Abstract
In a focal contingent capture paradigm (Ghorashi et al., 2003) we demonstrate that subjects search for familiar object categories on the basis of attentional control settings (ACS) that include information about associated contexts. Further, subjects attend to representations of these associated contexts automatically, that is, even when this harms performance in the primary search task. On each trial, observers responded about a cued target object in an RSVP stream of photographs. A wide range of target categories (approximately 70) was employed and the target category changed on each trial. After viewing this sequence of images, observers indicated whether the object was presented (Experiment 1) or which version of the object was presented (Experiments 2 and 3). Critically, on half of trials this target object (e.g. “camel”) was preceded by a photograph depicting an associated scene context (e.g. “desert”). Otherwise, the target object was preceded by an unassociated scene. Depictions of associated contexts captured attention and resulted in costs for detection and discrimination tasks. Manipulations of the relative positions of the associated context distractor and the target revealed the time course of this capture effect. Costs only emerge at lags greater than 1 and persist for approximately 500 ms. In addition to demonstrating that observers automatically use contextual information when selecting objects in a search task, these experiments are also a novel demonstration of both the flexibility and limitations of top-down attentional control. The target category changed on each trial and only repeated once in Experiments 2 and 3. Despite the highly variable and rapidly changing target criteria, observers were able to quickly establish representations that identify task relevant information. While observers could rapidly establish the ACS required to select a task (e.g. find a dog) they could not exclude associated, but task irrelevant, information (e.g. a doghouse).