Emmanuella Sanchez¹, Emily Martin¹, Fabian Soto¹; ¹Florida International University, Miami, FL

Human face perception is shaped by experience, and one of its most robust phenomena is the Other-Race Effect (ORE)—superior recognition of own-race faces relative to other-race faces. This asymmetry is typically attributed to greater experience individuating own-race faces, though experience with categorizing other-race faces may also contribute. Experimental tests that can causally isolate these learning components remain limited, in part because natural face categories confound perceptual and social factors. The present study introduces a controlled training paradigm using parametrically generated “alien” races, constructed from anthropometric principles to mirror the structural variability found across natural human groups, to test how specific types of learning give rise to ORE-like recognition patterns. Participants were randomly assigned to one of three learning histories: identity training, in which they learned to identify and name 25 faces from Race A; categorization training, in which they learned to classify faces as belonging or not belonging to Race B; or combined training, in which both learning tasks were completed with different races. After training, recognition memory was assessed for all races to determine how each learning history shaped participants’ ability to individuate faces. The alien races reliably induced an ORE-like pattern: participants in the combined-training condition showed enhanced recognition for the race they had learned to identify and reduced recognition for the race they had learned to categorize, demonstrating that distinct learning experiences can drive asymmetric recognition performance even in the absence of real-world racial categories. Identity-only and categorization-only groups showed patterns consistent with these learning contingencies, though the combined-training condition provides the clearest demonstration of opposing effects within the same observers. These results establish that controlled, type-specific experience with artificial races is sufficient to induce an ORE-like recognition asymmetry, providing a rigorous experimental framework for testing how different learning histories shape the structure of face representations.

Acknowledgements: This research was funded by NSF grant 2319234 awarded to Fabian Soto