Abstract
Ensemble perception allows us to rapidly derive summary statistical information from groups of similar objects. Ensembles are generated so quickly and efficiently that some researchers hypothesize that they can help guide visual search. If ensembles leak through the limits imposed by the attentional bottleneck, is it possible to filter irrelevant ensemble information? In the current study, we examined whether observers could selectively ignore similar but irrelevant ensemble information. In a series of experiments, observers viewed multiple sets of gabors, presented simultaneously, and had to report the average orientation of just one of the sets, post-cued, using method-of-adjustment. The sets of gabors were either two sets of four gabors (two ensembles), or a set of four gabors presented with an additional individual gabor. Arrangement of the gabor sets and cueing procedures varied across experiments. The average of the irrelevant set of gabors was oriented 45° away from the average of the cued set, and could feasibly be incorporated into an average representation. The results revealed that the number of trials in which observers’ average orientation responses were drawn in the direction of the irrelevant set was not different from chance. However, the magnitude of the pull on trials in which they were pulled toward the irrelevant set was significantly greater when the irrelevant set contained an ensemble as opposed to an individual. Interestingly, even on trials in which the response went in the unpredicted direction, the magnitude of deviation was smaller when the irrelevant stimulus was an ensemble, suggesting its presence kept observers more anchored. The overall pattern of these results was still present, albeit mitigated, when observers were pre-cued. We conclude that an irrelevant ensemble can be successfully ignored, but when observers fail in doing so the ensemble wields a larger influence than an individual item.