Figure 2.

(A) Surround suppression in more naturalistic scenes will be studied with a 2AFC perceived contrast task at a single pedestal contrast (unsurrounded reference at 25% RMS contrast on one side of the screen, and target with 33% RMS contrast, matched or non-matched surround on the other side) for stimuli composed of line segments, natural scene segments, and synthetic textures (first and second-order statistics derived from natural scene segments, generated by Simoncelli’s “steerable pyramid” toolbox; [73]). This task will be used to quantify surround suppression in uniform and segmented textures. The same underlying model will be employed, but the more sparse neuronal responses will probe a wider range of neural network states for representations of the central stimulus, as well as constrain terms representing long-range projections signaling grouping probability or scene segmentation cues. (B). An object recognition task using line-segment textures to depict either meaningful or meaningless objects will let us test the effect of object recognition on contrast discrimination, fMRI and MEG responses in early visual cortex, and the interaction between neurophysiological signals in early visual cortex and prefrontal cortex.