A model of object-based attention that guides active visual search to behaviourally relevant locations

  • Authors:

  • Linda Lanyon;Susan Denham

  • Affiliations:

  • Centre for Theoretical and Computational Neuroscience, University of Plymouth, U.K;Centre for Theoretical and Computational Neuroscience, University of Plymouth, U.K

  • Venue:
  • WAPCV'04 Proceedings of the Second international conference on Attention and Performance in Computational Vision
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

During active visual search for a colour-orientation conjunction target, scan paths tend to be guided to target coloured locations (Motter & Belky, 1998). An active vision model, using biased competition, is able to replicate this behaviour. At the cellular level, the model replicates spatial and object-based attentional effects over time courses observed in single cell recordings in monkeys (Chelazzi et al., 1993, 2001). The object-based effect allows competition between features/objects to be biased by knowledge of the target object. This results in the suppression of non-target features (Chelazzi et al., 1993, 2001; Motter, 1994) in ventral “what” stream areas, which provide a bias to the spatial competition in posterior parietal cortex (LIP). This enables LIP to represent behaviourally relevant locations (Colby et al., 1996) and attract the scan path. Such a biased competition model is extendable to include further “bottom-up” and “top-down” factors and has potential application in computer vision.