Toronto Ethnically Diverse face database: a multi-faceted stimulus set.

Abstract

Face stimuli are often used in psychological and neuroimaging research to assess perceptual, cognitive, social, and emotional processes. Many available face databases, however, have limited diversity in ethnicity, emotional expression, gaze direction, and/or pose, which constrains their utility to specific contexts. Having a diverse face database can mitigate these biases and may help researchers investigate novel topics that examine the effects of ethnicity on these processes. The Toronto Ethnically Diverse (TED) face database is designed to provide an open-access set of 271 unique White, Black, East-Asian, South-Asian, South-East Asian, Middle Eastern, Multi-racial, and Indigenous adult models. The TED database includes diversity in race, gender, pose, gaze direction, and three emotion variations (neutral, open-mouth happiness, closed-mouth happiness). Validation data of the stimuli based on judgments of the emotional expressions showed high inter-rater reliability and high accuracy as measured by proportion correct and Cohen's kappa scores. Intensity, and genuineness ratings are also presented for each model. The validation results for TED suggest that this face database consists of models displaying their intended emotions with high fidelity. This database will be useful to researchers seeking to study underrepresented groups and to other broad groups of researchers who are studying face perception.