Abstract
When performing race categorization, people respond faster to other- compared to own-race faces (Valentine et al., 1992; Levin, 1996; Caldara et al., 2004). The so-called other-race categorization advantage (ORCA) is also accompanied by a category boundary shift resulting in participants needing less other-race signal (i.e.prototypical characteristics) to categorize a face as such (Benton and Skinner, 2015). However, a large majority of ORCA studies have relied on a binary (i.e., own- vs. other-) race categorization task. In this study, we examined whether the typical category boundary shift still occurs when participants have to classify morphed faces into one of three categories. For each of the two gender profiles (i.e. Male, Female), 108 stimuli were generated by morphing triads of face images of different races (i.e. 2 Black, 2 White, 2 Asian), with race prototypicality ranging from 2% to 86%. A total of 286 participants (94 Black [South-African], 96 White [Canadian], and 96 Asian [Chinese]) each completed 432 trials on online platforms (Prolific and Pavlovia). Categorical boundaries (i.e., the threshold at which a face is categorized as e.g., Black, with a 50% probability) were defined separately for each stimulus and participant race. Other-race thresholds were averaged on a participant basis (e.g., [White + Asian]/2 for Black participants). A 2 (stimulus race: own, other) x 3 (participant race) mixed ANOVA was carried. Results show a main effect of stimulus race which interacted with participant race. Strikingly, there was a generalized inverse-ORCA: that is, participants were globally more sensitive to own- vs. other-race information. However, the interaction revealed that this effect was largely driven by the Black subsample, and also marginally by the White subsample. The Asian subsample, on the other hand, showed the typical categorical boundary shift. This highlights the influence of task parameters on social cognition measures.