What is Knowledge?
Let’s begin with the obvious. Suppose that the brain-in-a-vat hypothesis given in the previous post were true. In that case, a great many of our beliefs about the external world would be false, and this, without needing skeptical argument, would be sufficient for our lack of knowledge on these matters. So falsehood is a disqualifying condition for knowledge, and what is just the same thing, truth is a necessary condition. But, what is almost as obvious, truth is not sufficient for knowledge either. There are as many truths as there are falsehoods, perhaps even an infinite number of each, whereas the knowledge of such beings as yourself and I are partial and finite. Moreover, it is beings like you and I to whom that knowledge belongs, and thus is essential to knowledge that it be possessed by a knower. So we need some further condition which constrains knowledge in these respects, and it is customary to take this extra condition to be belief.
Still, our definition is not nearly sharp enough. For consider that our skeptical hypothesis (“a possible state of affairs that is both consistent with the evidence we have, and is inconsistent with our usual hypothesis based on that evidence”) manages to cast knowledge into doubt even if we insist that our beliefs about the external world are true. Such hypotheses may be laughable, yet it seems their mere possibility is enough to challenge our knowledge claims. And so we need to add to our definition a third condition, one which, even if it does not confute the skeptic, shows why skeptical arguments are effective.
The traditional proposal, dating all the way back to Plato in the 4th Century BC1, is that our third condition is justification. One motivation for this condition is the idea that if one knows, the connection between one’s belief and the truth of it is somehow more secure than in the case where one merely has true belief. Knowledge is steadfast, and so assertions of knowledge are more trustworthy than assertions of truth alone. Another motivation is that to say that one knows is to commit oneself to being able to show that their beliefs are true. Such ‘showing’ is quite naturally identified with the verbal practice of justifying oneself, and so our new theory of knowledge as justified true belief (hereafter JTB) is an obvious development. Indeed, it remained the orthodoxy for two millennia.
Or so it is commonly supposed. In reality, there was no such widespread consensus on the concept of knowledge, and different philosophers tended to pursue a conception of knowledge based upon their own metaphysics. What is true is that the most prominent analytic accounts of knowledge from the first half of the twentieth century each gave a very similar breakdown of knowledge to the JTB theory, and so it came as somewhat of a watershed when Edmund Gettier published his 1963 criticism of such theories. Gettier exploits the fact that the justification for a belief need not entail the truth of that belief, a subtlety which allows that knowledge on the JTB theory can be a matter of luck, where the justification for some reason or other fails to connect one’s belief to the truth. Here is an example2:
“Knowledge” is sometimes defined as “true belief”, but this definition is too wide. If you look at a clock which you believe to be going, but which in fact has stopped, and you happen to look at it at a moment when it is right, you will acquire a true belief as to the time of day, but you cannot be correctly said to have knowledge.
The example comes from Bertrand Russell, and though the target of it is the view that knowledge is merely true belief, it works just as well for the JTB theory. We would surely grant that the appearance of the clock is sufficient for justifying a belief as to the time, and so it looks as though all three conditions are met. Yet, intuitively, this is not knowledge.
Various proposals have been put forward to accommodate Gettier-style examples – that we should shore up the justification condition or replace it, that we should add some fourth ‘Gettier’ condition, that we should abandon the search for necessary and sufficient conditions in favor of a ‘family resemblance’ between cases of knowledge, or, what may or may not be the same thing, that we should recognize knowledge as inherently unanalyzable and thus a theoretically primitive notion3. Rather than adjudicate between these options, I’ll sketch the kind of theory of knowledge I tend to think in terms of, leaving the specifics to a later post. The theory I have in mind is called ‘Reliabilism’.
Reliabilism would have us replace the justification condition. In short, it is the view that knowledge is true belief produced by a means that is reliable in the circumstances. Of course, much rides on the phrase “in the circumstances”, and it is this which allows a treatment of Gettier cases. Taking Russell’s example above it is this qualification which explains why the passerby lacks knowledge, for although the clock is ordinarily a reliable time-keeper, obviously once it has stopped it is no longer reliable. It also explains the intuition that led us to the justification condition in the first place – when one claims knowledge, one is claiming that their beliefs originate in a reliable process, and providing justification is a verbal demonstration of the impeccable heredity of a belief.
Yet, despite the similarities between Reliabilism and the JTB account, Reliabilism heralds a serious wind-change for epistemology. Whereas the traditional JTB account requires that the security of one’s belief be transparent (or accessible) to the knower, Reliabilism allows that we can have knowledge without access to facts concerning the reliability of our beliefs. The philosophical terminology that goes with this distinction labels the former construal of the justification condition as ‘internal’ and the latter as ‘external’, corresponding to the relation that each describes relative to the human mind. Hence accounts such as the JTB theory are known as ‘internalist’, and accounts such as Reliabilism are known as ‘externalist’.
This conception of knowledge appears to me to be roughly correct. So in the next post I’ll concern myself with applying it to the problems raised by skepticism, showing both where we can resist the skeptic’s arguments, and what is right about them.
1 His Theaetetus dialogue.
2 Bertrand Russell, Human Knowledge: Its Scope and Limits (1948), p. 91. It is odd that Russell’s example should have been overlooked, despite it’s publication fifteen years prior to Gettier’s paper. It’s also odd that Gettier’s paper is held in such esteem when one of it’s principle targets, A.J. Ayer’s view in The Problem of Knowledge, is misrepresented. Ayer is much closer to the Reliabilist accounts in his conception of knowledge as “having the right to be sure” than he is to internalist JTB theories. It has been said that Gettier’s paper at once set the agenda for future work in epistemology and created a myth about the subject’s past. I can’t help but agree.
3 In whatever philosophical theory you like, there will be unanalyzable terms, since breaking down concepts into their component parts is not a process which can continue forever. In the end we must rest our theory on such notions, and so if the theory is to be illuminating as a whole, the unanalyzables at the bottom of it had better be concepts we intuitively grasp. Those who think we do have an intuitive grasp of knowledge seek to analyze other epistemic concepts in terms of it.