Abstract
Segmentation of an image can be accomplished via a range of cues, including colour, motion, time, and depth. We compared brain activity for two segmentation cues (motion and time) to investigate whether there may be one cue invariant central process for segmentation.
Participants performed a visual search task in a 3T Philips MRI scanner. Participants indicated whether an inverted T shape target (embedded in a display of rotated T shape distractors) was left or right of fixation. In the baseline condition there were no consistent cues to segmentation. In the motion segmentation condition, half the T's, including the target, moved down the screen. In the temporal segmentation condition, half the items (not including the target) appeared 2s earlier than the remainder (similar to the ‘preview search’ paradigm). Participants benefited from segmentation as evidenced by faster RTs in both segmentation conditions relative to the baseline.
Segmentation by motion was associated with increased BOLD activity in bilateral medial temporal areas, precuneus and superior parietal lobes. Segmentation by time was associated with increased activity in the bilateral precuneus. Part of the precuneus was significantly active in both segmentation conditions, and the magnitude of this activity increased with increasing behavioural benefit from the cue (in both conditions). In the time but not the motion condition, the behavioural benefit was inversely correlated with activity in early visual areas consistent with inhibition of early distractors.
Segmentation by both of the cues investigated here appears to involve both cue-invariant mechanisms mediated by part of the precuneus, in addition to cue-specific mechanisms operationalised by different parts of the lower level visual system.