Correction for different thresholds between face identities.

<p>Main analyses for both experiments after modelling responses to each face identity individually and shifting the thresholds to be at the respective group mean; a, results for experiment 1 (masking); b, results for experiment 2 (composite faces); note that a different number of data points is available for each bin on the x-axis, as functions may be partly shifted out of the displayed range, rendering estimates at the extremes less reliable; also, variability is modelled at the stimulus level, not at participant level; please refer to <a href="http://www.plosone.org/article/info:doi/10.1371/journal.pone.0134790#pone.0134790.s007" target="_blank">S6 Code</a> for more details on the analysis procedure.