Justification and Belief in the Gettier Problem — Epoche B2
What Zagzebski's Result Actually Rules Out: Gettier, Aptness and Objectively Sufficient Reasons A pessimism narrower than it looks Linda Zagzebski's 'The Inescapability of Gettier Problems' (1994 [1] ) is usually summarised as showing that no fourth condition can rescue the analysis of knowledge as justified true belief. Presented that way it looks like a counsel of despair, and it is regularly set against the optimism of virtue epistemology, as though the two were opposed camps. This review argues that the summary misstates the result and that the opposition is fictitious. What Zagzebski shows is that no analysis can succeed if it adds to true belief a property that is closely connected to truth but does not entail it [2] . That leaves two exits, and both are taken by real theories. Ernest Sosa takes one of them [3] ; so, in a different vocabulary, does the Mexican philosopher Luis Villoro [4] ; and so does Zagzebski herself [5] , who is not the sceptic in this story but the author of one of the accounts. The genuine disagreement is not between pessimists and optimists. It is about whether an analysis that gives up reduction has explained anything. The analysis and its failure The tripartite analysis says that a subject knows a proposition when she believes it, it is true, and she is justified in believing it. It is customary to attribute this to Plato [6] , and the attribution deserves a moment's care: the Theaetetus considers true judgement accompanied by an account, and rejects it, ending without a definition. What the dialogue supplies is the question, not the answer. Edmund Gettier's 1963 paper gave two cases in which the three conditions are met and knowledge is absent. In each, the subject reasons validly from a well-supported but false belief to a true conclusion, and the conclusion is true for a reason having nothing to do with the evidence. The belief is true; the justification is real; the connection between them is severed. Half a century of proposals followed, each adding a fourth condition designed to reconnect them: that the justification depend on no false premise, that the belief be caused by the fact, that it be produced by a reliable process, that it be sensitive to the truth. The recipe Zagzebski's argument is a procedure for defeating any such proposal. Take an analysis on which knowledge is true belief plus some property X, where X is a good-making feature closely tied to truth but not guaranteeing it. Then two steps suffice. First, since X does not entail truth, there must be possible cases in which a belief has X and is false. Construct one. Second, alter the case so that some accident makes the belief true, while leaving X in place and leaving the accident unconnected to X. The result is a belief that is true, that has X, and that is plainly not knowledge, because its truth came from the accident. The structure is a double dose of luck: bad luck that would have made the belief false, cancelled by good luck that makes it true. Any X that admits the first step is exposed to the second. That is the whole argument, and it is general because the only property of X it uses is the failure to entail truth. Zagzebski draws the consequence herself. If the recipe is to be blocked, either the good-making property must entail the truth of the belief, or knowledge must not have the form of true belief plus an independent component at all. These are not counsels of despair. They are specifications for a theory. The first exit: aptness Sosa's account meets the first specification. He distinguishes three appraisals of a performance, taking archery as the model. A shot is accurate if it hits the target, adroit if it is competently taken, and apt if it hits the target because it was competently taken. Belief is a performance with truth as its aim, so a belief is apt when it is true because it manifests the believer's competence, and knowledge is apt belief. Aptness entails accuracy. There is no such thing as an apt belief that is false, since aptness is success attributable to competence and there is no success to attribute. Zagzebski's first step therefore cannot be taken: there are no cases of a false but apt belief from which to build. Sosa's own treatment of interference cases shows this working rather than failing. Consider an archer whose shot is pushed off line by a gust and pushed back by a second. The shot is adroit and accurate, and it is not apt, because hitting the target is not attributable to the competence — the wind did it. Sosa uses the case to explain his distinction, and it is a mistake to press it against him as a counterexample. It is the distinction's clearest illustration. The second exit, and where Zagzebski stands The other specification is met by abandoning the additive form. Zagzebski's own Virtues of the Mind (1996) defines knowledge as a state of believing the truth arising out of acts of intellectual virtue, where the truth is reached because of those acts. Truth is not a separate conjunct sitting alongside a virtue condition; it is inside the condition, as its upshot. Her 1994 paper is the argument for building a theory this way, which is why reading it as scepticism about the project inverts its purpose. An older distinction, put to work The same manoeuvre appears in a tradition that the anglophone literature rarely consults. Luis Villoro's Creer, saber, conocer (1982) separates three things that English and Spanish both blur: believing, knowing that something is so, and knowing something by acquaintance. His condition on the second is that the subject possess reasons that are objectively sufficient, and the qualifier carries the weight. Reasons may be subjectively sufficient — enough, from where the subject sits, to settle the matter for her — without being objectively sufficient. The distinction is Kant's [7] , from the Canon of Pure Reason, where holding-something-true is graded as opinion when it is sufficient neither subjectively nor objectivel