ONORA O'NEILL CONSISTENCY IN ACTION l. UNIVERSALITY TESTS IN AUTONOMOUS AND IN HETERONOMOUS ETHICS Many recent discussions of universality tests, particularly those in English, are concerned either with what everybody wants done or with what somebody (usually the agent: sometimes an anonymous moral spectator) wants done either by or to everybody. This is true of the universality tests proposed in Singer's Generalization Argument, in Hare's Universal Prescriptivism and generally of various formulations of Golden Rules as well as of Rule Utilitarianism. Since universality tests of these sorts all make moral acceptability in some way contingent upon what is wanted (or, more circumspectly expressed, upon what is preferred or found acceptable or promises the maximal utility) they all form part of moral theories which are heteronomous, in Kant's sense of that term. Such theories construe moral acceptability as contingent upon the natural phenomena of desire and inclination, rather than on any intrinsic or formal features of the agent or his intentions. If we rely on any of these proposed criteria of moral acceptability, there will be no types of act which would not be rendered morally acceptable by some change or changes in human desires. By contrast Kant's proposed universality test, the Categorical Imperative, contains no reference either to what everybody wants done or to what somebody wants done either by or to everybody. Kant's first formulation of the Categorical Imperative, the so-called Formula of Universal Law, runs Act only on that maxim through which you can at the same time will that it should become a universal law. 1 We are invited here to consider that we can will or intend, what it is possible or consistent for us to 'will as a universal law' (not what we would will or would find acceptable or would want as a universal law). Since the principle contains no reference to what everybody or anybody wants, nor to anything which lies beyond the agent's own capacity to will, it is part of a moral theory for agents who, in Kant's sense of the term, act autonomously. The principle asserts that such agents need only to impose a certain sort of consistency on 159 Nelson Potter and Mark Timmons (eds.), Morality and Universality, 159-186. © 1985 by D. Reidel Publishing Company. 160 ONORA O'NEILL their actions if they are to avoid doing what is morally unacceptable. It proposes an uncompromisingly rationalist foundation for ethics. Nevertheless, Kant interpretation, particularly in English, is rich in heteronomous readings of the Formula of Universal Law and in allegations that (despite claims to the contrary) it is impossible to derive non-trivial, actionguiding applications of the Categorical Imperative without introducing heteronomous considerations. 2 Textual objections apart (and they would be overwhelming objections), such heteronomous readings of Kant's ethics discard what is most distinctive and challenging in his ethical theory. These are the features of his theory on which I intend to concentrate. I want to challenge the view that Kantian ethics, and non-heteronomous ethical theories in general, must be seen as either trivially empty or relying covertly on heteronomous considerations in order to derive substantive conclusions. To do so I shall try to articulate what seem to me to be the more important features of a universality test for agents who, in a certain sense of the term, act autonomously, that is without being determined solely by their natural desires and inclinations. 3 I shall take Kant's Formula of Universal Law as the canonical case of such a universality test, and shall argue that it is neither trivially formalistic nor requires supplementing with heteronomous considerations if it is to be action guiding. However, my main concern here is not to explicate Kant's discussion of his universality test, nor to assess the difficulty or adequacy of his various moves. 4 I shall say nothing about his vindication of the Categorical Imperative, nor about his powerful critique of heteronomy in ethics, nor about his conception of human freedom. By setting aside these and other more strictly textual preoccupations I hope to open the way for a discussion of some features of universality tests for autonomous agents which have an interest which goes far beyond a concern with reading Kant accurately. I hope to show that Kant's formula, taken in conjunction with a plausible set of requirements for rational intending, yields strong and interesting ethical conclusions which do not depend on what either everybody or anybody wants, and so that reason can indeed be practical. Over the last twenty years considerable light has been shed on the underlying structure of heteronomous ethical theories (as well as on other, particularly economic and political, decisions) by drawing on studies of the formal aspects of decision making under various conditions which have been articulated in various models of rational choice. In such discussions it is generally taken for granted that rational choosing is in some way or other contingent upon a set of desires or preferences. s I shall suggest that a similar CONSISTENCY IN ACTION 161 concentration on certain requirements of rationality which are not contingent upon desires or preferences can help to provide a clearer picture of the underlying structure and strength of an ethical theory for autonomous beings. The sequence of argument is straightforward. Section 2 provides an explication of Kant's Formula of Universal Law and of some of the ways in which this affects the character of an ethic for autonomous beings. Section 3 discusses some ways in which intentional action can fall into inconsistency even when the question of universalizing is not raised. Sections 4, 5 and 6 show how requirements for rational intending can be conjoined with Kant's universality test to yield determinate ethical conclusions. 2. MAXIMS AND MORAL CATEGORIES The test which Kant's Formula of Universal Law proposes for the moral acceptability of acts can be divided into two aspects. In the first place it enjoins us to act on a maxim; secondly it restricts us to action on those maxims through which we can will at the same time that it should be a universal law. It is only the latter clause which introduces a universality test. However, for an understanding of the nature of this test it is essential in the first place to understand what Kant means by 'acting on a maxim'. For, contrary to appearances, this is not a trivial part of his criterion of morally acceptable action. Because a universality test for autonomous beings does not look at what is wanted, nor at the results of action, but merely demands that certain standards of consistency be observed in autonomously chosen action, it has to work with a conception of action which has the sort of formal structure which can meet (or fail to meet) standards of consistency. It is only those acts which embody or express syntactically structured principles or descriptions which can be thought of as candidates either for consistency or for inconsistency. Mere reflexes or reactions, for example, cannot be thought of as consistent or inconsistent: nor can acts be considered merely instrumentally as means for producing certain outcomes. In requiring action on a maxim Kant is already insisting that whatever is morally assessable should have a certain formal structure. A maxim, Kant tells us, is "a subjective principle of action"; it is "the principle on which the subject acts". 6 A maxim is therefore the principle of action of a particular agent at a particular time; but it need not be 'subjective' in the sense that it seeks to fulfill that particular agent's desires. In speaking of maxims as subjective principles Kant is not adopting any sort of heteronomous standard, but means to propose a standard against which the 162 ONORA O'NEILL principles agents propose to act on, of whatever sort, may be tested. The Categorical Imperative provides a way of testing the moral acceptability of what we autonomously propose to do. It does not aim to generate plans of action for those who have none. While maxims are the principles of action of particular agents at particular times, one and the same principle might be adopted as a maxim by many agents at various times or by a given agent on numerous occasions. It is a corollary of Kant's conception of human freedom that we can adopt or discard maxims, including those maxims which refer to our desires. On the other hand acting on a maxim does not require explicit or con· scious or complete formulation of that maxim. Even routine or thoughtless or indecisive action is action on some maxim. However, not all of the prin· ciples of action that a particular agent might exemplify at a given time would count as the agent's maxim. For principles of action need only incorporate some true description of an agent and some true description of his act and situation, whether these descriptions are vacuous and vague or brimming with detail. But an agent's maxim in a given act must incorporate just those descriptions of the agent, the act and the situation upon which the doing of the act depends. An agent's maxim in a given act cannot, then, be equated simply with his intentions. For an agent's intentions in doing a given act may refer to incidental aspects of the particular act and situation. For example, in making a new visitor feel welcome I may offer and make him or her some coffee. In doing so there will be innumerable aspects of my action which are intentional - the choice of mug, the addition of milk, the stirring; and there will also be numerous aspects of action which are 'below the level of inten· tion' - the gesture with which I hand the cup, the precise number of stirs and so on. But the various specific intentions with which I orchestrate the offer and preparation of coffee are all ancillary to an underlying intention. Maxims are those underlying intentions by which we guide and control our more specific intentions. In this particular example, had I lacked coffee I could have made my visitor welcome in other ways: the specific intention of offering and making coffee was subordinate to the maxim of making a visitor welcome. Had I had a quite different maxim - perhaps to make my visitor unwelcome - I would not in that context have acted on just those specific intentions. In another context, e.g. in a society where an offer of coffee would be understood as we would understand an offer of hemlock, the same or similar specific intentions might have implemented a maxim of making unwelcome. CONSISTENCY IN ACTION 163 The fact that maxims are underlying or fundamental intentions has important implications. 7 It means in the first place that it may not be easy to tell on which maxim a given act was done. For example, a person who helps somebody else in a public place may have the underlying intention of being helpful or alternatively the underlying intention of fostering a certain sort of good reputation. Since the helpful act might equally well be done in futherance of either underlying intention, there may be some doubt as to the agent's maxim. Merely asking an agent what his underlying maxim is in such a situation may not settle the issue. The agent might himself be unsure. Both he and others can work out that if he would have done the action even if nobody had come to know of it, then his underlying intention would not have been to seek a certain sort of reputation. But he may after all be genuinely uncertain what his act would have been had he been faced with the possibility of helping, isolated from any effects on his reputation. Isolation tests can settle such issues 8 if we know their outcome; but since most such tests refer to counterfactual situations we often don't know that outcome with any great certainty. Further, isolation tests provide only a negative test of what an agent's maxim is not. Even the person who has used such a test to show that his maxim is not to acquire a reputation may still be unsure whether his maxim was just to be helpful. He may perhaps wonder whether his underlying intention was not to preserve a certain sort of self image or to bolster his own sense of worth. Kant remarks on the opacity of the human heart and the difficulty of self-knowledge; he laments that for all we know there never has been a truly loyal friend. 9 And he does not view these as dispellable difficulties. Rather these limits to human self-knowledge constitute the fundamental context of human action. Kant holds that we can know what it would be to try to act on a maxim of a certain sort, but can never be sure that what we do does not reflect further maxims which we disavow. However, the underlying intentions which guide our more specific intentions are not in principle undiscoverable. Even when not consciously formulated they can often be inferred with some assurance, if not certainty, as the principles and policies which our more specific intentions express and implement. On a certain view of the purpose of a universality test the fact that the maxim of a given action is neither observable nor always reliably inferable would be a most serious objection. For it would appear to render the outcome of any application of a universality test of dubious moral importance since we might mistakenly have applied the test to a principle other than the agent's maxim. Further, even if the maxim had been correctly formulated, whether 164 ONORA O'NEILL by the agent or by others, the maxim itself might reflect mistaken beliefs or self-deception in the agent, or the agent's act might fail to live up to his maxim. How then could any test applied to the agent's maxim be expected to classify acts into moral categories such as the right and the forbidden? For these categories apply to the outward and observable aspects of action. It is after all common enough for us to think of acts which are at least outwardly right (perhaps even obligatory) as nevertheless reflecting dubious intentions (I aim to kill an innocent, but mistakenly incapacitate the tiger who is about to maul him), and of acts whose intentions are impeccable as issuing tragically in wrong action (I aim for the tiger but despatch the innocent). The answer Kant gives to this problem is plain. It is that rightness and wrongness and the other 'categories of right' standardly used in appraisal of outward features of action are not the fundamental forms of moral acceptability and unacceptability which he takes the Categorical Imperative to be able to discriminate. 1o Since the locus of application of Kant's universality test (and perhaps of any non-heteronomous universality test) is agents' fundamental intentions, the moral disinction which it can draw is in the first place an intentional moral distinction, namely that between acts which have and those which lack moral worth. In an application of the Categorical Imperative to an agent's maxim we ask whether the underlying intention with which the agent acts or proposes to act - the intention which guides and controls his other more specific intentions - is consistently universalizable; if it is, according to Kant, we at least know that the action will not be morally unworthy, and will not be a violation of duty. The fact that Kant is primarily concerned with judgments of moral worth is easily forgotten, perhaps because he speaks of the Categorical Imperative as a test of duty, while we often tend to think of duty as confined to the outward aspects of action. It is quite usual for us to think of principled action as combining both duty and moral worthiness, which we regard as separate matters (e.g. showing scrupulous respect for others) or alternatively as revealing a moral worthiness which goes beyond all duty (e.g. gratuitous kindness which we think of as supererogatory). Correspondingly it is quite usual for us to think of unprincipled action as in any case morally unworthy but still, in some cases, within the bounds of duty (e.g. the case of a would-be poisoner who mistakenly administers a life-saving drug). This is quite foreign to Kant's way of thinking, which sees the central case of duty as that of action which has moral worth, and regards as derivative that which accords merely in external respects with morally worthy action. On Kant's view the would-be CONSISTENCY IN ACTION 165 poisoner who inadvertently saves life has violated a duty by acting in a morally unworthy way. By taking an agent's fundamental or underlying intention as the point of application of his universality test Kant avoids one of the difficulties most frequently raised about universality tests, namely that it seems easy enough to formulate some principle of action for any act, indeed possibly one that incorporates one of the agent's intentions, which can meet the criterion of any universality test, whatever the act. Notoriously some Nazi war criminals claimed that they were only 'doing their job' or only 'obeying orders' which are after all not apparently morally unworthy activities. The disingenuousness of the claim that such acts were not morally unworthy lies in the fact that these Nazis were not only doing this, and indeed that in many cases their specific intentions were ancillary to more fundamental intentions which might indeed have revealed moral unworthiness in the agent. (Such fundamental intentions might range from 'I'll do whatever I'm told to so long as it doesn't endanger me' to a fundamental maxim of genocide.) The fact that we can formulate some universalizable surface intention for any action by selecting among the agent's various surface intentions is no embarrassment to a universality test which is intended to apply to agents' maxims, and offers a solution to the problem of relevant descriptions. It is equally irrelevant to a universality test that applies to maxims that we may be able to find some non-universalizable intentions among the more specific intentions with which an agent implements and fills out any maxim. If in welcoming my visitor with a cup of coffee I intentionally select a particular cup, my specific intention clearly cannot be universally acted on. The very particularity of the world means that there will always be aspects of action, including intentional aspects, which could not be universally adopted or intended. Kant's universality test, however, as we shall see, construes moral worth as contingent not on the universalizability or otherwise of an agent's specific intentions but on the universalizability of an agent's fundamental or underlying intention. 11 For Kant, then, the Categorical Imperative provides a criterion in the first place for duties to act on underlying intentions which are morally worthy. It is only as a second and derivative part of his ethical theory that he proposes that the Categorical Imperative also provides a test of the outward wrongness and rightness of acts of specific sorts. He proposes in the Groundwork that acts which accord in outward respects with acts done on morally worthy maxims of action should be seen as being 'in conformity with' or 'in accord with' duty. The claim that we can provide a general account of which specific 166 ONORA O'NEILL actions conform to the outward expressions of morally worthy maxims is highly controversial. We have already noted that there are many ways in which ancillary intentions may be devised in undertaking action on a given maxim, and there may be no single specific intention which is indispensable in all circumstances for action on a given maxim. Hence it is not generally clear what outward conformity conforms to. Kant appears to accept that the notion of outward conformity to duty is empty in many cases of duties of virtue, which are not sufficiently determinate for any particular more specific intentions to. be singled out as required. He speaks of such duties as being 'of wide requirement'. But he also speaks of duties of narrow or strict requirement, and includes among these duties of justice and certain duties of respect to ourselves and to others.12 Hence he takes it that there could in principle be a merely outward conformity to these strict or 'perfect' duties. Whether this claim is justified depends on the success of his demonstration that the underlying maxims of justice and respect have determinate specific implications for all possible human conditions. If they do not, then there will be no wholly general account of the requirements of justice and respect for all possible situations. It is then at any rate not obvious that we can derive a standard for the outward rightness of acts from a standard for the moral worth of underlying intentions. This is a major problem which I intend to set on one side in order to explore the implications of a universality test which applies to underlying intentions and so aims, at least primarily, at a test of the moral worth rather than the outward rightness of actions. The fact that Kant's universality test focuses on maxims, and so on the moral worth of action, implies that it is a test which agents must seek to apply to their own proposals for action. This is not, however, because an agent is in a wholly privileged epistemological position with respect to his own underlying intentions. No doubt others may often have some difficulty even in discerning all of an agent's surface intentions, and may be quite unsure about his underlying intention. But Kant does not regard the agent's vantage point as affording infallible insight into his own intentions - selfconsciousness is not transparent - and would not deny that on occasion others might arrive at a more accurate appreciation of an agent's underlying intention than the agent could himself reach. The reason why a universality test in an autonomous ethical theory is primarily one for the use of agents rather than of moral spectators is rather that it is only an agent who can adopt, modify or discard maxims. While a test of the outward moral status of acts might be of most use and importance to third parties (legislators, judges, educators - those of us who pass judgment CONSISTENCY IN ACTION 167 on others), because it may be possible or indeed necessary to prevent or deter or praise or punish in order to elicit or foster outward action of a certain sort, it is difficult if not impossible for outward regulation or pressure to change an agent's underlying intention. Surface conformity can be exacted; intentional conformity is more elusive. 13 Precisely because we are considering what a universality test for autonomous beings must be like, we must recognize that the test is one which we can propose to but not impose upon moral agents. 3. INCONSISTENCY WITHOUT UNIVERSALIZING This account of acting on a maxim shows at least how action can be construed in a way which makes consistency and inconsistency possible, and provides some grounds for thinking that a focus on maxims may avoid some of the difficulties which have arisen in attempts to apply universality tests unrestrictedly to principles of action of all sorts. This opens the way for showing how action on a non-universalizable maxim is inconsistent and for considering whether such inconsistency constitutes a criterion of moral unworthiness. Before dealing with these topics it will be useful to run over some of the many ways in which action on a maxim may reveal inconsistency even when universalizing is not brought into the picture. It is, of course, true that any act which is done is possible, taken in itself. But it does not follow that the intentions which are enacted are mutually consistent. There are two sorts of possibilities here: in the first place there may be an internal inconsistency within an agent's maxim; in the second place there may be contradictions between the various specific intentions an agent adopts in pursuit of his maxim, or between some of these intentions and the agent's maxim. These two sorts of contradiction correspond closely to the two types of contradiction which Kant thinks may arise when attempts to universalize maxims fail, and which he characterizes as involving respectively 'contradictions in conception' and 'contradictions in the will'. 14 Since I am also interested in charting the inconsistencies which can arise independently of attempts to universalize, as well as in those which arise when universalizing fails, I shall use the rather similar labels 'conceptual inconsistency' and 'volitional inconsistency' to distinguish these two types of incoherence in action. A consideration of the different types of incoherence which maxims may display even when the question of universalizability is not raised provides a useful guide to the types of incoherence which nonuniversalizable maxims display. 168 ONORA O'NEILL A maxim of action may in the first place be incoherent simply because it expresses an impossible aspiration. An agent's maxim might be said to involve a conceptual inconsistency if his underlying intention was, for example, both to be successful and to be unworldly, or alternatively to be both popular and reclusive, or both to care for others and always to put his own advantage first, or both to be open and frank with everybody and to be a loyal friend or associate, or both to keep his distance from others and to have intimate personal relationships. Agents whose underlying maxims incorporate such conceptual inconsistencies do not, of course, succeed in doing impossible acts; rather then pattern of their action appears to pull in opposite directions and to be in various ways self-defeating. At its extreme we may regard such underlying incoherence in a person's maxim, and consequent fragmentation of his action, as tragic or pathological (and perhaps both), since there is no way in which he or she can successfully enact the underlying intention. In other cases we may think of the pattern of action which results from underlying conceptual incoherence as showing no more than ambivalence or presenting conflicting signals to others, who are consequently at a loss as to what they should expect or do, finding themselves in a 'double bind'. However, not all cases of disjointed action constitute evidence of an internally inconsistent maxim. For it may well be that somebody adopts some accommodation of the potentially inconsistent aspects of his underlying intention. For example, somebody may adopt the maxim of being competitive and successful in public and professional life but disregarding such considerations in private life; or of being obedient and deferential to superiors but overbearing and exacting with all others. Provided such persons can keep the two spheres of action separated, their underlying intention can be internally consistent. Hence one cannot infer an inconsistency in someone's underlying intentions merely from the fact that he or she exhibits tendencies in opposing directions. For these tendencies may reflect a coherent underlying intention to respond or act differently in different types of context or with different groups of people. A non-universalized maxim embodies a conceptual contradiction only if it aims at achieving mutually incompatible objectives and so cannot under any circumstances be acted on with success. A focus on maxims which embody contradictions in conception pays no attention to the fact that maxims are not merely principles which we can conceive (or entertain, or even wish) but principles which we will or intend, that is to say principles which we adopt as principles of action. Conceptual contradictions can be identified even in principles of action which are never adopted or acted upon. But a second and rather different type of incoherence CONSISTENCY IN ACTION 169 is exhibited in some attempts to will maxims whose realization can be quite coherently envisaged. Willing, after all, is not just a matter of wishing that something were the case, but involves committing oneself to doing something to bring that situation about when opportunity is there and recognized. Kant expressed this point by insisting that rationality requires that whoever wills some end wills the necessary means in so far as these are available. Who wills the end wills (so far as reason has decisive influence on his actions) also the means which are indispensably necessary and in his power. So far as willing is concerned, this proposition is analytic: for in my willing of an object as an effect there is already conceived the causality of myself as a working cause - that is, the use of means; and from the concept of willing an end" the imperative merely extracts the concept of actions necessary to this end. 15 This amounts to saying that to will some end without willing whatever means are indispensable for that end, insofar as they are available, is, even when the end itself involves no conceptual inconsistency, to involve oneself in a volitional inconsistency. It is to embrace at least one specific intention which, far from being guided by the underlying intention, is inconsistent with that intention. Kant, however, explicitly formulates only one of the principles which must be observed by an agent who is not to fall into volitional inconsistency. The 'Principle of Hypothetical Imperatives', as expressed in the passage just quoted, requires that agents intend any indispensable means for whatever they fundamentally intend. Conformity with this requirement of coherent intending would be quite compatible with intending no means to whatever is fundamentally intended whenever there is no specific act which is indispensable for action on the underlying intention. Further reflection on the idea of intending the means suggests that there is a family of Principles of Rational Intending, of which the Principle of Hypothetical Imperatives is just one, though perhaps the most important one. The following list of further Principles of Rational Intending which coherent intending (as opposed to mere wishing or contemplating) apparently requires agents to observe may not be complete, but is sufficient to generate a variety of interesting conclusions. First, it is a requirement of rationality to intend not merely all indispensable or necessary means to that which is fundamentally intended but also to intend some sufficient means to what is fundamentally intended. If it were not, I could coherently intend to eat an adequate diet, yet not intend to eat food of any specific sort on the grounds that no specific sort of food is indispensable in an adequate diet. 170 ONORA O'NEILL Secondly, it is a requirement of rationality not merely to intend all necessary and some sufficient means to what is fundamentally intended, but to seek to make such means available when they are not. If it were not, I could coherently claim to intend to help bring about a social revolution but do absolutely nothing, on the grounds that there is no revolutionary situation at present, so settling for rhetoric and gesture rather than politics. But if I do this, I at most wish for, and do not intend to help to bring about, a social revolution. it is a requirement of rationality to intend not merely all necesary and some sufficient means to whatever is fundamentally intended but to intend all necessary and some sufficient components of whatever is fundamentally intended. If it were not, I could coherently claim to intend to be kind to someone to whom, despite opportunity, I show no kindness in word, gesture or deed, merely because acting kindly is not the sort of thing which requires us to take means to an end but the sort of thing which requires that we act in some of the ways that are of kindness. 16 Fourthly, it is a requirement of rationality that the various specific intentions we actually adopt in acting on a given maxim in a ce.rtain context be mutually consistent. If it were not, I could coherently claim to be generous to all my friends by giving to each the exclusive use of all my possessions. Fifthly, it is a requirement of rationality that the foreseeable results of the specific intentions adopted in acting on a given underlying intention be consistent with the underlying intention. If it were not, I could coherently claim to be concerned for the well-being ofa child for whom I refuse an evidently life-saving operation, on the grounds that my specific intention perhaps to shield the child from the hurt and trauma of the operation - is itself aimed at the child's well being. But where such shielding foreseeably has the further consequence of endangering the child's life, it is clearly an intention which undercuts the very maxim which supposedly guides it. There may well be yet further principles which fully coherent sets of intentions must observe, and possibly some of the principles listed above need elaboration or qualification. The point, however, is to reveal that once we see action as issuing from a complex web of intentions, many of which are guided by and ancillary to certain more fundamental intentions under particular conditions, intending coherently and avoiding volitional inconsistency becomes a demanding and complex affair. Reflection on the various Principles of Rational Intending reveals a great deal about the connections between surface and underlying intentions to which a rational being must aspire. Underlying intentions to a considerable CONSISTENCY IN ACTION 171 extent express the larger and longer term goals, policies and aspirations of a life. But if these goals, policies and aspirations are willed (and not merely wished for) they must be connected with some set of surface intentions which express commitment to acts that, in the actual context in which the agent finds himself, provide either the means to or some components of the underlying intention, or at least take the agent in the direction of being able to form such intentions, without at any point committing the agent to acts whose performance would undercut his underlying intention. Wherever such coherence is absent we find an example of intending which, despite the conceptual coherence of the agent's maxim, is volitionally incoherent. In some cases we may think the deficiency cognitive - the agent fails despite available information to appreciate what he needs to do if he is indeed to act on his maxim (he may be stupid or thoughtless or calculate poorly). In other cases we might think of the deficiency as primarily volitional: the agent fails to intend what is needed if he is to will his maxim and not merely to wish for it to be realized. Each of these types of failure in rationality subdivides into many different sorts of cases. It follows that there are very many different ways in which an agent whose intentions are not to be volitionally inconsistent may have to consider his intentions. Perhaps the most difficult of the various requirements of coherent willing is the last, the demand that the agent not adopt specific intentions which in a given context may undercut his own maxim. There are many cases in which agents can reach relatively clear specific intentions about how they will implement or instance their maxim, yet the act selected, though indeed selected as a means to or component of their underlying intention, backfires. It is fairly common for agents to adopt surface intentions which, when enacted, foreseeably will produce results which defeat the agent's own deeper intentions. Defensive measures generate counterattack; attempts to do something particularly well result in botched performances; decisive success in battle is revealed as Pyrrhic victory. It is perhaps unclear how long a view of the likely results of his own action an agent must take for us not to think action which leads to results which are not compatible with his underlying intention is irrational. But at the least the standard and foreseeable results of his action should not undercut the underlying intention if we are to think of an agent as acting rationally. Somebody who claims to intend no harm to others, and specifically merely intends to share a friendly evening's drinking and to drive others home afterwards, but who then decides on serious drinking and so cannot safely drive, cannot plausibly claim to intend merely the exuberant drinking and bonhomie and not the foreseeable 172 ONORA O'NEILL drunkenness and inability to drive safely. Given standard information, such a set of intentions is volitionally incoherent. For it is a normal and foreseeable result of exuberant drinking that the drinker is incapable of driving safely. One who intends the drinking also (given normal intelligence and experience) intends the drunkenness; hence cannot coherently also intend to drive others home if his underlying intention is to harm nobody. I? This brief consideration of various ways in which agents' intentions may fail to be consistent shows that achieving consistency in action is a difficult matter even if we do not introduce any universality test. Intentions may be either conceptually or volitionally incoherent. The demand that the acts we do reflect conceptually and volitionally coherent sets of intentions therefore constitutes a powerful constraint on all practical reasoning. This provides some reason for thinking that when these demands for consistency are extended in the way in which the second aspect of Kant's Formula of Universal Law requires we should expect to see patterns of reasoning which, far from being ineffective or trivial, generate powerful and interesting results. 4. INCONSISTENCY IN UNIVERSALIZING The intuitive idea behind the thought that a universality test can provide a criterion of moral acceptability may be expressed quite simply as the thought that if we are to act as morally worthy beings we should not single ourselves out for special consideration or treatment. Hence whatever we propose for ourselves should be possible (note: not desired or wanted - but at least possible) for all others. Kant expresses this commonplace thought (it is, of course, not his argument for the Categorical Imperative) by suggesting that what goes wrong when we adopt a non-universalizable maxim is that we treat ourselves as special: ... whenever we transgress a duty, we find that we in fact do not will that our maxim should become a universal law - since this is impossible for us - but rather that its opposite should remain a law universally: we only take the liberty of making an exception to it for ourselves (or even just for this once) ... 18 It is evident from this understanding of the Formula of Universal Law that the notion of a community of other autonomous agents is already implicit in the Formula of Universal Law. It is not the case that Kant only introduces that notion into his ethics with the Formula of the Kingdom of Ends, which would imply that the various formulations of the Categorical Imperative could not be in any way equivalent. To universalize is from the start to CONSISTENCY IN ACTION 173 consider whether what one proposes for oneself could be done by others. This seems to many too meagre a foundation for ethics but not in itself an implausible constraint on any adequate ethical theory. Clearly enough whatever cannot be consistently intended even for oneself also cannot be consistently intended for all others. The types of cases shown conceptually or volitionally inconsistent by the methods discussed in the previous section are a fortiori non-universalizable. This raises the interesting question of whether one should think of certain types of cognitive and volitional failure as themselves morally unworthy. However, I shall leave this question aside in order to focus on the types of failure in consistent intending which are peculiar to the adoption of non-universalizable intentions. I shall therefore assume from now on that we are considering cases of maxims which are in themselves not conceptually incoherent, and of sets of underlying and surface intentions which are not themselves volitionally inconsistent. The task is to pinpoint the ways in which inconsistency emerges in some attempts to universalize such internally consistent intentions. The second part of Kant's Formula of Universal Law enjoins action only on maxims which the agent can at the same time will as a universal law. He suggests that we can imagine this hypothetical willing by working out what it would be like "if the maxim of your action were to become through your will a universal law of nature". 19 To universalize his maxim an agent must satisfy himself that he can both adopt the maxim and simultaneously will that others do so. In determining whether he can do so he may find that he is defeated by either of the two types of contradiction which, as we have already seen, can afflict action even when universalizing is not under consideration. Kant's own account of these two types of incoherence, either of which defeats universalizability, is as follows: We must be able to will that a maxim of our action should become a universal law - that is the general canon for all moral judgement of action. Some actions are so constituted that their maxim cannot even be conceived as a universal law of nature without contradiction, let alone be willed as what ought to become one. In the case of others we do not find this inner impossibility, but it is still impossible to will that their maxim should be raised to the universality of a law of nature, because such a will would contradict itseif.2o Kant also asserts that those maxims which when universalized lead to conceptual contradiction are the ones which strict or perfect duty requires us to avoid, while those which when universalized are conceptually coherent but not coherently willable are opposed only to wider or imperfect duties. 21 Since we probably lack both rigorous criteria and firm intuitions of the 174 ONORA O'NEILL boundaries between perfect and imperfect duties it is hard to evaluate this claim. However it is remarkably easy to display contradictions which arise in attempts to universalize maxims which we might think of as clear cases of violations of duties of justice and self-respect which Kant groups together as perfect duties; and it is also easy to show how contradictions emerge in attempts to universalize maxims which appear to exemplify clear violations of duties of beneficence and self development, which Kant groups together as imperfect duties. By running through a largish number of such examples I hope to show how groundless is the belief that universality tests need supplementing with heteronomous considerations if they are to be action guiding. 5. CONTRADICTIONS IN CONCEPTION Maxims which may lead to contradictions in conception when we attempt to universalize them often do not contain any conceptual contradiction if we merely adopt the maxim. For example, there is no contradiction involved in adopting the maxim of becoming a slave. But this maxim has as its universalized counterpart the maxim we must attempt to 'will as a universal law' the maxim of everybody becoming a slave. 22 But if everybody became a slave there would be nobody with property rights, hence no slave holders, hence nobody could become a slave. 23 Consider alternatively a maxim of becoming a slave holder. Its universalized counterpart would be the maxim that everybody become a slave holder. But if everybody became a slave holder then everybody would have some property rights, hence nobody could be a slave, hence there could be no slave holders. Action on either of the non-universalizable maxims of becoming a slave or becoming a slaveholder would reveal moral unworthiness: it could be undertaken only by one who makes of himself a special case. Contradictions in conception can also be shown to arise in attempts to universalize maxims of deception and coercion. The maxim of coercing whoever will not comply with my will has as its universalized counterpart the maxim that everybody will coerce others when they do not comply with his will: but this requires that each party coerce others, including those who are coercing him, hence that each party both complies with others' will (being coerced) and simultaneously does not comply with others but rather (as coercer) exacts their compliance. A maxim of coercion cannot coherently be universalized and reveals moral unworthiness. By contrast, a maxim of autonomous coordination can be consistently universalized. A maxim of deceiving others as convenient has as its universalized counterpart the maxim CONSISTENCY IN ACTION 175 that everyone will deceive others as convenient. But if everyone were to deceive others as convenient then there would be no such thing as trust or reliance on others' acts of communication, hence nobody could be deceived, hence nobody could deceive others as convenient. An argument of the same type can be applied to the maxim which is perhaps the most fundamental for a universality test for autonomous action, namely the maxim of abrogating autonomy. One whose maxim it is to defer to the judgment and decisions of others - to choose heteronomy24 - adopts a maxim whose universalized counterpart is that everyone defer to the judgments and decisions of others. But if everyone defers to the judgments and decisions of others, then there are no autonomous decisions to provide the starting point for deferring in judgment, hence it cannot be the case that everybody defers in judgment. Decisions can never be reached when everyone merely affirms 'I agree'. A maxim of 'elective heteronomy' cannot consistently be universalized. Interpreters of Kant have traditionally made heavier weather of the contradiction in conception test than these short arguments suggest is necessary. There have perhaps been two reasons for this. One is clearly that Kant's own examples of applications of the Categorical Imperative are more complex and convoluted than these short arguments suggest. 25 But while detailed analysis of these examples is necessary for an evaluation of Kant's theory, it is clarifying to see whether a contradiction in conception test works when liberated from the need to accommodate Kant's particular discussion of examples. But a second reason why the contradiction in conception test has seemed problematic to many of Kant's commentators is perhaps of greater importance for present concerns. It is that while many would grant that we can detect contradictions in attempts to universalize maxims simply of slaveholding or coercing or deceiving or deference, they would point out that no contradiction emerges if we seek to universalize more circumspect maxims such as 'I will hold slaves if I am in a position of sufficient power' or 'I will deceive when it suits me and 1 can probably get away with it' or 'I will defer in judgment to those 1 either admire or fear.' Still less do contradictions emerge when we aim to universalize highly specific intentions of deception or deference such as '1 will steal from Woolworths when 1 can get away with it' or 'I will do whatever my Parish Priest tells me to do.' However the force of this objection to the claim that the contradiction in conception test can have significant moral implications is undercut when we remember that this is a test which applies to agents' maxims, that is to their underlying or fundamental intentions and that as a corollary it is a test 176 ONORA O'NEILL of moral worth. For what will be decisive is what an agent's fundamental intention in doing a given act really is. What counts is whether the expression of falsehood expresses a fundamental attempt to deceive, or whether agreement with another (in itself innocent enough) expresses a fundamental refusal to judge or think for oneself. For an agent cannot truthfully claim that his underlying intent was of a very specific sort unless the organization of his other, less fundamental, intentions reveals that his intention really was subject to those restrictions. Precisely because the Categorical Imperative formulates a universality test which applies to maxims, and not just to any intention, it is not rebutted by the fact that relatively specific intentions often can be universalized without conceptual contradiction. Conversely, further evidence for the interpretation of the notion of a maxim presented in Section 2 is that it leads to an account of the Categorical Imperative which is neither powerless nor counterintuitive. However, for the same reason (that it applies to maxims and not to intentions of all sorts) the Categorical Imperative can most plausibly be construed as a test of moral worth rather than of outward rightness, and must always be applied with awareness that we lack certainty about what an agent's maxim is in a given case. This is a relatively slight difficulty when we are assessing our own proposed maxims of action, since we at least can do no better than to probe and test the maxim on which we propose to act (but even here we have no guarantee against self-deception). And it means that we will always remain to some extent unsure about our assessment of others' acts. Kant after all insists that we do not even know whether there ever has been a truly morally worthy act. But that is something we do not need to know in order to try to do such acts. Self-deception may cloud our knowledge of our own maxims; but we are not powerless in selfguidance. 6. CONTRADICTIONS IN THE WILL Just as there are maxims which display no conceptual incoherence until attempts are made to universalize them, so there are maxims which exhibit no conceptual incoherence even when universalized, but which are shown to be volitionally inconsistent when attempts are made to universalize them. Such maxims cannot be 'willed as a universal law'; attempts to do so fail in one way or another to meet the standards of rationality specified by the group of principles which I have termed Principles of Rational Intending. For to will a maxim is, after all, not just to conceive the realiZation of an underlying intention; that requires no more than speculation or wishing. Willing CONSISTENCY IN ACTION 177 requires also the adoption of more specific intentions which are guided by and chosen (in the light of the agent's beliefs) to realize the underlying intention, or, if that is impossible, as appropriate moves towards a situation in which such specific intentions might be adopted. Whoever wills a maxim also adopts more specific intentions as means to or constituents of realizing his underlying intention, and is also committed to the foreseeable results of acting on these more specific intentions. Since intending a maxim commits the agent to such a variety of other intentions there are various different patterns of argument which reveal that certain maxims cannot be willed as a universal law without contradiction. Clearly the most comprehensive way in which a maxim may fail to be willable as a universal law is if its universal counterpart is inconsistent with the specific intentions which would be necessary for its own realization. Universalizing such a maxim would violate the Principle of Hypothetical Imperatives. The point is well illustrated by a Kantian example. 26 If I seek to will a maxim of non-beneficence as a universal law, my underlying intention is to help others when they need it and its universalized counterpart is that nobody help no others when they need it. But if everybody denies help to others when they need it, then those who need help will not be helped, and in particular I will not myself be helped when I need it. But if I am committed to the standards of rational willing which comprise the various Principles of Rational Intending, then I am committed to willing some means to any end to which I am committed, and this must include willing that if I am in need of help and therefore not able to achieve my ends without help I be given some appropriate help. In trying to universalize a maxim of non-beneficence I find myself committed simultaneously to willing that I not be helped when I need it and that I be helped when I need it. This contradiction, however, differs from the conceptual contradictions which emerge in attempts to universalize maxims such as those considered in the last section. A world of non-benevolent persons is conceivable without contradiction. Arguments which reveal contradictions in the will depend crucially upon the role of the various Principles of Rational Intending - in this case on the Principle of Hypothetical Imperatives - in constraining the choice of specific intentions to a set which will implement all underlying intentions. It is only because intending a maxim of non-benevolence as a universal law requires commitment to that very absence of help when needed, to which all rational intending requires assent, that non-benevolence cannot coherently be universalized. A second Kantian example,27 which provides an argument to volitional 178 ONORA O'NEILL incoherence, is a maxim of neglecting to develop any talents. A world of beings who develop no talents contains no conceptual incoherence. The maxim of an individual who decides to develop no talents, while imprudent, reveals no volitional inconsistency. For it is always possible that he finds that others fend for him and so that there will be means available for at least some autonomous action on his part. (It is not a fundamental requirement of practical reason that there should be means available to any project he adopts, but only that he should not have ruled out all autonomous action.) However, an attempt to universalize a maxim of neglecting talents commits one to a world in which no talents have been developed, and so to a situation in which necessary means are lacking not just for some but for any sort of autonomous action. Any autonomous agent who fails to will the development, in himself or others, of whatever minimal range of talents is required and sufficient for some autonomous action is committed to internally inconsistent sets of intentions. He intends both that autonomous action be possible and that it be undercut by neglect to develop even a minimal range of talents which would leave some possibility of autonomous action. This argument shows nothing about the development of talents which may be required or sufficient for any specific projects, but only points to the inconsistency of failing to foster such talents as are needed and sufficient for autonomous action of some sort or other. It is an argument which invokes not only the Principle of Hypothetical Imperatives but the requirement that rational beings intend some set of means sufficient for the realization of their underlying intention. These two examples of arguments which reveal volitional inconsistencies show only that it is morally unworthy to adopt maxims either of systematic non-benevolence or of systematic neglect of talents. The duties which they ground are relatively indeterminate duties of virtue. The first of these arguments does not specify whom it is morally worthy to help, to what extent, in what ways or at what cost, but only that it would be morally unworthy to adopt an underlying intention of non-benevolence. Similarly the second argument does not establish which talents it would be morally worthy to develop, in whom, to what extent or at what cost, but only that it would be morally unworthy to adopt an underlying intention of making no effort to develop any talents. The person who adopts a maxim either of non-benevolence or of non-development of talents cannot coherently universalize his maxim. He or she must either make an exception of himself, and intend, unworthily, to be a free-rider on others' benevolence and talents, or will be committed to some specific intentions which are inconsistent with those required for action on his own maxim. CONSISTENCY IN ACTION 179 Another example of a maxim which cannot consistently be willed as a universal law is the maxim of refusing to accept help when it is needed. The universalized counterpart of this underlying intention would be the intention that everyone refuse to accept help when it is need. But autonomous beings cannot consistently commit themselves to intending that everyone forgo a means which, if ever they are in need of help, will be indispensable for them to act autonomously at all. A further example of a non-universalizable maxim is provided by a maxim of ingratitude, whose universalized counterpart is that nobody show or express gratitude for favours received. In a world of autonomous but non-self-sufficient beings a universal maxim of ingratitude would require the systematic neglect of an important means for ensuring that help is forthcoming for those who need help if they are to realize their intentions. Hence in such a world nobody could coherently claim to will that those in need of help be helped. Yet we have already seen that failure to will that those in need of help be helped is volition ally inconsistent. Hence, willing a maxim of ingratitude also involves a commitment to a set of intentions not all of which can be consistently universalized. The volitional inconsistency which overtakes would-be universalizers of this maxim arises in two stages: the trouble with ingratitude is that, practised universally, it undercuts benevolence; the trouble with non-benevolence is that it cannot be universally practised by beings who (being autonomous) have at least some underlying maxims, yet (lacking self-sufficiency) cannot guarantee that their own resources will provide means sufficient for at least some autonomous project. The hinge of all these arguments is that human beings (since they are autonomous adopters of maxims) have at least some maxims or projects, which (since they are not self-sufficient) they cannot always realize unaided, and so must (since they are rational) intend to draw on the assistance of others, and so must (if they universalize) intend to develop and foster a world which will lend to all the support of others' benevolence and talents. Such arguments can reveal the volitional inconsistencies involved in trying to universalize maxims of entirely neglecting the social virtues - benevolence, beneficence, solidarity, gratitude, sociability and the like - for beings who are autonomous yet not always able to achieve what they intend unaided. It follows from this that the social virtues are very differently construed in autonomous and in heteronomous ethics. An ethical theory for autonomous agents sees the social virtues as morally required, not because they are desired or liked but because they are necessary requirements for autonomous action in a being who is not self-sufficient. The content of the social virtues in an 180 ONORA O'NEILL autonomous framework cannot be spelt out in terms of the provision of determinate goods or service or the meeting of certain set needs or the satisfaction of a determinate set of desires. Rather the content of these virtues will always depend on the various underlying maxims and projects, both individual and collaborative, to which autonomous agents commit themselves. What will constitute beneficence or kindness or care for others will depend in great part on how others intend to express their autonomy, and to collaborate in exercising their autonomy, in a given context. 7. CONTRADICTIONS IN THE WILL AND FURTHER RESULTS The patterns of argument which can be used to show underlying anti-social intentions morally worthy make use of various Principles of Rational Intending in addition to the Principle of Hypothetical Imperatives. In particular they draw on the requirements that rational agents intend not merely necessary but sufficient means to or components of their underlying intentions, and that they also intend whatever means are indirectly required and sufficient to make possible the adoption of such specific intentions. However, the particular features of the fifth Principle of Rational Intending - the Principle of Intending the Further Results - have not yet been displayed. Attempts to evade this Principle of Rational Intending lead to a peculiar sort of volitional inconsistency. Good examples of arguments which rely on this principle can be developed by considering cases of maxims which, when universalized, produce what are frequently referred to as 'unintended consequences'. For example, I can adopt the underlying intention of improving my economic well-being, and the specific intention of doing so by competing effectively with others. The maxim of my action can be consistently universalized: there is no conceptual contradiction in intending everyone's economic position to improve. The specific intention of adopting competitive strategies is not inconsistent with the maxim to which it is ancillary; nor is universal action on competitive strategies inconsistent with universal economic advance (that indeed is what the invisible hand is often presumed to achieve). But if an agent intends his own economic advance to be achieved solely by competitive strategies, this nexus of intentions cannot consistently be willed as universal law, because the further results of universal competitive activity, by itself, are inconsistent with universal economic advance. If everyone seeks to advance by these (and no other) methods, the result will not put everybody ahead economically. A maxim of economic progress combined with the specific intention of CONSISTENCY IN ACTION 181 achieving progress merely by competitive strategies cannot be universalized, any more than the intention of looking over the heads of a crowd can be universally achieved by everyone in the crowd standing on tiptoes. 28 On the other hand, a maxim of seeking economic advance by means of increased production can be consistently universalized. It is merely the particular specific intention of advancing economically by competitive strategies alone that leads to volitional inconsistency when universalized. Competitive means are inherently effective only for some: competitions must have losers as well as winners. Hence, while it can be consistent to seek individual economic advance solely by competitive methods, this strategy cannot consistently be universalized. Once we consider what it would be to intend the consequences of universal competition - the usually unintended consequences we can see that there is an inconsistency not between universal competitive activity and universal economic progress, but between the further results of intending only universal competitive activity and universal economic progress. Economic progress and competitive activity might each of them consistently be universal; indeed it is possible for them to coexist within a certain society. (Capitalist economies do experience periods of general economic growth.) Nevertheless there is a volitional inconsistency in seeking to achieve universal economic growth solely by way of universal adoption of competitive strategies. This argument does not show that either the intention to advance economically or the intention to act competitively cannot be universalized, but only that the composite intention of pursuing economic advance solely by competitive tactics cannot be universalized. It does not suggest that either competition or economic progress is morally unworthy, but only that an attempt to achieve economic progress solely by competitive methods and without aiming at any productive contribution is not universalizable and so morally unworthy. Similarly there is no inconsistency in an intention to engage in competitive activities of other sorts (e.g. games and sports). But if such competition is ancillary to an underlying intention to win, then the overall intention is not universalizable. Competitive games must have losers. If winning is not the overriding aim in such activities, if they are played for their own sake, the activity is consistently universalizable. But to play competitively with the fundamental intention of winning is to adopt an intention which makes of one's own case a necessary exception. 182 ONORA O'NEILL 8. CONCLUSIONS The interest of an autonomous universality test is that it aims to ground an ethical theory on notions of consistency and rationality rather than upon considerations of desire and preference. Kant's universality test meets many of the conditions which any such universality test must meet. In particular it focuses on features of action which are appropriate candidates for assessments of coherence and incoherence, namely the maxims or fundamental intentions which autonomous agents may adopt and the web of more specific ancillary intentions which they must adopt in a given context if their commitment to a maxim is genuine. While Kant alludes specifically to conceptual inconsistencies and to those volitional inconsistencies which are attributable to non-observance of the Principle of Hypothetical Imperatives in attempts to universalize intentions, there is in addition a larger variety of types of volitional inconsistency which agents who seek to subject their maxims to a universality test (and so not to make an 'exception of their own case) must avoid. A universality test applied to autonomously chosen maxims and their ancillary, more specific, intentions can be action-guiding in many ways without invoking any heteronomous considerations. However, precisely because it applies to autonomously chosen intentions, a universality test of this sort cannot generally provide a test of the rightness or wrongness of the specific outward aspects of action. It is, at least primarily, and perhaps solely, a test of the inner moral worth of acts. It tells us what we ought to avoid if we are not to use our autonomy in ways which we can know are in principle not possible for all others. Such a test is primarily of use to agents in guiding their own moral deliberations, and can only be used most tentatively in assessing the moral worth of others' action, where we are often sure only about specific outward aspects of action and not about the maxim. This point will not be of great importance if we do not think it important whether an ethical theory enables us to pass judgment on the moral worth of others' acts. But specific outward aspects of others' action are unavoidably of public concern. The considerations discussed here do not reveal whether or not these can be judged right or wrong by Kant's theory. Kant no doubt thought that it was possible to derive specific principles of justice from the Formula of Universal Law; but the success of this derivation and of his grounding of Rechtslehre is beyond the scope of this inquiry. The universality test discussed here is, above all, a test of the mutual consistency of (sets of) intentions and universalized intentions. It operates by showing some sets of proposed intentions mutually inconsistent. It does CONSISTENCY IN ACTION 183 not thereby generally single out action on anyone set of specific intentions as morally required. On the contrary, the ways in which maxims can be enacted or realized by means of acts done on specific intentions must vary with situation, tradition and culture. The specific acts by which we can show or fail to show loyalty to a friend or respect to another or autonomy in our dealings with the world will always reflect specific ways of living and thinking and particular situations and relationships. What reason can provide is a way of discovering whether we are choosing to act in ways (however culturally specific) which we do not in principle preclude for others. The 'formal' character of the Categorical Imperative does not entail either that it has no substantive ethical implications or that it can select a unique code of conduct as morally worthy for all times and places. Rather than presenting a dismal choice between triviality and implausible rigorism, a universality test for autonomous action can provide a rational foundation for ethics and maintain a serious respect for the diversity of content of distinct ethical practices and traditions. Philosophy Department University of Essex NOTES 1 I. Kant, Grundlegung zur Metaphysik der Sitten, tr. H. J. Paton as The Moral Law, Hutchinson, London, 1953 (hereafter Grundlegung), p. 421 (Prussian Academy pagination). 2 Heteronomous readings of Kant's ethics include Schopenhauer's in On the Basis of Morality, tr. E. F. J. Payne, Bobbs Merrill, Indianapolis, 1965, Section 7, but are most common in introductory works in ethics. Recent examples include William K. Frankena, Ethics, Prentice Hall, New Jersey, 1963, p. 25; Gilbert Harman, The Nature of Morality, Oxford University Press, New York, 1977, p. 73 and D. D. Raphael, Moral Philosophy, Oxford University Press, Oxford, 1981, p. 76. Allegations that Kant, despite his intentions, must invoke heteronomous considerations if he is to reach substantive ethical considerations can notoriously be found in J. S. Mill, 'Utilitarianism', in Utilitarianism, Liberty and Representative Government, J. M. Dent & Sons, London, 1968, p. 4, but are now more common in the secondary literature on Kantian ethics. Examples include C. D. Broad, Five Types of Ethical Theory, Littlefield Adams and Co., New Jersey, 1965, p. 130 and (with respect to imperfect duties) M. Singer, Generalization in Ethics, Alfred Knopf, New York, 1961, p. 262. 3 In Kant's terminology such action is spontaneous (not determined by alien causes, negatively free) and it is a further claim that it is also Kantianly autonomous (determined by pure practical reason). The argument here will not hinge on this strong and distinctive conception of human autonomy, but only on the weaker and more common 184 ONORA O'NEILL conception of autonomy as independence from determination by outside forces including desires. 4 I have done so in O. O'Neill (0. Nell), Acting on Principle, Columbia University Press, New York, 1975. The account I offer here reflects some changes in my understanding of Kant's notion of a maxim. For this I am particularly indebted to Otfried Haffe, 'Kants kategorischer Imperativ als Kriterium des Sittlichen', Zeitschrift fiir Philosopische Forschung, 31, 1977, pp. 354-84. 5 Even a very wide-ranging and reflective recent discussion of rational choice theory such as Jon Elster's in Ulysses and the Sirens, Cambridge University Press, Cambridge, 1979, does not discuss any non-heteronomous conceptions of rational choosing. 6 Grundlegung, p. 421, n.; see also p. 401, n. 1; as well as I. Kant, The Metaphysics of Morals, of which Part II, the Tugendlehre appears as The Doctrine of Virtue, tr. M. J. Gregor, Harper and Row, New York, 1964 (hereafter Metaphysic of Morals), p. 225 (Prussian Academy pagination). 7 However, the claim that maxims are underlying or fundamental intentions should not be collapsed into the claim, which Kant makes in the Religion, that for any agent at a given time there is one fundamental maxim which underlies all his other maxims. I shall not here consider whether there are fundamental maxims in this sense. But see I. Kant, Religion within the Limits of Reason Alone, tr. T. Greene and H. Hudson, Harper and Row, New York, 1960, p. 16ff. 8 This is presumably why Kant often argues from isolation tests. E.g. Grundlegung, pp. 398-9 and 407. 9 See, for example, Metaphysics of Morals, pp. 440 and 445-6; Grundlegung, pp. 407-8 and Religion, p. 16. 10 See Grundlegung, 397-8 (" ... the concept of duty, which includes that of a good will"). The persistence of the view that Kant is primarily concerned with a criterion of right action perhaps reflects a modern conception that duty must be a matter of externals more than it reflects the Kantian texts. See also Acting on Principle, Ch. 4 and O. O'Neill, 'Kant After Virtue', Inquiry, 26,1983, pp. 387 -406. 11 The points mentioned in this and in the preceeding paragraph suggest why a focus on maxims may make it PQssible to bypass a variety of problems which are said to plague universality tests when applied to principles which are 'too general' or 'too specific'; these problems include those of invertibility, reiterability, moral indeterminacy, and those of the generation either of no results or of trivial or of counter intuitive results. 12 Kant then does not see all acts which are specific, strict requirements of duty as matters of justice. There are also strict or perfect requirements for some duties of virtue, such as refraining from suicide, mockery or detraction which he views as indispensable means to the rather indeterminate maxims that we act on when we act on a maxim of virtue. See Metaphysic of Morals, pp. 421ff. and 463ff. and Acting on Principle, pp.52-8. 13 Cf. Metaphysic of Morals, pp. 380-81. 14 Grundlegung, p. 424 and cf. Sections 5 and 6 below. 15 Grundlegung, p. 417. 16 Kant's discussions of duties of virtue suggest that he would in any case count the necessary constituents or components of some end - and not merely the instrumentally necessary means to that end - as among the means to that end. 17 The fIfth requirement of Rational Intending clearly deals with the very nexus of CONSISTENCY IN ACTION 185 intentions on which discussions of the Doctrine of Double Effect focus. That doctrine claims that agents are not responsible for harm which foreseeably results from action undertaken with morally worthy intentions, provided that the harm is not disproportionate, is regretted and would have been avoided had there been a less harm producing set of specific intentions which would have served the same maxim in that situation. (The surgeon foresees but regrets the pain unavoidably inflicted during life-saving procedures, and so is not to be held responsible for inflicting this pain.) While the Doctrine of Double Effect holds that agents are not to be held responsible for such results, it allows that agents do, if 'obliquely' rather than 'directly', intend the results. It is therefore quite compatible with the Doctrine of Double Effect to insist that an agent whose oblique intention foreseeably undercuts the fundamental intention for the sake of which what is directly intended is done acts irrationally. For where the fundamental intention is undercut the forseeable results of specific intentions were not proportionate; rather they defeat the very intention to which they are supposedly ancillary. 18 Grundlegung, p. 424. 19 This is the so-called Formula of the Law of Nature, cf. Grundlegung, p. 421, as well as p. 436, 'Maxims must be chosen as if they had to hold as universal laws of nature' and Metaphysic of Morals, p. 225, 'Act according to a maxim which can at the same time be valid as a universal law.' In this discussion I shall leave aside all consideration of the relationships between different formulations of the Categorical Imperative and in particular of the differences between versions which are formulated quite generally 'for rational beings as such' and those versions which are restricted to the human condition ('typics'). These topics have been extensively discussed in the secondary literature including in H. J. Paton, op. cit.; John Kemp, The Philosophy of Kant, Oxford University; R. P. Wolff, The Autonomy of Reason, Harper and Row, New York, 1973, Ch. 2; Bruce Aune, Kant's Theory of Morals, Princeton University Press, Princeton, New Jersey, Chs. II to IV. 20 Grundlegung, p. 424. 21 Ibid., as well as Metaphysic of Morals, Introduction, esp. p. 389. 22 For further discussion of the notion of the universalized counterpart of a maxim cf. Acting on Principle, pp. 61-3. 23 For an application of the Formula of Universal Law to the example of slavery see also Leslie Mulholland, 'Kant: On Willing Maxims to Become Laws of Nature', Dialogue, 17,1978. 24 To see why Kant thinks that the abrogation of autonomy would be the most fundamental of moral failings see I. Kant, 'An Answer to the Question "What is Enlightenment?"', tr. H. B. Nisbet, in Hans Reiss, ed., Kant's Political Writings, Cambridge University Press, Cambridge, 1970 as well as Barry Clarke's discussion of 'elective heteronomy' in 'Beyond the Banality of Evil', British Journal of Political Science, 10, 1980, pp. 17-39. 25 Cf. the various works of commentary listed under note 17 above as well as Jonathan Harrison, 'Kant's Examples of the First Formulation of the Categorical Imperative' and J. Kemp, 'Kant's Examples of the Categorical Imperative', both in R. P. Wolff, ed., Kant: Foundations of the Metaphysics of Morals, Bobbs Merrill, Indianapolis, 1969. 26 Cf. Metaphysic of Morals, pp. 447-64 for Kant's discussion of the various virtues of love and sociability. 27 Cf. Metaphysics of Morals, pp. 443-7 for Kant's discussion of the duty to seek to 186 ONORA O'NEILL develop talents (duty to seek one's own perfection). It is important to remember that 'talents' here are not to be understood as any particularly unusual human accomplishments, but rather as any human powers which (unlike natural gifts) we may choose either to cultivate or to neglect. The morally significant talents are powers such as those of self-mastery and self-knowledge. 28 Cf. F. Hirsch, Social Limits to Growth, Harvard University Press, Cambridge, Mass., 1976. PHIL 3020 Ethics Nov. 15 1. The role of reason: Hume vs. Kant - Practical reason vs. theoretical reason. o Practical reason: forming beliefs about how we have to act. o Theoretical reason: forming beliefs about how the world is. Hume: practical reason is subordinate to desires; it does not set goals; it had only an instrumental role to play in practical deliberation. Kant: practical reason can set goals; and it is required in order to set moral goals, which are independent of our desires. 2. Hypothetical vs. categorical imperatives - Hypothetical imperative: conditional command; it holds given some of our desires, ends, or goals. Categorical imperative: unconditional command; it holds no matter what our desires, ends, or goals are. 3. Acting in accordance with duty vs. acting from the motive of duty - - A shopkeeper, who gives the correct change to a young and inexperienced customer, because she is interested in keeping a good reputation, is acting in accordance with duty, but not from the motive of duty. Why? Because the motive of her action is self-interest. Good will = disposed to act from the (sole) motive of duty. 4. The Categorical Imperative - The Formula of the Universal Law (of Nature): Act only according to that maxim through which you can at the same time will that it become a universal law. A maxim is a subjective principle of action: I will A in C in order to realize or produce E. “Maxims are those underlying [or fundamental] intentions by which we guide and control our more specific intentions.” (O’Neill, p. 162) 5. Inconsistency without universalization - Conceptual inconsistency: a maxim is incoherent simple because it expresses an impossible aspiration. - o “A non-universalized maxim embodies a conceptual contradiction only if it aims at achieving mutually incompatible objectives and so cannot under any circumstances be acted on with success.” (O’Neill, p. 168) Volitional inconsistency: concerns (probably most of the time) the relations between (fundamental) intentions and the “Principles of Rational Intending.” o One commits a volitional inconsistency if she wills some end without willing whatever means are indispensable for that end. This is inconsistent with the “Principle of Hypothetical Imperatives”, according to which one should will the means indispensable for the end she wills. 6. Inconsistency in universalization - “We must be able to will that a maxim of our action become a universal law: this is the canon of moral appraisal of action in general. Some actions are so constituted that their maxim cannot even be thought without contradiction as a universal law of nature, far less could one will that it should become such. In the case of others that inner impossibility is indeed not to be found, but it is still impossible to will that their maxim be raised to the universality of a law of nature because such a will would contradict itself.” (Kant, p. 494) a) Contradiction in conception o Maxim: I will get money on a false promise whenever I am in need of money and I have no other way of getting it. o Universalized: Everyone will get money on a false promise whenever they are in need of money and have no other way of getting it. o “I then see at once that it could never hold as a universal law of nature and be consistent with itself, but must necessarily contradict itself. For, the universality of a law that everyone, when he believes himself to be in need, could promise whatever he pleases with the intention of not keeping it would make the promise and the end one might have in it itself impossible, since no one would believe what was promised him but would laugh at all such expressions as vain pretenses.” (p. 493) b) Contradiction in the will o Maxim: I will refrain from helping those in need whenever I am in a position to help and despite the fact that I am well off. o Universalized: Everyone will refrain from helping those in need whenever they are in a position to help and despite the fact that they are well off. o Adopting the universalized counterpart of the maxim consists in a “mistake of rationality”, because it is inconsistent with the following principle of rational agency: I will that I be helped whenever I am in need and others are in a position to help me. o “It is only because intending a maxim of non-benevolence as a universal law requires commitment to that very absence of help when needed, to which all rational intending requires assent, that non-benevolence cannot coherently be universalized.” (O’Neill, p. 177) PHIL 3020 Ethics Nov. 3 1. A problem for classical utilitarianism? - Classical utilitarianism may fail to be a determinate ethical theory. Principle of utility as the criterion of the morally right rather than as a decision procedure: “Instead, most consequentialists claim that overall utility is the criterion or standard of what is morally right or morally ought to be done. Their theories are intended to spell out the necessary and sufficient conditions for an act to be morally right, regardless of whether the agent can tell in advance whether those conditions are met. Just as the laws of physics govern golf ball flight, but golfers need not calculate physical forces while planning shots; so overall utility can determine which decisions are morally right, even if agents need not calculate utilities while making decisions. If the principle of utility is used as a criterion of the right rather than as a decision procedure, then classical utilitarianism does not require that anyone know the total consequences of anything before making a decision.” (Sinnott-Armstrong, “Consequentialism”, SEP). 2. Act / extreme utilitarianism vs. Rule / restricted utilitarianism - - Rule / restricted utilitarianism: “In general, the rightness of an action is not to be tested by evaluating its consequences but only by considering whether or not it falls under a certain rule. Whether the rule is to be considered an acceptable moral rule, is, however, to be decided by considering the consequences of adopting the rule. Broadly, then, actions are to be tested by rules and rules by consequences.” (Smart, p. 423) o We should accept the rule that will maximize utility: § R1: Whenever one has made a promise, one ought to keep the promise. § R2: Whenever one has made a promise, one may break the promise if one prefers to do so. Act / extreme utilitarianism: the rightness or wrongness of an action depends only upon the consequences of that action on that particular occasion. Rules could be useful, but they are only rules of thumb. o Smart’s objection to rule utilitarianism: § If we have worked out the consequences and if we have perfect faith in the impartiality of our calculations, and if we know that in this instance to break the rule R will have better results than not to break it, then we should break R (Smart, p. 425). § Doing otherwise would be “a form of superstitious rule-worship” (Smart, p. 425). 3. Singer’s argument 1. Suffering and death from lack of food, shelter, and medical care are bad. 2a. If it is in our power to prevent something bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it. 2b. If it is in our power to prevent something very bad from happening, without thereby sacrificing anything morally significant, we ought, morally, to do it. 3. It is in our power to prevent something bad from happening by giving money to famine relief. 4a. By giving money to famine relief we are not sacrificing anything of comparable moral importance. 4b. By giving money to famine relief we are not sacrificing anything morally significant. 5. Therefore, we ought to give money to famine relief. - Example that sustains 2a / 2b: “if I am walking past a shallow pond and see a child drowning in it, I ought to wade in and pull the child out. This will mean getting my clothes muddy, but this is insignificant, while the death of the child would presumably be a very bad thing.” (p. 467) - If Singer’s argument is sound, charity cannot be understood as a supererogatory act: “When we buy new clothes not to keep ourselves warm but to look ‘welldressed’ we are not providing for any important need. We would not be sacrificing anything significant if we were to continue to wear our old clothes, and give the money to famine relief. By doing so, we would be preventing another person from starving. It follows from what I have said earlier that we ought to give money away, rather than spend it on clothes which we do not need to keep us warm. To do so it not charitable, or generous. Nor is it the kind of act which philosophers and theologians have called ‘supererogatory’ – an act which it would be good to do, but not wrong not to do. On the contrary, we ought to give the money away, and it is wrong not to do so.” (p. 469) PHIL 3020 Ethics Nov. 8 Singer’s argument Stronger version of the argument: 1. Suffering and death from lack of food, shelter, and medical care are bad. 2a. If it is in our power to prevent something bad from happening, without thereby sacrificing anything of comparable moral importance, we ought, morally, to do it. 3. It is in our power to prevent something bad from happening by giving money to famine relief. 4a. By giving money to famine relief we are not sacrificing anything of comparable moral importance. 5. Therefore, we ought to give money to famine relief. Weaker version of the argument: 1. Suffering and death from lack of food, shelter, and medical care are bad. 2b. If it is in our power to prevent something very bad from happening, without thereby sacrificing anything morally significant, we ought, morally, to do it. 3. It is in our power to prevent something bad from happening by giving money to famine relief. 4b. By giving money to famine relief we are not sacrificing anything morally significant. 5. Therefore, we ought to give money to famine relief. - Example that sustains 2b: “if I am walking past a shallow pond and see a child drowning in it, I ought to wade in and pull the child out. This will mean getting my clothes muddy, but this is insignificant, while the death of the child would presumably be a very bad thing.” (p. 467) - If Singer’s argument is sound, charity cannot be understood as a supererogatory act: “When we buy new clothes not to keep ourselves warm but to look ‘welldressed’ we are not providing for any important need. We would not be sacrificing anything significant if we were to continue to wear our old clothes, and give the money to famine relief. By doing so, we would be preventing another person from starving. It follows from what I have said earlier that we ought to give money away, rather than spend it on clothes which we do not need to keep us warm. To do so it not charitable, or generous. Nor is it the kind of act which philosophers and theologians have called ‘supererogatory’ – an act which it would be good to do, but not wrong not to do. On the contrary, we ought to give the money away, and it is wrong not to do so.” (p. 469)
Purchase answer to see full attachment