Biased Competition Mechanisms for Visual Attention in a Multimodular Neurodynamical System

Abstract

Visual attention poses a mechanism for the selection of be- haviourally relevant information from natural scenes which usually con- tain multiple objects. The aim of the present work is to formulate a neurodynamical model of selective visual attention based on the “bi- ased competition hypothesis” and structured in several network modules which can be related with the different areas of the dorsal and ventral path of the visual cortex. Spatial and object attention are accomplished by a multiplicative gain control that emerges dynamically through inter- cortical mutual biased coupling. We also include in our computational model the “resolution hypothesis” in order to explain the role of the neurodynamics control of spatial resolution evidenced in psychophysical experiments. We propose that V1 neurons have different latencies de- pending on the spatial frequency to which they respond more sensitively. In concrete, we pose that V1 neurons sensitive to low spatial frequency are faster than V1 neurons sensitive to high spatial frequency. In this sense, a scene is first predominantly analysed at a coarse resolution level and the dynamics enhances subsequently the resolution at the location of an object until the object is identified.KeywordsSpatial FrequencyVisual SearchObject RecognitionVisual AttentionDorsal StreamThese keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.