Abstract
One can search for a familiar type of object, such as a hat, without knowing its shape or color but it remains unclear exactly whether naturalistic targets evoke rapid, contingent, covert shifts of attention in like fashion to targets defined by simpler stimuli such as color. To detect covert attention in such search, two experiments tested whether images containing conceptually defined targets such as dinner food or four-legged animal elicited behavioral and electrophysiological signifiers of visuospatial attention. The paradigm used dual RSVP streams of natural images. Learning and intrinsic salience confounds were minimized by presenting each target image only once per participant, and counterbalancing target sets across subjects so that an image that matched the target set for one subject would be a false target for another subject. The pre-registered behavioral analyses showed that when two targets were presented in rapid sequence, subjects reported the second target more often when it was in the same spatial location as the first. This finding suggests that the leading target evoked a spatial attention shift that affected processing of the second item. The second pre-registered experiment used EEG and showed that N2pc and P3 EEG components were elicited by images that matched the conceptually specified target set. Moreover, by comparing the latency of the N2pc evoked by targets in this task, with featural targets and singletons from previous experiments, it is observed that latencies become with increasing target complexity. However latencies for the natural image search is still within the bounds of typical N2pc components, suggesting that such attention is triggered by later stages of ventral stream processing, rather than a completely distinct process. These results indicate rapid, selective deployment of spatial attention to images that match current search goals at a conceptual level in the absence of featural signifiers.
Acknowledgement: National Science Foundation 1734220, 1331073