Bernard Williams once wrote, ‘Desiring to do something is of course a reason for doing it.’Footnote ¹ No philosopher, I think, would now second the ‘of course’. The relation between reasons for action and the agent's desires is now widely debated. Which is a ground for welcoming an intelligent and ingenious attempt to rehabilitate desires at least as sources of reasons. Mark Schroeder, a young philosopher at the University of Southern California, has won a reputation with a plethora of interesting papers (many accessible through his website).Footnote ² This book, his first, seems to me less successful, though it is still inventive, and deserving of attention.

Schroeder avoids any statement like Williams's. Instead, he follows Philip Pettit and Michael Smith to the extent of taking desires typically not to be reasons (though they may become reasons on occasion), but to background reasons.Footnote ³ Pivotal is a distinction between reasons and background conditions. Schroeder offers the following principle (here call it H) of what he terms ‘Hypotheticalism’:

Schroeder resists basing H upon some categorical rule like ‘Everyone has a reason to promote the satisfaction of any of his desires.’ If we were apply H to that very rule, we would need to suppose that everyone has a second-order desire, for each of his desires, that it be satisfied. Yet at best that would be contingently true. I don't think this is a problem for Schroeder: nothing about the form of H precludes its being, as he supposes, axiomatic and a priori.

H has the tactical advantage for the Humean that what he then cites as reasons will be things that are of a general kind (like facts about means and ends) to count as reasons on any view. H further helps Schroeder to evade what he calls ‘the Self-regarding objection’. He finds it a plausible constraint on what count as an agent's reasons for action that they are ‘the kinds of thing about which he should be thinking’ (26). Yet it is hardly attractive for him to be constantly considering what he wants. Schroeder relegates that to the background of his deliberations.

Does this disarm the objection? To my mind, this depends on the ambitions of the theory. Williams required, not implausibly, that a statement about an agent's reasons be ‘a distinctive kind of statement about, distinctively’ the agent; it cannot, he thought, be true that x has a reason to do a simply because x is ‘well-placed to do some good’ by doing a.Footnote ⁴ Suppose that we allow that x's doing a would have some point or value (say, in doing some good). We might still refrain from ascribing to x a reason to do a unless, in some way or other, that point or value connects positively with x's psychology. It needn't be that x already desires to perform such actions – and Schroeder initially (1) writes open-endedly about ‘some feature’ of x's psychology (here call it F). Whatever F may be (which might well vary from case to case), it doesn't follow that it would itself count as one of x's reasons, and so become something of a kind to which x should attend in deliberating.

However, Schroeder's theory is more ambitious, since – more austerely than Pettit and Smith, who permit desirability-characterizations to form major premises within deliberation – he casts desires and the like as x's only starting-points. He boldly offers his principle H as a reductive analysis of reasons for action, spelling them out in natural and non-normative terms; and he falls in with those who hold that ‘what it is to be normative is to be analyzed in terms of reasons’ (81). This implies that x's reasons for doing a cannot derive even in part from a's point or value. If values come in, this can only be later, and on the back of reasons. Hence F's being a feature of his psychology is all that makes a means-end relation a positive reason for x. Which risks becoming unattractive in cases where, for instance, x morally ought to do a. Even if backgrounding desire saves deliberation from being self-regarding in its focus, its basis becomes oddly bare and egocentric.

As we may put it, Schroeder makes F not just a conduit that translates a value into a reason, but the wellspring of the reason. This bears on another query. Take a case where the feature consists of a desire for an identifiable end: why should it then be inappropriate for x to say, especially when the relation of means to end is common knowledge, that his reason for realizing M is his desire for E? Suppose that I happen now to feel like a cup of coffee, which I can always secure, when working at home, by getting up to go to the kitchen. The question ‘Why are you getting up?’ would then be more naturally answered by ‘Because I want some coffee’ than ‘Because I can only get to the kitchen on my feet.’ Schroeder may respond that then there is a further desire in play. What might this be? To do whatever I want? But that seems too indiscriminate. To do this thing I want? But that hardly seems separate from wanting to do it. More prudent backgrounders can respond that ‘I want some coffee’ does not really, in their intended sense, foreground the desire that it avows; rather, it indicates in context that I ascribe some value to my drinking coffee now – and it is this value, and not the desire, that is a reason (one that, in their view, still depends for this status upon my desires). But it does not seem open to Schroeder to say this.

He concedes intuitive force to two opposite but compatible objections to a desire-based account of reasons. First, H may admit too many reasons (e.g. to act expensively on a whim). Schroeder replies in effect that, when we say that x has no reason at all to act in a certain way, even though it would satisfy one of x's desires, all we really are intuiting is that x has no reason to count (92-7). I would need to have a clearer grasp of the cogency of H before I was persuaded of this. Take the following, which is at least a conceptual possibility: x has a desire to do a, though (whether he is aware of this or not) actually doing a would be gratifying to no degree whatever, even as a relief from the desire; rather, it would be nothing for x but dust and ashes.Footnote ⁵ It seems to me plain that x has, so far, no reason at all for acting so, however intense the desire.

Secondly, H may admit too few reasons (e.g. to meet a disagreeable obligation). Schroeder contests this in two ways. He takes a very weak view of the promotion relation: ‘x's doing a promotes p just in case it increases the likelihood of p relative to some baseline’, which ‘is fixed by the likelihood of p conditional on x's doing nothing’ (113). (In which case there is a reason for any action, since acting, as Dr Johnson remarked, is ‘driving on the system of life’.) And he speculates, in a manner perhaps rather promissory than promising, that good actions have reasons that are so massively over-determined that ‘they are reasons for anyone, no matter what she desires, simply because they can be explained by any (or virtually any) desire’ (109). Such reasons he calls ‘agent-neutral’.

Schroeder realizes that he needs to add an account not just of the existence of practical reasons, but of their force. A characteristically Humean view is what he calls ‘Proportionalism’ (97-102). This makes the force of a practical reason for an agent be a function of the intensity of his desire for the end, and the strength that he can assign to the promotion-relation. However, this is liable to concede too much to perversely intense desires, and too little to regrettably feeble ones. Worse for Schroeder, it has to foreground the desire in order to assess its intensity. So he instead jumps ship into what is certainly not Humean, proposing this principle: ‘Set of reasons A is weightier than set of reasons B just in case it is correct to place more weight on A than on B’ (140). As what can only be a partial criterion of correctness (and yet he suggests no other), he offers a principle defining ‘right kind of reasons’: ‘The right kind of reasons to do A are reasons that are shared by everyone engaged in the activity of doing A, such that the fact that they are engaged in doing A is sufficient to explain why these are reasons for him’ (135). He gives playing chess as an example: anyone who is playing chess must have some reason not to castle out of check, because that is against the rules (ibid.). No doubt chess players also have reasons to respect moral considerations; yet, ‘the fact that someone is a chess player is not sufficient to explain the existence of these reasons’ (ibid.).

I find the example unhelpful: the rules of chess are constitutive, for playing chess is playing by the rules of chess; so no one needs a reason not to castle out of check while playing chess. Schroeder hopes to derive an argument that he can apply to a case where Katie needs help, and Ryan is in a position to help her, but may feel disinclined. How weighty is his reason to help her? I can only quote in extenso:

After which display of verbal prestidigitation one may return with relief to Hume's honest engagement with human contingencies.

Schroeder appears to be trapped in a quandary of his own making that comes of backgrounding desire without foregrounding anything that could feed into a weighing of reasons. His only expedient is to cast a veil of words over the vacuity. But the remedy, as it seems to me, must be not a ruse, but a rethinking