The control of overt and covert attention across two nodes of the attention-control network

Kavli Affiliate: Winrich Freiwald

| Authors: Pablo Polosecki, Sara C Steenrod, Heiko Stemmann and Winrich A. Freiwald

| Summary:

Attention is a central cognitive capability whose focus is thought to be directed by a spatial map coding behavioral priority. Here we tested the three defining properties of priority map theory with electrophysiological recordings from two attentional control areas. Both areas, lateral intraparietal area LIP and dorsal posterior inferotemporal cortex PITd, selected behaviorally relevant locations even in the absence of visual stimuli (cognitive sustainment principle). Second, priority signals are thought to arise from the summation of multiple spatial signals (superposition principle). LIP approximated linear summation for visual stimuli, spatial attention, and eye movements, while PITd did not. Third, the same priority signal should guide different behaviors (agnosticity principle). LIP, instead, used separable processing channels for representing attentional focus and eye position, while PITd coded attentional focus only. Thus primate attentional control circuits implement multiple priority maps, whose functional diversity and dimensionality increase the computational capacity of attentional selection. SIGNIFICANCE STATEMENT A central theory of attention poses that the brain computes a priority map to highlight spatial locations of relevance for behavior. Here we tested the hypothesis and its key predictions about how priority signals are assembled and used, through electrophysiological single-unit recordings from two nodes of the attention control network, localized by functional magnetic resonance imaging. We found both areas to highlight locations even in the absence of a stimulus, and that each assembled spatial signals differently and provided spatial information in different forms. These findings force a revision of how and where spatial attention is controlled in the brain.

Read More