Abstract
Visual attention allows one to voluntarily orient the focus of their attentional spotlight toward behaviorally relevant information. Many fMRI studies have shown that the medial superior parietal lobule (mSPL) robustly increases activation when subjects are cued to shift attention. However, when cued to hold attention, mSPL shows a well-defined, small-amplitude BOLD response increase, suggesting that mSPL may perform computations of cue interpretation, rather than exclusively attention shifting. We investigated this alternative using a behavioral attention task involving two target RSVP streams (one in each hemifield), each flanked by three distractor RSVP streams. Cues (red letters embedded within a homogeneously white letter stream) instructed participants to Shift or Hold their attention on the current stream. Participants discriminated the parity of target digits via button press. Simultaneously, a nontarget digit was either absent (50% of trials) or present in the opposite stream: either congruent (25%) or incongruent (25%) with target identity. Critically, three subject groups experienced different proportions of Shift/Hold trials (70%/30%, 50%/50%, 30%/70%). Experiments in which Hold trials are more prevalent have not previously been conducted. During Hold trials, if mSPL activation simply reflects cue interpretation, we would predict no effect of nontarget congruency; however, if mSPL reflects a brief erroneous attention shift, performance could be affected by the nontarget in the noncued stream. We observed a statistically significant nontarget congruency effect during Hold trials in the 70/30 condition (when most trials cued a Shift) and 50/50 condition, but not in the 30/70 condition (when most trials cued a Hold). We saw this across two cue-target intervals (212ms and 318ms). Thus, our results suggest that the BOLD increase observed in mSPL following a Hold cue may not reflect cue interpretation, but rather a brief shift of attention encouraged by the large proportion of shift cues commonly used in these experiments.