Reasons for Rule Consequentialists

This paper explores what a Rule Consequentialist of Brad Hooker


WOODARD
normative reasons for action) appears infrequently-and many of the uses describe others' views (Hooker, 2000a, p. 7, p. 15 n. 22, pp. 24-25, 47-48, 65, 67, 117, 120, and 182).In some other places Hooker makes claims about normative reasons for action-for instance, claiming 'There is always a pro tanto moral reason to promote welfare, but it is distinct from the moral reason not to break one's promises ' (2011, p. 242; see also Hooker, 2000b).
He also uses the language of 'general moral duties' and 'general moral considerations' in ways that suggest he is using the concept of normative reasons for action (Hooker, 1996, p. 534 and2000a, pp. 106-107, 128).So it is not that Hooker never discusses reasons for action in explaining or defending his version of Rule Consequentialism.However, he does not discuss normative reasons for action at length, nor focus attention upon them.This is somewhat surprising.Talk of reasons is widespread, and not just among professional philosophers.
When people deliberate, or discuss some dilemma in the abstract, they often note what reasons or considerations there are and how they conflict with or reinforce each other.They disagree with each other about whether some feature of a situation provides a reason.They distinguish between cases in which someone acts wrongly for a good reason and cases in which someone acts wrongly for no good reason, and treat that distinction as making a difference to judgements of blameworthiness.In short, judgements about what reasons people have, and why, seem to be common and to play a very important role in ethical thought and discussion.That provides a rationale for any theory in normative ethics to address them.For those who adopt the method of reflective equilibrium, as Hooker does (2000a, pp. 4-19), that rationale seems especially strong.
In this paper I will explore what someone who accepts Hooker's Rule Consequentialism can and should say about normative reasons for action ('reasons', for short).Reasons, I shall assume, are genuine considerations in favour of, or against, performing an action.I shall claim that Rule Consequentialists can provide a plausible account of reasons, but that to do so they must distinguish between different roles played by rules in the ideal code.

| A S ERIE S OF C A S E S
Let us consider some very simple cases, to illustrate some features of Rule Consequentialism.In all of these cases you can prevent a harm, but only by breaking a promise you have made to your friend to celebrate her birthday.
The cases differ from each other only in the size of the harm that you can prevent.At the start of the series the preventable harm is very small, while at the end it is a disaster.In considering these cases we can ask both what you ought to do, and what Rule Consequentialism implies that you ought to do.I assume that in cases with small harms you ought to keep your promise, but that in cases with large harms or disasters you ought to break it.In some intermediate cases it is less clear what you ought to do.As the harm increases, we will reach a point where it is hard to judge whether you ought to prevent it or to keep your promise.
Perhaps there is a sharp cut-off between the last case in which you should keep your promise and the first in which you should break it.If so, it is hard to identify where this cut-off falls.Alternatively, perhaps there is a zone in which it is vague or indeterminate what you ought to do.
Rule Consequentialism has similar implications.It claims that what you ought to do depends on what the ideal code permits.As Hooker argues, it seems likely that the ideal code contains a rule requiring promise-keeping, and a rule that licences breaking other rules when necessary to prevent disasters (2000a, pp. 98-99, 126-134).When tiny harms are at stake only the first rule applies, and so Rule Consequentialism implies that you ought to keep your promise.When disasters are at stake both rules apply, but the disaster-prevention rule takes precedence-so Rule Consequentialism implies that you ought to break your promise to prevent the disaster. 2Hooker claims that the amount of harm required to justify breaking another rule varies with features of the case.As he puts it, 'It takes 2 The overridingness of the 'prevent disasters' rule is not an ad hoc feature of Rule Consequentialism.Hooker argues persuasively that a code containing such an overriding rule would be better (in the way he specifies as relevant for codes) than one lacking it (2000a, pp. 98-99).This is one of the impressive features of Rule Consequentialism: other non-absolutist views about constraints do not offer similar explanations of why harm prevention sometimes takes precedence over abiding by constraints.I return to this point briefly in Section 5.
|3 WOODARD the threat of a small disaster to justify breaking a small promise, a larger disaster to justify stealing, and a much larger disaster to justify inflicting serious physical harm on innocent people ' (2000a, p. 135).He also notes that the concepts employed in rules in the ideal code require judgement to apply, and that this allows for the possibility that what you ought to do is indeterminate (Hooker, 2000a, p.133, pp.135-136). 3  In sum, the implications of Rule Consequentialism seem broadly to match our judgements about this series of cases.They agree that you ought to keep your promise at the start of the series, but break it at the end, and that there is a point or a zone in the middle of the series where the verdict either flips sharply or is indeterminate.For all we have said, Rule Consequentialism might imply that this point or zone is not where we think that it really is; so, we should not overstate the claim that the implications of Rule Consequentialism match the judgements we would like to explain.But they are, as it were, at least on the same page.
Suppose we change the question, however, and ask about what reasons you have, rather than what you ought to do.Here again I will make some assumptions.I assume that you have a reason to keep your promise in all of these cases, not just those cases in which you ought to keep your promise.Even when you ought to break your promise to prevent a disaster, you seem to have a reason to keep your promise.If you do indeed keep your promise, the fact that you had a reason to keep it explains why we would treat your failure to prevent the disaster differently than if you could have prevented it without breaking any promises or incurring any other morally significant cost.We might think that you have done the wrong thing, and that you are blameworthy, but we would think that at least you had a reason for acting in this way-that is, a normative reason, not only a motivation.
Similarly, I assume that you have a reason for preventing the harm in all of the cases in the series.Near the start of the series the harm you could prevent is small, and not large enough to justify breaking the promise.Again, though, it does not seem right to say that you have no reason to prevent the harm.If you wrongly break the promise to prevent a small harm, you at least had a normative reason for doing so.
If these assumptions are correct, you have a reason to keep your promise in all cases in the series together with a reason to prevent the harm in all cases.What changes as we go along the series, in either direction, is the relative importance of these reasons, and the deontic status of your two options.4Can Rule Consequentialists of Hooker's sort explain these claims about reasons?

| RE A SON S AND RULE S
Act Consequentialists are likely to try to explain the existence of conflicting reasons in terms of the existence of different values.Though Rule Consequentialists could also try to do that, they are more likely to try to explain conflicting reasons in terms of different rules in the ideal code.Like the strategy of appealing to different values, the strategy of appealing to different rules offers to preserve two important features of the concept of a reason.
On one hand, we have the relationship between reasons and rightness.When an action is right there is sufficient reason to perform it, even if there is one or more reasons not to perform it.When an action is wrong there is not sufficient reason to perform it, even if there is one or more reasons to perform it. 5If we think of reasons as associated with rules we seem to have the right structure to explain these observations.Whether an action is right or wrong, according to Rule Consequentialism, depends on whether the whole ideal code permits it.Since rules, like reasons, can conflict, we can say that there is a reason to do something that the code as a whole forbids, or a reason against doing something that the code as a whole permits.6 Second, associating reasons with rules seems to allow explanation of the phenomenon of feeling torn by conflicting reasons when deliberating.On Hooker's view, a virtuous agent is someone who has internalized the rules in the ideal code.When such an agent faces a decision in which different moral reasons conflict, they will have conflicting aversions-each one associated with the prospect of breaking a different rule they have internalized.He writes: To illustrate, suppose people who have the pattern of aversions that rule-consequentialism prescribes face a situation in which their breaking a promise is the only way to prevent a loss in well-being.Would they break the promise?They would if the choice is between preventing a huge loss to someone and keeping a small promise.They would not if the choice is between preventing a small loss to someone and keeping a solemn promise about an important matter.(2000a, p. 132) That is, a virtuous agent would be averse to breaking the promise and to failing to prevent the harm, and the strength of these aversions would vary according to the details of the case.In deliberating, they would be pulled in different directions by these conflicting aversions.So, whether we focus on the connection between reasons and rightness, on one hand, or on the psychological role of reasons in the deliberations of virtuous people, on the other, the association between reasons and rules in the ideal code makes sense.7 However, this explanatory strategy must be developed with care.We can see this by asking which rules in the ideal code might explain the reasons you have in the cases we considered in the previous section.Fairly obviously, the rule that is relevant to your reason to keep your promise in all of these cases is the rule requiring promisekeeping.But what about your reason to prevent harm?In some of the cases-the ones in which the harm is large enough to justify breaking your promise-the disaster-prevention rule seems relevant.But I claimed that you have a reason to prevent the harm in all of the other cases, too.Now as we have already noted, Hooker claims that what counts as a disaster will vary with details of the case, including which other rule is at stake (2000a, p. 135).But no matter how we define 'disaster', there will be cases in which the harm at stake does not meet the threshold but in which you seem to have a reason to prevent the harm.
If we accept that there is a reason to prevent the harm in these sub-disaster cases, one possibility would be to say that the disaster-prevention rule explains the reason to prevent disaster-sized harms, but that some other rule explains the reason to prevent smaller harms.Some of Hooker's remarks suggest that this might be the picture.
For example, he refers to a general motivation of beneficence that is weaker than the disaster-prevention motivation, and these motivations presumably correspond with rules in the ideal code (2000a, p. 98 n. 7).So perhaps the reason to prevent smaller harms is derived from a separate reason to be beneficent, and is quite distinct from the reason to prevent larger harms.
The trouble with this picture is that it is hard to believe that the nature of the reason to prevent a harm changes as we descend the series from cases in which the harm is just large enough to justify breaking a promise to cases in which it is not quite large enough to do that.Note that this claim is independent of whether we think that change in deontic status occurs at a precise point, or across some zone.Wherever that change occurs, it is more plausible to think that it is the result of the changing weight or importance of the reason to prevent harm as compared with | 5 WOODARD the weight or importance of the reason to keep promises, rather than as a result of one kind of reason to prevent harm being supplanted by a different kind of reason to prevent harm.If there is just one kind of reason to prevent harm we can try to explain why, as the magnitude of harm changes, it becomes permissible to break the promise.
If instead there is one kind of reason to prevent disastrous harms, and a different kind of reason to prevent smaller harms, we have two coincident changes to explain: one kind of reason supplanting another, and a simultaneous change in the deontic status of breaking the promise.
It seems to me to be more plausible to claim that what changes as we ascend or descend the series of cases is not the kind of reason at play, but whether it is outweighed.For Rule Consequentialists the key to capturing this intuition is to distinguish two different roles of rules in their theory.The ideal code both specifies what reasons there are and encodes information about their relative strength-and different rules in the ideal code may perform one of these roles without performing the other.The rule 'keep promises', for example, seems to identify a fundamental kind of reason, associated with the kind of action 'promise-keeping'. 8It does not appear to encode any information about the relative strength of reasons of this kind, though the ideal code as a whole will encode this information elsewhere, and this will be reflected in the aversions of virtuous agents.In contrast, the rule 'prevent disasters' is explicitly overriding: it is formulated as requiring agents to break other rules when necessary to prevent disaster (Hooker, 2000a, p. 98).
The point is not that some rules come with strengths attached and others do not.Virtuous agents will have aversions to each kind of prohibited action of an appropriate strength.Instead, the point is that we can be misled, when thinking about the relationship between rules and reasons, if we assume that all rules are formulated with both roles equally in mind.The rule 'keep promises' seems to specify a fundamental kind of reason.In contrast, the rule 'prevent disasters' seems to be playing a different role.It identifies a class of cases in which a certain kind of reason is overriding.The fact that this rule features in the ideal code does not imply that there is a fundamental kind of reason associated with acts of preventing disasters.
One way to put the point is that the idea of the ideal code plays several roles in Hooker's Rule Consequentialism.It is supposed to show what makes actions right or wrong, by articulating an ideal moral consciousness with good consequences.It is also supposed to be of practical use to real, imperfect, agents in their deliberations.These roles mean that it must be verdictive-it must lead to conclusions about the rightness or wrongness overall of actions.If we also assign it the theoretical task of helping us to identify fundamental kinds of reasons, we must do so with care.We should not assume that all rules in a helpfully-formulated ideal code-helpful, that is, given the various roles it has to play-will stand in a one-to-one relationship with a fundamental kind of reason.Some rules may be formulated instead with an eye to fulfilling one of the other functions, such as making clear which verdicts an ideal agent would reach or assisting real agents with their dilemmas.
Once we recognise this, we can see that it might be a mistake to assume that Rule Consequentialism implies that the reason to prevent disasters is of a different kind than the reason to prevent lesser harms.Even though it distinguishes cases involving disaster from other cases, this need not reflect a division in kinds of reason; it might instead reflect a division between cases in which this reason is overriding and cases in which it is not.

| RULE CONS EQUENTIALIS T E XPL ANATIONS OF RE A SONS
One might object that talk of fundamental kinds of reason is out of place in a discussion of Rule Consequentialism.
Such talk might be thought to misconstrue the idea of an ideal code in the theory, as playing a merely epistemic role in helping us to identify reasons that have an independent existence.This would be to overlook the fact that, according to Rule Consequentialism, the existence of a reason is explained by features of the ideal code.If there is a reason to keep promises, for example, this is (according to the Rule Consequentialist) because a code containing a rule requiring promise-keeping has better consequences than alternative codes.The idea is that we explain the existence of this reason in terms of the ideal code, not that we use the ideal code as a tool to identify independently existing reasons.This is an important point, but it is consistent with the claims made in the previous section.Rule Consequentialism implies that every fundamental reason is to be explained in terms of features of the ideal code-but it does not imply that every rule in the ideal code specifies a fundamental kind of reason.If we claim that some rule fails to identify a fundamental kind of reason, we should mean that any reason it is associated with is already explained by some other part of the ideal code.In the case of the disaster-prevention rule, this might be a reason to prevent harm.It is quite plausible to think that an ideal code would contain a (not always overriding) rule 'prevent harm', if only because there are cases in which it is possible for agents to prevent harm without incurring any cost, and a code that tells them to make use of these opportunities will do better by Rule Consequentialist lights than one that does not.
So the important lesson from this possible objection is that all reasons should be specified somewhere in the code.This is compatible with the idea that not all rules in the code specify reasons.Some rules may have some other function, such as specifying how different reasons interact to determine what is morally required.
If that is correct, we should distinguish between reason-specifying rules and other rules in the ideal code.
We can then say that the reasons that exist, according to Rule Consequentialism, are those specified by the reason-specifying rules in the ideal code.There are reasons in favour of actions of all of the kinds that feature in positive reason-specifying rules, and there are reasons against actions of all of the kinds that feature in negative reason-specifying rules.To check whether a rule is a reason-specifying rule, we should ask whether the reason for or against actions of the kind the rule is about can be explained in terms of some more fundamental reason specified elsewhere in the code.If it can, this rule is not reason-specifying, but is playing some other role.
I claimed earlier, in effect, that the rule requiring promise-keeping might be regarded as a reason-specifying rule.Suppose the reason to keep promises cannot be explained in terms of some more fundamental reason specified elsewhere in the code.In that case, what exactly is your reason to keep your promise in the cases we considered, according to Rule Consequentialism?One possibility is that your reason to do whatever would keep your promise is, simply, that it would keep your promise.This would be to take the reason-specifying rule to fully specify the reason.A different possibility is that this specifies only part of the reason.A fuller specification of it would add that a rule to this effect is part of the ideal code.Your reason to do whatever would keep your promise would then be that this would keep your promise, and the ideal code contains a reason-specifying rule requiring promise-keeping.
Which of these things should the Rule Consequentialist say? Hooker writes: An essential feature of the rules establishing promissory obligations is that the particular obligations one has because of the combination of a general duty to keep one's promises and one's autonomous commitments are self-imposed.In this way, the source of a duty to do X can be one's autonomous promise to do X.But, at a deeper level, rule-consequentialism explains why an autonomous promise can create an obligation, that is, because of the expected utility of having a social practice enabling the creation of self-imposed obligations.(2011, p. 249) Here Hooker is noting that Rule Consequentialism can explain the sense in which obligations to keep promises are self-imposed-not imposed by facts about the utility of each act of promise keeping that are, as it were, imposed by the circumstances.But the distinction he draws between levels of explanation is relevant to our question.Rule Consequentialists could claim either that the value of the ideal code is part of the explanation of your reason to keep | 7 WOODARD your promise, or that it is the answer to a different question.Perhaps the full specification of your reason is that doing this thing is necessary to keep your promise, and the value of the ideal code answers a question instead about why there is a general duty to keep promises.Similar issues arise for other moral theories.For example, a virtue ethicist may claim that reasons are to be explained in terms of virtues, and that one of the virtues is to keep promises.She then faces a similar choice about how to answer the question about the nature of the reason to keep promises.Should she say that your reason to do what is necessary to keep your promise is simply that it would keep your promise?Or should she add that one of the virtues is promise-keeping-together, perhaps, with her account of what makes a character trait virtuous?Parallel questions face the contractualist.
It seems to me that Rule Consequentialists could go either way on this issue.They could cite the value of the ideal code as part of the explanation of the reason, or they could insist that it answers a different question.9Which way they go might depend on other features of their theory.The value of the code might more naturally be regarded as part of the explanation of the reason if codes are evaluated according to the consequences of compliance than if they are evaluated according to the consequences of acceptance.On the compliance construal, each individual act of promise-keeping can be regarded as part of a pattern of promise-keeping that would have good consequences-and these consequences might then be regarded as part of the full specification of the reason (Woodard, 2019(Woodard, , ch. 5, 2020, pp. 187-188, pp. 187-188).An acceptance Rule Consequentialist like Hooker could say, instead, that the reason is fully specified by the rule, and the existence of the rule is explained by the consequences of people accepting the code.Their views about which reasons there are would then be like those held by the Rossian pluralist.The difference between this kind of Rule Consequentialist and the Rossian would not lie in their views about the nature of these reasons, but in their views about whether there is an overarching explanation of why these reasons exist, and not others (Hooker, 1996, 2000a, pp. 19-23 and104-107).

| WHAT IS THE RE A SON TO PRE VENT D ISA S TER S?
The rule requiring prevention of disasters does not seem to be a reason-specifying rule.What is distinctive about disasters is the magnitude of harm they involve, and this seems apt to explain why a reason to prevent harm that exists in other cases tends to be overriding when a disaster is at stake.It would also explain why whether some harm counts as a disaster depends on what else is at stake.This is enough to explain why it seems that you have the same kind of reason in all of the cases in which you must choose between breaking a promise and preventing harm.But perhaps we should go further and think of the reason to prevent harm as derived from a more fundamental reason to promote the good.If we extended the series of cases, to include new cases in which you must choose between keeping your promise and bestowing a benefit on others, this new hypothesis would imply that you are still subject to conflict between the same basic reasons: a reason to keep promises on one hand, and a reason to promote the good on the other.Admittedly, Hooker says that 'The overriding commitment to prevent disasters must not be confused with an overriding commitment to maximize the good ' (2020, p. 458, emphasis in original).But here he is discussing overriding commitments, not reasons that can be outweighed.And elsewhere he writes that 'the fact that an act would benefit others … always counts as a moral reason to do it.This reason can be overridden by other moral reasons.But it is a moral reason always anyway ' (2000b, pp. 8-9).So it is at least possible to interpret his Rule Consequentialism as claiming that the fundamental reason at issue in the disaster-prevention rule is a reason to promote the good.This could be true even if the reason to promote the good is generally stronger in cases of preventing harm than WOODARD it is in cases of bestowing benefits, and even if it does not typically outweigh other reasons unless prevention of a disaster is at stake.
It might be objected that this is to interpret Rule Consequentialism as recognising a reason of the very same sort that Act Consequentialism recognises.That seems true-but it would also be true if we thought that reasons to prevent harm are not explained by a more general reason to promote the good.Under both interpretations the reason has to do with the agent's ability to bring about an outcome that is better in some way than it would have been.The difference between the two construals is a matter of axiology-whether we focus narrowly on harm, or more broadly on the good.Both construals posit equally 'Act Consequentialist' reasons, in the sense that both posit reasons that derive from the agent's power to affect outcomes as an individual.
However, the idea that the reason to prevent disasters is a special case either of a general reason to prevent harm or of a general reason to promote the good does not threaten to make Rule Consequentialism unstable or prone to collapse into Act Consequentialism.The threat of collapse would arise if recognising these general reasons entailed treating them as overriding in all cases.But there is no such entailment.We do not accuse Ross's view of collapsing into Act Consequentialism just because it recognises a general duty of beneficence, since Ross makes clear that this duty often gives way to other, independent duties (Ross, 2002(Ross, [1930]], ch.2).The same goes for Rule Consequentialism-at least for those of us persuaded by Hooker's general defence against the collapse objection (Hooker, 2000a, ch. 4).Recognising a general reason to prevent harm, or a general reason to promote the good, introduces no new threat of collapse, since Rule Consequentialism has the resources to explain why these reasons are often outweighed by other, independent reasons.

| RULE CON S EQUENTIALIS M VS . AC T CON S EQUENTIALIS M
We can compare the different theories of reasons offered by Rule Consequentialism and Act Consequentialism. 10   As I have just argued, we can interpret Rule Consequentialism as recognising a general reason to promote the good alongside other general reasons such as a reason to keep promises.These are general reasons, in the sense that they are reasons associated with kinds of action.Act Consequentialism can also recognise reasons that are general in this sense.It seeks to explain these reasons by claiming that certain kinds of action have noninstrumental value.
For example, it can claim that instances of promise-breaking have negative noninstrumental value, and so try to explain a general reason to keep promises.Where Rule Consequentialism appeals to a plurality of rules to explain the plurality of reasons, Act Consequentialism appeals to a plurality of values.On what grounds should we prefer one approach to the other?This is a complex issue, and I do not pretend to offer decisive arguments here.However, there are several simple points that we can make.One is that, while both appeal to a kind of plurality, Rule Consequentialism seeks to explain the plurality of rules, and the interaction between them, while Act Consequentialism must treat the plurality of values, and their importance in comparison to each other, as basic.This makes Rule Consequentialism in that respect a more ambitious theory.To the extent that it is successful, and if all else is equal, it has greater explanatory power-as Hooker has emphasised (2021, p. 223). 11  On the other hand, Act Consequentialism offers a picture of the general nature of reasons that many find compelling, and Rule Consequentialism offers nothing similar.According to Act Consequentialism, all reasons are reasons to make the world better.This expresses an appealing general view about the purpose of action.As Ewing put it: 'That it tends to produce a good seems not only a reason but the only ultimate reason for doing anything … 10 For the purpose of this discussion I will assume that neither theory faces a fatal objection.
11 Of course, Rule Consequentialism may also appeal to a plurality of values-but we would expect it to place less weight on that, since it has other resources to explain the plurality of reasons.
| 9 WOODARD This is at least a very plausible position to take up ' (1948, p. 103).12Rule Consequentialists either deny that all reasons have something in common, or must claim that what they have in common is that they are specified by rules in the ideal code.This is a less compelling claim, in the abstract, than the claim made by Act Consequentialists.So far we have noted one respect in which each theory seems superior to the other.It is a matter of judgement where the balance of advantages lies.However, there is a further possible advantage of the Rule Consequentialist theory which could tip the balance for some.This is that Rule Consequentialism is in a much better position to explain moral constraints than is Act Consequentialism.Constraints are reasons for or against performing an action of a certain kind that make it wrong, on some occasions, to make the outcome as good as possible impartially speaking (Kagan, 1989, p. 4, pp. 24-32).For example, if there is a constraint against breaking promises, then it is sometimes wrong to break a promise even when doing so would make the outcome impartially better.Rule Consequentialism explains such constraints in terms of the rules in the ideal code.If, as we have been supposing, there is a rule requiring promise-keeping in the ideal code, then Rule Consequentialism claims there is a reason to keep your promises that is independent of reasons to make the outcome better.So long as this reason is not always outweighed by reasons to make the outcome better, Rule Consequentialism can then explain why there is a constraint against promise-breaking.
Act Consequentialists can also try to explain this, but their efforts to do so involve distorting their claims about value.To capture the agent-relative and time-relative features of constraints, they need to make agent-relative and time-relative claims about value.For example, they must claim that, so far as your reasons are concerned, your breaking a promise now is worse than someone else's breaking the same kind of promise, or your breaking the same kind of promise at a different time (Broome, 1991, ch. 1).These claims about value seem to me to be, on their own merits, highly implausible (though for a defence see Portmore, 2011Portmore, , pp. 111-116, 2022)).They also threaten to undermine the advantage Act Consequentialism enjoys in presenting a plausible general account of the nature of reasons.That advantage is undermined if we must distort our theory of value in these ways in order to give that account.
So far as these considerations go, then, a key issue in choosing between the Act Consequentialist and Rule Consequentialist theories of reasons is whether we wish to explain constraints.For those who do not wish to do so, Act Consequentialism may win out.For those who do wish to do so, Rule Consequentialism may be more plausible overall.I should emphasise that this judgement is made with respect to a limited range of considerations, and so is not a decisive argument one way or the other.

| CON CLUS ION
Rule Consequentialists of Hooker's sort can and should give an account of reasons, alongside their accounts of other matters of ethical importance.Doing so increases the power of their theory, by enabling it to explain more of our ethical judgements.It also helps to crystallise some of the issues involved in the choice between different forms of consequentialism.The main trap to avoid in thinking about Rule Consequentialist theories of reasons is the assumption that all rules in the ideal code perform the same function.Rule Consequentialists can develop a theory of reasons by claiming that reasons are specified by a subset of the rules in the ideal code-the reasonspecifying rules, as I have called them.Other rules in the code perform other functions-telling us, for example, about the relative strength of different reasons.

ACK N OWLED G EM ENTS
It is an enormous pleasure to contribute to this volume of essays on Brad's work, since it is one way of acknowledging my huge debt to his writings and to him personally.I am very grateful to Luke Elson and Charlotte Newey for the opportunity.