Epistemology, or the Theory of Knowledge, is that part of philosophy concerned with knowledge: what it is, how we acquire it, how far it extends. As usual, such an account is pretty unilluminating. You have to dive in to the subject to see exactly what issues are being discussed.

To the extent that philosophy is a matter of giving a reasoned, synoptic account of how things are, one would expect epistemology to figure prominently, and from early on. If someone says "everything is really water" then a reasonable question to ask him is "how do we tell?" And any answer that is offered to that question is likely to provoke further queries itself, and so we have started into epistemology.

A very potted history

I think it is fair to say that something like this happened in the ancient Greek tradition with which philosophy in the West begins. Early contributors made astonishing claims about what the world is like and soon found that they had to give some account of why their particular version should be believed in preference to other views. This reflective activity led to a concern not only for what knowledge is like but also for what reasoning is like (and thus led to logic and associated issues). 

From very early it was seen that there are two major channels by which anything like knowledge reaches us: the senses (you can diversify a bit beyond the traditional five to include introspection, memory, and the testimony of other people) and our capacity to reason. The relative importance and varying contributions of these two channels provide a continuing topic of debate throughout the history of philosophy. We label friends of the senses "empiricists", those who stress our reason "rationalists". 

One of the earliest epistemological treatises to have survived is Plato's Theaetetus. It is interesting that Plato's focus there is on what one might call everyday knowledge: he argues, for instance, that knowing is more than simply believing truly by reference to the ability of a persuasive lawyer to get jurors to believe what the accused did or didn't do. Plato's point is that since they didn't see what happened, the jurors can't be said to know, even if their verdict turns out to be the true one. It is interesting because in his own positive philosophy Plato offers us an account of things that downplays ordinary everyday realities and surely deserves more explicit epistemology than he ever provides.

Plato's major successor, Aristotle, offers an account of scientific knowledge and a theory of perception, a theory that includes what we now think of as psychology or physiology, as much as philosophical approaches to these matters. But he is little troubled by the questions that typically occupy epistemologists. These were highlighted by the sceptical schools of thought that developed in opposition to various views enunciated mainly after Aristotle's time. The sceptics queried all claims to knowledge and many of them rejected any kind of endorsement of appearances. They encouraged the search for arguments against both commonsensical and philosophical positions and left a legacy of problems that have recurred throughout most of the subsequent history of the subject.

Jumping over more than a thousand years, the intellectual ferment of the Reformation, Renaissance, and later Enlightenment produced not only a blossoming of science that had a marked impact on how we construe knowledge but a long-drawn-out debate between empiricists and rationalists. Among the former we may include scientists such as Galileo (though Galileo's practice may not have been as empiricist as scientific propaganda supposes - he is here more for his work with telescopes than for reflections on motion on an inclined plane) and Boyle as well as theorists like Bacon, Hobbes, and Locke. Traditionally the line of British Empiricists continues with Bishop Berkeley, a keen critic of Lockean notions, and culminates in David Hume, whose humane naturalism and mitigated scepticism may be about as close as we have come to the truth (Quine has said that the Humean condition is the human condition, though, as Gellner has stressed, it is not what most people recognise as such). The line of European rationalists starts with Descartes, who perhaps more than anyone else made epistemology, and in particular the problem of scepticism, central to philosophy in the succeeding centuries, and goes on through Spinoza and Leibniz. At the end of the eighteenth century we come upon Kant who attempted a synthesis of both traditions.

It is worth stressing the comparative novelty of the theoretically based knowledge that came with these developments: everyone has a range of everyday knowledge of the world around them, of how people behave, and so on, but, apart from spectacular achievements in some areas of mathematics, and a quantity of descriptive knowledge of astronomy, natural history, and so on, we did not have much that could be called successful scientific knowledge of the world before the sixteenth century.  (We had extensive technology, but that is another matter.)   Our concerns in this course reflect these two large areas of knowledge: epistemology will mainly deal with any kind of knowledge; methodology will mainly deal with issues that arise more clearly for our theoretically grounded scientific knowledge.

The main areas in epistemology

Williams suggests that these issues interact.  He offers an attractive picture of how we should deal with them: by aiming at a reflective equilibrium among our views on the various issues.  Since this brings with it an epistemological position (indeed Williams' own preferred contextualism) he perhaps ought to indicate that it is a loaded picture, rather than the only sensible possibility.

Description and norm

The way I have characterised our problems it may seem that our task is to describe a particular extensive and deeply rooted type of thing or human activity: our knowledge and our cognitive activities.  One may wonder why we shouldn't leave this to the psychologists.  But Williams wants us to see that knowledge is not so straightforward.  'Knowledge' and its cognates are honorific terms; they embody evaluations or standards, telling us not merely what we have done but what we ought to do.  To say that Mary knows X is not merely to describe (Mary, or Mary and the world, or what-have-you) but also to pass judgment, to endorse.  So Williams wants to claim that epistemology is in a similar boat to ethics: both disciplines are concerned, not with what we do do, but with what we ought to do.

Some preliminary contrasts

Analysing concepts is a somewhat mysterious business and we can often usefully begin by turning to the language we use to express them. The hope is that regular differences within the language will reflect significant differences at a conceptual level. (For more on this approach to analysis, see my remarks in Argument Analysis.)

Let us begin by noting some grammatical or syntactic facts about the English verb, know. Philosophers nowadays tend to invoke three distinct constructions with this verb:

1. John knows NP [a nominal phrase] - John knows Mary, Paradise Lost, Pythagoras' theorem, a stockbroker, ....
2. John knows how to VP [a verbal phrase] - John knows how to swim, to calculate the area of squares on a right-angled triangle, ....
3. John knows that S [an indicative sentence] - John knows that Bridgetown is the capital of Barbados, that the square on the hypotenuse is equal to the sum of the squares on the other two sides of a right-angled triangle, that the sun will rise tomorrow, ....

The main point for us now is that traditional epistemology has had virtually nothing to say about practical knowledge and tends to regard acquaintance as a combination of 3 with perception, so that most of the time we should think of knowledge (as far as epistemology goes) as of type 3.  (Things get more complicated here because Russell, from whom I have taken the terminology knowledge by acquaintance, used it to mean a somewhat more mysterious relation between a person and something else.)

The next contrast to note is not obviously enshrined in our language, but is the venerable philosophical distinction between a priori knowledge and a posteriori knowledge. The latter is knowledge that we can only acquire through relevant experience or observation, the former is knowledge that we can arrive at without any reliance on "relevant" experience (though, of course, if you don't have any experience at all you wouldn't be there to enjoy any knowledge). To discover whether most bachelors had domineering mothers you would need to undertake surveys, to discover whether bachelors are unmarried you have only to think (assuming you grasp the relevant bits of English or whichever language you are using). To discover that the sun is about 93 million miles away we had to make extensive observations and calculations; to discover that there are an infinite number of prime numbers we have only to reason. [For a brief exposition of this last proof, you can check the Wikipedia article on primes.]

Traditionally there is another contrast, not of the basis for our knowledge but of kinds of truth (or falsehood): analytic and synthetic. The terminology is due to Kant, who explained it in terms of the Aristotelian logic of his time, by saying that in an analytic truth the predicate was already contained within the subject (all bachelors are unmarried - unmarried is a component of what we mean by bachelor) whereas in a synthetic truth the predicate adds something new (all metals expand when heated - their behaviour when heated is not part of what we understand by the term metal). Pojman (What Can We Know?) retains this Kantian account of the distinction but I think most other philosophers who want to use it have altered it so that analytic truths include any that are logically true. They then usually explain the rest by saying that they are true in virtue of the meaning of the terms involved. Anything not analytic is synthetic. (As we will see, Quine has argued many people out of an acceptance of the utility or even meaningfulness of this distinction.)

I have explained analytic and synthetic with reference to truths. It ought to be obvious how the terms can be applied equally well to falsehoods. All bachelors are married is analytically false.

Although the distinctions are different, you can see from my use of the same example that they seem to be closely related. One way to characterise the difference between empiricism and rationalism is to say that for an empiricist there are really only two kinds of item: analytic truths known a priori and synthetic truths known a posteriori. A rationalist, and in particular Kant, holds that there is a third category: synthetic truths known a priori. [Traditionally no one would have bothered about the fourth possibility - analytic items known a posteriori - since they would have indicated a failure of clear thinking. Would anyone pay for a survey to discover whether bachelors are unmarried? No one would have thought that there could be analytic items that could only be known a posteriori.]