Contextual Learning in Visual Search Tasks

Contextual cueing refers to the ability of the visual system to learn the invariant information useful to behaviour. Since visual regularities are not confined to spatial information, other information from the environment can be used to cue our behaviour. For example, item identity, or motion trajectory are often important features to make search easier and to guide vision and action (Biederman, Mezzanotte, Rabinowitz, 1982). The association between target and distractors in a visual search display can also be made by semantic properties of stimuli or motion trajectory as well as by color features of stimuli.
In 1999 Chun and Jiang modified their original contextual cueing paradigm (1998) to show the great ability of the visual system to learn non-spatial regularities. In their experiment, half trials were composed by old displays, in which the identity of the items was paired and repeated across blocks, whereas spatial locations were randomly generated. In the other half of trials the identity association as well as the spatial association between items in the display were randomly chosen over blocks. Crucial was that spatial properties of target and distractors did not create a context, so that it could not be learned. Nonetheless, an RT advantage for the old configurations only emerged when the identity information in the old pattern of configurations was repeated, showing that learning the target shape, the distractors shape and their associations made the target more detectable.