Understanding the conditions and content of face learning underlying race perception biases

People are bad at recognizing faces when they belong to people from another race, tending instead to group them together (“they all look alike to me”), in what is known as the other-race effect. Such racial biases in face perception play a role in a number of problems in society. For example, more than a third of wrongful convictions involve misidentification of suspects for another race, and implicit racial biases that have been linked to the other-race effect lead to discrimination in the legal, medical, and educational systems. The evidence clearly shows that we learn the other-race effect from experience with some faces but not others, but scientists don’t know how that learning happens and its effects in perception. This project’s goal is to understand what types of experiences lead to development of the other-race effect, and how those experiences modify how we perceive faces. We approach this goal by testing people using computer-generated faces and races, including new artificial races. Generating faces this way allows us to precisely control all their relevant features (for example, shape of the eyes, nose, etc.) and the experience people have with them. We also measure people’s face perception with high precision and use mathematical modeling to formalize our hypotheses. Knowledge gained from this project leads to a better understanding and measurement of the other-race effect, setting the stage for the development of training procedures to eliminate it. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.