In the face of diversity: revealing the influence of ethnicity and culture on social trait face perception

Gosetti, Valentina (2025) In the face of diversity: revealing the influence of ethnicity and culture on social trait face perception. PhD thesis, University of Glasgow.

Abstract

On a daily basis, people make spontaneous judgements about who to trust or avoid based on facial appearance. Because of their considerable downstream consequences, a longstanding goal has been to understand which facial features drive social trait judgements. Despite emerging evidence showing that ethnic and cultural diversity influence social trait perception, most knowledge remains centred on White Western observers perceiving White faces. This bias questions the generalizability of prominent theories and feature-based models. In this thesis, I combine a data-driven reverse correlation approach with a generative model of 3D human faces to model the specific facial features of 3D shape and 2D complexion that drive perceptions of trustworthiness and dominance from three face ethnicities—Black African, East Asian, and White European—in two observer cultures—White Western and East Asian. Using information-theoretic analyses, I show that both White Western and East Asian observers perceive trustworthiness and dominance from a core set of facial features which are shared across face ethnicities, and map onto previous findings, plus novel face ethnicity-specific variations. These variations challenge the generalizability of prominent feature-based models and characterize the causal influence of face ethnicity on social trait perception. Further, while conceptually similar, the results for White Western and East Asian observers comprise different facial features. To formally test these differences, I next examine the cultural specificity of the modelled facial features across face ethnicities. Results show that, while White Western and East Asian observers provide similar social trait ratings for each face ethnicity, the features they base their ratings on differ. This questions previous claims of universality based only on rating comparisons. Further analyses reveal that the facial features specific to Western culture resemble specific emotion cues (e.g., smiling, frowning), whereas those specific to East Asian culture do not. This contrasts prominent theories such as emotion overgeneralization and highlights the Western-centric bias of current knowledge. Finally, I use a machine-learning approach and information-theoretic analyses to examine how face ethnicity, observer culture, and their synergistic interaction causally influence social trait perception. Results show that, across face ethnicities and observer cultures, social trait perception is driven by four feature sets: those that are shared, those that are face ethnicity-specific, those that are culture-specific, and those that are synergistic. Subsequent examinations of these feature sets confirm that they represent key sources of variance in social trait perception. These findings extend current efforts to quantify the relative contributions of the face, the observer, and their interaction and offer direct empirical support for modern theories of social trait perception. Together, this thesis responds to mounting calls to diversify psychological science by showing that ethnic and cultural diversity systematically alter the causal facial features for perception of key social traits, with direct implications for current knowledge and theory development.

• Item Type: Thesis (PhD)
• Qualification Level: Doctoral
• Subjects: B Philosophy. Psychology. Religion > BF Psychology; R Medicine > RC Internal medicine > RC0321 Neuroscience. Biological psychiatry. Neuropsychiatry
• Colleges/Schools: College of Medical Veterinary and Life Sciences > School of Psychology & Neuroscience
• Supervisor's Name: Jack, Professor Rachael E. and Schyns, Professor Philippe G.
• Date of Award: 2025
• Depositing User: Theses Team
• Unique ID: glathesis:2025-85597
• Copyright: Copyright of this thesis is held by the author.
• Date Deposited: 19 Nov 2025 10:03
• Last Modified: 19 Nov 2025 12:00
• Thesis DOI: 10.5525/gla.thesis.85597
• URI: https://theses.gla.ac.uk/id/eprint/85597

Related URLs

• Book section
• Pre-print
• Article DOI