Nozick, Robert (1938–2002)

3. Philosophical Explanations

Nozick’s second book, Philosophical Explanations, ranges unconfined over topics in metaphysics, epistemology and the philosophy of value. While here he makes important contributions to the issues of personal identity and the foundations of ethics, his treatments of knowledge and scepticism have attracted the greatest attention.

What is it to know something, as distinct, for example, from guessing correctly? The traditional analysis of knowledge asserts that knowledge is justified true belief. Yet this seems to have been refuted by ‘Gettier examples’: once we appreciate that some beliefs, while justified, are nevertheless false, it is a short step to realize that a justified belief might be true by accident or fluke (see Gettier problems).

To avoid such problems Nozick presents the ‘truth-tracking’ theory. This is an externalist theory, and Nozick’s own approach is a variety of reliabilism (see Reliabilism), in which, broadly, one has knowledge if one’s true beliefs are generated by a reliable mechanism. Nozick claims that, if ‘S knows that p‘, then four conditions are individually necessary and jointly sufficient:

Conditional statements (3) and (4) replace the appeal to justification in the traditional account. These are to be interpreted as subjunctive or counterfactual conditionals rather than as material implications. Using possible worlds semantics for counterfactual conditionals (see Counterfactual conditionals §3), the third condition can be read as: ‘In close possible worlds where p is not true, S does not believe p‘. If your belief is true by accident, there will be a close possible world in which you still have that belief, even though it is false. The belief fails therefore to meet the third condition, and so does not count as knowledge.

We can now see why Nozick calls his theory the ‘truth-tracking’ theory: knowledge requires not just that you believe the truth, but that your belief ‘tracks’ the truth among close possible worlds. Together the third and fourth conditions require that, not only do you believe the truth, but in reasonably similar situations where it is false you would not believe it, and where it is true you would.

Nozick’s analysis gives him a reply to the sceptic about knowledge. If we grant that you cannot know that you are not now a brain in a vat on Alpha Centauri, then it seems to follow that you cannot know anything else: for example, that you are now on Earth, reading. Nozick’s reply is to point out that the sceptic’s argument presupposes that ‘knowledge is closed under known logical implication’, and that his own analysis of knowledge shows that this assumption is false.

The closure principle is a relatively simple idea. If you know that p, and that p logically entails q, then according to the closure principle it follows that you know that q. The sceptic uses this principle (in reverse) to argue that we have no knowledge. If I am reading on Earth, then it logically follows (and I know that it follows) that I am not a brain in a vat on Alpha Centauri. Hence, according to the closure principle, if I know that I am reading on Earth, then I know that I am not a brain in a vat. But, the sceptic claims, as I do not know that, I do not know that I am reading on Earth. Indeed, I cannot know anything known to be inconsistent with my being a brain in a vat on Alpha Centauri.

The closure principle is central to the sceptic’s case. But on Nozick’s analysis the principle is false. Nozick points out that whether or not an individual knows a proposition depends on how their beliefs vary over a set of close possible worlds. But the possible worlds to take into account differ according to the proposition under consideration. Hence, I can know that p, and know that p entails that q without knowing that q, and the sceptic’s argument falls.

While an ingenious response, critics argue that Nozick’s analysis of knowledge should be rejected. For it has the peculiar consequence that I can know a conjunction ‘p and q’ without knowing one of the conjuncts q for the set of possible worlds in which ‘p and q’ is true is not identical to the set of possible worlds in which ‘q’ is true. Although this is a consequence Nozick points out himself, for many it is a reductio ad absurdum of his approach.