Unconscious evaluation of faces on social dimensions

Abstract

It has been proposed that two major axes, dominance and trustworthiness, characterize the social dimensions of face evaluation. Whether evaluation of faces on these social dimensions is restricted to conscious appraisal or happens at a preconscious level is unknown. Here we provide behavioral evidence that such preconscious evaluations exist and that they are likely to be interpretations arising from interactions between the face stimuli and observer-specific traits. Monocularly viewed faces that varied independently along two social dimensions of trust and dominance were rendered invisible by continuous flash suppression (CFS) when a flashing pattern was presented to the other eye. Participants pressed a button as soon as they saw the face emerge from suppression to indicate whether the previously hidden face was located slightly to the left or right of central fixation. Dominant and untrustworthy faces took significantly longer time to emerge (T2E) compared with neutral faces. A control experiment showed these findings could not reflect delayed motor responses to conscious faces. Finally, we showed that participants' self-reported propensity to trust was strongly predictive of untrust avoidance (i.e., difference in T2E for untrustworthy vs neutral faces) as well as dominance avoidance (i.e., difference in T2E for dominant vs neutral faces). Dominance avoidance was also correlated with submissive behavior. We suggest that such prolongation of suppression for threatening faces may result from a passive fear response, leading to slowed visual perception.

Figure 1. (A) Two-dimensional (Trust × Dominance) space of social evaluation of faces. (B) The sequence of events in a typical trial of Experiment 1. T2E = Time to emerge; CFS = continuous flash suppression.

Figure 2. Experiment 1 results. Time to emerge (T2E) averaged across subjects (N = 23) are plotted against dominance (A) and trustworthiness (B) of the suppressed faces. Error bars = standard error of mean difference between the specific condition and neutral face. (C) Gray scale heat map illustrates the T2E for all 7 × 7 combinations of trust by dominance. Brighter shades correspond to larger T2E.

Figure 3. The results of Experiment 2. Each plot shows the average time to emerge (T2E) of subjects (N = 23) plotted against dominance (A, C) and trustworthiness (B, D) of the suppressed faces. The impact of each social dimension on responses is plotted separately (collapsing on the other dimension). Top (A, B) and bottom (C, D) panels correspond to suppressed and visible faces, respectively. Error bars = standard error of mean difference between the specific condition and neutral face. ns: p > .05. * p < .05. ** p < .01.

Figure 4. The relationship between self-reported personality traits of participants (x axis) and psychophysical performance (y axis) (N = 50). Each plot shows the size of dominance avoidance effect (A, C, E) or the size of untrust avoidance effect (B, D, F) plotted against scores of personality trait questionnaires. Each subject is represented by a data point, and dashed lines represent lines of best fit.