Figure 2. Contents of mental representations of familiar faces.

(A) Mental representation of ‘Mary’ (a typical participant). Ground truth: 3D vertex positions deviate both Inward (-) and Outward (+) from the categorical average to objectively define the shape of each familiar face identity. Greyscale values reported on the flanking faces color-code the normalized magnitudes of inward and outward deviations from the categorical average. Mental representation: Inward and Outward colored faces highlight the individual 3D vertices whose position faithfully deviate from the categorical average in the GMF (p < 0.05, two-sided). Blue to red colors represent the normalized magnitudes of their deviations. 2D scatter plots: Scatter plots indicate the relationship between each vertex deviation in the ground truth (normalized scale on the X-axis) and the corresponding vertex in the memory representation (normalized scale on the Y-axis). The white diagonal line provides the reference of veridical mental representation in the GMF—i.e. a hypothetical numerical correspondence between each shape vertex position in the ground truth face and in the mental representation of the same face. White dots indicate vertices that were not faithfully represented. (B) Mental Representations (group results). Same caption as Figure 2A, except that the colormap now reflects the number of participants (N = 14) who faithfully represented this particular shape vertex.