MIT CogNet, The Brain Sciences ConnectionFrom the MIT Press, Link to Online Catalog
SPARC Communities
Subscriber : Stanford University Libraries » LOG IN

space

Powered By Google 
Advanced Search

 

A Neurocomputational Model of Object-based Attention by Disinhibition

 Frank van der Velde and Marc de Kamps
  
 

Abstract:
Abstract: Different features of objects in the visual field (e.g., shape, color, motion, location) are processed in different areas and pathways in the visual cortex. Our model shows how the selection of one feature of an object (e.g., its shape or color) results in an attentional enhancement of activity for all neurons that represent the features of the object in all areas of the visual cortex. In our model, feedforward activation interacts with feedback activation in the retinotopic areas of the visual cortex. The feedforward activation results from the stimulus (bottom-up), and the feedback activation results from the selected object feature (top-down). The interaction occurs in local microcircuits, which produce gain and gating of activation by means of disinhibition. This selectively enhances the activation of only those neurons that are activated by the selected object. It also retrieves the location of the selected object in the retinotopic areas. As a result, (spatial) attention can be directed to the object as a whole, which results in an enhanced activation of the neurons responsive to the other features of the selected object. In this way, the activation of neurons responsive to the selected object are enhanced in all areas of the visual cortex, which integrates the features of the object and selects the object from other objects in the visual field.

 
 


© 2010 The MIT Press
MIT Logo