapes the Geometry of Distributed Semantic Representation

Abstract

Humans prioritize different semantic qualities of a complex stimulus depending on their behavioral goals. These semantic features are encoded in distributed neural populations, yet it is unclear how attention might operate across these distributed representations. To address this, we presented participants with naturalistic video clips of animals behaving in their natural environments while the participants attended to either behavior or taxonomy. We used models of representational geometry to investigate how attentional allocation affects the distributed neural representation of animal behavior and taxonomy. Attending to animal behavior transiently increased the discriminability of distributed population codes for observed actions in anterior intraparietal, pericentral, and ventral temporal cortices. Attending to animal taxonomy while viewing the same stimuli increased the discriminability of distributed animal category representations in ventral temporal cortex. For both tasks, attention selectively enhanced the discriminability of response patterns along behaviorally relevant dimensions. These findings suggest that behavioral goals alter how the brain extracts semantic features from the visual world. Attention effectively disentangles population responses for downstream read-out by sculpting representational geometry in late-stage perceptual areas.

Representational organization of novel task sets during proactive encoding

“…Stokes, Thompson, Nobre, & Duncan, 2009) have been reported in 688 perceptual and motor regions during preparation. Importantly, these changes have been linked to the representational organization in regions along the visual pathway is dynamically adapted to 692 task demands (Nastase et al, 2017). Our current results add to these findings by showing that 693 representational space tuning could be a mechanism of preparatory bias, which could reflect 694 predictive coding principles where iterative loops of feedback and feedforward communication 695 shape cognition (Friston, 2005).…”

A computational model of shared fine-scale structure in the human connectome

“…In primary visual cortex, retinotopy is multiplexed with ocular dominance columns, edge orientation, spatial frequency, motion direction, and motion velocity, among other low-level visual attributes [36,37]. Primary visual cortex sends coherent projections to other visual areas where these topographies are recapitulated and transformed, affording the emergence of more complex features, such as curvature, texture, shape, color constancy, and biological motion; and, subsequently, even higher-order attributes such as object categories, view-invariant face identity, and species-invariant attributes of animals such as action categories and dangerousness [12,15,[26][27][28][38][39][40]. Similar transformations of multiplexed topographies characterize other sensory modalities and, undoubtedly, supramodal cognitive operations.…”

“…Hyperalignment of movie data started with hyperalignment of data in 40,968 overlapping searchlights of 20 mm radius centered on cortical nodes with 1.9 mm average spacing between the nodes. Cortical nodes were defined in a standard cortical surface from FreeSurfer (fsaverage)(https://surfer.nmr.mgh.harvard.edu) and resampled into a regular grid using AFNI's MapIcosahedron [42,43] with 20,484 nodes in each hemisphere.…”

Neural responses to naturalistic clips of behaving animals in two different task contexts

“…(B) Two example design matrices for predicting hemodynamic responses to the clips over the course of two runs with the taxonomy attention task. In the condition-rich design, each of 80 visually unique stimuli receives a separate predictor (following Kriegeskorte et al, 2008a;upper), while in the category design, the four exemplar clips per taxonomy-behavior condition are collapsed to form 20 category predictors (following Nastase et al, 2017;lower). Hypothesized neural responses are convolved with a simple hemodynamic response function (Cohen, 1997).…”