1. Pragmatism and Practical Reasoning

A proper understanding of pragmatism can only come, I think, by looking at the views of pragmatist writers about practical reasoning, specifically about the relationship between prudential and moral reasoning. Without exception, they objected to rationalist theories of ethics that erect a sharp division between prudential reasoning based on inclinations (or wants or desires) that are in some sense "given" and moral reasoning based on rational apprehension of moral principles that if followed restrain our inclinations. The solution for the pragmatists to overcoming the apparent conflict between inclinations and duties in accordance with these principles is to regard such conflicts as arising from competing inclinations or wants, some of which may be shared among the community to which an individual belongs.

This is most clearly expressed by Ralph Barton Perry in his General Theory of Value. Perry notes the traditional view that values reside on the side of moral duties, while inclinations or wants must reined in by Reason as the charioteer of Plato’s metaphor. He concludes that "The conflict between inclination and the sense of duty is a genuine conflict between one interest and another, and there seems to be no good reason for imputing value to the object of the latter while denying it to the object of the other" (p. 101). The categorical and authoritative character attributed by Kant to moral "ought" judgments arises, Perry says, from ignorance of the background reasoning based on inclinations used in earlier times to derive through argumentation these judgments. The "irreducibility" and "impressive and categorically authoritative character" of an "ought" judgment, he says,

With Perry as our representative, we can identify pragmatism with the view that practical reasoning has as its basis inclinations or desires, and this feature extends to moral reasoning. Dewey seems to be in basic agreement with this view, for he rejects Kant’s assigning to reason the role of evaluating our desires. "Rationality," he says, "is not a force to evoke against impulse and habit. It is the attainment of a working harmony among diverse desires."1 Instead, reasoning is assigned the subsidiary Humean role of tracing the consequences of action. It "puts before us objects which are not directly or sensibly present, so that we then may react directly to these objects, with aversion, attraction, indifference or attachment."2 But we also attribute to Dewey his much-discussed distinction between the desired and the desirable. That there is such a distinction seems to imply that there is some rationally apprehended principle in terms of which desired ends can be evaluated, but Dewey never endorses this solution. It is certainly not for him the principle of maximizing utility, for he explicitly criticizes utilitarianism as setting up a "fixed end."3 His attempts to state the desired/desirable distinction have been criticized by a number of writers, including Perry. It remains one of the least satisfactory parts of his otherwise admirable discussions of practical reasoning.

As still another representative of pragmatism who rejects the Kantian inclination/duty we can point to Richard Rorty in his Contingency, Irony and Solidarity. There he criticizes attempts to derive an unconditioned rational foundation of ethics, and restates the distinction between the moral and the prudent in a way consistent with Perry. We can keep the distinction, Rorty says, "if we think of it not as the difference between an appeal to the unconditioned and an appeal to the conditioned but as the difference between an appeal to the interests of our community and the appeal to our own, possibly conflicting, private interests" (p. 59). Those desires shared within communities that have contingently evolved within the human species become for Rorty the principal basis for moral judgments. This is the reason that in his brave new world the arts and literature as their primary means of expression assume a central role in moral deliberation, supplanting philosophic attempts to persuade us of rationally derived principles from which moral conclusions can then be inferred.

With our respected pragmatist authorities duly lined up in advance on our side, we can now proceed to discuss our central problem. This is to give an account that shows the similarities and differences between prudential and moral reasoning in a way that avoids the Kantian inclination/reasoned duty contrast. We have the writings of the early pragmatists, and we have Rorty’s suggestive remarks. But these are vague and inconclusive, and in themselves insufficient to establish the pragmatist view of practical reasoning against persistent rationalist criticisms, most recently by writers such as Thomas Nagel. One remedy is to exploit a quite obvious feature of all reasoning, including practical reasoning. It is that to reason is to use an inference with premisses and conclusion. This is obvious for deductive reasoning, where we do specify premisses and conclusions, and trace the logical relations between them. It is perhaps less obvious, but no less real, for practical reasoning. How we specify the premisses of the practical inferences used in this reasoning and their logical relations to conclusions will determine the solution to our problem of relating the prudential and the moral. As we shall see in the final section, this solution also hinges on our being able to successfully relate practical inferences used to justify actions to explanations of the actions being justified.

2. First-Person Prudential Practical Inferences

The obvious starting point is an account of the familiar Aristotelian "practical syllogisms" in which normative "should" conclusions are justified on the basis of premisses. Prudential forms of these inferences prove to be more tractable than their moral cousins, which we will turn to shortly, and provide a useful basis for comparison and contrast. Since they have been discussed in some detail, their features can be summarized briefly.4

What we can refer to as a positive basic pattern prudential inference has a premiss expressing in the first person a want for an end E, another describing an action A necessary to bring about in a given situation or circumstance S what is wanted, and a statement to the effect that S does in fact obtain. Its form in the first person is roughly

P1 P2 P3		I want E My doing A is necessary to attain E if S S
C		Therefore, I should do A.

The want or desire for E is a special type of valuation, and E can be regarded as a type of value. The end E can be a personal ideal, with the want being the agent’s valuation of what he or she regards as a virtuous life. Much prudential deliberation is directed towards the realization of our self-image, and if this self-image is that of a saint, it is certainly not egoistic or what we call "selfish." But E can also be a mundane condition for creature comfort. For example, someone might reason that she wants to keep warm this winter, that her buying a coat is necessary to keep warm if the winter will be cold, which it will be, and therefore she should buy a coat. The ‘should’ or ‘ought’ of the conclusion is only a prima facie requirement, since the conclusion could be defeated if the costs of performing A outweigh the perceived benefit of E (the agent prefers to be cold over the expense of the coat). These costs are other values that are sacrificed by the attainment of E. Consideration of them can lead to a change of the initial desire for E. Indeed, the use of ‘should’ in the conclusion implies that some aversion must be overcome, or in Kantian terms, that there is some inclination not to perform A, some cost attending its performance.

There are other complications that can be only briefly summarized here. Left implicit is the assumption that the agent has both the ability and opportunity to do A. There also may be a range of alternatives used to specify A as a general type of action (a parka, lined rain coat, and dress coat may be alternatives between which the agent must choose). In such cases, the conclusion is asserted "all things considered" relative to preferences between the costs and benefits of the alternatives. How the action is specified (whether buying a coat in general, or a type of coat) will vary with situation and interests. Further, corresponding to the positive basic pattern inference just outlined, there is a negative pattern whose first premiss expresses an aversion towards a "bad" state of affairs E* and includes a premiss stating that an action A is sufficient to bring about E*. Its form is

P1 P2 P3		I don’t want E* My doing A is sufficient to bring about E* if S S
C		Therefore, I shouldn’t do A.

The "bad" E* is often encountered in past bitter experience, and this provides a definiteness often lacking in wants directed towards unattained states of affairs. This is but the familiar truth that we are aware more of our aversions than our wants, and insures that the negative pattern is at least as commonly used as the positive.

It should be obvious that we seldom use a full-blown version of the basic pattern inferences just described. Typically one or more of the premisses is suppressed as a background assumption. It may be P1, the expressed want for E, or it may be P2, P3, or P4, or some combination of them. The agent may reason, for example, ‘Since I want to keep warm, I should buy a coat’. P4 may be the focus of attention, and appear by itself in the foreground, as in ‘Since this winter will be cold, I should buy a coat’. In such prudential cases, agents can sometimes make explicit the want or accepted information suppressed in the background. But even for simple prudential reasoning of the kind being considered, much that we take for granted can be below the threshold of consciousness and be difficult or impossible to verbally formulate. As for theoretical reasoning, to make explicit the premisses used is a task of reconstructing what agents are often only dimly aware of.

Provided the agent forms an intention to act based on a ‘should’ conclusion and actually performs the action, for every prudential inference used by an agent X in justifying an action A in the first person there will potentially be a corresponding third-person explanation of why A was performed. This feature of prudential reasoning we can refer to as justification/explanation correspondence. This explanation will be in the past tense, and will substitute descriptions of beliefs for what are asserted in the practical inference. Thus, corresponding to the positive inference we have the explanation,

P1 P2 P3		X wanted E X believed that doing A was necessary to attain E if S X believed that S obtained
C		Therefore, X did A.

The justification/explanation correspondence relation is obviously asymmetrical, for there will be explanations citing unconscious motivations not articulated within a justifying prudential inference. What is in the foreground of the explanation need not have been in the foreground of the corresponding prudential inference used to justify the action. Indeed, the agent herself may become aware of her own desire at the time of deliberation only when she begins to explain later why she performed the action. Knowing her beliefs and realizing what she has done, she infers her past desire.5 For the negative pattern there will be a corresponding explanation why A was not performed, and again this explanation may cite an aversion that was not in the foreground of the prudential inference used to justify the non-performance of A.

As for the practical inference, an explanation may highlight some premisses and suppress others as assumed background. Thus, an explanation why A was done might be simply ‘X did A because she believed S’ (bought the coat because she believed there would be a cold winter), with X’s desire and other beliefs omitted as not relevant to what happened to be in question. In such cases, we would regard the explanation as incomplete, as an ellipsis for one with a fuller complement of premisses. It is doubtful, however, we can give a clear account of the contrast between what is an elliptical explanation and what we regard as "complete." All explanations seem to be made relative to varying interests, and it would seem all of them could be supplemented by additional premisses. To the explanation given above, for example, we could add that X preferred the advantages of E to the costs consequent on the performance of A, or we could introduce specific alternative ways of A-ing (buying a parka or raincoat) for which preferences must be specified.

Granted that we can’t specify sufficient conditions for completeness, we do seem to be able to specify the presence of some desire or preference as a necessary condition. It is reasonable to think that at least for explanations of prudential actions for which the interests of others are for the most part irrelevant a complete explanation of an action must include mention of some form of desire as a motivating reason. The citation of beliefs only is invariably an ellipsis for a more complete explanation in which desires of the agent are included. That desires are necessary conditions for completeness of explanation we can refer to as the desiderative condition for prudential action. This condition seems to be implicitly accepted by those stating the Kantian inclination/duty distinction, since such a distinction concedes for the prudential the explanatory role of inclinations.

There are familiar rationalist objections to the use of desires for justificatory reasons, even in prudential reasoning. Many have insisted (Dewey for one, as we have seen) that only an end E worth wanting or desirable enables us to derive a normative "should" conclusion, and this requires asserting that attaining E satisfies certain conditions, e.g., is in the agent’s best interests or conducive to his greater happiness in the long run. Such an objection implies that the belief that E satisfies these conditions can itself explain why A was performed, contrary to the desiderative condition for a complete explanation. A close look at such evaluation of a desire reveals, however, that it invariably takes place relative to some other desire, just as Perry maintained. I have a craving to eat a box of chocolates, and know that buying a box is a necessary means. But I also know that eating the chocolates contributes to obesity and bad health, which I don’t want. Though I want to eat the chocolates, I thus realize that eating is not worth wanting. We can be said to endorse a want for an end E that survives such consideration of costs, and is used to justify action. This endorsement is a type of evaluation of a want, but it is made relative to desires or aversions directed towards other consequences of the means, not some cognitive hedonistic standard to which wants are irrelevant.

Another objection arises from Thomas Nagel’s observation that for prudential reasoning there is always the possibility of a person’s desire for E at some time t1 at which the reasoning takes place changing at the later time t2 at which the action A is to be performed. The agent must believe that what she desires at t1 persists until t2 in order to conclude that A should be done later.6 Her justifying reason for A is then not the desire for E itself, but instead the belief in its persistence. It would seem to follow that the explanation of her having done A could omit mention of the desire and instead cite only this belief. But this objection seems to conflict with our practice of both justifying and explaining. Desires do change, and beliefs about them can be part of the informational basis of justification. I may believe that, though I want to travel now, I probably won’t want to travel when an octogenarian. As a result, I judge that I shouldn’t save now for travel in the distant future. The decision not to save is based, however, not simply on the belief in a change of wants, but on a desire not to forego present use of money for a future in which it will not be gainfully used. This other desire guides justification, and should be cited in any explanation of why I don’t now presently save.

There are also, of course, second and third-person versions of basic and option pattern inferences where evaluation of wants becomes relevant, though here again there is no desire-free rational standard that is brought to bear. From ‘You (or he) want(s) E’ and ‘Doing A is necessary for your (his) bringing E about’ the speaker would infer a normative "should" conclusion only if he or she endorsed the subject’s want for E’. If we disapprove of the subject’s attaining E (as for the satisfaction of a harmful craving), our aversion to this attainment may outweigh for us the agent’s own want and preferences, with the result that we conclude that the subject shouldn’t do A.

It can also be objected that often our actions are guided by prudential maxims such as ‘One ought to save for the future’ or ‘One ought never to put off now what must be done later’. Such maxims we learn from others or from our own experience, and then apply to situations in which we find ourselves. The application is made by means of a deontic inference of the general form,

Everyone ought to do A in situation S I am now in S

I ought to do A.

One using such an inference can be regarded as justifying A (for example, saving some money) on the basis of the normative first premiss (everyone should save when having surplus money). The argument is then that any explanation of why A was performed could cite only the belief in the truth of this normative proposition.

The reply to this objection should be obvious. One using the deontic inference would want whatever advantages come from following the normative prudential maxim, though he may not be aware of what these exactly might be if following blindly what has been told to him by others. Trusting the maxim’s source that these benefits will accrue, whatever they may later prove to be, he complies with it. The fact that prudential maxims have been passed down through the generations should be at least some indication to him that following them has had beneficial consequences and these will be likely be repeated in the future. Again, a desire guides the justification, and should be cited in any complete explanation.

Despite such objections, then, let’s concede with Kant that for prudential reasoning of the kind just discussed wants or desires (inclinations) are essential components of both a justifying practical inference for an action A, though possibly only as part of the background, and of a complete explanation of why this action A was performed. The central question for us now is whether these conclusions can be extended to inferences used to both justify moral obligations to do A and to explain A’s performance.

3. Moral Inferences and Shared Desires

The pragmatists’ answer is that reasoned moral "ought" conclusions are to be based, at least in part, on shared desires within a community. Much of the mystery surrounding such desires and their association with Rousseau’s "general will" should have been dispelled by recent discussions of joint intentions and cooperative actions by a number of writers.7 A shared desire is simply one for which there is overlap in the content of desires of individual members. Sufficient overlap becomes the basis for the forming of joint intentions and cooperative action, and the justification for such intentions and action would seem to be derived from a practical inference. For example, the members of a neighborhood may share a desire for their block to be cleared of snow after a storm (E). Realizing that hiring a plow (A) is necessary for this if one is not supplied by the city (S) and that the city won’t supply one, they conclude that they should hire a plow. The inference as used by some speaker representing the neighborhood is thus of the form:

P1 P2 P3		We want E Our doing A is necessary to attain E if S S
C		Therefore, we should do A.

They then form the joint intention to hire a plow, and act accordingly. The inference used is simply a positive form basic pattern inference with a plural ‘we’ subject.

Despite the introduction of a plural ‘we’ subject, however, the ‘ought’ of this inference’s conclusion would seem to have only prudential force, for we would not regard the group as under a moral obligation to hire the plow. Indeed, all the principal features of first-person prudential forms seem to carry over: the inference is agent-homogeneous for the plural ‘we’, the conclusion is defeasible, and there are option and negative forms parallel to those for singular prudential inferences. Also, justification/explanation correspondence holds: if members of the group do indeed use the plural form to justify their action of hiring a plow, then there will be a corresponding explanation with a third-person ‘they’ subject that cites the shared wants and beliefs of the justifying inference. The reason why they hired the plow was that they wanted the streets cleared and believed that the hiring was a necessary means given the city’s lack of help. As before, from the fact that we are able to produce such an explanation doesn’t guarantee that there was a corresponding justifying inference used within the group. Its members may have not have been aware of the wants and beliefs they shared in common, nor be able to reconstruct after the fact their justification of what they did.

The reason that the above plural form fails as a moral inference is that because it is agent-homogeneous, it cannot apply its "ought" conclusion to individuals or to sub-groups within the larger group. As Christopher Kutz notes, for joint action by members of a community there are expectations for cooperation accompanied by temptation for free-riding on the part of individuals.8 The need to impose moral norms is introduced by this possibility of individuals (or sub-groups) attaining a special advantage for themselves by opting out of a general activity that contributes to an end valued by all. Often legal sanctions against such opting out are imposed that alter the preferences of individuals and reduce the temptation. Penalties for avoiding taxes is an obvious example of this. But general compliance to laws seems to require at least some moral justification to the legal system of which they are a part, and there are areas of conduct that are not effectively legislated. Here moral reasoning is relevant. Inferences used in this reasoning would seem to have the feature of the expression of a shared want for an end as a premiss and a conclusion that requires individuals to contribute to attaining what is of general benefit. Since mutual trust is a necessary condition for cooperative action, maintenance of this trust seems to be an especially central moral obligation placed on members.

As a consequence of this, truth-telling and promise-keeping are regarded as paradigm actions to which moral norms apply. It is possible to reconstruct from our practice of imposing these norms a plural form of practical inference that provides their justification. Its first premiss is a shared desire for mutual trust within the group (E). Then there is the realization that mutual trust requires nearly everybody to keep their promises (A). We say "nearly everybody," since some breaking of promises would not itself destroy mutual trust in the word of others; only widespread defection would accomplish this. But we regard it as unfair that some individuals gain advantage for themselves by breaking a rule complied with by others that advances the mutual trust that these individuals themselves want. As Kant noted, promise-breaking itself presupposes mutual trust established by nearly everybody keeping their promises. Hence we need a universalization principle preventing free-riding to the effect that if nearly everybody’s doing A is necessary for the mutually desired E, then everybody should do A. With these premisses, the obligation to keep promises can be applied to a specific member X of the group. The form of the agent-heterogeneous inference used in this reasoning is thus:

P1 P2 P3		We want E Only if nearly all (or most) of us do A will E be attained If nearly all (most) of us should do A, then everyone should
C		Therefore, X should do A.

There is an implicit intermediate conclusion inferred from the three premisses to ‘Everyone should do A’, and the assumption is made that X is a member of the group whose representative expresses the shared want in the first premiss. As for prudential inferences, the use of ‘should’ in the conclusion implies some aversion on the part of X to doing A. The performance of A can thus be described in Kantian terms as an overcoming of an inclination.

Notice that a situation S in which A is to be performed is not part of this positive moral inference, as general obligations are typically imposed that are not situation-specific. Instead, there are often a number of possible types of situations that if present are understood as releasing X of the obligation. P3 could thus be understood as ‘If nearly all of us should do A, then everyone should do A unless S1 or S2 or … Sn obtain’, with the additional premiss added that none of the extenuating situations do in fact obtain to release X of his moral obligation. If one or more of S1,S2,… Sn do obtain, we may infer ‘It is permissible for X to forbear from A’. As for the conclusions of prudential inferences, the implied intermediate conclusion ‘Everyone should do A’ is defeasible, as the costs of doing A may outweigh the advantage of E. Also, there is a negative form in which the inference is made from a shared aversion and the sufficiency of most doing A for bringing about the object of this aversion to the conclusion that an individual X shouldn’t do A. Here the universalization principle is that if most shouldn’t do A, then no one should. Because the members of a community are more likely to be aware of their shared aversions than their positive desires, moral prohibitions based on these aversions are more likely to be conclusions of moral reasoning than are obligations.

The effect of a universalization principle is to enable conclusions applied to specific individuals to be inferred from group wants and activities. This represents a complication of premisses that seems inconsistent with the ease with which we usually make moral judgments. Though we are able to philosophically construct a justification of moral requirements, the question arises whether such moral inferences are actually used in moral deliberation, and is answered in the same manner for the prudential case. As for the prudential, there is no need that all premisses be in the foreground, no more than this was necessary for prudential forms. A community may have not articulated its shared wants, and its members may be unaware of their common contents. There is some awareness of the immorality of an individual free-riding and taking advantage of others following rules in a way that benefit him or her. This awareness is vague, and certainly not articulated in a way that would allow inferring to a moral "ought" conclusion. But such lack of articulation of premisses is also a characteristic of prudential inferences, and there seems little doubt that these are used in agent-centered deliberation.

It is also true that for the great majority of moral "ought" judgments there is a direct inference from some moral maxims such as ‘Everyone ought to keep their promises’, ‘Everyone ought to tell the truth’, or ‘Everyone ought to come to the aid of those in distress’. Typically we apply these maxims of the form ‘Everyone ought to A’ to particular situations facing ourselves and others. These are inculcated during moral education and reinforced by praise and scolding. For those with fundamentalist religious beliefs, the maxims can be recorded in authoritative sacred texts. The tendency to act in accordance with them is also undoubtedly part of our innate inheritance through processes of selection. Both this inheritance and education confer on them that intuitively "self-evident" character that has been appealed to by moral theorists. Such immediate moral judgments need not be based on some articulated moral generalization, moreover. In many situations, we undoubtedly also directly apply moral exemplars derived from story telling or other imaginative sources, as argued by Mark Johnson.9 Just as we instinctively classify relative to prototypes, so these exemplars seem to guide our judgments of what should be done in a given situation, but not on the basis of any reasoning from shared wants and means/ends premisses.

All of this can be conceded without threatening the thesis that moral practical inferences are used in moral deliberation. As we have seen, all the features just noted are present in the use of prudential inferences. Inculcated maxims such as ‘Everyone should save for the future’ discussed in Section 2 are applied in prudential reasoning, but we recognize this as an economical short-cut that avoids the need to subject every "should" judgment to full-fledge deliberation. In the same way, past experience and moral education furnish those generalizations allowing us to guide conduct without rethinking the basis for them. There is therefore no fundamental contrast with prudential reasoning, and the response to those questioning the use of practical inferences for the prudential case can be carried over to the moral. Judgments made on the basis of previously accepted generalizations or exemplar application are essential for the purposes of most conduct, as Simon Blackburn has so convincingly argued.10 But they can be challenged, and both the criticism and defense argued within the moral deliberation that follows will appeal to ideals that members of the community want to have realized. Circumstances may change that bring about change in the shared wants on which earlier moral and prohibitions were based. Such changes, which are accompanied by debate and differences in weighting of ends and costs of means, are the result of moral deliberation in which moral practical inferences would seem to be used.

Moral inferences thus share these features with prudential inferences, and if we grant the latter are used to justify, in light of these similarities there seems no reason to deny that the former are too. There is, however, an important difference between prudential and moral inferences, and this threatens to undermine the claim that practical inferences are actually used for moral justification.

4. The Apparent Breakdown of Justification/Explanation Correspondence

The difference is brought about by the agent-heterogeneity of moral practical inferences, their combination of a plural ‘we’ premiss and a singular conclusion of the first, second, or third person form ‘I (you, he) ought to A’. This conclusion in itself seems to guarantee that there will be no justification/explanation correspondence of the kind present in prudential reasoning. When we explain why an individual X performed a certain action A we invariably cite that individual’s desires and beliefs, not the desires and beliefs of the community of which X is a part. Corresponding to a positive moral inference justifying A we have the following explanatory inference:

P1 P2 P3		They desired E They believed that only if nearly all of them did A would E be attained They believed that if nearly all should do A, then everyone should
C		Therefore, X did A.

Even granted that X is a member of the community referred to by ‘they’ and shares the desire and beliefs described in the premisses, we would not explain X’s action in this way. Such complex shared desires and beliefs seem to have little relevance to the motivations and beliefs with which we explain individual behavior.

The so-called "Humean view" of morality developed by Michael Smith presents a thesis of individual psychology.11 It is that beliefs cannot in themselves explain individual behavior, that they must be supplemented by other motivational states such as desires. For actions based on agent-centered prudential reasoning, this thesis seems entirely plausible, as here we do cite desires and aversions in a variety of forms (greed, ambition, status-seeking, fear, etc.) in explaining behavior. As noted above, the desiderative condition for explanation is thus satisfied. For a wide variety of actions we label as "moral" - including acts of kindness, charity, and more generally, considerateness of others - we also cite motivations such as Hume’s sympathy and feelings of benevolence as altruistic feelings we recognize as basic parts of our evolutionary inheritance. To act directly on the basis of them, however, is not to act on the basis of a reasoned moral justification. ‘I empathize with Smith’ does not itself justify an action of kindness towards Smith, though the feeling would probably be interpreted as the mark of a virtuous person and could be used in explaining the action.

There is also a wide variety of other actions for which Humean emotions don’t seem to provide plausible explanations, and it is appeal to these that provides the basis for rationalist criticisms of the Humean view. Why did Jones keep his promise to Smith? It may have been out of sympathy for Smith, and a desire not to hurt his feelings. It may also have been because he feared social disapproval if his action were publicized, or if he was a fundamentalist, perhaps fear of divine punishment. But the most plausible explanation in most cases is simply that Jones believed that everyone should keep their promises, and believed that this principle applied to his situation. In this case, the belief by itself would seem to explain the action, and we have a violation of the desiderative condition. It is not that the belief happens to be in the foreground of the explanation because of our interests, but that no other motivational states seem at all relevant. For the Kantian paradigm actions like promise-keeping and truth-telling - those actions following from the categorical imperative that Kant said we had "perfect duties" to perform - the beliefs themselves provide a complete explanation.

These considerations appear to provide a basis for restating the Kant’s inclination/duty contrast that the pragmatists have criticized. In prudential reasoning we use a practical inference with a premiss expressing a desire in order to justify a certain action, and corresponding to this inference there will be a complete explanation of the action citing this desire. This occurs despite the fact that agents may justify their action on the basis of prudential maxims. The correspondence between prudential justification and explanation establishes that the practical inference is actually used to justify. But moral reasoning seems to provide us with a breakdown in justification/explanation correspondence. Since explanation is of an individual’s action, it seems that it must cite only that individual’s motivations. These motivations seem to be of the two kinds just discussed: altruistic emotions such as feelings of empathy and benevolence and agent-centered desires such as those for social approval and aversion to criticism. Neither are "shared" in the sense of being directed towards some common content. And since neither explain satisfactorily the performance of many actions such as promise-keeping we classify as moral, philosophers in the Kantian tradition have persisted in maintaining the inclination/duty contrast.

Agent-heterogeneity of moral practical inferences may in this way have contributed to the persistence of the Kantian contrast by apparently forcing us to resort to implausible explanations of moral actions in terms of individual motivational psychology. Now there are two alternatives open to us to avoid these difficulties. One is to abandon the assumption of justification/explanation correspondence. Without this assumption, we can accept the fact that explanation of an individual’s actions must appeal to individual motivations, and still retain moral practical inferences with ‘we’ premisses as the principal model of moral justification. We should be reluctant to accept this way out, however, as correspondence to an explanation seems to provide our best means to distinguish those justifications actually in use from pretenders. If in fact we use moral practical inferences in deliberation, then there should be an explanation that cites in some form the premisses of these inferences. If there were no correspondence, this would undermine any claim that these inferences are in actual use and not our artificial, contrived rationalizations.

The second alternative is much more plausible. It is to recognize that explanation of an individual X’s moral action can cite the fact that X shared a desire with members of a group. Individual motivational psychology need not provide the only basis for all forms of explanation. A simple example of a two-member group borrowed from Gilbert’s On Social Facts illustrates this. Why did Jane take a walk yesterday with Herbert? One explanation is that Jane didn’t want to hurt Herbert’s feelings. Another might be that she wanted to be seen with him to improve her status with her friends. But another possible explanation is that both of them wanted to be together then and believed that a walk together was the best way to achieve this. In this latter explanation, it is neither Jane’s sympathy nor her desire for status that explains her action, but the fact that she shared a want with her friend. Here we have, then, an explanation citing a shared want that corresponds to a justifying practical inference.

Such an explanation can be transferred to moral actions. Why did Jane keep her promise to meet Herbert yesterday, though she preferred to stay at home? As for the walk, one answer is that she was sympathetic to her friend’s feelings if stood up; another is that she feared scolding if she broke the promise. But another possible answer is that she shared with her friend a desire to preserve mutual trust, and believed that both must keep promises to preserve this trust. This last answer has the effect of certifying Jane’s action as one performed by someone possessing what Kant called a "good will," one who wants what those in her community want (here a community of just two), and in a derived sense wants to do her duty in the face of costs she may incur.

This pattern of explanation can be expanded to groups of indefinitely large size. Jane may make a promise to someone who is a perfect stranger, and thus one for whom no specific ties have been established. Nevertheless, Jane and the stranger may be recognized as members of a wider community sharing a want for mutual trust. The fact that Jane keeps her promise can be interpreted as an implicit acknowledgment of this sharing, and what is necessary to maintain it. More generally, for any individual X an explanation of a moral action A will have the form

To be sure, there will be a very restricted class of actions to which such an explanation applies; most actions will be explained by Humean other-regarding emotions and agent-centered desires and aversions specific to the individual X. The explanation fits only those actions with what Kant called "moral worth" performed by those of good will. But it is precisely for these actions that the inclination/duty contrast was constructed. That we have a form of explanation corresponding to a justifying practical inference thus serves to undermine the contrast.

It is understandably rare for justifications in terms of shared wants, relations of means to ends, and a universalization principle to be explicitly acknowledged by agents. Such justification must be reconstructed from an analysis of prudential inferences and of the distinctive features introduced by moral inferences, and the reconstruction is not obvious. Explanation of moral acts usually cite attitudes such as empathy or belief in moral principles, thus neglecting what is essential for understanding certain actions. One of the important tasks philosophy should set for itself is that of educating the general public about the grounds for moral justification and the nature of explanation of actions performed by those of good will. All too often we have grudging submission to what are regarded as oppressive moral restrictions imposed by theological and rationalist traditions. The cause of morality will surely be advanced if philosophy adopts the role of helping others see themselves as what Phillipa Foot once called "volunteers banded together" for ideals they share in common.

I know that many will have misgivings about this pragmatist view of moral reasoning. What of the treasured objectivity and necessity that we think is the mark of moral conclusions? What of the dangers of moral relativism? Haven’t we abandoned the important goal of adjudicating between the moral conclusions of one community and those of another? Doesn’t this require rationally derived moral principles? The very short answer to such questions is simply to say that objectivity and necessity should be recognized as idols of a discredited theological and rationalist search for foundations. What is required in their place are some means by which sympathies and shared ideals of the community referred to by ‘we’ are extended to increasingly wider communities. The arts and literature surely have a role to play in this. Philosophy also contributes by stating through its social contract theories certain counterfactual wants. Conceiving what each of us would want if ignorant of our special circumstances that create bias is an important means of determining what all of us actually do want in common and thus the first premisses of moral reasoning. But the way to this answer is fairly involved, and we have run out of time. It is sufficient for present purposes to have established certain basic similarities between prudential and moral reasoning.

1. Human Nature and Conduct (New York: Henry Holt, 1922), p. 196.

2. Ibid., p. 200.

3. See Reconstruction in Philosophy (New York: Henry Holt, 1920), p. 181.

4. See D. S. Clarke, Practical Inferences (London: Routledge and Kegan Paul, 1987), Robert Audi, Practical Reasoning (London: Routledge, 1989), and Douglas Walton, Practical Reasoning (Totowa: Rowman and Littlefield, 1989).

5. This inference would be based on the behavioral definition of a desire for E as disposition to perform A when A is believed to be a necessary means to E. That background desires are known in this way is argued by Philip Petit and Michael Smith in "Backgrounding Desire," Philosophical Review 91 (1990): 565-92.

6. See The Possibility of Altruism (Oxford: Clarendon Press, 1970), Ch. VIII. Practical reasoning is directed not only towards what should be done in Nagel’s view, but also towards what should be wanted.

7. See Margaret Gilbert, On Social Facts (Princeton: Princeton University Press, 1992), Ch. 4; Michael Bratman, "Shared Cooperative Activity" in Bratman, Faces of Intention: Selected Essays on Intention and Agency (New York: Cambridge University Press, 1999); Raimo Tuomela and Kaarlo Miller, "We-Intentions," Philosophical Studies 53 (1988): 367-90; and Christopher Kutz, "Acting Together," Philosophy and Phenomenological Research 61 (2000): 1-31. The focus of these studies has been on the problem of stating conditions for mutual awareness and common content of intentions that qualify an intention as jointly held within a group and an action as cooperative.

8. Kutz, "Acting Together," p. 12.

9. Moral Imagination (Chicago: University of Chicago Press, 1993). See especially pp. 189-192 for a discussion of exemplar application in moral judgments that is analogous to prototype perceptual identification.

10. See Ruling Passions (Oxford: Clarendon Press, 1998), pp. 40-43 for Blackburn’s example of a referee calling a game. The game may be designed to entertain spectators, but a good referee applies its rules strictly on each occasion. Deliberation with reference to the game’s larger purpose would diminish the sport itself.

11. Michael Smith, "The Humean Theory of Motivation," Mind 96 (1987): 36-61. For a useful summary discussion of the debate for and against this view see Philip Clark, "What Goes without Saying in Metaethics," Philosophy and Phenomenological Research 60 (2000): 357-79. The issue as stated by Clark is whether beliefs to the exclusion of desires can be regarded as "complete motivators" of behavior.