An Examination of the Epistemology of Prejudiced Belief

In “Racial Prejudice and Friction,” John Dewey writes: “Too often we try to discuss race prejudice morally before we have dealt with it scientifically. That is, we justify or condemn it before we understand it.” Dewey’s remark applies well to contemporary work on the epistemology of prejudiced belief, in which there are dueling tendencies either to condemn or to justify these beliefs without sufficient care. In this dissertation, I chart a middle path between these extremes. I first critique Amia Srinivasan’s radical epistemology, according to which judgments about the epistemic status of morally and politically problematic beliefs should be guided by considerations of moral and political utility. I argue instead that we must uphold the priority of the epistemic, keeping separate epistemic evaluations of problematic beliefs and concerns about the utility of our epistemological judgments. I then examine the role of testimony in the formation of prejudiced belief. I defend testimonial reductionism, which holds that hearers must have sufficient positive reasons to regard speakers as trustworthy in order to be justified in believing what they say. I then argue that while prejudiced testimonial beliefs can be epistemically justified, this is true most often when these beliefs are formed in highly isolated and evidentially impoverished environments. Most prejudiced agents in societies like our own, however, are not in such dire epistemic straits. So, although perhaps not all prejudiced beliefs are epistemically unjustified, there is prima facie reason to regard such beliefs with suspicion. Finally, I discuss prejudiced beliefs based on biased perceptual experiences, disputing Susanna Siegel’s position that biased experiences can be irrational and, when they are, this irrationality transmits to beliefs formed on their basis. Drawing on Anil Gupta’s Conscious Experience, I present a coherentist account on which the epistemic status of an agent’s perceptual beliefs depends o