Abstract
Under what conditions, and at what cost, can multiple visual targets be selected at the same time? We investigated spatiotemporal properties of attentional selection using rapid serial visual presentation (RSVP) tasks. Two streams of letters were presented simultaneously, offset either horizontally (presented in different hemifields) or vertically (same hemifield), with target letters cued by a surrounding ring. In single-target conditions, a target could appear in either of the streams; in dual-target conditions, two targets appeared simultaneously, one in each stream. Observers reported the identity of each target, and a mixture model fitted to the distribution of serial position errors estimated the efficacy (probability of reporting a relevant item), latency, and temporal precision of selection. In dual-target conditions, latency was similar in both streams, suggesting that selection was simultaneous even when both streams were in the same hemifield. Furthermore, the latency and precision observed for single targets were unaffected by requiring selection of a second, simultaneous target. However, efficacy was lower when both streams were in the same hemifield than when streams were positioned in different hemifields, consistent with greater resource independence in the different-hemifields configuration. When targets were presented simultaneously to both hemifields, we found higher efficacy for selection of targets in the left hemifield than in the right; yet for single-target conditions, efficacy was higher in the right hemifield than in the left. These findings are consistent with a model—previously proposed on the basis of neuropsychological evidence—in which attentional resources of the left cerebral hemisphere are restricted to the contralateral hemifield, while resources of the right hemisphere can be flexibly allocated to processing contralateral or ipsilateral stimuli. While some tasks may require serial shifts of attention (e.g. Holcombe, Linares, & Vaziri-Pashkam, 2011), the present results suggest that selection of targets in multiple RSVP streams can occur simultaneously.
Meeting abstract presented at VSS 2013