What is it that is justified? Not propositions: they are true or false, but it would be strange to say that a proposition is justified independently of anyone believing it to be true (or probable). Was the heliocentric theory justified before anyone had any beliefs about it? It was true, but it wasn’t justified. So is it beliefs that are justified—or claims or statements? This is the way people usually talk in epistemology (“knowledge is true justified belief”, “the belief that ghosts exist has no justification”), but it is subtly wrong. This kind of talk makes it sound as if the relation of justification holds between evidence and beliefs without any mediation by a rational subject—as if the subject plays no essential role in the epistemic set-up. Evidence justifies belief and that is all. The belief can be justified or not, rational or not, independently of the epistemic subject. Evidence transmits justification to belief directly. The justification relation holds between evidence and belief.
But that is not how we normally talk about justification. The canonical form of an ascription of justification is, “S is justified in believing that p”. It is the subject that is justified not the belief he or she forms. At any rate, the primary bearer of justification is the rational subject, with the belief’s justification following as a secondary and derivative matter. Just as it is the agent that is justified in acting in certain ways, so it is the agent that is justified in holding certain beliefs. You can see this from the fact that if you say of someone, “Her belief that p is justified” this could be true even though she is not justified in believing that p. The belief that p might be justified in virtue of evidence possessed by other subjects, while this subject has formed the belief irrationally. In order to determine if her belief is justified, in the intended sense, we need to ask whether she is justified in holding that belief. Generally, to say that the belief that p is justified is shorthand for saying that there are people who are justified in believing that p. When people are justified in holding certain beliefs we can say that those beliefs are justified, but not otherwise. It is not the belief state itself that is justified; it is a subject’s holding that belief. Similarly, it is not the assertion itself that is justified when someone makes an assertion; it is the speaker’s making that assertion. We cannot abstract the subject away from the relation of justification, as if justification merely involves relations between evidence and mental states or acts of speech. Justification is a triadic relation between evidence, belief, and person, not a dyadic relation between evidence and belief.
Thus the person enters epistemology essentially. The basic thing that is justified is the person: I am justified in believing certain things. What is it that is justified? Me, you, him, her—we are justified in how we have formed our beliefs. There is something odd in the idea that a belief state itself could be justified—just as it is odd to suppose that an action, construed as a bodily movement, could be justified. That is like supposing that a state or act could be reasonable or rational; people are reasonable or rational. Compare: words don’t refer—people do. If I am rational in forming a belief, then we can say derivatively that my belief is rational, but we must not forget that such rationality flows from the epistemic subject. It is the self that rational or irrational, reasonable or unreasonable, justified or unjustified. We can say derivatively that an assertion is justified too, construed as a vocal utterance, but it would be strange to say that a pattern of sound is the primary bearer of justification, not the subject who produces that sound. The canonical form of ascription here is “S is justified in asserting that p”. This is why I am responsible if I lapse from good epistemic practice—not my speech act or my belief. The critic will say to me, “Youwere not justified in believing or asserting that p”. In such a case I need to be more careful, not my beliefs or assertions. It would be a category mistake to criticize my mental states or speech acts—they are not the delinquents.
This has a bearing on how we talk about verification and falsification in the sciences. It is not that theories, construed as sets of propositions or sentences, are verified or falsified, as if this can take place in a personal vacuum; rather, people are justified in believing certain theories, given the evidence, or in believing the negation of those theories. To say that a theory has been falsified is just to say that people are justified in believing the negation of that theory; theories don’t get to be falsified independently of mental acts by persons. It is not that evidence somehow confronts theories by itself and renders them verified or falsified; the evidence has to go via persons who evaluate that evidence. Scientific theories are verified or falsified only because scientists have justifiably formed various beliefs about them. That is the structure of an epistemic fact: E is evidence for T if and only if E provides a justification for believing T in subjects S. Hence epistemology must be personalized: it cannot just deal in sense data and beliefs or sensory stimuli and behavioral assent—it must recognize persons as indispensable components in the process of justification. Foundationalism, for example, must be formulated as the doctrine that all justification depends on rational agents being justified in holding a set of foundational beliefs; and coherentism must be formulated as the doctrine that justification consists of rational agents forming their beliefs in a coherent manner. It is not a matter of beliefs per se having solid foundations or cohering with other beliefs; it is a matter of persons basing their beliefs on a foundation or ensuring that their belief systems are coherent. This is the right way to think about the structure of justification—not as something abstracted away from persons. Similarly, when considering skepticism we should ask whether people are justified in believing in the external world or other minds, not whether the corresponding propositions are justified or even whether such beliefs are justified. The skeptical question is whether I am justified in believing in the external world or other minds: am I being rational in holding commonsense beliefs of these kinds? If I am not, that is a fault of mine. The skeptic is criticizing me—I am the one failing to live up to the requirements of rationality. It isn’t that formulating the questions in this way helps to solve them—it may even make them more difficult—but this is conceptually the correct way to do it. Wherever there is justification there is a subject justified. Epistemology may or may not be naturalized, but it should it personalized. 
 Why has the self not figured more prominently in modern epistemology? I reckon it is because of an assumed empiricism about the nexus of justification: the self has not been amenable to empiricist treatment and is generally regarded with suspicion by empiricists, beginning with Hume. When empiricism began to take a more materialistic form, spearheaded by Quine, the self was even less on the list of approved entities—especially the self as conscious, reflective, and norm-sensitive. Justification began to be conceived as a mere triggering of internal states by physical stimuli, not as a person rationally evaluating evidence. Evidence must be received by an epistemic subject and then used by that subject as a justification for belief or assertion; it doesn’t just feed directly into belief or assertion.