What can we learn from Gettier?

Edmund Gettier, in 1963, published a short paper  (1,000 words) which stirred up and keeps stirring up debate in epistemology. In it he gave what have been thought to be counter-examples to the plausible-sounding idea that knowledge is to be defined as justified true belief. (I discussed his paper in an earlier entry.) In these examples someone is said to hold a true belief, to be justified in holding it, and yet not be said to know it. In response, philosophers have tried to formulate ways of getting around the problem, others have proposed different examples, etc. (See http://www.iep.utm.edu/gettier/.)

Many of the examples used in the debate are utterly artificial or farfetched – that, I would argue, is a problem in itself. I wish to get back to that issue in a later entry. However, the following example seems to be fairly straightforward: as I hurry past the big clock on the railway station, I notice that it shows six. However, without my knowing it, this clock, which I have always relied on in the past, actually stopped at six this morning. On the other hand, it actually is six p.m., and I believe it is. So I have grounds for believing it to be six, it is six, and yet one might question whether I know that it is six.

So far so good. The question that intrigues me is what contribution this debate is making to an understanding of knowledge. What would happen if we found ourselves at a loss for a definition of knowledge? First of all, what purpose would a definition of knowledge serve? Clearly, we don’t need one for pedagogical purposes: it is not a question of conveying the use of the words “know” or “knowledge” to people who are not familiar with these words. (Proposal for a thought experiment: try to imagine life without these words or equivalent words in some other language.)

Sometimes definitions are created in order to formulate standards for the application of a word. We do this when we introduce technical terms, or give a specialized application to a word currently in use. A “child”, for the purposes of buying a train ticket, is anyone under the age of 12, etc. But neither can this be what we are after here. The discussion is not about reforming the use of these words, nor about sharpening the criteria; rather, it is concerned with giving an account of how they are actually used. Remember: what seems to make Gettier’s examples counter-examples is the fact that they seem to fulfil the standard we presumably accept, yet we are not ready to accept them as cases of knowledge. (Hanfling, however, has argued that Gettier-type examples might qualify as knowledge in certain cases.)

Another way of thinking about the task of definitions in philosophy is to think about them as providing an overview of the use of some expression in the language. On this view of the task, the problem is not that speakers are at a loss when it comes to using certain expressions; the difficulty arises when we try to give an account of the use. The Gettier examples have been taken to show that the picture we had of the use of “know” and “knowledge” was erroneous, and philosophers have been trying to amend the picture ever since.

The conventional way of thinking about this task is to assume that we have to identify some condition, some state of things, to which the words refer: “knowledge = X”. To say that I, or somebody else, knows something is to maintain that the condition X is fulfilled with respect to the matter in question. But would anyone seriously wish to defend the definition Gettier was trying to undermine?

When we speak about knowledge, would we ever be inclined to substitute “justified true belief” for the word? First of all, the whole discussion doesn’t seem to fit first person present tense uses of “know”. That’s evidently because there’s something strange about first person present tense uses of “justified”. Just imagine the following conversation:

“Are you sure your sister will find the way?” – Suppose instead of answering (1) “I know she will” I said, (2) “My sister will find the way, I believe she will, and I am justified in so believing”, or for short, (3) “I’m justified in believing she’ll find the way.”

The suggested paraphrases sound bizarre. Perhaps if I said (3) I meant something like “I have grounds for believing … “; but if that’s what I mean, it would be misleading for me to say “I know…” (I hope get back to the sense of “justified” in a later entry.)

What then about third person uses? A to B: “She knows that those mushrooms have to be parboiled.”

(1) “So you don’t need to tell her.” The question whether they need to be parboiled doesn’t arise. Both A and B take it for granted that they need to be parboiled to get rid of the poison. The question how Sue came to know it doesn’t really matter.

(2) Paraphrase: “Sue knows how to cook mushrooms. She says they have to be parboiled. So we can rely on that.” They assume that Sue has the competence, and from that they infer that the mushrooms have to be parboiled, arguing from knowledge to truth, as it were.

(3) “So we’ll have to think of another method of poisoning her.” Again, the truth is taken for granted, her justification doesn’t matter, the main thing is that A and B know how Sue will act.

(These examples were suggested by Oswald Hanfling’s chapter “Knowledge and the uses of ‘knowledge’”, in his book Philosophy and Ordinary Language: The Bent and Genius of Our Tongue, 96 ff.)

One point to note here is that there is a shift between foreground and background in these examples. In (1) and (3) the facts are taken for granted, what is foregrounded is Sue’s relation to the facts. In (2) we argue from Sue’s assertion to the facts. So we might say, in some cases the truth of the judgment in question is being presupposed, in other cases the truth is what we wish to get at. The definition which is the target of Gettier’s critique fails to acknowledge this distinction. Another related point is that we need to distinguish between the question of justification and truth. Suppose it is discovered that those mushrooms don’t really have to be parboiled; the story of their being poisonous is really a myth, though one that is generally believed at the time. After learning this, we might say that people used to believe they had to be parboiled, not that they knew it. But that doesn’t mean that A and B were misusing the word “know”, since they believed the judgment – and its truth was not at issue.

I believe that most attempts to define “know” miss both these points. Indeed, this is probably what is bound to happen if you start out looking for a definition. A definition is supposed to give you the conditions for a word truly to apply to some object or circumstance. This presupposes that the conditions for its application can be determined independently of a context of use. When it comes to the widely ramified use of a word like “know”, this blinds us to the dynamic of the word’s actual roles in the context of human interaction. The problem may not be so evident when it comes to words or expressions like “titanium”, or “king’s gambit”, or “referendum”, where the purpose of asking for a definition may be taken for granted. (Cp. Avner Baz on “the theorist’s question”, in his book When Words are Called for: a Defense of Ordinary Language Philosophy, chap. 3).

So I suggest the lesson from Gettier should be, not that some other definition should be found, nor that the lack of a credible definition constitutes a crisis in epistemology, but rather that the search for a definition is not the right way of trying to get a clear understanding of how we operate with the word “know”.