ABSTRACT
A computational model of learning in visual attention is described, and it is shown how it can integrate information over time to form a representation of the visual context. This visual context is subsequently used to guide the focus of attention in a visual search task. Computer simulations shows that the model can learn to use contextual cues when they are available. The performance using contextual cues is compared to a situation where they are not used. The result shows that contextual cueing is only useful when the visual scene is sufficiently complex to warrant the extra time needed to establish context identity.
