Thursday, 31 January 2008


Last night I found myself listening to a cd by Stars of the Lid (And Their Refinement of the Decline), a gloriously minimal and abstract piece of music (attractively described as ‘drone-based ambient’ but very well reviewed last year) whilst at the same time reading a paper by Julian Dodd called ‘There is no norm of truth’. I was struck by the common explanation I would offer of their aesthetic appeal. Both are fine pieces of minimalism. And yet the nature of what is minimal is so obviously different in the two cases. What can ambient drone music have in common with an elegant philosophical defence of a slimmed down notion of truth? At what ‘organisational’ level is something common there to see? Is it more than that ‘minimal’ rightly applies to both?

I've no idea. But given that Dodd’s solution is so attractive, I’ll try to summarise it here for its own sake. Technology prevents me similarly conveying Stars of the Lid.

Minimalism about truth starts from an assumption about the nature of truth on which almost everyone agrees. Aside from paradoxical cases, it seems trivial that there is an equivalence between asserting some sentence p (using the sentence) and asserting that that sentence is true (mentioning the sentence). Given an assertion that a mentioned or quoted sentence is true, one can make the same claim by stripping it of its quotation marks (the way we mention sentences) and using what remains as an assertion. Truth is exemplified in this role as aiding disquotation in the disquotational schema thus:

DS. ‘p’ is T iff p.

Minimalism argues that this is all there is to truth. There is, eg, no substantial property (such as correspondence to facts) that all truths have in common.

But, as Dummett argued years ago, it seems that if one knew only this about truth one would be ignorant in an important way. One also seems to need to know that truth is the aim of assertion (just as knowledge of the rules of a game shorn of the concept of winning as its aim would be only partial knowledge). Aiming at truth is what normatively governs assertion. (We'll see what is right and wrong about this in a mo.)

One might deal with this point by saying that asserting is a practice and thus like any practice has standards of correctness, standards according to which one is warranted in making a move, in this case an assertion. (Any practice not respecting such a norm is not recognisable as, and is not, a practice of asserting.) But then given the equivalence / disquotational schema, the standards which warrant asserting p, warrant asserting ‘p’ to be true. So it is tempting to argue that the normative standards for truth are just the same as the norms implicit in the idea of warranted assertibility. (This move concedes that there are norms in play. So Dummett was partially right. But so far, it does not follow that these are norms of truth rather than of assertion.)

But as Crispin Wright argued a few years ago [Wright 1992], this does not work to disarm Dummett's worry.

Consider again the DS:
1. ‘p’ is T iff p
Letting p be –p, we can derive
2. ‘-p’ is T iff -p
But we can also negate both sides of the biconditional in 1 and derive:
3. –(‘p’ is T) iff –p
From 2 and 3 we can derive
4. ‘-p’ is T iff –(‘p’ is T)

This, however undermines the hypothesis that the norm of truth is the same norm as the norm of warranted assertibility. Even if aiming at one is aiming at the other (a prescriptive agreement), two norms are nevertheless distinct if failing to achieve one is not necessarily the same as failing to achieve the other (a descriptive disagreement). So if truth and warranted assertibility differ extensionally, they are distinct norms. And now if one replaces T with warranted assertibility in 4 one gets:

5. ‘-p’ is warrantedly assertible iff –(‘p’ is warrantedly assertible)

Which unlike 4 is not true (because just because p is not warrantedly assertible it doesn’t follow that its negation is. There may be no evidence either way).

Wright goes on to outline a variety of options for more substantial notions of truth whose normative aspects go beyond warranted assertibility. But it seems that a minimalist and deflationary view of truth has to inflate under this pressue.

Dodd’s delightful paper suggests that, whilst Wright has shown something about the mismatch between the norms of truth and of warranted assertibility, his argument to inflate truth can be undermined by arguing that this mismatch results from the fact that truth does not have any distinct normative features itself. Given that truth has merely the role displayed in DS any normative properties apparently accruing to truth, really belong to each particular truth in question. Thus in each case, one should aim to

assert that snow is white iff snow is white, and
assert that grass is green iff grass is green
These can gathered together by saying that one should assert p iff 'p' is T . But in doing this, one is not adding a distinct norm of truth, one is simply codifying the prior norms.

There is no extra norm in truth, just the norms thus listed. And now it is obvious why whilst aiming at expressing a truth and aiming at warranted assertibility might offer the same advice, they can diverge. Minimalism is thus redeemed, for the moment. Lovely!