Émilie Désaulniers¹, Marie-Pier Plouffe-Demers^1,2, Marie-Claude Desjardins¹, Justin Duncan^1,3, Daniel Fiset¹, Caroline Blais¹; ¹Université du Québec en Outaouais, ²Université du Québec à Montréal, ³University of Ottawa

The other-race categorization advantage (ORCA) is a well-established effect where individuals are faster and better at categorizing other-race than own-race faces. This effect is typically found in binary racial categorization tasks, where participants categorize faces of two races, usually their own and another. However, recent work raised concerns about the robustness of ORCA, showing that relatively minor changes in task parameters can substantially alter the observed pattern. In a previous ternary categorization task including Black, White and Asian faces, we tested Chinese, South African and Canadian participants (n = 648) and observed a robust reverse-ORCA, meaning a better categorization of own-race than other-race faces, especially in South African participants. This unexpected reversal suggested that task structure may critically shape whether an ORCA emerges, possibly by shifting the categorization strategies used. The aim of the present study was to test whether the ORCA would re-emerge when using the same stimuli but a different task structure. We used the original 216 morphed stimuli (ranging from 2 percent to 86 percent racial prototypicality of White, Black and Asian faces) in a binary categorization task, in which participants judged whether each face belonged to their own racial group or not . 216 participants (108 Canadians and 108 South Africans) completed the task and reported their contact with people of same and other racial groups. Perceptual sensitivity (d’) were calculated and analyzed in mixed 2 (groups: Canadians and South Africans) x 2 (condition: same vs other race) ANOVAs, revealing a significant interaction. Canadians showed a clear ORCA, while no significant differences were found in South Africans after Bonferroni correction. No significant correlations emerged between the contact measure and d’. These results indicate that the classic ORCA depends on a binary task context: introducing a third racial category abolishes the effect and can even invert it.