Justification in the Gettier Problem — Epoche B2
What Reliabilism Is Missing: Access, Defeaters, and the Case of Norman The problem For a long time philosophers analysed knowledge as justified true belief: S knows that p just in case p is true, S believes p, and S is justified in believing p. In 1963 Edmund Gettier published a three-page paper [1] , 'Is Justified True Belief Knowledge?', which showed that the three conditions are not sufficient. Take his second case. Smith has strong evidence that his colleague Jones owns a Ford: he has often seen Jones driving one, and Jones has just offered him a lift in it. Smith has no idea where another acquaintance, Brown, is; he has never given the matter a thought. From his well-evidenced belief about Jones he infers a disjunction, choosing the second disjunct at random: either Jones owns a Ford, or Brown is in Barcelona . A valid inference from a justified belief yields a justified belief, so Smith is justified. As it happens Jones does not own a Ford — he is driving a rented car — but Brown, by coincidence, is in Barcelona. So Smith's belief is true, believed, and justified. He plainly does not know. The moral is that justification, as traditionally conceived, does not connect a belief to its truth in the right way. The truth can arrive by accident from a direction the justification knew nothing about. Reliabilism, stated accurately The most influential response came from Alvin Goldman [2] , and it is worth separating two of his proposals that are often run together. The first, in 'A Causal Theory of Knowing' (1967), is a theory of knowing : S knows that p only if the fact that p is causally connected in an appropriate way with S's believing it. In Gettier's case there is no such connection between Brown's being in Barcelona and Smith's belief, so the case is excluded. The second, in 'What Is Justified Belief?' (1979), is a theory of justification , and it is what 'reliabilism' now standardly means. A belief is justified if it is produced by a reliable cognitive process — one that tends to produce a high proportion of true beliefs. Perception in good conditions, memory, and valid inference from justified premises are reliable; wishful thinking and guessing are not. Reliabilism is an externalist theory: what makes the belief justified is a fact about how it was produced, and the believer need not have any awareness of that fact. Internalism , by contrast, requires that whatever justifies a belief be accessible to the believer by reflection. One consequence is regularly misreported and should be stated plainly: reliabilism about justification does not by itself solve the Gettier problem. Goldman did not claim that it did [3] . A reliably formed belief can still be true by luck. Goldman's own example [4] , in 'Discrimination and Perceptual Knowledge' (1976), makes this vivid. Henry drives through a district and sees what is in fact a barn. His eyesight is excellent and the conditions are good, so the process is reliable and the belief is true. But unknown to him the district is full of barn façades, and had he glanced at any of the others he would have believed the same thing falsely. He does not know. Goldman credits the case to Carl Ginet. The internalist complaint The objection that presses hardest on reliabilism is not about Gettier cases at all. It is that a process can be reliable while the believer has no business believing what it delivers. Laurence BonJour put the case memorably in The Structure of Empirical Knowledge (1985 [5] ). Norman is a completely reliable clairvoyant. He receives, from time to time, accurate beliefs about matters he could not have learned about in any ordinary way. He has no evidence that he has this power. He also has no evidence against it; he has simply never considered the question. One day he finds himself believing that the President is in New York. He has no idea why he believes it. The belief is true, and it was produced by a reliable process. Reliabilism therefore says Norman is justified. That verdict is hard to accept. From Norman's own point of view, the belief has simply arrived. He cannot cite anything in its favour. If he were to ask himself why he holds it, he would find nothing. Believing something in that condition looks less like knowledge than like a lucky hunch that happens to be correct — and Norman's situation is not improved by a fact about his reliability that he has no inkling of. The tempting repair, and why it fails The obvious response is to add an access requirement: a belief is justified only if the believer can, on reflection, produce the grounds on which it rests. This is the repair the internalist tradition suggests, and it is the one this essay's original thesis proposed. It is too strong, and the standard objection to it is decisive. Small children know that the dog is in the room and that their sister is angry. Non-human animals have knowledge of their surroundings. Ordinary adults, most of the time, believe what they see without formulating any grounds at all; asked why they think there is a cup on the table, they can only repeat that they can see it. If justification required an articulable ground, almost none of our perceptual beliefs would qualify, and the theory would have solved the Norman problem by abolishing knowledge. Here a case makes the point better than an abstraction. In Japanese craft workshops a shokunin — a master of a trade, a joiner or a swordsmith or a plasterer — commonly acquires discriminations that resist statement. A joiner of forty years' experience picks up two boards of the same species and grade, and says one will move and the other will not. He is right far more often than chance, and far more often than an engineer with instruments. Asked how he can tell, he may say only that the wood tells him, or that one feels dead. Does he know? Surely yes. He is right, reliably, about a matter within his competence, and his workshop is full of the evidence: the joints that held, the doors that still close.