Abstract: 

The orthodox view in epistemology is that belief is constitutively evidence-responsive: if you can’t change what you believe in response to evidence, then you don’t believe it. Why should we think this? I propose a new argument for this view.

In line with recent work on function-first epistemology and conceptual engineering, I focus on what the/our concept of belief is for. I argue that the concept of belief functions to mark attitudes that are in the space of (epistemic) reasons. Specifically, it marks attitudes that we appropriately attempt to regulate by offering evidence and argument. When we ascribe beliefs, we are not merely describing a state of the agent, but also attempting to regulate agents into compliance with epistemic norms, generating rationality loops. I argue that the attitudes that fill this role in our practices of regulation are evidence-responsive, in the sense that subjects have the capacity to revise them in response to evidence.

I then explore additional benefits of holding that belief is constitutively evidence-responsive. These include benefits for taxonomizing attitudes and handling borderline cases; for the empirical study of belief; and for relating to others in cases of deep divergence. For these reasons, we should preserve this notion of belief as opposed to reforming it.