Fig. 5
Stimulus contrast does not explain blur selectivity. a Profile schematic of how a blurred stimulus (red) has a decreased mean foreground intensity relative to a sharp (β = 0.005) stimulus (black). Low/high values correspond to background/foreground image intensities, respectively. An intensity control is constructed from a sharp shape with an identical mean foreground intensity (blue). b Example blurred stimulus and intensity-matched controls. Stimuli are shown in black but were presented at either positive or negative luminance contrast. c, d Responses of preferred (red) to non-preferred (blue) shapes that were presented either blurred (left) or as intensity-matched controls (right). While blur and contrast control tuning curves are remarkably different in c, they are quite similar in d. e Center-of-Mass analysis (see Methods: ‘Analysis and model fitting’) reveals that for a majority of cells (n = 31 of 34) blur and contrast-control tuning curves have significantly different (black) tuning profiles (t-test, p < 0.05); other cells (gray) exhibited blur and contrast-control tuning curves with CoM values that were not significantly different. For these neurons, blur tuning may be explained in the context of intensity tuning. Neurons above the diagonal typically exhibited a tuning preference for intermediate intensities while remaining largely invariant to all but the highest levels of blur; a more conservative estimate that discounts these cells (n = 6, ≈ 18%) still finds the majority of neurons unexplained by tuning for stimulus contrast (n = 25, ≈ 74%)
