Red shape, blue shape: political ideology influences the social perception of body shape

In democratic societies, voting behavior is often assumed to be rational and based on candidates' political agendas. However, research demonstrates that physical traits, particularly facial features, significantly influence voting decisions. This study investigates whether body shape also plays a role in political trait attribution and whether this perception is modulated by voters' political identification. Previous research has established links between physical appearance and character judgments, including studies linking facial features to election outcomes and showing how political affiliation biases the perception of candidates' faces. While some research explored the relationship between body type and personality, the link between body shape and political trait attribution remains largely uninvestigated. This study aims to address this gap by examining if body shape elicits perceptions of political traits and whether individual political identification modifies these perceptions. The researchers hypothesize that, similar to facial cues, body shape influences the attribution of political traits, and that these attributions are further influenced by the individual's own political leanings.

The relationship between physical appearance and character has been a topic of interest for centuries. Early attempts like those by Kretschmer and Sheldon tried to correlate body types with personality and mental illness, although these attempts have largely been discredited. However, the concept of somatotypes (endomorph, mesomorph, ectomorph) has continued to be relevant in research investigating social biases in body perception. More recent studies have focused on the impact of body shape on various aspects of social perception, including the perception of emotions, personal identity, attractiveness, self-esteem, gender, and personality. These studies show that trait inference from appearance relies on automatic and involuntary cognitive processes which affect various life aspects, such as employment, medical treatment, legal decisions, and political elections. The cognitive system uses stereotypical mental representations to match visual input with specific social categories; hence, it is crucial to understand and visualize these representations to comprehend social judgments based on visual features. Previous research used statistical methods and 3D face models to visualize mental representations of stereotype faces, reconstructing synthetic faces that vary in social trait dimensions such as trustworthiness and dominance. While statistical face models have been utilized extensively, comparable research on 3D body perception is limited, despite the recent development of high-quality statistical body models made possible by advances in machine learning and body scanning technologies.

The study employed two online experiments using the SMPL (Skinned Multi-Person Linear Model), a statistical 3D body model. Experiment 1 aimed to determine whether body shape elicits political trait attribution. 492 participants (227 male, mean age 34.3 years) from Amazon Mechanical Turk (AMT), residing in the USA, were randomly divided into two groups (male and female body shape observation). Participants rated 256 randomly generated 3D body shapes (128 male and 128 female) using 25 political trait descriptors (including 'Republican' and 'Democrat') and 5 body shape descriptors (as controls) on a 5-point Likert scale. Linear regression was used to model the relationship between the 3D body shape parameters (principal shape components from SMPL's PCA) and the political trait ratings. A leave-one-out cross-validation approach was employed to assess model accuracy. Experiment 2 investigated the effect of political identification on this relationship. 487 new participants (190 male, mean age 34.7 years) from AMT, excluding participants from Experiment 1, rated 4 male body shapes (average male body from SMPL and stereotype bodies for 'Democrat', 'Republican', and 'Trustworthy' from Experiment 1) using 6 descriptors (4 political, 2 physical). Participants were categorized into Republican, Democrat, and Independent groups based on their self-reported political affiliation. Independent t-tests were used to compare ratings between groups. Data cleaning procedures, including catch trials, were implemented in both experiments to ensure data quality. In both experiments, participants provided demographic information, including political affiliation.

Experiment 1 demonstrated that body shape significantly predicts the attribution of political traits (p<0.001 for both male and female models). The model demonstrated statistically significant predictive power. Reconstruction error (RE) for the male model was 15.41 mm (SD 7.05mm) and 14.49 mm (SD 6.43mm) for the female model. These values are significantly lower than the reconstruction error for random models, indicating a clear relationship between body shape and trait ratings. Stereotypical 3D body shapes were visualized for each political trait. Analysis revealed significant correlations between political and physical trait ratings, highlighting associations between body shape perceptions and physical descriptions. For example, the male Republican stereotype was described as 'Heavyset', 'Stocky', while the male Democrat stereotype was 'Masculine', 'Lean'. Gender differences also emerged, indicating that political traits were assigned to bodies differently depending on gender. Experiment 2 confirmed that political identification significantly affected trait perception. The manipulation check validated that the stereotype bodies elicited corresponding political trait perceptions. Republican raters rated the Republican stereotype body as significantly less 'Republican' than Democrat raters. Conversely, Democrat raters rated the Democrat stereotype body as significantly more 'Democrat' than Republican raters. This difference was not attributable to differences in height, weight, or gender between the groups. Visualizing stereotype bodies for Democrats and Republicans revealed differences in their mental representations of political traits, highlighting in-group favoritism.

The findings demonstrate that body shape is utilized to make inferences about political traits, influencing how individuals perceive and interact with others in political contexts. This echoes findings on the influence of facial features on social judgments. The significant gender differences observed in trait perceptions may be attributed to social role expectations and existing gender bias in leadership perception and attractiveness judgments. Experiment 2's results corroborate the hypothesis that political identification significantly modulates the perception of political traits from body shape, a novel finding extending prior research on face perception. The observed in-group favoritism suggests that individuals attribute more positive characteristics to their own group, leading to potential biases in judgment. The study highlights unconscious biases, calling for awareness and mindful decision-making in political and other social contexts.

This research established a mathematical relationship between body shape and the perception of political traits, showing that this relationship is significantly influenced by the observer's political views. The study introduces a novel methodology for analyzing the perception of political traits from 3D body shapes, demonstrating that electoral decisions may be influenced by a candidate's body shape. Future research could explore the cross-cultural generalizability of these findings and investigate the interplay of other factors (e.g., age, socioeconomic status) in modulating body shape perceptions and political trait attribution.

The study primarily focused on US participants from AMT, limiting the generalizability to other populations and cultural contexts. The use of synthetic body shapes might not perfectly capture the complexity of real-world body variations and perceptions. Future studies could use more diverse populations and real-world images to improve generalizability and ecological validity.