Abstract
The human lateral occipital complex (LOC) responds more strongly to objects than scrambled objects and patterns. However, objects differ from patterns in several ways: they have a shape, contain a salient region and are delimited by contours. Here we examined the role of LOC in processing each of these types of information: salient regions, contours and shape.
We conducted two fMRI experiments (16 subjects) on a 3T GE scanner. Subjects viewed stimuli created by random dots (Stereo experiment: random dot stereograms; Motion experiment : random moving dots) which were matched for low-level visual information. Experiments contained five conditions: (1) random stimuli (random disparity dots, or randomly moving dots); (2) Two surfaces, creating two salient regions, but no contours; (3) Contours, generated by randomly scrambling object silhouettes, with no salient region; (4) Shaped holes; (5) Object silhouettes. Conditions 4 and 5 contained shaped salient regions with identical contours. Subjects were instructed to indicate whether the fixation was on the front or back surface (Stereo experiment) or if the motion was to the right or left (Motion experiment).
In both motion and stereo experiments, the object and hole conditions elicited significantly higher responses in LOC than contours, two surfaces or random conditions. In contrast, LOC activation to two surfaces without contours was not significantly higher than random noise. In posterior LOC, contours produced significantly greater activation than surfaces or random noise but less than objects and holes, suggesting that contours (even when they do not define an enclosed shape) contribute to posterior LOC activation. Anterior LOC did not show significant activation to contours vs. surfaces or random noise, suggesting that anterior regions process shape. These results provide evidence for a hierarchy of processing stages in LOC in which posterior regions processes both contours and shapes and anterior regions process shapes and objects.
Whitehall Foundation grant #2005-05-111-RES