If something like Reformed Epistemology is correct, an agent is innocent in regarding certain ways of forming beliefs to be reliable until those ways have been proven guilty. An important species of argument purporting to show guilt (1) identifies the ways of forming beliefs at the core of our cognitive activity, (2) isolates the features of our core practices which account for their reliability, and (3) determines whether or not peripheral practices which ought to have those features enjoy at least (...) their functional equivalents. An example. Sense perception is at the heart of our cognitive activity; a feature of sense-perception which provides us with confidence in its reliability is that we can subject sense-perceptual beliefs to intersubjective criticism - others can check our beliefs. Beliefs about God formed on the basis of religious experience cannot be so checked and therefore lack positive epistemic status.An important response to such criticism consists of arguing that the difference between two ways of forming beliefs is just what we should expect given some relevant difference between the subject matters of those two ways of forming beliefs. This species of response employs what I call ‘the Ontological Principle,’ viz., that the nature or characteristics of an object constrain the way an agent ought to form beliefs about that object.In this paper, I attempt to provide a rationale for the Ontological Principle. I argue as follows. Any epistemic norm which requires of an agent that she enter into causal relations with an object which she cannot in the ‘nature’ of the case enter lacks epistemic merit - it violates the ought implies can dictum. Because the epistemic norms properly governing the cognitive activity of a given agent are constrained by the causal relations possible between an agent and an object of belief, and because the causal relations possible between an object of belief and an agent are determined in part by the characteristics of the object of belief, the epistemic norms properly governing the cognitive activity of a given agent are determined in part by the characteristics of the object of belief. That is, the Ontological Principle is true. (shrink)