Context effects on border assignment in the target stimulus in visual search

Using a modified version of Rauschenberger & Yantis' (2001) search task, Rauschenberger, Peterson, Mosca, & Bruno (VSS '02) presented evidence for a dynamically evolving influence of the non-target stimuli on the representation of the target stimulus (spatiotemporal context effect). Here, we extend this finding, building a case for the generality of spatiotemporal context effects in visual search. In displays that were masked after either 100 or 250 ms, subjects searched for an enclosed symmetric novel outline target whose vertical edges sketched a 'standing woman' on the ground side. This target was presented among one of two types of non-targets (in fully mixed blocks): One type of non-target had the same vertical edges as the target; but they were each mirror reflected so that the 'standing woman' was sketched on the figure side of the edge rather than the ground side (experimental condition). The other type of non-target was a scrambled version of the 'standing woman' non-targets (control condition). The control condition serves as a baseline against which the experimental condition may be compared, as it is less subject to context effects. Because spatiotemporal context effects require time to evolve, we predicted that the non-targets would exert a greater influence on the perception of the target in the 250 ms SOA condition than in the 100 ms condition (as compared to the control condition). Results were consistent with this prediction. In the experimental condition, the search slopes were significantly larger in the 250 ms SOA condition than in the 100 ms SOA condition, whereas in the control condition, the slopes in these two SOA conditions were not statistically different. Moreover, the 100 ms condition showed the typical pattern of more efficient search among familiar non-targets than among unfamiliar non-targets. This pattern disappeared in the 250 ms condition, however. We attribute this disappearance to context effects.