Attention to Objects and Perceptual Organization: A neurally plausible computational model for intermediate vision

27 June 2014
Begin time: 
CITEC 2.015


Visual attention is often understood as a modulatory field acting at early stages of processing, but the mechanisms that direct and fit the field to the attended object are not known. It is unlikely that neuronal structures that control the deployment of attention (like prefrontal cortex) contain detailed representations of object features. How can they then direct attention to specific objects? We show that a purely spatial attention field propagating downwards in the neuronal network responsible for perceptual organization will be reshaped, repositioned and sharpened to match the object's shape and scale. Key features of the model are neurons integrating local features into coherent (proto-)objects, excitatory feedback to the same local feature neurons which caused a proto-object's cell activation, and inhibition between incompatible interpretations bothat the level of local features and at the level of proto-objects.