Automatic integration of social information in emotion recognition.

Abstract

This study investigated the automaticity of the influence of social inference on emotion recognition. Participants were asked to recognize dynamic facial expressions of emotion (fear or anger in Experiment 1 and blends of fear and surprise or of anger and disgust in Experiment 2) in a target face presented at the center of a screen while a subliminal contextual face appearing in the periphery expressed an emotion (fear or anger) or not (neutral) and either looked at the target face or not. Results of Experiment 1 revealed that recognition of the target emotion of fear was improved when a subliminal angry contextual face gazed toward-rather than away from-the fearful face. We replicated this effect in Experiment 2, in which facial expression blends of fear and surprise were more often and more rapidly categorized as expressing fear when the subliminal contextual face expressed anger and gazed toward-rather than away from-the target face. With the contextual face appearing for 30 ms in total, including only 10 ms of emotion expression, and being immediately masked, our data provide the first evidence that social influence on emotion recognition can occur automatically.

Formalizing emotion concepts within a Bayesian model of theory of mind

“…In these ways, dynamic expressions may lead to more specific or more confident inverse inferences (though observers can also be surprisingly insensitive to dynamic information per se [36,57]). On the other hand, when temporal change in the face coincides with temporal change in the external event structure, dynamics support forward inference by highlighting the emotionally-relevant aspect of an event [58]. For example, observers are generally quite insensitive to elements of surprise (“wide-eyed”) in mixed expressions [19,42].…”

Emotion Recognition in Simulated Social Interactions

“…pairing of anger and fear proved to be strong in a simulated social situation, when the contextual face looked at the target [4], [5]. However, in these studies the two avatars did not 'interact,' in the sense that there was no mutual gaze.…”

“…Sixty-seven undergraduate students (64 females, 3 males; mean age: 19.2 ± 1.1 years) at Cardiff University, UK, participated in partial fulfillment of a course requirement. The sample size was defined in advance on the basis of results from previous experiments [4], [5].…”

“…Each displayed four blends of shame-sadness expressions and four blends of anger-disgust expressions. A pilot experiment conducted to control the ambiguity of the four expression blends of shame-sadness is presented as online material, and information about the four expression blends of anger-disgust was published as supplementary materials in [5]. The other four photofits were used for the contextual face and these displayed expressions of contempt, disgust, sadness or a neutral state.…”

“…Recent research using synthesized facial expressions shows that emotional reactions apparently directed towards a target avatar exert specific influences on the perception of the target's facial expressions [4,5]. This socioaffective inferential mechanism is particularly strong when two emotions share a functional relationship.…”