Feature-based attentional interference revealed in perceptual errors and lags.

According to a limited-resource account of feature-based attention, dividing feature-based attention by selecting targets on the basis of different features dilutes its power. Multiple-feature costs have been documented previously, but it is not clear whether the multiple-feature cost arose at the selection (segregating targets from non-targets) stage predicted by the limited-resource account. The cost might instead result from a post-selection difficulty in processing or accessing the contents of the targets. By defining the targets with a selection attribute (color) that is very distinct from the attribute participants must access and report (spatial period), we were able to manipulate the selection process independently from the access stage. We still found a cost for different selection features (colors), suggesting that multiple-feature costs can arise at the selection stage. The cost was only significant however when distracters were present that shared the selection features. The cost manifested not only as greater errors or less precision in reporting the access attribute (spatial period), but also as an increased temporal lag between the physical stimuli and the reported percept. In summary, splitting selection among different features incurred little or no penalty by itself, but selection interference by distracters sharing target features could be large and could slow processing.