Abstract
In a recent study on mechanisms of the preview effect (Belopolsky, Theeuwes, Kramer, submitted) we have demonstrated that participants are able to search up to 14 objects defined only by luminance transients and ignore up to 14 physically interleaved irrelevant objects. Since the target always appeared in the flashed subset, both top-down and bottom-up factors could have contributed to this effect. In the present study we attempted to de-convolve top-down and bottom-up factors. In Experiment 1 the target was equally likely to appear in either the flashed or non-flashed subset of objects. Participants exclusively searched the flashed objects first when the target was presented 300 ms after the flash, but not 800 ms after the flash, suggesting bottom-up prioritization. In Experiment 2 the influence of bottom-up factors was examined further by always presenting the target in the non-flashed subset of objects. Results indicated that participants could not ignore the flashed objects and prioritize selection for the non-flashed objects. However, with practice subjects were able to begin to prioritize search in the set of non-flashed objects. We concluded that although initially the prioritization of multiple objects by luminance transients is under bottom-up control, it can be overridden with sufficient practice. We propose that grouping by luminance transients plays an important role in this effect. Discrepancies with other studies, showing that attentional capture under certain conditions is limited to about 4 elements (Burkell & Pylyshyn, 1997; Yantis & Jones, 1991) will be discussed.