shared perceptual basis of emotional expressions and trustworthiness impressions from faces nikolaas...
TRANSCRIPT
Shared Perceptual Basis of Emotional Expressions and TrustworthinessImpressions from FacesNikolaas N. Oosterhof and Alexander Todorov
Presented by: Patricia SayeghMarch 15th 2010
Background• dsa
http://webscript.princeton.edu/~tlab/members/
• Social cognition and social neuroscience lab:• His research is focused on the cognitive
and neural mechanisms of person perception, in particular on face perception
• An assistant professor of psychology and public affairs at Princeton University,
Nikolaas Oosterhof:
Alexander Todorov:
Introduction•People use information about the dynamic
changes in the face to understand the mental state of others.
•We use facial features to help us identify a persons identity.▫Judge trustworthiness, aggressiveness,
dominance.▫Thus facial features will affect important
social outcomes.•Previously, Blair et al. (2004) observed
that Afro-American features predict criminal-sentencing.
Cesare Lombroso (1835-1909): An Italian criminologist and founder of the School of Positivist Criminolgy.• Believed that criminals could be identified
based on their physical defects, specific physiognomic attributes.
• large jaws, • high cheekbones• flattened/ squashed nose• thin beards and hair• sloping foreheads• fleshy lips• expressive faces• small wandering eyes • thick and close eyebrows
Introduction
•Previously, Blair et al. (2004) observed that Afro-American features predict criminal-sentencing.
Introduction
• Based on our interpretation of a single, static aspect of ones appearance, we make spontaneous and rapid options about their character.▫ Likability, trustworthiness, competence
and aggressiveness ( just on appearance!)
• Suggests that our judgments are based on subtle facial cues.
• Emotion overgeneralization hypothesis:▫ The emotional expressions we perceive from neutral
faces affect the traits or behavioural tendencies we attribute to that face.
Introduction
• Based on their past work, they found that happiness and anger ratings were significantly correlated with trust and distrust ratings.
• Angry and happy ratings are based on subtle facial cues like:▫ Angry: eyebrows become more U-
shaped and the mouth more ^-shaped. ▫ Happy: eyebrows become more ^-
shaped and the mouth more U-shaped.
• Our judgments are based on subtle facial cues.
• Emotion overgeneralization hypothesis:▫ The emotional expressions we perceive from neutral
faces affect the traits or behavioural tendencies we attribute to that face.
If our judgments are based on subtle facial cues then how
will angry versus happy faces affect our rating of trust?
Hypothesis/Prediction
•Hypothesis: expressions of anger and happiness are related to perceptions of trustworthiness.
•Prediction: ▫a) untrustworthy angry faces should be
perceived as angrier than trustworthy faces expressing the same emotion.
▫b) trustworthy happy faces should be perceived as happier than untrustworthy faces expressing the same emotion.
Methods
•Participants:▫60 undergraduate students
21 for selection of trustworthiness of neutral faces
39 participated in an animation study•Facial stimuli:
▫96 computer generated faces▫Emotionally neutral bald male Caucasian
faces ▫Rated on a scale from 1 to 8 on
trustworthiness▫Picked the top 5 trustworthy and 5
untrustworthy faces for the animation study.
Methods•Animation study:
▫51 frames of computer generated faces▫Emotional expressions were linearly added
to each face Used Facegen’s Anger and SmileOpen happy
expression control
Methods•Animation study:
▫Started with neutral face (frame 1) and emotion was added to either 25% or 50% (frame 51) Condition1: weak happy (25%) Condition2: medium happy (50%) Condition3: weak angry (25%) Condition4: medium angry (50%)
• Incongruent
Emotion added to same face
40 faces (10 faces x 2 emotions x 2 strength)
Methods•3 animations:
• Face stimuli• Congruent
Same trust level at start and end.
Face identity changes
Different trust level at start and end.
Face identity changes
Participants were asked to judge how happy or angry end face became.
Rated expression on a continuous slider scale
•Difference in perceived emotion was augmented when identity and trustworthiness changed. F(1, 38) = 114.9, p <.001)
•Incongruent versus congruentt(38) = 6.55, p < .001 / t (38) = 10.89, p < .001
• Incongruent versus same face
t(38) = 8.26, p <.001 / t(38) = 12.13, p < .001
•Expression of trustworthy faces were perceived as happier than untrustworthy faces.(M = 3.3, SD = 4.5, F(1, 38) = 158.19, p < 0.001 )
Results: Perception of happy emotions
•Incongruent versus congruentt(38) = 5.62, p < .001 / t (38) = 8.22, p < .001
• Incongruent versus same face
t(38) = 9.23, p <.001 / t(38) = 8.62, p < .001
•Difference in perceived emotion was augmented when identity and trustworthiness changed. F(1, 38) = 85.73, p <.001)
•Expression of untrustworthy faces were perceived as angrier than trustworthy faces.(M = -.4, SD = 4.9, F(1, 38) = 329.81, p < 0.001 )
Results: Perception of angry emotions
•Perception of trustworthiness increases perception of an emotion.
•Same emotion was added to all faces BUT:▫A) perceived trustworthy faces as happier
than untrustworthy faces.▫B) perceived untrustworthy faces as
angrier than trustworthy faces.
Discussion and Conclusions
•Perception of trustworthiness increases perception of an emotion.
•Same emotion was added to all faces BUT:▫A) perceived trustworthy faces as happier
then untrustworthy faces.▫B) perceived untrustworthy faces as
angrier then trustworthy faces.
Discussion and Conclusions
Changes along the trustworthiness scale intensifies
emotion perception, and are based on subtle facial cues.
•The researchers concluded that:▫Changes in expression of emotion affect
trait impressions.▫Valence is the main dimension along which
emotionally neutral faces are evaluated. Trustworthiness judgements as valance
evaluator for neutral faces. Valance evaluation triggers approach/avoid
behaviour. Happiness/ anger expression trigger
approach/ avoid responses.
Discussion and Conclusions
•Focused on angry and happiness because previous work suggested that these emotion are affected by trustworthiness (Todorov, 2008).▫Researchers acknowledge that other
emotions may also affect trust judgements.▫Which other emotions?
Maybe emotions where similar facial features are expressed.
Gender, facial maturity, similarity to self and facial textures have all been observed to interfere with trustworthiness.
Discussion and Conclusions
•Previously : trustworthiness is based on subtle facial cues of neutral faces (angry and happy expressions).
•In the current study, judgements on face trustworthiness affect perception of angry and happy emotions.
Discussion and Conclusions
Shared basis of perceptions of face trustworthiness and expressions of
anger and happiness!
•Hypothesis can account for dissociation between identifying a face and judging its trustworthiness.▫Prosopagnosics can not identify/remember
facial identity but can judge trustworthiness Todorov & Duchaine (2008).
▫Normal recognition of anger and happiness can preserve trustworthiness judgement.
Discussion and Conclusions
Shared basis of perceptions of face trustworthiness and expressions of
anger and happiness!
•Can you perceive emotion (happy or angry) and trustworthiness independently? ▫Is trustworthiness just a by-product of
experiencing an emotion?•What would be a good study to perform to
try to distinguish shared or distinct neural pathways for happy, angry and trustworthiness?▫To test this hypothesis more directly .
Questions for discussion
Questions for discussion
Winston et al. (2002) Nature Neuroscience
Some of the regions activated in the study by Winston and colleagues. -Fusiform gyrus (FG, green) and superior temporal sulcus (STS, red) process features of the face stimulus. Amygdala (AM, blue) associates perception of the face with an emotionalresponse to the face. Insula (INS, purple) participates in representing this emotional response as a feeling about the person whose face we view. Activation in STS can also be modulated by the task, demonstrating top-down influences and suggesting that most information flows in both directions along this circuit.
Questions?
Analysis
•Design :▫2 (emotion) x 2 (strength of emotion) x 2
(trustworthiness) x 3 (Morph).•2 x 2 x 2 x 3 Repeated ANOVA was
performed.