Abstract
When directed to multiple spatial locations, attention has traditionally been believed to be simultaneously distributed among them. Rhythmic presentation of spatially disjoint targets at optimal frequencies improves their detection, however, suggesting that covert attention may in fact be rapidly cycled among attended locations. This work has relied on behavioral measures of target detection to assess the location of visual attention at any given time. Such findings may indeed reflect the serial allocation of attention during attentional search, but may also reflect a serial limitation on the ability to encode target items such that they are available for report. We used highly time-resolved fMRI (TR = 88 ms) to investigate whether directing attention to multiple items results in serial rather than simultaneous enhancement of their representations in visual cortex. Critically, we examined only trials in which behavioral targets were not present; any "serial" timing effects observed under these conditions necessarily reflect serial allocation of attention rather than serial target encoding. We measured extrastriate signal evoked by stimuli in the four quadrants under three conditions. A simultaneous, 400 ms, 25% increase in luminance of all four items served as a model for simultaneously distributed attention. A sequential, 100 ms, 100% increase in luminance for each item served as a model for sequentially allocated attention. We compared these with an attended condition, in which participants monitored the four items (whose luminance did not change). Although the peak of the evoked BOLD response did not change significantly across visual fields under simultaneous stimulation (p =.22), it did change significantly across the visual field under sequential stimulation (p=.04) and attended conditions (p = .04). The results undermine the notion that attention may be simultaneously diffused across multiple items; instead, attended items appear to undergo serial sampling.
Meeting abstract presented at VSS 2014