Man, scepticism sucks. It's so hard to get out of. I wrote some stuff about what I think is the best solution, but now I read back through it I realise it's maybe not successful at all.
The problem of whether we can know we have knowledge is illustrated by this simple form of the sceptic’s argument:
(1) S can know P (some item of empirical knowledge) only if S can know Y (that I am not dreaming/a brain in a vat etc).
(2) S cannot know Y.
(3) Hence, S cannot know P.
The unsatisfactory nature of ‘justified true belief’ as knowledge has led to many attempts to add conditions or totally replace the notion all together. In this writing I will be positing Nozick’s as the most successful of these. Appealing to just common sense, many are prone to just not accepting the sceptic’s radical conclusion, because logically one would just want to advance in an anti-sceptical way. However, conditional accounts such as Nozick’s (1981), seek to show how knowledge can exist despite the sceptic’s argument, and therefore show that S CAN know P without knowing whether or not they are a brain in a vat. This in turn leads to the denial of the closure principle: If S knows P and knows that P entails Y, then S knows (or can be in a position to know) Y. This leads to the denial because if you accept the closure principle, you should be in a position to know that you’re not dreaming or a brain in a vat etc.
Nozick is looking for an account of knowledge that gives you necessary conditions, that is to say any case that fails them would not be an instance of knowledge, and jointly sufficient conditions, that is to say any case that satisfies all of them will be an instance of knowledge. The subjunctive conditionals he formulates are as follows:
(N 1)p is true
(N2) s believes that p
(N3) if p were not true, s would not believe that p
(N4) if p were true, s would believe that p.
How condition (N3) works is illustrated by the Gettier example of believing one person in the office has a Ford, when he doesn’t, but the stranger also in the office does, so I am justified in my belief that someone in the office owns a Ford, but surely I do not know that someone does. If no one in the office owned a Ford, however, according to condition (N3), to have knowledge I would have to believe that no one did own a Ford. However, if the stranger did not own a Ford in this example, I would still believe the first person did, as I did before and thus condition (N3) makes this case not a case of knowledge. Many have added a fourth condition to ‘justified true belief’ different to (N3), that is that a belief in p does not rest on any false beliefs. However the barn example mentioned by Nozick himself; stopping by a real barn in an area where there are a lot of facsimiles of
barns and not knowing the barn stopped outside is real, poses a problem for this clause.
However condition (N3) is not enough, as it only shows belief being sensitive to falsity, and not to truth. Condition (N4), then, completes the account. In addition to p’s truth meaning s would believe it, it also entails that if P were true in sufficiently similar circumstances, i.e. in the ‘close’ worlds where p is true, s would still believe that p. For Nozick, this answers the problem of the brain in the vat, as it does not satisfy condition (N4), for it could be true someone was a brain in a vat, but they could be stimulated to believe they were not (or just not stimulated to believe they WERE). So for Nozick, ‘to know that p is to be someone who would believe it if it were true, and who wouldn’t believe it if it were false’.
What is unique about Nozick’s conditional account is the notion of ‘tracking’ that he introduces. When a person truly believes p, and conditions (N3) and (N4) are met, his belief for Nozick tracks the truth that p: ‘to know is to have a belief that tracks the truth’. This is to say that our knowledge has some sort of relation to the world. For Edward Craig, however, ‘as a weapon against scepticism, the [tracking] analysis is either impotent or redundant’. He argues this with regards to Nozick’s idea of close possible worlds. Craig calls the worlds where we are subject to some illusion ‘sceptical worlds’, for the purpose of his argument.
Craig claims if any sceptical world is a close possible world, then Nozick’s account fails and the sceptic will win. Therefore Craig thinks that for Nozick to succeed with his account, there will have to be no sceptical world close to the actual world, and therefore the actual world is not a sceptical world, for it is close. The problem with this, Craig thinks, is that to posit that the actual world is not a sceptical world (which is essential for Nozick’s account to succeed, as explained) then the sceptic must have already been overcome some how without any reference to the ‘tracking’ analysis.
Craig’s claim seems plausible, however Brueckner in his answer to Craig questions whether Nozick can show that closure is false without assuming that the actual world is not a sceptical world, as if he can, Craig’s critique causes no problem. He thinks that Craig is mistaken. In Brueckner’s example, he shows that it can be possible for ‘Moore knows he has hands’ to be true when ‘Moore knows that he is not in a sceptical world’ is false. This argument is outlined below:
Moore is inhabiting world w, a non-sceptical world. In this world w, Moore has a belief Q – that he has hands. In worlds where Moore does not have hands, Moore does not believe he has hands. So this meets condition (N3). It also satisfies (N4) as in worlds where Moore has hands, he correctly believes he has hands. So on Nozick’s account of knowledge, Moore knows he has hands.
Furthermore, Moore knows that Q entails R - he is not in a sceptical world being deceived into thinking he has hands. With respect to R, it does not satisfy (N4) since in ¬R worlds Moore erroneously believes that R: in these sceptical worlds he believes he is not in a sceptical world. So, on Nozick’s account Moore does not know R. This example shows that, as stated before, s CAN know p without knowing y. This counter-example makes no reference to the actual world, only a different world ‘w’ which could be disparate to the actual world. So Brueckner concludes that now the sceptic can no longer appeal to the closure principle to maintain the premise (1). This seems to successfully eliminate the objection put forward by Craig.
Nozick very clearly points out that he is not trying to refute the sceptic, but only to explain how knowledge is possible even if we must accept it is logically possible we are a brains in vats etc. Nozick believes the sceptic’s argument is ‘playing’ on condition (N3), to show the condition is not satisfied, so they seem to be saying: ‘even if p were false, S would still believe that p’. However, Nozick claims that his subjunctive conditionals are very different to the entailment that the sceptic is assuming is implied. For Nozick, the existence of a potential circumstance in which p is false yet S believes that p does not falsify (N3).
B. J. Garrett puts forward an objection to Nozick’s conditional account of knowledge that also seems to fail. He cites an example where X believes that A’s father is a philosopher, based on being told that B’s father is a philosopher. A and B are brothers. If we apply Nozick’s theory to this situation, X knows that A’s father is a philosopher, even though he does not know A and B are brothers. Garrett wants to refute this; on the grounds that it was just ‘good luck’ that they happened to be brothers that X’s method worked. However, the necessity of brothers always being brothers renders this objection somewhat void. As David Gordon argues, X is using a reliable method, whether or not he knows he cannot use it for unrelated people. It is not, as Garrett says, a matter of luck, it in fact could not have come out otherwise. As Gordon says, ‘why is it a requirement of knowledge that one have good grounds for thinking one’s method reliable?’. Gordon goes further and says even if X has an unreliable method, is Garrett’s example still a counter-example to Nozick? Nozick’s conditional knowledge account is not trying to add to the justified true belief thesis of Gettier, but is seeking a replacement. Due to this, Garrett’s example is not a true counter-example, as it employs X using an unreliable method, when Nozick is not really concerned with justified true belief.
Gordon finally suggests the difference between Nozick and Garrett is as follows in Garrett’s concluding paragraph, when he says that Nozick must be able to solve ‘the problems which arise for his theory from certain situations in which a belief that Ø is based on a belief that Ψ (where the believer is unaware of the logical or conceptual relation which holds between Ø and Ψ)…’ For Nozick, it is not required to be aware of these logical or conceptual relations to know that Ø. These relations show why the belief is true, but you don’t have to know this to have a method that resulted in knowledge. The essential conclusion of Gordon’s article is that Nozick does not include justified belief in his conditions, so any objections made on the grounds of unreliable methods pose no problem for Nozick’s conditional account of knowledge.
Nozick’s account has had many objections, and I have shown two major ones and how they do not really pose a problem for Nozick. Nozick’s two subjunctive conditions seem to hold in cases of knowledge. As an attempt to show how one can have knowledge despite the sceptic’s possibilities, I think Nozick has succeeded. There may indeed be problems with Nozick’s account, but I find his account the most persuasive. For Nozick, s can have knowledge without knowing they are not a brain in a vat, which satisfies the former ‘justified true belief’ thesis that posed such a major problem for Gettier.
əˈfɪz