Gettier Cases and the Traditional Analysis
I. The Traditional Analysis of Knowledge (TAK)
A. According to TAK, propositional knowledge is justified, true belief.
1. It would appear that I cannot know P if I do not even believe P. Take ‘belief’ here to pick out a variety of pro-attitudes, all of which require a commitment to P.
2. It seems that one cannot know P unless P is true. Thus, I cannot know that the Vikings won the 2008 Super Bowl because, well, they didn’t. (Note, though, that I can know that the claim, “The Vikings won the 2008 Super Bowl”, is false, due to the fact that it is true that this claim is false.)
3. Finally, to rule out lucky guesses where I have a true belief that doesn’t qualify as knowledge, it is important that my true belief be justified. That is, I must have some compelling story or account to give of P that is itself grounded in reality. This account should go quite a distance toward establishing that P is true. Justification is what closes the gap between relying on luck and being confident. (Not that confidence is sufficient for knowledge, as we all appreciate.)
B. This is a way of thinking about knowledge that goes back at least to Plato, and was a largely unchallenged part of Western intellectual culture for more than two millennia. But why?
1. Some of this has to do with the power of persuasion wielded by folks like Plato, Aristotle, and after them, Descartes. Smart guys.
2. Most of it has to do with the fit between this analysis and the paradigm cases of propositional knowledge that are often considered by analysts. If you pick a proposition—say a scientific claim, or perhaps a claim about the perceived world—it seems like you can say that if you believe it, you’re justified in believing it, and it’s true, then you know it; conversely, if you know it, people have a right to expect it to be true and you to believe it and have justification for your belief. Very little seems to be missing here.
3. While it isn’t the only show in town (see the “On the Analysis” handout), it has received top billing for most of the history of western philosophy.
II. The Gettier Challenge
A. It’s important to understand the centrality of TAK, as well as the unchallenged sway it held over epistemology, when turning to the Gettier cases. It is said that Gettier won tenure and promotion with his small paper, and given the reaction to it, that is not surprising. These cases are counterexamples that seem clear to many in this game, and they can be reported so simply and quickly that they it surprises many that they were not identified before.
B. The Gettier Cases
1. These are counterexamples to TAK, as they demonstrate that one can have justified, true belief without that qualifying as knowledge.
a. These counterexamples show that justified, true belief is not sufficient for knowledge—you need something more, or perhaps something different. (The latter would be a real surprise, as it would involve jettisoning the other necessary pieces, but perhaps that is the way to go.)
b. TAK is supposed to be universal and apply to all instances of knowledge, so a counterexample’s job is only to show that it fails to be general by failing to cover all the examples of knowledge.
2. Counterexample #1 (from Gettier): Smith has strong evidence that Jones has 10 coins in his pocket and he also has strong evidence that Jones will get the promotion, so Smith has strong evidence for the conjunctive proposition, “Jones is the man who will get the job, and Jones has ten coins in his pocket”; Smith believes this, and deductively infers from it the equally justified conclusion that the man who will get the job has ten coins in his pocket; this is true, but as it turns out, Smith gets the job and he has (by a stroke of epistemic luck!) exactly 10 coins in his pocket; thus, Smith has a justified, true belief that does not count as knowledge.
3. Counterexample #2 (from Gettier): Smith is justified in believing that Jones owns a Ford, and proceeds to conclude from this the claim, “Jones owns a Ford, or Brown is in Barcelona”, which is valid and so justified even though Smith is just guessing on Brown’s whereabouts; turns out, Jones is just driving a rental but Brown is in fact in Barcelona, so Smith’s disjunctive belief is true and justified, even though it doesn’t count as knowledge.
4. Counterexample #3 (modified from Chisholm): Smith is justified in believing that there is a deer in a field, having just seen one of those lookalikes that Fish and Game put up to catch drive-by hunters; his son asks him if there is a deer in the field, and he responds affirmatively; as it turns out, there is a deer in the field—one Smith didn’t see—so he has a justified, true belief that doesn’t count as knowledge.
C. It is fair to say that these caused a panic in epistemology, encouraging the creation of a cottage industry devoted to figuring out how to deal with the counterexamples. This was something we thought we got right, so this challenge was a shock.
1. A conservative response would be to reject the claim that these are truly counterexamples. This could be done either by arguing that you can’t have justification for the false claim each of these purported counterexamples begin with, or that justification isn’t preserved when you generalize away from the false claim to the true one. The former strategy would appear to make justification otiose, while the latter seems to fly in the face of the idea that deductive inference just pulls out what’s already there.
2. One way is to embrace justification, truth, and belief as necessary conditions, but to argue that they are not the only necessary conditions. Something else (perhaps itself complex) must be added to yield a jointly sufficient condition. (Remember that the goal is to generate a definition of ‘knowledge’ with this analysis, and that requires a necessary and sufficient condition.) Thus, many epistemologists ran screaming into the street in search of the ever elusive “fourth condition”.
a. Feldman points to the condition according to which one cannot have false grounds for what one claims to know. In these examples, there is a false ground for the conclusion that is true. However, this would appear to be too strong due to the fact that we might have more than one ground for a particular claim (pp. 31-33).
b. Another condition would be that there can be nothing such that being justified in believing it would undermine one’s justification for the proposition in question. But this is also too strong, due to examples involving subjunctive conditionals and misleading defeaters (pp. 33-36).
c. Feldman opts for the idea that justified true beliefs must not depend essentially on any falsehood. This shores up the first suggestion, ensuring that the falsehoods at issue are those that are involved in justifying the proposition one claims to know.
3. Another response, though, is to jettison the whole mess (or at least part of it) and start to look elsewhere. One could argue that the empirical approaches (beginning with the causal approach of Goldman) caught fire precisely because of a crisis of confidence in TAK. These include approaches that wish to avoid the problem of bringing justification and truth together by leaving justification out in the cold.
III. Analyzing the Gettier Challenge
A. The bottom line would appear to be that because justification isn’t what makes a belief true, it is always possible to have justified beliefs that are true but not because of what justifies them. The thinking was that the more and better justification you had, the closer you got to the truth. Gettier cases reveal that justification can be more and better without getting you any closer to the truth or any farther away from just gettin’ lucky.
B. In these cases, you start with a justified belief in a false proposition and then use deductive generalization (existential generalization in (1) and (3), disjunctive generalization in (2)) to infer a second proposition that is true. This second proposition is also justified because it is a valid conclusion of a deductive inference from a justified premise. But the second premise is true for reasons unbeknownst to Smith. Thus, it is just a matter of luck that Smith’s belief is true, given his justification. (Another person, Williams, could believe the same thing and be justified in believing it, and thereby be in the know even though Smith isn’t.)
C. Feldman’s Analysis
1. The cases rest on the following principles:
a. The Justified Falsehood Principle (JF): It is possible for a person to be justified in believing a false proposition. (p. 28)
b. The Justified Deduction Principle (JD): If S is justified in believing p, and p entails q, and S deduces q from p and accepts q as a result of this deduction, then S is justified in believing q. (p. 28)
2. Both of these principles seem correct—justification understood as evidence can be had in truckloads for propositions that turn out to be false, and deductive inference would seem to be justification preserving if anything is. (Remember: a valid, deductive inference guarantees the truth of the conclusion given the truth of the premises, and so if the premises are justified, the inference should do nothing to threaten that justification; put another way, the justification of the premises and the validity of the inference should suffice to guarantee the justification of the conclusion.)
D. How to Make a Gettier Case
1. One way to understand Gettier cases involves knowing how to make them. What follows is a recipe for taking any proposition whatsoever and constructing a Gettier case out of it.
2. Step 1: select any false proposition, P, for which some believer A has ample justification.
3. Step 2: generalize away from P using a principle of deductive logic to a claim Q that is true but not for the reasons adduced by A in support of P.
a. The principles of choice are deductive generalization and existential generalization. These support transfer of the justification from P to Q, but the reasons are no more the truthmakers for Q than they were for P.
b. These principles produce generalizations that have the following property: that only one claim of the possibly many over which the generalizations they produce range need be true in order for the generalization to be true. Thus, while the original claim would make the generalization true if it were true, so would other claims at the same logical level (i.e., the same level of abstraction) as the original claim. Thus, if you have a true disjunction, for example, M or N, it might be true because N is true even though M is false.
c. Other generalization principles include conjunctive generalization and universal generalization, but both of these require for truth that all of the claims they generalize over be true. Thus, they would not be appropriate for generating Gettier cases.
3. Step 3: Have A believe Q on the basis of the justification that was transferred deductively from P.
4. Step 4: Q will be a justified, true belief for A that will not count as knowledge.