Abstract
There is substantial evidence both for a broad focus of attention that shrinks/expands (e.g., Eriksen & St.James, 1986) and for a narrow focus that shifts from one location to another (Weichselgartner & Sperling, 1987). We hypothesized that either mode of attentional control (shrink/expand vs. shift) can be implemented, depending on task demands. We used the attentional blink (in which perception of the second of two targets is impaired at short inter-target lags) and Lag-1 sparing (spared perception of the second target when it is presented directly after the first) to test this hypothesis. In the present work, two concurrent streams of random-dot-pattern distractors were presented one on either side of fixation. Lag-1 sparing is known to occur to targets (letters) in different streams onlyif the second target falls within the attentional focus. Thus, the magnitude of Lag-1 sparing provides an index of the spatial extent of attention. One group of participants knew which stream would contain the first target, encouraging a narrow focus that is shifted between the two streams. For a second group, the targets appeared in either stream unpredictably, encouraging a broad focus of attention encompassing both streams. To manipulate the time available for changing the extent/location of the focus of attention, we varied the stimulus-onset-asynchrony between successive items in the stream. If participants did not know which stream would contain the first target, there was a gradual, linear change from Lag-1 sparing to Lag-1 deficit with increasing SOA, indicating a broad focus of attention that shrank to the location of the first target. If participants knew which stream would contain the first target, Lag-1 deficit changed to Lag-1 sparing, indicating a narrow focus of attention shifted from the first-target stream to the opposite stream. These results also speak to the relative time course of shrinking/expanding and shifting the focus of attention.
Grants to LNJ and VDL from the Natural Sciences and Engineering Research Council of Canada.