What is knowledge? To know something is to correctly understand the world, so knowledge must involve the coming together of truth and belief.
Unfortunately, a definition of knowledge as “true belief” doesn’t work, because truth and belief can coincide by coincidence. For example, suppose that I have a high fever and hallucinate that my sister is in front of me talking to me, but in reality she is sitting quietly in a chair behind. I believe that she’s in the room with me, and it’s true that she’s in the room with me, but I don’t know that she’s in the room with me, because my belief that she’s in the room is based on bad reasons, reasons that are completely disconnected from the fact that she actually is in the room.
Thus, in epistemology, the traditional definition of knowledge was as “justified true belief.” The point of adding justification is to deal with cases in which I believe something for bad reasons but it coincidentally happen to be true.
Unfortunately, philosophers have recognized that this definition fails, by giving examples in which justification doesn’t do the work it’s supposed to. These examples are most often called “Gettier cases” or “Gettier problems,” after Edmund Gettier, who wrote a famously concise and highly influential article on the issue in 1963. Consider the following scenario (my paraphrase of an example posed by Roderick Chisholm):
I am standing in a field and I believe there is a sheep in the field with me because I can see what looks like a sheep. It turns out, however, (a) that I am looking at a dog disguised as a sheep, and (b) there is a real sheep behind the hill in the middle of the field. Thus:
- I belive that there is a sheep in the field. (I really do believe it.)
- My belief is justified. (Seeing what looks like a sheep is a good reason to believe that it is one.)
- My belief is true. (There is a sheep in the field, behind the hill.)
The problem here is that once again, my belief is true for the wrong reasons. Seeing what looks like a sheep is sufficient to justify my belief, but not sufficient to actually make it true. This failure, however, is perfectly offset by the coincidence that there is a sheep behind the hill. What makes the belief true is not the same as what justifies it.
The range of scholarly responses to Gettier cases is immense. Some philosophers use them to argue that the problem goes away on a fallibilist theory of knowledge, one on which one’s knowledge is always subject to refutation or revision. These approaches attack the truth element of knowledge, because they are open to the possibility that one can “know” false things. Other philosophers have tried to formulate a more rigorous justification element, one that rules out Gettier-style coincidences. (These efforts often seem to run afoul of Gödelian incompleteness; any definition of justification provides a roadmap for evading it.)
But the present author is by no means a philosopher, and it seems to me that that there is a much simpler response. If knowledge hinges on the nexus between what is actually true in the world and what justification we have for our beliefs, and the two can coincide or come apart by happenstance, then perhaps knowledge is not a philosophically important concept. Truth matters, and we can reason about what is actually true in the world. Belief matters, and we can assign moral weight to what a person does and doesn’t believe. Justification matters, and we can debate what evidence is sufficient to justify a belief. But knowledge, by itself, is too contingent to matter.
To me, then, the real lesson of Gettier cases is that the entire attempt to define knowledge in terms of true belief is a dead end. Introducing justification was important not because it patched up the definition—or could be made to patch it up with suitable revisions—but because it introduced the thing we actually ought to care about when we ask what people know. What are we justified in believing? is a great question, and one that is far more useful to ask than What do we know?