Body Weight Can Change How Your Emotions Are Perceived

Abstract

Accurately interpreting other's emotions through facial expressions has important adaptive values for social interactions. However, due to the stereotypical social perception of overweight individuals as carefree, humorous, and light-hearted, the body weight of those with whom we interact may have a systematic influence on our emotion judgment even though it has no relevance to the expressed emotion itself. In this experimental study, we examined the role of body weight in faces on the affective perception of facial expressions. We hypothesized that the weight perceived in a face would bias the assessment of an emotional expression, with overweight faces generally more likely to be perceived as having more positive and less negative expressions than healthy weight faces. Using two-alternative forced-choice perceptual decision tasks, participants were asked to sort the emotional expressions of overweight and healthy weight facial stimuli that had been gradually morphed across six emotional intensity levels into one of two categories-"neutral vs. happy" (Experiment 1) and "neutral vs. sad" (Experiment 2). As predicted, our results demonstrated that overweight faces were more likely to be categorized as happy (i.e., lower happy decision threshold) and less likely to be categorized as sad (i.e., higher sad decision threshold) compared to healthy weight faces that had the same levels of emotional intensity. The neutral-sad decision threshold shift was negatively correlated with participant's own fear of becoming fat, that is, those without a fear of becoming fat more strongly perceived overweight faces as sad relative to those with a higher fear. These findings demonstrate that the weight of the face systematically influences how its emotional expression is interpreted, suggesting that being overweight may make emotional expressions appear more happy and less sad than they really are.

Fig 1. Experimental stimuli. A. Exemplar facial stimuli used for the neutral-happy judgment task. B. Exemplar facial stimuli used for the neutral-sad judgment task.

All facial stimuli were computer-generated and no actual faces were used in our study. A total of four computer-generated identities (two males and two females) were used in each experiment. Emotional expression and bodyweight of facial stimuli were systematically manipulated by using a morphing software. Faces have emotion gradients ranging from 0% (neutral emotion) to 100% (full emotion; either happy or sad) by increments of 20%.

Fig 2. Affective perceptual judgment task. A. Sample screen of the neutral-happy judgment task (Experiment 1). B. Sample screen of the neutral-sad judgment task (Experiment 2).

Participants were asked to make perceptual judgments about the emotion of faces (neutral vs. happy; neutral vs. sad) in a two-alternative forced-choice procedure. The spatial locations of emotion category labels were counterbalanced across participants.

Fig 3. Psychometric response curve modeling by Naka-Rushton contrast response function.

X-axis represents stimulus intensity level and Y-axis represents response probability. The stimulus intensity in this study represents the incremental increase of emotional intensity of facial expressions. The response represents the proportion of happy (experiment 1) or sad (experiment 2) decisions in a two-alternative forced choice task. The C₅₀ or PSE (Point of Subjective Equality) parameter indicates the perceptual decision threshold. A leftward shift of the psychometric curve (red arrow) would constitute evidence for a decreased perceptual threshold for condition A compared to the control condition, and a rightward shift of the psychometric curve (blue arrow) would constitute evidence for an increased perceptual threshold for condition B compared to control condition, respectively.

Fig 4. Behavioral findings. A. Average probability of happy decisions as a function of bodyweight and emotional intensity of faces (Experiment 1). B. Average probability of sad decisions as a function of bodyweight and emotional intensity of faces (Experiment 2). C. Response times of happy decisions (Experiment 1). D. Response times of sad decisions (Experiment 2).

Error bars denote the standard error of the mean. * p < .05, ** p < .01.

Fig 5. Psychometric curve fits. A. Happy decisions psychometric curves by body weight (Experiment 1). B. Sad decisions psychometric curves by body weight (Experiment 2).

For each bodyweight condition, psychometric curves were separately fitted by using the Naka-Rushton response function. Compared to healthy weight faces (gray dashed line), a leftward-shift of the psychometric curve of overweight faces (red line) in the neutral-happy judgment task and a rightward-shift of the psychometric curve of sad faces (blue line) in the neutral-sad judgment task were observed. A dotted horizontal line represents the 50% probability of a happy or sad decision.

Fig 6. Correlation result.

A scatter plot of the relationship between AFA (Anti-Fat Attitude)–fear scale and C₅₀ parameter differences (C₅₀ Overweight − C₅₀ Healthy weight) in the neutral-sad judgment task. Higher C₅₀ difference represents larger bodyweight modulation (the rightward shift of the psychometric curve) on sad decisions. The solid line represents a linear fit.