Spatio-Temporal Dynamics of
Face Recognition in a Flash:
ItÕs in the Eyes
Figure 1. This figure illustrates the twelve frames of a sample movie stimulus. Each frame was presented for about 24 ms, and thus the movie lasted about 282 ms. A film version of this figure is available here.
Figure 2. (a) These pictures depict the Z-scored regression coefficients of human observers. They were obtained by regressing space-time information samples with response accuracy per observer (click on the following initials to see a film version of these static frames: AN, CA, CL, DE, GY, JF, MA, ME, MY, and ST), and across observers (click on MEAN to see a film version of these average static frames). These coefficients (varying from black, ² -1.65, to white 1.65) indicate the relative usefulness of each region of the face through time, for judgements of face identity – coefficients above 1.65 (p < 0.05) were replaced by a red face marker to facilitate reading. The pictures also plot the average Z-scores in the regions of the left eye (circle), the right eye (rectangles), and both eyes (ŌXÕx), as well as the regions most used by a super-ideal observer (diamonds). (b) This panel shows the evolution of the Pearson correlations between the coefficients of all human observers (average) and that of the super-ideal observer (dotted curve) as well as that of the eye-region model (solid curve), for each one of ten experimental blocks.
Figure 3. The plots in (a) depict the average Z-scored coefficients of regression per observer (e.g., DE, AN, GY) over time with corresponding Fourier coefficients (b).
Click here to access a folder with all the faces used in our experiment.