Abstract

I defend a novel account of the justification of testimonial beliefs within a general reliabilist framework. I argue that an agent's testimonial belief is justified only if the agent is discriminative about testimonial reports. To be discriminative about a testimonial report requires that an agent not form a testimonial belief, had the testimonial report been different in relevant ways. In developing this account, I provide a way of determining the relevant testimonial reports an agent must be capable of discriminating that appeals to the evidence available to the agent, and I discuss the different mechanisms that allow agents to be discriminative about testimonial reports. ;My account treads a middle ground between two prominent accounts, Weak Individualism and the Blind-Trust account. According to Weak Individualism, the justification of a testimonial belief must be grounded in non-testimonial evidence, while according to the Blind-Trust account, a testimonial belief is justified if it is formed on the basis of a disposition to blindly believe others. I show that Humean Weak Individualism leads to skepticism, and the Blind-Trust account licenses gullibility. On my account, since an agent has to be discriminative about testimonial reports and since an agent can be discriminative without possessing the non-testimonial evidence required by Weak Individualism, my account avoids both of these consequences. ;Because I adopt a general reliabilist framework, I also provide solutions to two serious problems raised for reliabilism, the Generality Problem and the New Evil Demon problem. I offer a solution to the Generality Problem, the problem of identifying the unique process that produces epistemic results in accordance with our intuitions. If the reliabilist appeals to the process that reflects the actual dynamics of belief formation, the account will produce the correct epistemic results. I also develop a solution to the New Evil Demon problem, a problem involving cases in which agents form justified beliefs that are produced by unreliable processes. I argue that if the reliability of a process is evaluated in worlds that are fixed by the evidence that is available to agents, a reliabilist account will capture our intuitions in these cases