Figure 3. Preferential Responses to Natural or Disordered Face Motion within the Face Patch System.

(A) Schematics of stimuli used for analyses of natural motion selectivity. For faces and non-face objects, picture, natural movie, and jumbled movie stimuli were derived from the same 60 fps (frames per second) source videos. Each video was downsampled to 2 fps in the picture condition and 15 fps in the movie conditions. By randomizing the order of each natural movie’s frames, a matched jumbled movie was created. Each exemplar stimulus lasted 3 s; a 1 s period is shown here for demonstration. (B) Preference for natural face motion over jumbled face motion across six face patches and two control regions. Dorsal patches MD and AF show a significant preference for natural face motion, while, conversely, ventral patches PL, ML, and AL significantly prefer the rapidly changing jumbled face movies. (C) Preference for either natural face motion (red) or jumbled face motion (blue), as calculated in panel B, across face-selective cortex. Opacity reflects strength of face selectivity (Figure 2A). (D) Preference for natural object movies over jumbled object movies. (E) Strength of interaction between form (face or object) and frame ordering (natural or jumbled): (natural face movies - jumbled face movies) - (natural object movies - jumbled object movies).

* = p < 0.05 and ** = p < 0.01, corrected using Holm–Bonferroni method for 24 tests (8 ROIs × 3 measures). Dots on bar plots represent the values for individual subjects. Error bars represent standard error. Signal change in bar plots is normalized per ROI. Signal change in maps is normalized per subject.