Fig. 8.

CSS model is consistent with tolerance to changes in object size. We tested how well the CSS model, estimated from responses to spatial contrast patterns, predicts responses to objects that vary in size. Note that the model predicts only the relative pattern of responses, not the overall gain in responses (see main text). A: sample stimulus frames. Objects were presented on textured backgrounds within circular apertures of different sizes. B: model accuracy in different visual field maps (format same as Fig. 6A). The accuracy of the CSS model is close to the noise ceiling. C: data and model predictions. For visualization purposes, we normalize the responses from each voxel to unit length and plot the median response to each object size as a black bar (error bars indicate the interquartile range). In anterior maps, responses are relatively unaffected by changes in object size, and the CSS model successfully reproduces this effect.