Abstract
Cross-attribute attentional effects — i.e., attentional modulation of task-irrelevant attributes of an object when one of its attributes is attended — have been interpreted as supporting object-based attentional mechanisms. Alternatively, these effects can be explained by mechanisms that bind spatiotemporally collocated attributes, before object formation. We showed cross-attribute attentional effects between luminance and motion, and found that these effects indeed occur at the stage of object representation.
During adaptation, observers viewed two populations of random dots. Half of the dots (“effectors”) moved along 0°. The other half (“distractors”) alternated direction every 4 s between +90° and −90°. The two groups of dots were also colored differently, red or green. Observers' task was to detect occasional brief luminance increases of either red or green dots, thereby directing attention to the luminance of effectors or distractors. In the ‘transparent’ condition, dots were randomly located, so observers perceived two transparent surfaces moving orthogonally. In the non-transparent condition, each effector dot was locally paired with a distractor dot at close proximity [Qian et al. 1994], so observers perceived one bi-colored surface moving along the vector average of the two motion vectors. We measured the duration of the MAE using a group of static dots.
We found strong cross-attribute attentional effects, but only when dots were perceptually segregated into two surfaces: In the transparent condition, the MAE duration after attending to the luminance of effectors was longer than after attending to that of distractors. By contrast, in the non-transparent condition, the two MAE durations were not significantly different.
These results imply that cross-attribute attentional modulation is based on object-level properties of the stimuli; it appears that there is no direct attentional modulation between visual attributes at a local pre-object processing stage.