Internal and external reasons

Bernard Williams's article 'Internal and external reasons' in his Moral Luck begins with a double definition of an internal reason:

(1) The sentence 'A has a reason to do X' asserts the existence of an internal reason if and only if it implies that there is some motive (i.e. desire, loyalty, disposition of evaluation etc.) M of A's which will be served by her doing X. Otherwise it asserts the existence of an external reason.

and

(2) The sentence 'A has a reason to do X' asserts the existence of an external reason if and only if the truth of the sentence depends on there being some motive M of A's which will be served by her doing X. Otherwise it asserts the existence of an external reason. (101)

Presumably we must interpret (2) as making it clear that the 'implies' in (1) is that of material implication, so the two definitions coincide. Williams goes on to present:

(3) 'A has a reason to do X' if and only if there is some desire D of A's which will be served by her doing X (101)

and to say that he will  'work it up into something more adequate' by 'addition and revision' (102). In fact the 'addition and revision' that follows (102-5) does not seem to alter (3) at all except in that it replaces 'desire' by 'motive', and with that replacement it becomes the claim that all reasons are internal reasons by teh above definition. For all that Williams does in these pages is clarify that what matters is that A's doing X will in fact serve some motive M, rather than whether A thinks it will, and describe some deliberative processes through which A might discover what reasons to act she has.

At one point he says that such deliberative processes might give A new desires and new reasons to act (104-5). His meaning is obscure but the idea seems to be that if, for example, I want an entertaining evening and come to the view that going to the opera would make for an entertaining evening, then this will give me a desire to go to the opera. Since buying opera tickets will serve that desire I now have a reason to buy opera tickets. But is this really a case of a creating a new reason? Surely if it was the case from the start that I wanted an entertaining evening, and going to the opera would make for an entertaining evening, then I had a reason from the start to buy opera tickets. I just hadn't seen that I had a reason to. And even if my deliberative process does create a new reason, as opposed to revealing an existing one, this does not affect the truth of (3).

Therefore it seems that by the time we get to Williams' argument against external reasons his position is simply a slightly weakened version of (3), namely:

(4) 'A has a reason to do X' only if there is some motive M of A's which will be served by her doing X.

This is just the view that all reasons are internal reasons. Why does Williams think there is something incoherent about external reasons? His argument seems to go:

(5.1) If P is a reason for A to do X then it could be a reason for which A does X on a particular occasion, and it would then help explain why A did X (106). That is, P (or the belief that P) could provide part of the motivation for A's doing X (could be a motivating reason for A's doing X).

(5.2) The only way this could happen with an external reason is through A coming to believe that P is a reason for A to do X. (107)

(5.3) Furthermore A would have to come to believe that P is a reason for A to do X as a result of deliberating rationally. (109)

(5.4) Furthermore A would have to come to be motivated to do X just as a result of acquiring this belief. (109)

(5.5) But 'I see no reason to suppose that these conditions could possibly be met.' (109)

No doubt an external reasons theorist could deny (5.3) on the grounds that the route to coming to see that P is a reason for A to do X need not itself be a rational route; it might be some process of 'awakening'. But even if (5.3) is omitted the argument looks quite strong. For (5.2) and (5.4) seem to offer the most plausible way that (5.1) could be true for an external reason (at least the onus is on the external reasons theorist to offer an alternative). And (5.4) contradicts the very plausible belief-desire theory of motivation. This is probably what is behind (5.5), given Williams's earlier reference to Hume's assertion that reason - by which Hume means the faculty of discovering truths -  cannot by itself motivate (108). So it looks as if an external reasons theorist needs to deny belief-desire theory.

But I think there is a way of denying Williams' conclusion without abandoning belief-desire theory. For what is the basis of (5.1)? In later writings, Williams says it is that unless (5.1) is true the practice of offering reasons for action to each other would be pointless. (This echoes one of Hume's arguments for 'moral internalism', the claim that to have a moral belief is to be somewhat motivated to act on it: if it were not, Hume says, what would be the point of arguing about morality with each other?) But this seems to assume that either (5.1) is true for everyone or it is true for no-one. Surely all we need to ensure that the practice of offering reasons for action to each other has a point is something weaker than (5.1), such as:

(5.1a) If P is a reason for A to do X, it could be a reason for which A does X on a particular occasion, and it would then help explain why A did X, as long as A is motivationally normal.

For it is to, and about, motivationally normal people that we make the vast majority of our assertions about reasons for action. For a motivationally normal agent A (for example, with a some minimum capacity to identify with the pains and pleasures of others), it follows by the belief-desire theory and Williams' argument (or something like it) that all the reasons for A to act are internal. But if someone is sufficiently motivationally abnormal ( for example, if they completely lack this capacity) then it doesn't follow. All that would follow in general is:

(4a) 'A has a reason to do X' only if there is some motive M which A would have if she were motivationally normal, which will be served by her doing X.

On this 'normal-agent internalist' view of reasons for action, when we assert that A has a reason for action we tacitly treat her as motivationally normal. If it turns out that she is not, to the point where she has no motive which will be served by doing X, we do not conclude that has no reason to do X, only that she is incapable of doing what she has reason to do. So in this case we assert an external reason. But (4a) still holds of this assertion. The link between reasons for action and motivation is at the level of what is socially (or humanly) normal, and it is at this level that all reasons are internal.

(Revised 3 February 2010)