by Ted Honderich

This is a piece that tries to put right what seems to me to be a lot of confusion about consequentialism in morality and moral philosophy -- the doctrine that actions and the like are made right by their consequences. It defends consequentialism, and in fact maintains that morality consists in nothing but consequentialism. The only moral reasons are consequentialist reasons. Certainly the piece swims against the tide, or anyway a current and a lot of eddies. What is maintained is very relevant to a particular morality that recommends itself to me -- the Principle of Humanity. A revised version of this piece, and one on the Principle of Humanity, and related papers, are in the collection On Political Means and Social Ends (Edinburgh University Press, 2003).

Abstract: (1) Non-Consequentialism and Consequentialism are best identified ostensively, by giving two groups of reasons for the rightness of actions. Concentration on Utilitarianism distorts understanding of the second group. (2) Traditional causal conceptions of the two groups are inadequate and unenlightening. Conceptions are inadequate if they do not cover all the reasons of only one group. They are unenlightening if they do not convey the supposed general recommendation of each group of reasons. (3) Reasons of the second group are in fact best understood in terms of the satisfaction and frustration of desires, and giving a strict priority to reducing frustration. They come from moralities of concern. Reasons of the first group either do not have to do with satisfaction and frustration at all, or do not give a strict priority to reducing frustration. (4) Reasons of the first group, including agent-relative reasons, cannot be defended by collapsing the question of right action into that of the goodness of an agent. (5) Reasons of the first group, if taken as not involving satisfaction and frustration at all, are in fact not reasons for action and hence not reasons for the rightness of action. (6) Reasons of this group, if taken as involving satisfaction and frustration, face a judgement of selfishness.  

Two Groups of Reasons

Here are some kinds of reasons for taking an action to have been morally right. 

(1) It was done out of a good intention or a pure good will on the part of the agent, or was owed to a virtue of hers. 

(2) It issued from the agent's moral perception or intuition with respect to a situation, not from the application of a general principle or from calculation of the consequences of possible actions. 

(3) Although it would give rise to distress or worse, the action was one of integrity, autonomy, or self-concern, perhaps in accord with the agent's aversion to killing by her own hand, or true to her life-hope to achieve a success on her own. 

(4) It did not violate the agent's natural, unique, or established relationship with someone, say her child, loyal friend, or benefactor. 

(5) It was true to the agent's identification with or first concern for her own people, class, or group, perhaps within a homeland or society. 

(6) Despite being likely to have bad or awful consequences, the action was in accord with a moral rule or law such as those against torture, lying, and promise-breaking.  

(7) It respected the constraint of someone's moral or legal right, perhaps to a piece of private property, a job, or a pension.1 

Here is a second group of reasons for taking an action to have been morally right.  

(8) It saved the life of someone, or saved someone from injury, or stopped the suffering of an animal. 

(9) It made someone feel better, or gave them the means to grow, or a job, or hope, or self-respect, or a decent life. 

(10) It treated everyone the same, taking into account only their human needs.  

(11) It compensated someone for a loss, deprivation, or injury, or gave them help with a physical disability. 

(12) It secured, from among the possible distributions of goods in a society that would satisfy the reward-demands of producers, the particular distribution that would give most goods to the people with least. 

(13) It was likely to alleviate the condition of someone badly-off in terms of the satisfaction of fundamental human desires rather than improve the condition of someone already well-off in this way. 

(14) It gave rise to or contributed to a more equal distribution of incomes, freedoms, opportunities, other particular goods, or well-being. 

(15) It was, of the possible actions, the one likely to produce the greatest balance of happiness or the smallest balance of unhappiness.2 

Traditional Conceptions

The first group of seven kinds of reasons is a mixed bag. Also, the reasons can be taken in different ways. Further, several particular ones assigned to different kinds might instead be put together in a new kind, and several particular reasons might be taken as inclusive of others or maybe as being the same reason. It is also true that several may conflict. None of this will affect my inquiry. If the first group of kinds of reasons is a mixed bag, they are not in the bag by chance. They can all be taken and have been taken in such a distinctive way as to share a certain nature. They may then be labelled as non-consequentialist or deontological reasons. There are also conflicts and other relations within the second group of reasons, but it is also possible to take these reasons in a distinctive way, as sharing a nature, and hence to label them consequentialist or teleological reasons.  

The two lists of reasons give us definite subject-matters. Not all discussions in this neighbourhood of moral philosophy have had this advantage. Also, the second list reduces the prominence given to Utilitarianism in recent discussions of this group of reasons. It figures only in (15). To my mind the attention of opponents of these reasons to Utilitarianism has greatly distorted understanding of them. This has to do with the fundamental objection to Utilitarianism, the fact that it sanctions victimization, injustice, and the like, as moral philosophers have long recognized. 

To come to a critical overview of the two groups of reasons, which is my aim, it is not enough to take them in an unreflective way to have different natures, to have just a sense of their different natures. What are these natures? What articulated conceptions are we to have of the two groups of reasons? What really is the general difference between them? 

Shall we conceive of the first group, taken in the distinctive way, as not being about the consequences of an action? And the second as being about only the consequences of an action? Many philosophers still have the habit of doing so.3 But these two conceptions, which may be called the wholly non-consequential conception and the wholly consequential conception, are inadequate. That is, neither covers reasons in just one of the two groups. This can be shown.  

Consider (4) in the first group, about an agent's close relationships. Consider in particular a woman who makes the life of her own child better when she could instead improve the life of another far less fortunate child. Her reason, as she somehow expresses it, is to the effect that she is benefitting her child. It is impossible to take the reason as not being at all about the consequences of her action. So it fails to fall under what we are contemplating as a conception of the first group of reasons, the purely non-consequential conception. There is the same upshot with reasons of all the kinds (3)-(7).  

Shall we then conceive of the first group of reasons as being about not only the consequences of an action? We can call this the partly non-consequential conception. Consider the woman again. We are now contemplating the idea that her reason has a part which does not have to do with the consequences of her action. We can register this idea by an emphasis in stating her reason. She is benefitting her child. Or we could go further and expand her reason. We could say she is benefitting the child to whom she has a natural relationship, or to whom she has a unique obligation. 

There are problems about this. For a start, any version whatever of her reason can be reformulated in a certain way. Thus we say that her reason is that her action has the consequence that the child to whom she has a natural relationship is benefitted, or that her action has the consequence that the child to whom she has a special obligation is benefitted. Nor can the causal prefixes be put aside as somehow adding no relevant fact -- as some suppose the prefix "It is true that" adds nothing to a proposition. What makes the causal reformulations possible, it seems, is a certain fact. It is that it is part of the resulting state of affairs that the benefit goes to the child to whom she has a natural relationship or a unique obligation. This seems as unquestionable as, say, that it is part of the state of affairs that a person is benefitted. So the partly non-consequentialist conception, which now appears to be misnamed, fails. In fact the reasons of the first group seem to be not partly but entirely about consequences. We thus have no distinction between the two groups of reasons. 

Will someone resist this by saying that it wasn't her action, the one we are talking about, that made the child her child? And that the consequence of her action precisely conceived is to be taken just as a benefitting, say a feeding, or, better, a property of something's being benefitted, say being fed? There is a lot to be said for this precision, a focussing on properties or events as effects, not states of affairs.4 If we are precise in this way, we can certainly say that a part of her reason did not have to do with consequences. But then there will be no relevant distinction between this reason and a related wholly consequential reason, where someone benefits a child to whom they are unrelated. The latter reason will also have a part which does not have to do with the consequence of the action. It cannot fail to be a part of the latter reason that what is benefitted is a person or indeed a child. So this precise understanding of the partly non-consequential conception makes it cover reasons of the second group. Again we have no distinction. 

There is the same story, although I will not spell it out, with all of the reasons (3)-(7). So the partly non-consequential conception of the first group of reasons fails or at least faces serious difficulties. 

Consider the second group of reasons and the idea that they fall under the wholly consequential conception -- that they are reasons about only the consequences of an action. We can doubt this already, by way of the precise idea of effects. But that is not all. In particular, consider (11), which justifies an action by the fact that it will benefit someone for a loss or the like in the past. This evidently does not have to do only with consequences of the action. It also has to do with the person's previous life.5 What of (13) alleviating the condition of someone badly-off in terms of fundamental human desires rather than improving the condition of someone better-off? It is again plain that the antecedent state of the person is part of the reason. So with (12), about people with least goods, owed to Rawls, and some other reasons in the group. Thus the consequential conception of the second group of reasons fails or at least faces serious difficulties. 

There are the same difficulties with what is a kind of elaboration of the wholly consequential conception of the second group of reasons. One philosopher who has been much concerned with the two groups of reasons takes it up. He may be taken to characterize reasons of the second group, or rather the related theories, which comes to the same thing, in a certain way. 

...we may think of a consequentialist coming in two parts. First, it gives some principle for ranking overall states of affairs from best to worst from an impersonal standpoint, and then it says that the right the one that will produce the highest-ranked state of affairs.... Obviously, there can be as many different theories of this type as there are criteria for ranking overall outcomes.6 

That is to say that reasons of the second group involve ranking the outcomes or consequences of possible actions in some way or other, in terms of happiness produced or whatever else, and also involve a rule specifying the right action as the one producing the best outcome, say the most happiness. Plainly this elaborated conception also runs afoul of the fact that some reasons of the second group have to do not only with consequences of actions but also with antecedent facts. 

Shall we now try to pile on epicycles, try to refine the partly non-consequential and the wholly consequential conceptions so as to make them adequate? This would be the endeavour of saying more about the causal relations of actions to other things, but, so to speak, without looking into the very nature of the terms of those relations. I am not the first to think this would be pointless, and hence to take as misleading the persistent labels "Non-Consequentialism" and "Consequentialism".7 In my view there is something more promising to think about. 

If we want to inquire into the relative worths of the two groups of reasons, we not only want adequate conceptions, conceptions true of just the right things, but conceptions which are enlightening. That is, we want conceptions which display the supposed general recommendations or creditable general natures of the two groups of reasons, the grounds for sympathy with them. Suppose we can somehow refine or enlarge the two causal conceptions, so that they nevertheless remain of the same kind, somehow having to do with causal relations, the categories of non-consequences and consequences, but so that each is clearly true of all the reasons of only one group. This will not help us much. It will not give us enlightening conceptions. 

This is so, to put the matter one way, because the new causal conceptions, like the old ones, will almost certainly be true of many reasons so far uncontemplated, and, crucially, certain unwelcome or intolerable reasons. They will licence including these reasons in one or the other of the two groups. Take the following reason. 

(16) The action was in accordance with the folk-law that because it is midsummer day the life of a maiden must be sacrificed. 

That, understood in a certain natural way, will go into the first group. It will accompany or be included in (6), about moral rules or laws. So will a lot of terrible reasons actually held by people today, some of them having to do with races or ethnic groups. Or take the following reason having to do with personal integrity, which will go into (3) in the first group. 

(17) The torturing was true to the agent's deep commitment to a political movement or regime, not just done as part of a policeman's job. 

And take three other reasons. 

(18) The action was the one most likely to increase the amount of the colour mauve, taken as an intrinsic good, and hence not as a means to well-being, or to fairness, or to anything else whatever. 

(19) The action gave someone what she deserved -- a penalty or distress, a reward or satisfaction, or some other fitting return -- for her action or activity.  

(20) The action, although it was likely to improve lives less than another, had as a likely consequence more actions in the future arising from a good intention, a pure good will, or a virtuous disposition. 

These will go into the second group according to a causal conception.8 So will certain counterparts of (20), similar reasons derived in the same way from the other reasons in the first group. 

But sympathizers with each group of reasons will reasonably resist the proposed additions to their group. Perhaps mere taxonomists of some sort might not mind the additions, but my concern is with the two groups of reasons taken as advanced for our real moral consideration. 

It is not clear to me what general ground could be given by sympathizers with the first group of reasons for resisting the inclusion of (16) and (17). But sympathizers with the second group will resist (18), about mauve, on roughly the ground that the action in question would not improve lives at all. They will also resist (20), about more morally-motivated actions in the future, on roughly the ground that such actions may not improve lives. They will resist (19) on roughly the ground that lives will not be improved at all, but worsened, in the cases where what is said to be deserved is a penalty or distress. Reasons of retribution and the Retribution Theory of punishment do not fit into and have never been taken as fitting into the moralities of sympathizers with the second group of reasons. Indeed, these sympathizers will wish to consign retribution to the first group of reasons. It can be said to consort with (6) and (7), about rules and rights. Also, reasons of retribution will actually be accepted by some sympathizers with the first group of reasons.9  

The plain fact is that sympathizers with each group of reasons have some conception of it, an enlightening conception, which cannot possibly be caught by the generality of talk of just causation -- talk of just non-consequences and consequences. What they are sympathizing with are what they regard as good or at least contemplatable reasons. You are very unlikely to be able to distinguish good from bad consequences by sticking to thinking about causation itself. So with non-consequences.  

More might be said about the unenlighteningness of the traditional conceptions of our two groups of reasons and of other related conceptions of them, some of which concentrate on or start from the first group rather than the second.10 But, to come instead to a main contention of mine, implied in what has been said of the need for enlightening characterizations, there is a large fact about the two groups of reasons which has so far been overlooked. 

Moralities of Concern

The second group of reasons, to recall a few of the descriptions already used, has to do with stopping suffering, giving hope, improving lives, not causing distress, alleviating the condition of the badly-off, treating everyone the same, securing equality of liberty or whatever, compensating someone, helping those with least, increasing happiness and decreasing unhappiness. We might also say, as others have, that the reasons of the second group have to do with well-being, utility, preferences, interests, or choices.11 It is best to persist in what, despite its great generality, is the strongest and perhaps the most traditional characterization of these reasons. They have to do with the satisfaction and frustration of desires. 

All the reasons, to speak in this general way, indeed too general way, recommend an action by the fact that it is likely somehow to reduce frustration or increase satisfaction, usually taking into account the time before the action as well as after. Such a fact about an action is one whose moral relevance and force is uniquely indisputable. Nothing else, save perhaps certain facts having to do with very simple kinds of fairness, has the same dominance. It is mainly this character of the second group of reasons which is the ground for rejecting the inclusion of (18)-(20), about mauve, penalties, and more morally-motivated actions in the future. 

Talk of satisfaction and frustration in general needs more clarification than can be given here. Certainly a desire of mine is not satisfied in the relevant sense just if the proposition I desired to be true becomes true, but unknown to me. Nor is the desire satisfied if no more is the case than that I know that the proposition has become true. Satisfaction is not merely a cognitive fact, but involves emotion and conation -- and a valuable behavioral criterion. So satisfaction in the right sense is not something different in kind from truth-based happiness. Related things are to be said about frustration, which will turn out of be more important than satisfaction. None of this implies, by the way, that all we should value or aim at in our lives is happiness or other states of feeling. 

There is the question of what satisfactions and frustrations are to be counted in by the moralities from which the reasons of the second group come. The answer is that all satisfactions and frustrations without exception, of whatever kind, ought to be included. None of these moralities needs to weaken itself by placing any socially-concerned or even widely agreed limitation, let alone any high-minded or delicate one, on the satisfactions or frustrations to be considered. That would be a limitation very likely to have less moral relevance and force than satisfaction and frustration in themselves.  

This is not to say that these moralities cannot organize themselves in one way or another, say by using fundamental categories of satisfaction and frustration, perhaps defined by length of life, freedom and power, respect and self-respect, and so on. Nor, of course, does the inclusion of all satisfactions and frustrations mean that the moralities in question will take the fact that an action will satisfy or frustrate some one desire as being itself a sufficient reason for the action. Each action will be judged in relation to the satisfaction or frustration of other desires affected by it. Thus the fact that an action will satisfy or frustrate some single desire will often be overborne by the fact that it will frustrate or satisfy other desires, then or later. That does not take away from the indisputable moral dominance of satisfaction and frustration. 

There is something else quite as important about all the reasons of the second group except for the Utilitarian reason (15), that an action was likely to produce the greatest balance of happiness or the smallest balance of unhappiness. All the other reasons, (8)-(14), give a general priority to reducing frustration as against increasing satisfaction, and of course to reducing greater as against lesser frustrations, and to increasing lesser as against greater satisfactions. That is a very general description, of course. The moralities in question will necessarily include particular decisions and rules giving further expression to this commitment, which I shall henceforth refer to just as the commitment to giving priority to reducing frustration.  

The commitment is explicit in (8), about saving life and so on, and (9), about making someone feel better and so on, and (11), about compensation and the like, and (12), about a distribution that favours those who have least goods. It is also explicit in (13), about alleviating the condition of someone badly-off, which reason derives from what has good claim to the name of the Principle of Equality.12 In brief, this principle is that we should seek to remove people from a defined class of the badly-off, partly by having certain practices of equality.  

The priority given to reducing frustration also enters into the reasons in (10), (12), and (14), all having to do with equality, although it is not made explicit. The various egalitarian moralities in question are all of them fundamentally concerned with raising up those below rather than bringing down those above. Despite what has sometimes been said in opposition to these egalitarianisms, it is impossible to suppose that they would take their goals to be perfectly achieved if they attained an equality of educational opportunity owed to there being no schools at all, or an equality in liberty which involved minimal liberty when more was possible, or equal low incomes for all rather than equal higher ones for all. 

To return to the Utilitarian reason (15) for a moment, after which we will no longer be concerned with it, the same concern can be argued to have entered into Utilitarianism as a principal motive, perhaps the principal motive -- despite the disastrous failure to express it in the classical Principle of Utility. The failure, as already remarked, opens Utilitarianism to the fundamental objection that in some cases it sanctions victimization and the like as a means to the greatest balance of happiness. 

In sum, reasons of the second group have to do with satisfaction and frustration, and (8)-(14) give priority to the reducing of frustration. It is not easy to find a label or characterization for (8)-(14) and the associated moralities, one which conveyes the fact of priority. Without supposing that I have yet found any recommendation of them, I shall call them moralities of concern, and also speak sometimes of reasons of concern.13 

What is to be said of the first group of reasons? We have concluded so far only that the traditional conceptions of them are not adequate and enlightening.  

It has seemed that reasons of kinds (1) and (2), about good intentions and the like and moral perception and the like, can be taken as not having to do with satisfaction or frustration at all, including the agent's. This has often been insisted upon. Certainly the agent with the good intention or the moral perception may have in mind that her action will give rise to certain satisfactions. But, we have been told, the given reason for the rightness of the action is not these satisfactions. We are to understand that someone else can both accept such a reason for the rightness of an agent's prospective action and believe that the action will in fact not give rise to the satisfactions. With such a reason it is the intention or the perception alone that recommends the action. 

What of reasons of kind (3), having to do with an agent's preserving her integrity or the like? It has seemed possible to think of these reasons in the same way as with (1) and (2). That is, the recommendation of preserving one's integrity does not have to do with satisfactions or frustrations. On the other hand, it seems possible to think of reasons of kind (3) differently, as having to do with preserving self-respect, where self-respect is a satisfaction to the agent and its loss a frustration. I do this action rather than that mainly because the result of doing that would be the distress of my being unable to live with myself. 

With the remaining reasons of this group, (4)-(7), it is difficult to suppose they do not have to do with satisfaction and frustration. It has been done, no doubt, but it is difficult. With respect to (4), about close relationships, and in particular the case of the woman again, her concern is surely for the satisfaction of her child. So with (5), about one's own people or nation or the like. With respect to (6), about moral rules or laws, what must surely come to mind are the pains of being deceived, the hurt of being let down by a promise-maker, the agony of torture, and so on. Something of the same sort is surely true with (7), about rights. 

What we have then is that some of the reasons of the first group have seemed not to have to do with satisfaction or frustration at all, and some, under a certain natural understanding, do have to do with it. We can leave it unsettled whether reasons of kind (3), about integrity and so on, go into the first sub-group or the second, or, by way of two different understandings of them, into both. If you wish, leave open the same possibilities with (4)-(7).  

But there remains a salient fact about the first group of reasons. It is that those of them so understood that they do involve satisfaction and frustration do not involve giving a strict priority to reducing frustration. Consider (3), about integrity and the like. Someone who uses such a reason to defend an action does of course have the intention of defending the action against the consideration that another action would reduce frustration. In particular, to speak of a central case, there is the intention of defending the action when on the whole it gives rise to frustration and another action on the whole would give rise to satisfaction. The reason in question, which is advanced against reasons of the second group, would lose its point if understood otherwise. The same is true of all the remaining reasons of the first group, (4)-(7), and is given natural expression in several of them. 

In sum, reasons of the first group either (i) do not have to do with satisfaction or frustration at all or (ii) they do not give strict priority to reducing frustration. I do not mean, of course, that the moralities in question never take the reducing of frustration as decisive. Certainly the mother in (4) is not likely to choose any degree of satisfaction for her child as against any degree of frustration for another. What is nevertheless true of reasons (3)-(7) of the first group is that they do not involve a commitment to reducing frustration, and often go against such a thing. They are not associated with moralities of concern. There remains the possibility of coming to a positive rather than a negative characterization of them, and also an enlightening one. (3) and (4) in particular, about integrity and the like and close relations, have sometimes been called agent-relative moralities. 

Before taking a closer look at the first group of reasons, let me put in a traditional reminder. We have it that reasons (3)-(7) can be taken as having to do with satisfaction and frustration, but not with the priority of reducing frustration. That is not the only thing to be said about satisfaction and frustration, and priority, in connection with these reasons. Consider (6), the reason for an action that it was according to a moral rule or law such as those against lying, promise-breaking, and torture. Those three subject-matters will also fall under certain prohibitions in moralities of concern. They will fall under prohibitions derived from the main principles of those moralities. To speak of the morality which I am inclined to support, based on the Principle of Equality, it has in it a rule, in practice as good as absolute, against torture. It also contains severe rules against lying and promise-breaking.  

The situation is the same with (4) and in particular the mother and child. Certainly moralities of concern, consistently with what has been said of the strict priority to reduce frustration, will not only allow agents to give a certain precedence to the satisfactions and frustrations of certain other individuals, but will actually place an obligation on agents in this regard. These moralities will allow for and call for what can be named, perhaps irritatingly to philosophers of a certain sensibility, a division of moral labour. Each of us is likely by nature and situation to be well-placed, perhaps uniquely well-placed, to act on a morality of concern with respect to a few other individuals. There is the general conclusion, then, about reasons (3)-(7) as we have understood them, that to reject them would certainly not leave nothing in their place.14  

An Argument For Reasons of the First Group?

Let us now look more closely at the first group of reasons. Is there a general argument in favour of them? Is there a promise of an argument for them overlooked by their proponents or perhaps assumed but not articulated? 

Reasons of kind (1), about a good intention, a good will, or a virtue, are offered by philosophers as, and we are taking them as, reasons for the rightness of an action. In fact there is everything to be said for taking them differently, as reasons, or the ordinary reasons, or indeed the only possible sort of reasons, for something else. That is moral approval of the agent with respect to that particular action, or, as we can as well say, crediting her with responsibility for it. Also, of course, that an agent did not act out of a good intention or the like is a reason for morally disapproving of the agent with respect to the action, holding her morally responsible for it. In fact, this is how the reasons in (1) would naturally be taken if they had not been introduced as being reasons for the rightness of actions. It is also possible to take them as reasons among others for approving of the agent morally in a more general way, not having to do only with the action but rather with respect to a longer period of time or even her whole life -- approving of her as a good, human, or decent person. 

Can it be supposed that the issue of the rightness of actions must be regarded as identical with or somehow reducible to either of the two issues of moral approval of agents, more likely the first? That there is some kind of conceptual reason or reason of logic for doing this, as distinct from a general moral inclination? And hence that the reasons in (1), although not best described as reasons for the rightness of an action, have everything to be said for them as just such things? Might something of the same sort be true with other reasons of the first group, above all those in (3) having to do with integrity and the like?  

Presumably any such argument must fail. The first issue, the rightness of actions, is not conceptually or logically identical with or reducible to the others. Admittedly there is a logical connection between right actions and the first and fundamental kind of moral approval of agents. Right actions, to speak generally, are necessarily those done by agents who (i) earn this moral approval, and (ii) are agents of the best available knowledge and judgement. But this logical connection casts no doubt on the fact an action obviously can be held wrong while the agent is morally approved of, and an action can be held right while the agent is morally disapproved of.15 Obvious cases of the first sort will involve honest or honourable mistakes by the agents as to the likely consequences of their actions. There is no possibility, then, of moving from reasons for moral approval to a conclusion about the rightness of action.  

An Argument Against Reasons of the First Group Understood As Not Involving Satisfaction or Frustration

As remarked, it has seemed that reasons (1) and (2) in the first group, about good intentions etc. and moral perception etc., can be taken as not having to do with satisfaction or frustration at all, including satisfaction or frustration of the agent. It has seemed that (3), about integrity or the like, can also be taken in this way. We allowed the same thing to be a remote possibility with the remaining reasons of the first group. Is there a general argument against these reasons so understood? Is there anything that stands in the way of their actually being significant reasons for the rightness of action? 

Let us detach for a moment from our particular subject and indeed from moral philosophy. What is a reason for an action? I am not talking about a reason for the rightness of an action. Rather, the question is one from the philosophy of mind: What is a reason of whatever kind for an action? In what can be called the primary sense of the term, a reason for an action is what it is in the agent, a mental state or event, that actually gives rise to the action. Speaking very generally indeed, such a mental state or event consists in a propositional content and a desire -- a thought of something, and a desire with respect to that thing. A reason for action in this primary sense is something had or experienced.  

A reason for action in a secondary sense is something given to another person or oneself. It is, in brief, linguistic rather than mental. It is evidently to be understood in much the same way as a primary reason. A reason in the secondary sense is a propositional content and a representation having to do with desire. In short, what I do in giving a reason for action is to represent something as desirable.16 

This view, while rightly the subject of elaboration of various kinds, is and always has been taken as truistic in the philosophy of mind, for good reason. It is a ruling idea owed to the most fundamental of our beliefs about ourselves, that it is wanting something that gives rise to our actions.17 To return now to moral philosophy, what are reasons for the rightness of actions? What is the general nature, that is, of claims such as (1)-(20) if they are given as reasons for the rightness of actions? So far we have not reflected on that question.  

One part of the answer, obviously, is that necessarily they are in a certain relation to reasons for action in the secondary sense. Each of them is something that could be given to someone as a reason for an action. What I do in saying of a certain possible action that it would be right since it would make someone feel better is necessarily to give a reason for the action. What I do in saying of a past action that it was right -- as in the case of (1)-(20), is to give a reason for present or future actions of the same kind. Again this is truistic. The fact in question is bound up with the general truths that the point of morality is the guidance of action, and that its subject-matter is that of how things ought to be in so far as we can affect them.  

What is to be taken from this lightest of sketches, firstly, is that a reason for action in either the primary or the secondary sense necessarily involves a propositional content and a desire -- if anything purports to be a reason for action, and in particular does not involve desire, it is not what it purports to be. Secondly, anything that is a reason for the rightness of action is itself a possible reason for action -- anything which purports to be a reason for the rightness of action, and which is not a possible reason for action, is not what it purports to be. 

A certain general conclusion follows. To repeat, it has seemed that reasons (1) and (2), about a good intention or the like and moral perception or the like, do not have to do in any way with satisfaction or frustration, and in particular any satisfaction of the agent. It has seemed that (3), about integrity or the like, can also be understood in this way. We allowed the same possibility with other reasons of the first group. But if these various reasons do not have to do with satisfaction or frustration, they clearly do not have to do with desire. And if they do not have to do with desire they cannot be reasons for action and hence cannot in fact be reasons for the rightness of actions. The conclusion, again, is that if these supposed reasons for the rightness of an action are understood as not having to do with satisfaction or frustration, this actually precludes them from being what they are supposed to be.  

It may be useful to state the argument even more generally, as follows. Anything that is a reason for the rightness of an action must possibly be motivating; some so-called reasons for the rightness of an action include no reference to any desire, and hence cannot possibly be motivating; hence they cannot really be reasons for the rightness of an action. It may also be useful to approach the conclusion by way of (18), about mauve, which is unfamiliar and without confusing associations. If we exclude from our minds any thought of mauve's being satisfactory, to the agent or anyone else, we have no reason at all, and hence no reason for the rightness of an action. The same is true, by parity of reasoning, with all the reasons of the first group if they are taken in the satisfaction-excluding way. 

So the option of taking reasons (3)-(7) in the satisfaction-excluding way turns out not to exist. If they are reasons for the rightness of an action at all, they have to be understood in the other way, as having to do with satisfaction.  

What is to be said of (1) and (2), about good intentions and moral perception, where some have in fact gone so far as to insist that they are not understandable in terms of satisfaction? The short answer is that despite protestation they can only be understood in this way. It seems indisputable that they have been and can be reasons for the rightness of actions, and hence are motivating. The truth must be along the lines that to give a reason from (1) or (2) is in fact to depend on satisfactions associated with the action. Very likely it is to depend on satisfactions and frustrations involved in reasons (3)-(7) -- it is to give such reasons in a tacit way. Such a thing is familiar outside of morality. We value something for its dispositional character, which is tacitly to value it for its effects. 

A View of Reasons (3)-(7) Taken As Involving Satisfaction and Frustration

Consider the reasons in (3)-(7) understood in the satisfaction-involving way. Thus taken, they recommend actions by way of satisfactions and frustrations, but do not give a strict priority to reducing frustration. Thus the reasons in (3) may take an agent's own satisfactions in an action to count for more than the frustrations of someone else. The reasons in (4) may do this for an individual closely related to the agent, say her child. The reasons in (5) may do so for other persons, say her nation. Relatedly, one reason in (6) may not only place a bar on torture, as moralities of concern may also do, but also go further and somehow exclude from any consideration the situation of others than the agent and the immediate victim. It may be that such a reason against torture would prohibit it even in the case where it could be known without doubt that it would prevent more torture of other victims. The reasons in (7), finally, may put the satisfaction of rights-holders above the frustration of others.  

Do these propositions support a certain judgement which only deference or delicacy can keep from coming to mind? It has to do with selfishness, taken in the ordinary way as too great a concern for one's own satisfactions and frustrations, too little for those of others. The judgement is that the reasons in question, whatever else they may be, are reasons of selfishness. They are reasons of selfishness, more precisely, since they depart from giving a strict priority to frustration, in favour of the agent or related persons, and if, as is common enough, (i) they are advanced by the agent, or (ii) they are advanced by persons whose circumstances are like those of the agent, or (iii) they are advanced by those who in fact are the related persons.18 The second case is exemplified when, to think of reason (7), the holder of a piece of private property defends a claim of someone else, the agent, to the agent's piece of private property.  

Moral philosophers seem to have supposed that since reasons (3)-(7) are reasonably presented as reasons for the rightness of actions, and do seem to be such reasons, and thus do seem to be within the general category of moral reasons, they are unlikely also to be reasons of selfishness or cannot also be such reasons. This supposition seems to have been supported by another, that there is some test for being a moral reason, and that this test is passed by the reasons in question. 

One test that may be thought of is impartiality or consistency. To revert to (4), this impartiality is exemplified by the mother who not only favours her own child, but allows that other mothers have as much reason to favour their children. It is as much exemplified by (3) and in particular the reason having to do with one's own integrity, and by (5) and in particular the reason having to do with the agent's own people. In each case, others are allowed to have the same reason for the rightness of their actions. So with (6) and (7).  

Consistency has of course entered into definitions of moral judgements and morality. It is necessary but it could not possibly be sufficient by itself for regarding something as a moral reason or in particular no more than a reason for the rightness of an action. This is clear from just the fact that factual or inductive reasons have the same character. Suppose I think that putting something on my table will make it collapse, and give as my reason that the thing weighs 1000 pounds. I accept, as I must, that another thing of a like weight will make a like table sag. Indeed some reasons of selfishness satisfy the test of impartiality. If I defend putting myself first, I may grant that in consistency you too can do so. 

What is the relation between moral reasons and selfish reasons, or, better, the relation between reasons as moral and reasons as selfish? Consider the following ordered list of reasons for the rightness of actions, some of them now familiar. An action (i) satisfied a starving man's need for food, (ii) cured a child of a disease, (iii) gave someone a fair share of something, (iv) was in the fullest accord with the agent's own rights, (v) preserved an agent's integrity, maybe as an egalitarian, (vi) advanced one's socially useful career, (vii) preserved the dignity of one's family, (viii) gave one's son a somewhat unfair advantage in an examination, (ix) caused hatred of someone who injured the agent, (x) cleansed one's country of an ethnic group or race, (xi) allowed someone to suffer agony or die in order to avoid some bearable loss to oneself.  

I have arranged the reasons (i)-(xi) roughly in descending order, or rather, as it turns out, roughly in two descending orders. 

As you go down the list, as seems evident, the reasons become less moral and more selfish. Several reasons at the top of the list are paradigmatic reasons for the rightness of an action. Several reasons at the bottom of the list are, to say the least, paradigmatic reasons of selfishness. That is the first thing of importance about the list. 

The proposition that the reasons taken in descending order are less moral and more selfish need not be owed to a prior definition of morality. Rather, it can be owed to particular convictions, intuitions, or responses with respect to the reasons -- convictions and the like that must guide any attempt to define morality. Also, and quite as important, the proposition need not be owed at all to an articulated morality of concern, or even an informed sympathy with such moralities. We do not need a moral theory in order to make the given responses to such a list of reasons. We need what we have, particular convictions. There need be no circularity or question-begging in the imputations of selfishness.  

There is a second thing of importance suggested by such a list, connected with the first. The reasons are also roughly in a descending order with respect to giving priority to reducing frustration. Those at the top will be taken as clearly coming from moralities of concern. Those at the bottom are nothing of the sort. Those in the middle, (iv)-(viii), involve a compromise. What is suggested by this, since as the reasons do also fall into a descending order from paradigmatically moral to viciously selfish, is that morality is bound up with the priority of reducing frustration. 

A third thing illustrated by such a list, the main thing for present purposes, is that it is a mistake to regard all reasons as falling into exclusive categories, the moral and the selfish. This is just to overlook the common fact of mixed motivation. Some reasons are both moral and selfish. Reasons are in different degrees moral and selfish. Arguably this is true of (iv)-(viii). And, to come to my main conclusion here, it seems that an ordinarily reflective person unused to recent moral philosophy would also say just this of what is our main concern, our original reasons (3)-(7). Such a person would say that they partake of both morality and selfishness. They are instances of mixed motivation.  

Given this, what is to be said of reasons (3)-(7)? Let me limit myself to one thing. The dispute between sympathizers with these reasons and sympathizers with the reasons of concern (8)-(14) of the second group is best seen in a certain light. It is not best seen, so to speak, as a dispute or difference internal to morality. It is i fact a kind of moderated continuation of another dispute. This is the ancient dispute between those described as altruists and egoists, or moralists and amoralists. Defenders of reasons (3)-(7) understood in the way we have been considering do in fact stand in a relation, not a distant relation, with traditional egoists or amoralists. 

There is another connection between selfishness and the reasons (3)-(7) understood in the satisfaction-involving way. It has to do with what was noticed earlier, impartiality or consistency.  

Consider (4) again and in particular the mother and her child. She allows, we suppose, that another mother has a like reason. What if they meet, in a situation where there is food for only one child, and each mother demands it for her child? We can contemplate the same situation of conflict with respect to (3) and in particular personal integrity. Each of the preservation of my integrity and the preservation of your integrity depends on the possession of some one thing. So with (5) and (7), say about one's own people and legal rights, and, although imagination is required, with (6) and torture.  

To stick to the mothers, each of them seems to have two choices. Her first choice is to seek to find some new reason for the rightness of her action in demanding the food. Very likely, if she seeks a new reason, she will come to what was earlier called a reason of concern. She will at least approach a morality of concern. Her child, she says, is more hungry or less nourished. Her second choice is to persist in her demand and, crucially, to refuse to allow that the other mother has a reason to favour her child. But then, in her inconsistency, she doesn't have one either. She cannot claim to have a reason for the rightness of her action. 

The moral of the story seems to be that in situations of conflict reasons (3)-(7) are likely to be abandoned in favour of a reason of concern, or the agent must give up attempting to justify her action. If she takes the latter course, her position is one of pure amorality or pure selfishness. We can draw a related moral, of course, without imagining actual conflict. Thinking critically about one's own reasons of the kinds (3)-(7) is enough for this.19  

I end with a comparison. It was said earlier that moralities of concern, despite giving a strict priority to reducing distress, can have in them rules which are counterparts of those involved in reasons (3)-(7), including a rule against torture.20 Thus they can include rules which allow agents in a way to have a regard for themselves and persons to whom they are especially related. Supporters of the Principle of Equality, say, can defend those self-regarding actions of the two kinds which are in accord with their principle. They, and supporters of fairness or justice generally, can certainly go further in this regard than Utilitarians. Is there then less difference than may be supposed between moralities of concern and the ones involved in the first group of reasons we have been considering? 

In a way that is clearly true. But there remains a difference. The moralities of concern are in a certain sense principled. That is, each sets a specific limit to self-regarding actions, a limit fixed by its own fundamental principle. That is something which is not done by the other moralities with which we have been concerned. It is for this reason that the moralities of concern are not open to the judgement of selfishness. 



1. I am grateful to James Cornwell especially, and to Shahrar Ali, Ingrid Coggin, Roger Crisp, Nicholas Dent, Brad Hooker, Joel Kupperman, Jane O'Grady, Hayley Roberts, Michael Slote, Michael Targett, Alan Thomas, and Catherine Wilson, who commented on earlier drafts of this paper. We are not in agreement. For advocacy of or sympathy with reasons of the first group, see Bernard Williams, "A Critique of Utilitarianism", in Williams and J. J. C. Smart, Utilitarianism For and Against (Cambridge: Cambridge University Press, 1973); Williams, Moral Luck (Cambridge: Cambridge University Press, 1981); Amartya Sen and Williams, eds., Utilitarianism and Beyond (Cambridge: Cambridge University Press, 1982), especially the Introduction; Williams, Ethics and the Limits of Philosophy (London: Fontana, 1985); Samuel Scheffler, The Rejection of Consequentialism (Oxford: Oxford University Press, 1982); Scheffler, ed., Consequentialism and Its Critics (Oxford: Clarendon Press, 1988) and in particular the included papers by Thomas Nagel, Robert Nozick, and Philippa Foot; Scheffler, Human Morality (Oxford: Oxford University Press, 1992); John McDowell, "Virtue and Reason," The Monist, 62, 1979; David Wiggins, Needs, Values, Truth (Oxford: Blackwells, 1987). 

2. For advocacy of or sympathy with reasons of the second group, see John Rawls, A Theory of Justice (Cambridge, Mass.: Harvard University Press, 1971); articles by Derek Parfit, Peter Railton, T. M. Scanlon, and Scheffler in Consequentialism and Its Critics; articles by R. M. Hare, John C. Harsanyi, and J. A. Mirrlees in Utilitarianism and Beyond; Shelly Kagan, The Limits of Morality (Oxford: Clarendon, 1989); Honderich, "The Problem of Well-Being and the Principle of Equality," Mind, 1981 or Violence for Equality: Inquiries in Political Philosophy (London: Routledge, 1989), Ch. 2. However, the latter egalitarian morality has not been adequately worked out in personal as distinct from social or political terms. 

3. For example, Kagan, The Limits of Morality, p. xi; Anne Maclean, The Elimination of Morality (London: Routledge, 1993), p. 81; Philip Pettit, "Consequentialism." Companion to Ethics (Oxford: Blackwell, 1991), ed. Peter Singer, p. 231. 

4. See the discussion of the nature of effects and causes in my A Theory of Determinism: The Mind, Neuroscience, and Life-Hopes (Oxford: Clarendon Press, 1988), pp. 14-16, or Mind and Brain (Oxford: Clarendon Press, 1990), pp. 14-16.  

5. This strong point is owed to James Griffin, "Consequentialism", in The Oxford Companion to Philosophy (Oxford: Oxford University Press, 1995), ed. Ted Honderich. I disagree, however, with Griffin's wide use of "consequences", such that they are not confined to what follows an action. 

6. Scheffler, Consequentialism and Its Critics, p. 1. Cf. Sen and Williams, eds., Utilitarianism and Beyond, pp. 11-14. 

7. See for example Roger Crisp, "Deontological Ethics", The Oxford Companion to Philosophy.  

8. It needs to be kept in mind, with respect to (19) and its going into the second group, that we are contemplating a clarified or supplemented version of the wholly consequential conception of the second group, one which does allow that reasons of the group have to do not only with consequences of actions but also antecedents -- a version which escapes the earlier objections, based on (11) and (12), to the unrevised wholly consequential conception.  

9. The situation is complicated, but not much affected, if the attempt to justify punishment by retribution must be regarded, unusually, as resting on the satisfaction of retributive or grievance desires had by victims of offences and others. See my Punishment, The Supposed Justifications (Oxford: Polity Press, 1989).  

10. Williams' "A Critique of Utilitarianism", seminal for the whole subject of consequentialism, includes a good discussion in its second section. For brisk criticisms of the adequacy of other conceptions than the traditional ones, see Crisp, "Deontological Ethics". 

11. Cf. Sen and Williams, Utilitarianism and Beyond, pp. 3-4. 

12. "The Problem of Well-Being and the Principle of Equality". 

13. Sen and Williams speak of such moralities as involving a "drastic obliteration of useful information" (Utilitarianism and Beyond, p. 5). They have in mind other ways of comparing people, those indicated in the first group of reasons. Sympathizers with (8)-(14), of course, insofar as their fundamental principles are concerned, regard this "drastic obliteration" as precisely the great moral strength of their principles. However, the fundamental principles also give a qualified role to other ways of comparing people. See the last two paragraphs of this section and note 14. 

14. As will be understood, I do not agree with Williams (Moral Luck, pp. 51-3) that there are large problems in the way of combining, say, the Principle of Equality, with injunctions related to (3)-(7). On the contrary, the principle has one source in such injunctions. Nor do I agree that supporters of such general principles of justice or fairness, or Utilitarians, are at all committed, by their principled support of moral rules, to support for something else, a moral elite which arranges for ordinary people to abide by the rules unthinkingly. One form of this would be the "Government House utilitarianism" referred to by Sen and Williams, Utilitarianism and Beyond, p. 16. 

15. The matter is more fully discussed in my A Theory of Determinism, sections 7.6, 7.7 or The Consequences of Determinism (Oxford: Clarendon, 1990), sections 1.6, 1.7. For a rapid dismissal of the distinction between "judging the act and judging the agent", see Williams, Moral Luck, p. 53. (Cf. p. 84.) The dismissal rests, I take it, on refusing to distinguish between a good man's dispositions and his deliberations -- in effect, refusing to judge a good man's deliberations as mistaken. No reason is given for this.  

16. A reason for an action in a third sense, used elswhere in this paper but not relevant here, is simply the thing thought of or represented either in a primary or, in a way, a secondary reason -- in short, a property of an action.  

17. I am aware that what is a truism in the philosophy of mind has also been subjected to interesting examination in moral philosophy. That examination is one of several things that must go unexamined in this overview of the two groups of reasons. See Thomas Nagel, The Possibility of Altruism (Oxford: Oxford University Press, 1970); John McDowell, "Are Moral Requirements Hypothetical Imperatives?", Supplementary Proceedings of the Aristotelian Society, 52, 1978; Philippa Foot, "Reasons for Actions and Desires," Supplmententary Proceedings of the Aristotelian Society, 61, 1972. 

18. The suspicion of selfishness is clearly not the suspicion that agents who act on these reasons are "morally self-indulgent," where this is to be taken up with a self-image, engaged in a kind of self-esteem. The latter suspicion, more easily dealt with, is considered by Williams. See "Utilitarianism and Moral Self-Indulgence" in Moral Luck. Cf. Thomas Nagel, Mortal Questions (Cambridge: Cambridge University Press, 1979), p. 63. In Human Morality, Scheffler considers relations between "self-interest" and reasons of the first group, and touches on the matter of the politics and political philosophy associated with them. For the use of some of these reasons in one political tradition, see my Conservatism (Boulder: Westview, 1991). 

19. See, for example, R. M. Hare, The Language of Morals (Oxford: Oxford University Press, 1952), Freedom and Reason (Oxford, 1963), and Moral Thinking (Oxford: Oxford University Press, 1981). 

20. See above, p. 000 ["Before taking a closer look...."]


For  later and further thoughts on this theme, and on the fact and nature of ordinary morality, see After the Terror -- but Ch. 2 in particular.