3/08/2011

A question about coherence

For the record, I am not sympathetic to the coherence theory of truth. But I would like to understand what it is, especially since some logical empiricists (Neurath in particular) entertained it. So here's the question.

The coherence theory of truth maintains:
p is true if and only if p coheres with S
where p is an arbitrary proposition, and S is a special set of propositions. Many people criticize the coherence theory on the grounds that there is no principled way to pick out this special set, but let's bracket that (very important) issue for the moment.

So the obvious question to ask about the above statement of the coherence theory is: What is coherence? There are a number of answers out there, but everyone agrees that logical consistency is a necessary condition for coherence (but insufficient*). In other words:
If p coheres with S, then the set of propositions {p∪S} is consistent.
And now it looks like we're making some progress in understanding what the coherence theory is committed to, because logical consistency is a notion that we have a clear and independently-motivated handle on.

But wait -- our 'clear and independently-motivated' notion of logical consistency depends, of course, on the notion of truth: a set of sentences is consistent iff it's possible that they all be true. But the coherentist's notion of truth is exactly what we were originally trying to explicate here. So it seems like we've got a circular definition.

(Note: one can characterize consistency in purely syntactic terms; for example, in logics where Ex Falso Quodlibet holds: A set of sentences is consistent iff there is a sentence that cannot be derived from that set. Perhaps that is what the coherentist might do?)


-------
* Given certain reasonable instances of S, there will be two propositions q and not-q that are each individually consistent with S. But both can't be true.

6 comments:

Protagoras said...

The fact that two propositions are consistent iff it is possible for them both to be true together does not establish that truth is some more primary notion that consistency depends on. Iff does not show anything about such dependency (especially as it is bi-directional), so there is no obvious threat to what is surely the coherentist's view, that in fact it is consistency which is primary, and truth depends on consistency (and whatever the other factors in coherence are).

You seem to be thinking of something like that when you admit that consistency can be characterized syntactically, but the point is surely more general; at the very least there's always the option of taking consistency as primitive, and no doubt there are other candidates for notions that could be taken as primitive which the coherentist could take consistency to be derived from.

Greg Frost-Arnold said...

Thanks, Aaron.
I guess my question is then: how does the coherence theorist identify when a set of propositions is consistent?

Even if 'consistent' is taken as a primitive, it seems like the coherence theorist must have some way of separating the consistent from the inconsistent. You say there's "no doubt" that the coherence theorist could appeal to some other notion to explain consistency, but I'd actually like to hear what those are: the coherence theory just feels so strange to me, I have trouble figuring out what the coherentist's account of consistency would(/could/should) be.

Protagoras said...

Well, a coherence theorist could take a notion of inconsistency as primitive and definite consistency as the absence of that primitive. I wasn't thinking they'd be able to come up with anything that wouldn't sound circular and unhelpful to a skeptic; that's a rather high standard (I don't know if I'm a coherence theorist, but I do find attempts to explain correspondence truth always strike me as circular and/or otherwise unhelpful). And of course the coherence theorist could take a combined approach; say consistency is primitive, but syntactic methods can help detect it, or something of that sort.

Jonathan Livengood said...

I have no idea if this could work, but what if the coherentist opted for subjective Bayesianism?

Then the coherentist could avail himself of Bayesian measures of coherence and need not say anything about a world standing behind his beliefs. The question is only whether some sets of propositions and assignments of probabilities to those propositions are more coherent than others.

Nathan Coppedge said...

I take issue with the definition of coherentism, I feel it is too ill-conceived to be representational.

I feel Coherentists are those who hold that for n equivalent statements, organization S is valid when n is ennumerated.

Thus the statement of coherency is not a statement of contingent validity, but rather a statement of contingent organization.

If S is not valid, it is because it is disorganized, the more so if n are not ennumerated.

So before a statement is determined to be invalid or subjective, it may be seen that it is either:

A. A difficult claim in a particular organization

B. A difficult claim based on the perfection of evidence

Stated in these terms, coherentism can be a form of objective philosophy.

--Nathan Coppedge

Nathan Coppedge said...

Maybe this is the beginning of coherent objectivism