2. Neopragmatism and Necessities

According to Robert Westerbrook (Reference Robert2005, p. 1), the classical American pragmatism of Dewey, James, and Peirce was ‘less a coherent philosophical school or movement than a philosophical family – often a contentions family – of thinkers holding distinct if related positions on the “workmanlike” nature of knowledge, meaning and truth’. Contemporary neopragmatism can be described in similar terms. Still there are a few key commitments, shared by neopragmatists as diverse as Robert Brandom, Richard Rorty, and Huw Price. One commitment is that neopragmatism is a fully general approach to language. True, neopragmatists typically direct their attention to the sorts of vocabularies that have historically worried naturalistically-minded philosophers. These vocabularies include the normative, the numerical, and the necessary, just to take three examples from the ‘N’s. But the general neopragmatist idea is that the same practical, use-focused approach is applicable to all our linguistic practices – including those in which we make assertions about ordinary objects like oranges or ovens.Footnote ⁸ A further commitment is that for no linguistic practice do we need to understand the relation between the talk itself and the objects of that talk in terms of any substantive relation of reference or truth. We can refer to numbers in just the same sense of ‘refer’ as we can refer to physical objects. And the claim that murder-for-hire is wrong is true in just the same sense of ‘true’ as is the claim that the sun is mostly hydrogen. These senses will, of course, be deflationary, though the proper sort of deflation is not a matter of agreement among neopragmatists.Footnote ⁹

The type of neopragmatism I will be adopting in this paper is the naturalistic version advocated by Huw Price, and in what follows, ‘neopragmatism’ should be understood in that way.Footnote ¹⁰ Price himself often describes his view as a sort of big brother to more well-known, but more restricted, expressivist or quasi-realist views. But local expressivists or quasi-realists contrast what they regard as ‘genuinely representational’ talk with the sort of talk at which they direct most of their explanatory efforts. That is, as Robert Kraut puts it, they endorse a ‘bifurcation’ thesis.Footnote ¹¹ Neopragmatists reject bifurcationism, largely because they do not see any explanatory role, in a theory of how language works, for a notion of ‘genuine representation’. Of course – and as already mentioned – neopragmatists are perfectly content with our everyday use of the language of representation, reference, truth, and falsity. It’s just that they give deflationary, neopragmatist explanations of this sort of talk. For example, to say that ‘x’ refers to y may simply be another way of saying that x is y: a way that conveniently allows quantified sentences such as ‘All these terms here refer to the same object’.

The neopragmatist’s linguistically-focused strategy for dissolving philosophical worries is directed primarily at two sorts of philosophers. The first are those who hold that philosophically perplexing entities or properties – the numerical, the normative, and so on – must either be located in the natural world or regarded as fictions. The second sort are those who feel no compunction about locating these things in a non-natural realm – a realm to which our access is quite mysterious, though somehow also quite reliable. Neopragmatists hold that all forms of language are to be understood as social practices with marks, sounds, and signs, and that no substantive notion of reference to external things is required in order to explain these practices. Again, that goes as much for talk of gold as it does for talk of goodness. Of course, in the case of gold, our talk may be explained partly in terms of a causal relation to a certain substance. But it would be a mistake, according to the neopragmatist, to identify any such causal relation with reference. We can refer to numbers too, and to unrealized possibilities. And there is no question about causal relations to those things.

This is not the place to defend neopragmatism in general. Nor is it the place to defend Price’s version, which forms the background of the present paper. But it is important to see what it amounts to and how one might motivate it. The principal motivation is simply the naturalistic impulse: the twin thoughts that (i) natural science, broadly understood, is the right place to find explanations for all of the things that happen in the natural world, and (ii) human language is a naturalistic phenomenon, just as much as photosynthesis or the dances of bees. One insufficiently broad application of this thought ends up landing us with the view that all the genuine objects and properties about which we can talk must be locatable in the natural world. But this view simply assumes, in an a priori way that is at odds with its professed naturalism, that language must work in a certain way: by enabling reference – understood as a substantial relation between words and things – in such a way that we can represent the arrangements of things in the natural world. Price (Reference Huw2013, pp. 5, 14–15, 21) calls this view object naturalism, and opposes it to his own subject naturalism. A neopragmatism that incorporates subject naturalism takes language to be a naturalistic phenomenon: a social practice that enables group behaviours of various sorts. And it takes talk of truth and reference – or, better, use of words like ‘true’ and ‘refers’ – to be understood in the same practice-based way.Footnote ¹²

One worry about neopragmatism stems from the fact that it regards our concepts as the result of contingencies not only in human nature, but in the actual cultural evolutionary processes that yield our linguistic practices. If this is right, it seems we would have very little right to suppose that these concepts will ever fit together in any neat way. And even if they do, perhaps they will cease to do so next week. In some cases, I think this is right – though I think it is often a very good thing to recognize this. Recognizing that racial categories have changed in this way in response to local political dynamics is a nice antidote to the wrongheaded thought that they correspond to anything on which one could even try to base the sorts of normative views that underly racist attitudes and policies. Still, there are also more enduring features of human nature, and of the world we live in, so that when certain concepts arise at all, they arise in certain forms. A simple example of this comes from the realm of colour. Not all communities have the concepts of red and green. But whenever they do, those concepts are mutually exclusive. Moreover, we can even say, truly, that it is a necessary truth that any surface that is uniformly red has no green in it. This is not to confuse merely universal truth with necessity. Rather, it is the effect of features of the world on the shape our concepts take. I will return to this point in the conclusion. But for now let me return to the main line of the paper, and a case in which I think our concepts do not fit together in any neat way.

3. Morality and Prudence: No Neat Fit

In order to set the stage for a discussion of morality and prudence let me first make some remarks about the relation between a utilitarian’s evaluation of action and their evaluation of desire, since a clear view of the lack of fit within this sort of mainstream moral theory should make my later claims about morality and prudence easier to see and accept. In ‘Motive Utilitarianism’, Robert Adams (Reference Robert1976) pointed out that, for a utilitarian, there is a question about the proper level of focus. It is simply not possible for a single agent to have the optimific motives and also always perform the optimific acts. So, a utilitarian cannot have it all, in the sense of having it be possible for a single agent to be optimific in all respects. Even within a single domain – morality – one sort of theoretical neatness looks to be impossible.

As Adams (Reference Robert1976, pp. 472–3) suggests, a desire for clean theoretical lines might push someone to hold that the optimific motive is simply the disposition to perform optimific acts. Then both sorts of optimality would line up neatly. But on reflection, it is not a surprise that the motives that have optimific consequences do not always lead to actions that have optimific consequences. The mistake of thinking that they would is in some ways similar to the fallacy of composition. The fact that an act is causally produced by a certain motive is no reason to think that the act will have the same properties as the motive. A motive that is optimific, in comparison with other motives an agent might have, may yield some acts that are not optimific, in comparison with other acts that the agent might perform. On reflection it is not hard to see this. Still, what gives Adams’ paper its interest is, in part, the existence of the same sort of temptation that makes the fallacy of composition count as a fallacy, rather than a totally bizarre leap; it is a faulty move that we are sometimes tempted to make.

Now let us move on to the main focus of this section: the relation between prudence and morality. The notion of prudence at issue is the everyday one, about which we might be expected to have reliable intuitions. That is, I am in agreement with James Brown (Reference James2023, p. 395) that although ‘some might wish to define prudence in terms of some theoretical role it plays, its proper home is in ordinary ethical and practical thought and discussion’. In that context, it is not a matter of maximizing anything or a matter of the shape and quality of an agent’s whole life. Nor is it the Aristotelian notion of practical wisdom, despite the fact that ‘φρόνησις’ (phrónēsis) is sometimes translated as ‘prudence’. Keeping this ordinary notion in mind, reflection on Adams’ explanation of a lack of fit between act- and motive-utilitarianism should make it easier to see a similar explanation at work when we consider morality and prudence. Why is that? Here are two reasons.

First, despite disputes about the precise content of commonsense morality, there is widespread agreement that it is nicely modelled as a sort of legal code, containing a fairly standard set of prima facie duties: don’t lie, don’t cheat, don’t steal, don’t injure, don’t kill, keep your promises, and so on. This is why contractualism is attractive to many people, why rule-consequentialism is so much closer to commonsense morality than is act-consequentialism, and why it makes sense to think of immoral action as deserving something like punishment. But when it comes to prudence, the legal model is nowhere near as attractive. There is nothing like a standard set of prima facie requirements of prudence. Nor is the role of prudence to manage interpersonal interactions. It is more focused on the interests of the agent.

The second reason has to do with the primary purposes of prudential talk, as against those of moral talk. What are those purposes? The neopragmatist will look for a social purpose in both cases. This is because language, being a social practice, is – quite generally – to be explained from the social point of view. In the case of prudence, the suggestion I would like to make is related to Stephen Darwall’s (Reference Stephen2002) theory of welfare in terms of rational care. The idea, transposed from welfare to prudence, is that our talk about prudence is more closely related to the interpersonal activity of advising someone we care about than it is to purely individual decision-making. This gets at the social nature of prudential discourse. The point of moral discourse is, of course, also social. But it is still quite different. Morality functions largely to minimize harms in a society. It does this partly by forcing us to consider the welfare of other people by connecting moral infractions to liability to punishment or blame. And our interest in having people respect the rules of morality are sufficiently widespread that we are willing to endow political authorities with the power to mete out that punishment for sufficiently significant infractions. But, insofar as we are rational, we do not need to be forced to consider our own interests via liability to punishment. Nor do other people have the same worries about us failing to consider our own interests as they have about us failing to consider theirs. Relatedly we do not need to worry about the attitudes of our future selves to our past selves, since conflicts between these selves can never come to blows. Finally, the notion of consent is not even relevant to judgments of prudence, since one’s future self can never complain that they failed to give consent. For all of these reasons, we should never have expected prudential talk to have the same sort of structure as moral talk.

The main problem prudential talk seems to serve is to correct is a kind of lack of perspective – the result of a strong emotion or temptation – that can be mitigated from an external viewpoint. Being prudent in such situations is a matter of getting one’s priorities right. As a result, it is not surprising that prudence has less to do with rules and more to do with appropriate attitudes – more do to, that is, with caring about the right things to the right degrees at the right times. This is one reason to think of it as belonging more to the sphere of character than the sphere of action. Of course we can, and often do, focus our advice on particular actions. But the recipe for happiness – or, perhaps better, for avoiding unhappiness – has more to do with being the right sort of person than with obeying certain rules. And there is another reason to think of prudence more on the model of character: it often does seem to have to do with the consequences of policies and habits, rather than the consequences of a particular action. The paradigm of prudence is the ant who saves, even though each decision of the fun-loving grasshopper seems perfectly rational when considered in isolation. Finally, if Dawall is right, and advice to someone we care about is a good heuristic for determining what is prudent, then we should be focusing on overall character. We are, after all, advising a temporally-extended person. We are trying to help them – not just to prevent certain kinds of actions.Footnote ¹³

Of course I do not want to deny that there is more to morality than rules and punishment. There are also moral virtues. Nor do I want to deny that we sometimes punish people for being imprudent. But both of these points are perfectly consistent with the contrasts I am trying to draw. One standard way of thinking of moral virtue is Kant’s: strength of the will in the performance of one’s duty. If this is right, our interest in promoting the virtues is that they make people more reliable in acting in the ways required by morality. So, on this way of thinking, moral virtue is conceptually secondary, dependent on the prior notion of duty. As to the connection between prudence and punishment, the primary case of punishment for imprudent behaviour is the punishment of children by their parents. But it should be obvious that the point of that sort of punishment is very different from the point of the punishment that is merited for moral transgressions. It is part of the training of the young person and is intended to help them develop certain sorts of habits and attitudes for their own good. This latter fact helps explain why it is only imposed by a small group of people – typically parents. It would be completely inappropriate for anyone else – much less some kind of political authority – to punish someone for even a drastic failure of prudence. And it is inappropriate even for parents, once their children have become adults.

If I am right that our normal talk of prudence is better explained on the model of a virtue theory than as a quasi-legal set of norms, it follows that we should not expect a neatly specifiable relation between moral status and prudential status. We already saw, at the very beginning of this section, that even within the moral domain, there will be a lack of neat fit between optimific acts and optimific desires. That was because acts and desires are very different things, and their causal effects are a contingent matter. We should never have expected them to line up neatly. But when it comes to prudence and morality, things are even messier, and any expectation of a neatly specifiable relation is correspondingly less reasonable. Now let’s move on to a case in which it is more plausible that we might find a neat fit.

4. Rationality and Morality: Explicable Fit

In the case of rationality and morality, it is easy to see the attractions of the view that there will be at least some interesting necessary connection between the two normative systems. A comparatively weak version of such a connection is that morally required action is always rationally permissible, though not always rationally required. We can call this thesis ‘moral permissibilism’. The extreme plausibility of this comparatively weak thesis is associated with the idea that it would be theoretically disastrous to have morality and rationality pointing in opposite directions. After all, calling an action ‘irrational’ is, in part, saying that there is insufficient reason to perform it. And many people will think it is absurd to say this about an action that they also take to be morally required. A comparatively stronger version of the rationality/morality connection is that all morally required action is rationally required. We can call this thesis ‘moral rationalism’. The temptations to this view may stem from the optimistic idea that there is – so far undiscovered – some argument that would persuade all rational people to act morally.

In laying out the two ‘neat’ positions of moral permissibilism and moral rationalism, I have relied on intuitive understandings of both rationality and morality. This might, rightly, be a cause of some concern. Admittedly, most people have sufficiently strong pre-theoretical intuitions about morality to justify the thought that we know what we are talking about, even if we have important substantive disagreements about what, in particular, morality requires. But rationality is a different matter altogether. Unlike the everyday notion of prudence, rationality is a technical notion. Because of this, it is much easier to endorse, unreflectively, one of the neat views of what it amounts to. A notion of rationality unconstrained by firm shared intuitions can take on many forms, and neatness itself may seem like an attractive feature. For example, one might recognize the plausibility (indeed, the truth) of the claim that there are reasons to avoid harming other people, and – in the absence of pre-theoretical constraints – one might also be attracted to a formal understanding of rationality as a matter of responding appropriately to reasons. This in turn makes it easier to theorize one’s way to the view that rational people will respond impartially to reasons, and from there to the idea that moral requirements are a subclass of rational requirements: moral rationalism

What is completely unclear is what precisely the import of moral rationalism would be, if we took rationality to be a matter of responding to reasons in an impartial way. That is, it is completely unclear what practical import assessments of rationality – in that technical impartial sense – would have. Is being irrational, in that sense, connected in any way to mental illness? What is the connection between irrationality and mitigation of responsibility, on such a view? Can it ‘make sense’ to act irrationally, if the personal benefits are very great? Most generally: why care about the rationality of one’s own actions, or the actions of others, at all? Similar remarks go for the intuitively plausible view that what is rational to do depends on one’s contingent desires. The impulse to neatness can lead from there to the idea that rationality is a matter of maximizing the satisfaction of one’s preferences. But again, it then becomes very unclear what practical import assessments of rationality will have. Should one always act rationally? What if one’s preferences are wildly immoral, or self-destructive, or both?

The neopragmatist, whose theoretical target is the genealogy of our linguistic practices, is unlikely simply to focus on the actual word ‘irrational’, precisely because it is a technical term that admits of various interpretations. Rather, the focus will be on the everyday notions that a (responsible) philosopher might use in order to explain the importance of whatever particular technical sense of ‘irrational’ they have in mind. What might that sense be? Philosophers use the technical notion of rationality to do a number of things. One of them is to limit the sort of agents we need to consider in certain thought experiments. For example, when we think about hypothetical agreements that are intended to shed light on our moral system, or on the notion of justice, we have to limit the parties to the agreement in certain intuitive ways. That is, we do not want to have to take into account people whose overwhelming desire is simply to obstruct agreements no matter what the cost, or who desire that any potential agreement be extremely harmful to everyone, including themselves, or who cannot connect means to ends. What we want are what we will call rational parties, and we will explain this notion by reference to ordinary notions of sanity, or making sense. That is, the choices of such people should be ones that we can see make sense, even if we ourselves might not make the same choice.

Philosophers also use the technical notion of rationality in other contexts for other reasons. There is a sense of rationality on which being irrational mitigates moral responsibility. And there is sense that is relevant to classifying mental conditions as illnesses, rather than simply as deviations from the norm. And, finally, there is an important sense of irrationality that is meant to function as the end of the road for practical argument. By this I mean that if one shows someone that their intended action is irrational in this last sense, this should have the same sort of effect as showing them that a belief they hold is self-contradictory: reasoned argument has come to an end. Each of these practical consequences of irrationality help fix a notion that actually matters, so that the corresponding versions of moral permissibilism or moral rationalism would be correspondingly important. It is an interesting question, whether there is one notion of irrationality that underlies all of the roles described in the previous two paragraphs. But we need not settle that here.

One notable feature of rationality – no matter which of the theoretical roles just described one emphasizes – is that it give a special status to reasons of self-interest while allowing other-regarding considerations to mitigate a charge of irrationality. For example, the sort of agent important to the moral or political theorist’s thought experiments cannot be someone who is indifferent to their own welfare. But they may (or may not) care enough about other people to make some sacrifices for their sake. And the sort of irrationality connected to mental illness, and that undermines moral responsibility, has much more to do with self-harming behaviour than with mean or malicious behaviour. But, again, benefits to other people can make sense of behaviour that harms the agent. Finally, the pattern is repeated when we consider irrationality in the ‘argument stopper’ sense. If someone uses ‘But that would be immoral’ as an attempted argument-stopper, there is a perfectly comprehensible rejoinder: ‘Yes, but I just don’t care about those other people’. On the other hand, ‘But that would be irrational’ does not admit of any such response, strongly suggesting that irrational action – in the relevant sense – goes against the interests of the agent.

If one does have a genuinely practical role for a notion of rationality in mind, it is possible to mount arguments for or against the various neat accounts of it that are on offer. But if one does not explicitly consider the practical role, it is correspondingly possible to be drawn to artificially neat accounts. Consider, for example, the following pair of claims by Michael Smith (Reference Smith2013, pp. 12–13), as he criticizes a purely instrumental view of practical rationality:

1. (1) [I]t now looms as a serious possibility that people may be morally required to act in certain ways even though we could rationally criticize them for acting in those ways.

2. (2) Absent skepticism about both moral requirements and moral responsibility, it seems that moral requirements must, in some way, reduce to rational requirements.Footnote ¹⁴

Now, the force of the worry expressed in (1) implicitly depends on the argument-stopper view of irrationality. That is, (1) expresses the idea that there should not be unassailable arguments against acting in morally required ways. But all this idea actually entails is that morally required action should not be irrational. That is, it should be rationally permissible. That is moral permissibilism, not the moral rationalism expressed by (2). But because Smith does not work out the actual practical shape of the argument-stopper view of rationality, he does not see this and leaps directly from (1) to (2).

Let me summarize the point I mean to be making here. There are two neat views of the relation between morality and rationality: moral permissibilism and moral rationalism. Any view that is not neat in either of these ways will allow for moral requirements that are irrational. Since this possibility offends against what seem like strong first-order intuitions about morality and rationality, there is at least some reason to think one of the neat views may be correct. Of course, at this point the neopragmatist may suggest that the supposedly first-order intuitions are not in fact first-order, but are expressions of that same old traditional philosophical outlook: the one that focuses on metaphysics and conceptual analysis and is too easily persuaded to adopt neat views. But that suggestion needs to be vindicated. And in fact, as I will now argue, there are neopragmatist-friendly reasons for thinking that one of the neat views is correct: moral permissibilism.

Here is a sketch of a neopragmatist argument in favour of moral permissibilism, where the relevant notion of rationality is the argument-stopper one. To describe an action as irrational, in this sense, is to say that nothing can be said in its favour that would make sense of sincerely recommending it. Now, when an action is morally required there are things that can be said that make sense of sincerely recommending it. This is true whether or not these reasons would always be sufficient to move the person to whom they were addressed. After all, we can often understand perfectly well why someone might choose to act contrary to our sincere recommendation: it happens all the time.Footnote ¹⁵ That the reasons we can offer in favour of any morally required action are sufficient to make sense of recommending it is guaranteed by a plausible genealogy of moral practices. Morality is a public system: it functions as it does because – at least in most everyday cases – people know what is morally required, and know that other people know this as well. But we cannot expect people to behave in ways that are irrational in the relevant sense – in ways, that it, that no one could sincerely recommend. So when moral requirements get in the way of what we would like to do, or when they involve some sacrifice, it nevertheless has to be the case that the benefits to others are at least sufficient to underwrite a sincere – but perhaps ineffective – recommendation to do it anyway. That’s enough, by itself, to underwrite moral permissibilism.

What about going further, to moral rationalism? In examining this question, neopragmatists will look at plausible genealogies of our moral thought and talk. Such genealogies will emphasize the social role of these linguistic practices, and the way they function to control certain kinds of harmful behavior via threats of punishment. The connection between moral requirements and punishment is obvious when one looks at actual practices, though it sometimes disappears from view in philosophical theories. Still, it is implicit in many contractualist theories. And even Mill (Reference John and Sher2002 [1861], p. 48), the arch-utilitarian, gives it a central role. He distinguishes the most basic sense of ‘ought’, according to which one ought to perform the act that maximizes utility, from a derivative sense, expressed by ‘morally ought’, according to which to say that one morally ought to perform an action is shorthand for saying that failure to perform that action ought – in the more basic sense – to be punished. If there is an important role for liability to punishment in the aetiology of our moral concepts, the best explanation seems to be that it is needed to dissuade. That is, the bare facts to which we’d appeal to describe an immoral action itself – a description of what the agent is doing, such as lying or hurting someone – will not be enough, in many cases, to dissuade a rational agent from performing it. So, when a rational agent believes that the possibility of detection and punishment are sufficiently low, or that the personal benefits sufficiently high, they may well act immorally. If that is right, moral rationalism is false.

The upshot here is that the neopragmatist can indeed defend a neat theory of the connection between morality and rationality: moral permissibilism. The defence requires that rationality be taken in the argument-stopper sense. But this does nothing to undermine its neatness. And the content of the neopragmatist’s version of moral permissibilism is clearer than many other neat theories, since the sense of ‘rational’ is a clear and useful one. Moreover, the neopragmatist’s genealogical method not only underwrites an argument in favor of moral permissibilism, it also provides reasons to think that the stronger thesis – moral rationalism – is false. Finally, it may be worth mentioning that while moral permissibilism has not been totally invisible in the literature, the only view generally opposed to moral rationalism is its simple denial.Footnote ¹⁶ But the denial of moral rationalism encompasses two importantly different views. One – moral permissibilism – we have seen. The other – hinted at in our discussion of Smith’s discussion of moral rationalism – is one we might call ‘moral anti-rationalism’. This is the view that on at least some occasions it would be irrational to do what morality requires. That seems to be a theoretically disastrous view. The neopragmatist’s twin arguments allows us to avoid falling into moral anti-rationalism even if we reject moral rationalism.