Peter Murphy Rewrites the a Priori/a Posteriori Distinction

Dual Epistemic Justification

Peter Murphy’s simple point in this paper is that nonbasic beliefs can be justified by both experiential and nonexperiential justifiers; not just by either one or the other. He's concerned with the cases

“in which neither the nonexperiential justifier nor the experiential justifier would suffice on its own, in the absence of the other, to justify the non-basic belief” (2).

What would happen if one of the justifiers were absent? Murphy writes:

“In the absence of either justifier, one of the basic beliefs would be unjustified and, as a consequence, so too would the non-basic belief. The two justifiers function as cocontributors…” (2)

Despite that, Murphy’s prime purpose is to “rewrite” Kant’s strict binary distinction between a priori and a posteriori beliefs. He argues that in the case of the justification of a nonbasic belief which has both a nonexperiential and an experiential justifier, Kant classified the resultant belief a posteriori.

                                     A Kripke Case

Murphy gets his point across by offering us an example which is taken from Saul Kripke. Take the following argument-frame:

(1) (H = P) ⊃ □(H = P).

(2) (H = P)

(C) ∴ □(H = P)

Murphy says that Kripke takes (1) above to be known a priori. That is, if H and P are identical, then they're necessarily identical. The identity of H and P in this case, however, was only known (or discovered) a posteriori.

So what about (C) - the conclusion? Is that known a priori or a posteriori? More relevantly, is the whole argument a priori or a posteriori? Murphy argues that Kripke follows Kant on this. He takes (C) to be justified or known a posteriori because the identity of H and P (in (2) above) was only known experimentally (or through observation). Thus the whole argument, as well as the conclusion, is classified a posteriori, despite it dealing with a necessary identity.

Clearly Murphy has a problem with Kripke’s classification. The basic point is that although the necessary identity of H and P only came to be known a posteriori, the modal conditional - (H=P) ⊃ □(H=P) - still holds true without experience. That is, it's known to be necessarily true a priori. So in the Kripke case, and in the case of other nonbasic beliefs, we have what Murphy calls epistemic co-contributors.

Murphy gives another example of the phenomenon of epistemic co-contribution. Firstly we have the a priori part:

i) A child’s belief that 4 + 3 = 7 might be partly a priori justified by her intellectual insight.

We also have an a posteriori part:

ii) The same belief is partly a posteriori justified by her recent experience counting and recounting groups of blocks

Murphy also gives us a similar example.Again, firstly we have the a priori part:

i) A new logic student might have a marginally reliable a priori insight into DeMorgan’s Rule.

Then the a posteriori part:

ii) He might also base his belief in DeMorgan’s Rule on his marginally reliable roommate’s testimony that DeMorgan’s Rule is true.

A Priori Inference & Dual Justification

Murphy also highlights the case of inference.

Laurence BonJour makes much of inference in his apriorist criticisms of the Quinean “web of belief” thesis. (The beliefs in this web may well be empirical; though what about the links between them?) According to Murphy, BonJour also “proposes that acts of inferring are a distinct kind of justifier” (3).

What Murphy (or BonJour) is interested in is this move (or link) from belief to belief. Even if the move is from nonbasic and experiential beliefs to other beliefs, that link still requires an epistemic description, explanation and justification. More relevantly, the inference from belief to belief (or from beliefs to beliefs) needs to be justified. This will result in dual justification. Murphy argues that 

“there will be cases where the inference is justified one way and the relevant premise-beliefs are justified another way” (3). 

That is, the premise-beliefs may be justified a posteriori; though their links to - or the inferences from - other beliefs will be a priori in nature. Again, Murphy makes the conclusion that if we

“[t]ake away either the a posteriori justified belief or the a priori justified inference… the person’s conclusion-belief would be unjustified” (3).

In Kant’s book, according to Murphy, we can say that

“[s]ince the conclusion-belief is dependent on an experiential justifier for its justification, Kant has us put it in the a posteriori category” (3).

That is, Kant says that the conclusion-belief is experiential or a posteriori in nature, despite the fact that it was derived from a process which is a priori (even in the case when the premise-beliefs were also a posteriori).

Murphy also detects an "asymmetry" in Kant’s account. Thus:

i) To be a posteriori justified, a belief only needs to partially depend on experiential justifiers.

However, an a priori justified belief doesn't get the same treatment from Kant. Thus:

ii) “To be a priori justified, it is not enough that a belief partially depend on nonexperiential justifiers for its justification – it must exclusively depend on nonexperiential justifiers.” (3-4)

Murphy’s Weak Rationalism

Where does Murphy stand on the general empiricist/rationalist debate which underpins the issues just discussed?

He detects four positions:

Two Radical Positions:

1. Radical empiricism: this position denies that there are any non-experiential justifiers and insists that all justified beliefs are justified a posteriori.
2. Radical rationalism: this position denies that there are experiential justifiers and insists that all of our beliefs are justified a priori. (4)

Two Moderate Positions

3. Strong rationalism: this position argues that some of our justified beliefs are a priori justified.

4. Weak rationalism: this position argues that there are non-experiential justifiers, but restricts their justifying power to beliefs that are justified in a mixed manner.
   
   

Clearly, Murphy opts for position 4) above.

Reference

Murphy, Peter, ‘Rewriting the A Priori/A Posteriori Distinction’, Journal of Philosophical Research (2008)