What are degrees of belief? A defence of comparativism

Abstract What are degrees of belief? Accounts of credence often assume that they can be represented by a numerical function and focus on defending various norms or constraints on credence given this. Without focusing on the question of what degrees of belief are, however, we see that various idealising assumptions that go into understanding credences as numerically representable are ignored. In this paper, I defend comparativism as an interpretation of what degrees of belief are. I outline the comparativist position and show how it is possible to represent an agent’s comparative beliefs using a probability function. I consider problems that have been presented for comparativism that suggest comparativism cannot meet the desiderata I have outlined. Going through these objections in turn, I argue that comparativism is a viable account.