Consider three structurally similar cases of social bias. Mary’s application for graduate school in mathematics is rejected by the traditionalist Mr. T, an evaluator who has written a series of books arguing that women have a natural disposition toward being worse at abstract, logical thinking than men. Her application for a different program is rejected by oblivious Ms. O, an evaluator who avows egalitarian principles but finds that Mary just seems less suitable for the program, for reasons that go unarticulated and would not pan out under pressure. Her application for a third program is rejected by Hal, an automated program that is trained on past admittance data about which students, when accepted, have gone on to successful careers in the field.
This dissertation argues that there is a natural kind social bias that all three cases fall under and defends a theory of what that kind is. My theory explains how the cases are unified, how they differ, and why the differences between the cases matter. Within a computational theory of mind, the tasks of unification and differentiation can appear to be at odds with one another. The more we highlight differences among how Mr. T, Ms. O, and Hal were processing informational states, the harder it is to use those same computational resources to say what they have in common. My analysis reconciles these tasks within a cognitive science framework by shifting to a higher level of abstraction.
I argue that social bias is a functionally defined mental entity that takes propositional mental states as inputs and returns propositional mental states as outputs in a way that mimics inductions made on the basis of social kind membership. All three cases of bias relate the input that Mary is a woman to the output that she’s not suitable for a mathematics program. Like functional analyses of other mental states, my analysis of bias entails that it is multiply realizable by a variety of computational systems and decision-making processes. For instance, biases could be realized by an explicit belief that women are ill-suited for mathematics (as Mr. T has), by an unconscious, automatic association between women and the stereotypical property of being bad at math (as Ms. O has), or as patterns in how informational states are organized, even when those states are not about specific values or stereotypes, but instead reflect systematic patterns in how our society is organized (as is happening in the case of Hal). Throughout the dissertation, I explore the implications of this explicated notion of bias for the organization of the mind, theories of consciousness, and the system-dependence of biases.
