In western countries, people report holding fewer racist and sexist attitudes than a decade ago, yet discrimination and inequalities persist (Charlesworth and Banaji 2022). This issue can be examined at two interrelated levels: the systemic roots of racism and sexism, well-documented by sociological research, and the individual-level reproduction of inequalities through implicit associations between appearance or group membership and traits, beliefs or attitudes, which can bias behavior. To fight discrimination and inequalities, it is crucial to address both levels. At the individual level, experimental evidence suggests that people can exert some control over the expression of their implicit biases, yet the factors that facilitate or hinder this control are poorly understood. Our project aims at elucidating the cognitive mechanisms underlying implicit bias expression through the investigation of key contextual factors influencing bias control. We conduct experimental studies on first impressions of competence, using behavioral measures (choices and RTs) alongside computational modeling (drift-diffusion models) and physiological recordings (pupillometry and possibly EEG).