Nozick's Knowledge and Skepticism

Nozick’s paper, Knowledge and Skepticism is divided between those two areas:

Section 1, Knowledge:

In Nozick’s account, knowledge is still

1) p

2) Bp

However, the simplest form of the truth tracking replaces Jp with 2 clauses:

3) ¬pà ¬Bp. Where à is subjunctive as opposed to material. Material conditional is satisfied if right now when ¬p then ¬Bp, à is satisfied only if both right now, and in other possible situations ¬p then ¬Bp. This clause is a way of formalizing a casual connection between facts and belief. It shows stipulates that knowledge varies according to the facts.

4) pà(Bp&¬B¬p) (the extra clause to take care of where there are contradictory beliefs) This clause also shows that connection: it shows that knowledge adheres to the facts.

Clause 3) takes care of the Gettier counterexamples, but why 4)?

-Brain in Vat, whose knowledge of BiV is induced by scientist.

-news of dictator’s death by assassination suppressed.

[Would this cover a case of belief instilled by a god? (see next section, but how does it relates to Brain in Vat case as excluded above?)]

However, this is not Nozick’s final formula:

2) Bp via M(ethod), or BMp

3) (¬p & BM’p’)à ¬BMp. Where ‘p’ indicates whether or not p

4) (p&BM’p’)à BMp

The new clauses are to clarify cases where had you not used M you would not have believed P, such as where he only gets information via a book or the grandmother case.

However, what do we do in the case that something is a) over-determined b)situational multiple methods (i.e. p->BM₁p _, ¬p->BM₂p).......?

Section 2, Skepticism:

This section is not aimed at refuting the skeptic, rather showing that knowledge is possible even given what he says. But, he divides the sceptical issue into 4 sections:

A. Denying condition 3

This is the normal sceptical issue: how are we justified in believing p if we may be BiV. Put in Nozick’s terms there is a world in which even if p were not true, I would still believe it. But he objects that we needn’t rule out all possible worlds.

Nozick claims that the skeptic’s aim to say that we do not meet condition 3 (maybe it’s just to deny p?), because even if someone was a BiV they still might believe that they were not. To this Nozick retorts that 3 need not be true in every case that p is false, rather just in the closest cases (again, as all condition 3 needs to do is to show that there is a connection between p’s being true or false, and your believing it)

B. Denying p

But, this is not the skeptic’s only problem: What if we truly don’t know that we are not BiV, then surely we don’t know p either because K(p->q) [closure principle]

Nozick’s answer here is straightforward in content, but tricky in terms of explanation. He denies the closure principle. Why? Because knowledge varies according to facts, and the closure principle does not. When p is in the question the issue is not only if one does believe p, but also if one would believe p. However, we are not asking if one would believe q (i.e.¬BiV) and we are therefore willing to separate knowledge of q from p. This is the case because condition 3 for knowledge is not closed, thereby making knowledge in general unable to be closed.

Shouldn’t the closure principle serve as sort of a litmus test for knowledge at all? But if so, then we are adding another condition to knowledge, and this is exactly Nozick’s objection- that the conditions of K vary according to P..... and closure does not....

C. Denying condition 4

Nozick offers another issue: what if the skeptic’s arguments could sway you at any time? Then surely you don’t fulfil condition 4, as your knowledge ought to be of sterner stuff. Nozick concedes the point that if a person were able to be convinced by such arguments, then right now he does not have K.

D. Questioning the method

How do you know that your method is good? Suppose you want to use a certain method to question that method. You couldn’t do so where trying to fulfil condition 3, as you can’t use a method to know what would happen if you weren’t using that method. As Nozick clarifies, however, this would happen (probably) only with a very broad M. Otherwise, one could simply use another specific m to verify.

Isn’t this problematic though- if we need to verify that we are using M, then surely this triggers another infinite regress...? Why not just say we Know p via M even where we do not ‘know’ in the strong sense that M.