Motives, Manners, Reasons and Consequences

Robert Audi’s Of Moral Conduct  (2023) is an unusually substantial, wide-ranging and ambitious work of moral theory. The book has four parts. In Part I, Audi sets out the case for thinking of moral conduct as made up of three main dimensions: what is done, why it is done (its motivation) and how it is done (its manner). The first of these dimensions – the action itself – is the target of moral obligation; the second determines evaluations of the agent; and the third has both kinds of moral relevance (18–19, 31–37, 51–54). Part II sets out a substantive theory of moral obligation, enumerating a set of principles of moral obligation, discussing their application and explaining their relationship to each other. Part III supplies an account of the epistemology and metaphysics of morality covering those principles of moral obligation, but also singular moral judgements, understood as exercises of moral perception. Part IV provides a theory of reasons for action, a theory of value and finally an overall theory of right action, ‘consequencism’, which seeks to combine and reconcile the insights of both consequentialism and deontology. The book is an impressive achievement: few philosophers are equipped to attempt work in moral theory on this scale, let alone to carry it out with such uniform attention to detail – drawing as Audi does on his other work in the philosophy of action, the philosophy of practical reason and epistemology, as well as his earlier Kant-, Ross- and Aristotle-inspired contributions to moral theory.

A critical commentary of this length therefore has to be highly selective. In what follows, I comment on four of the topics covered in Parts I and III: the distinction between action and motive, the distinction between action and manner, the treatment of reasons for action and the book’s ‘consequencist’ normative moral theory.

1. Action and motivation

The book’s framing thought is that conduct has ‘a three-M structure determined by the matter of the “central” action, its motivation, and its manner’ (39). Each of these dimensions can be morally significant, and a fully adequate treatment of morality must be attentive to all three. This seems importantly right. In particular, as Audi says, the moral significance of the manner in which an action is performed is an under-discussed topic in moral theory, and his book makes a valuable contribution to correcting that.

This framing raises the following pair of questions, which Part I is concerned with addressing. First, what are the criteria for distinguishing action, motive and manner? After all, when you do something it can usually be described in different ways: describing an action as ‘harming’ or ‘helping’ leaves the motive and manner unspecified, but describing the same action as ‘revenge’ specifies a motive, and describing it as ‘kindness’ specifies both a motive (concern for welfare) and manner (gentleness). So, given a particular token action, is there any stable way of distinguishing the action that is performed from the accompanying motive and manner? Or is there always a plurality of act types to which any token action belongs, some of which can be motive- or manner-involving, while others are not? The second question is: once we have supplied a criterion for making the distinction, what moral significance does it have?

We can start with the first question, as applied to the action/motive distinction. Audi accepts that act descriptions can be motive-involving, but he insists that we should not let this obscure the difference between what is done and why it is done (26). This distinction, he maintains, is delivered to us by the theory of action. What distinguishes an action from a mere bodily movement is that there must be some end towards which control is exercised by the agent in doing what they do (14). One can then distinguish between, on the one hand, the end towards which the exercise of agential control makes the agent’s behaviour an intentional action and, on the other, the agent’s reason for pursuing that end. This reason might be that the action enables the attainment of some further end lying causally downstream, or perhaps because the proximal end – the end that makes the agent’s behaviour into an action – is seen by the agent as being itself good or as being the realization of some other good (as when an agent performs the action of harming from the motive of revenge).

So that supplies an answer to the first question, about how to draw the action/motivation distinction. Although motives do generate further act descriptions, we can still distinguish what is done from the agent’s motives for doing it by reference to the end with respect to which the agent’s exercise of control makes their behaviour into an action. Since there is a difference between the agent’s having that end and the reason for which they have it, that generates a criterion for distinguishing between what is done and why it is done.

That works well as a strategy for answering the first question. However, with this in hand, let us turn to the second. Having made the distinction, what is its moral significance? The answer presented in Chapter 2 is that action (as distinguished in this way) is the target of obligation, whereas motivation determines creditworthiness (31–32). This is an updated version of the distinction Ross makes between an ‘act’, which is ‘the thing done, the initiation of change’, and an ‘action’, which is doing something ‘from a certain motive’ – with acts being evaluable as right or wrong, but actions being evaluable as good or bad (1930: 7). Motives, since they are not subject to one’s own voluntary control and therefore not proper objects of an agent’s deliberation, are not candidates for obligation. Audi refines this Rossian view by saying that what is obligatory is an act type, whereas what is creditworthy is a token action, with the creditworthiness being dependent on the quality of the agent’s motivation (32).

This attractive picture seems right in many cases. If you save a drowning child from a pond solely out of the desire to get a reward, you could succeed in discharging an obligation (to perform an act of the type: saving the child) although your bad motivation – your defective quality of will – deprives your token action of any creditworthiness.

However, there are apparent exceptions to the general thesis that obligation is determined by motive-independent act types and only creditworthiness by motivation (Sverdlik 1996). Some of those apparent exceptions, it is true, can be handled by distinguishing motives from intentions. For example, it might be thought that criminal attempts provide counterexamples to Audi’s view: my harmless action of striking a match can become morally wrong if my reason for striking it is to burn down the building. Here, though, Audi can invoke his criterion for distinguishing action from motive – appealing to the end towards which the agential control is exercised – and say that the agent’s guiding intention determines what is done: attempting to burn down the building is something that there is an obligation not to do, irrespective of one’s motives. Intentions can be subject to obligations; motives cannot.

However, other cases are not so easily handled. Even when we carefully distinguish motives from intentions, there are cases where it is hard to deny that obligations are motive-dependent. Cases of wrongful discrimination provide compelling examples of this.¹ Suppose that two job applicants, Alice and Burt, are equally well qualified: since neither applicant has a greater claim to the job, it would be permissible to offer the job to either applicant and reject the other application. However, if Alice is White and Burt is Black, and the employer rejects Burt’s application out of racist animosity, the employer wrongs Burt. To do that is to subject Burt to racist job discrimination, and the employer has an obligation not to do that. However, this implies that the obligation, in this case, is an obligation not to act from a certain (racist) motive.

It is an interesting further question why we find this variation – why motivation is often relevant only to creditworthiness but not to obligation, whereas sometimes it is relevant to obligation. Scanlon’s treatment of the discrimination case is interesting in this connection (2008: 69–74). Although, in general, he maintains that the reasons for which an agent acts do not affect the permissibility of the action, he recognizes cases of discrimination as an exception to the generalization. To explain this, he appeals to the way in which an agent’s willingness to act for the reasons he or she does can carry a significance for the agent and others, and that because of this, racist motives can contribute to sustaining an unjust practice of publicly marking the members of a group as inferior (73). Further questions can be asked about this: just how is it that the agent’s motive itself, rather than the publicly accessible features of the action (whatever the motives for performing it), make that contribution? However, while that gives us further explaining to do, it does not call the phenomenon itself into doubt. The employer’s motives do evidently make his action into racist employment discrimination, and thereby wrong Burt. He can permissibly reject Burt’s application, but only as long as he does not do so from racist motives. While he may not be able to choose his motive at will, he has an obligation not to act from that motive. If he faces this decision and finds himself with racist motivations, he should toss a coin.

If so, cases of this sort pose a problem for Audi’s answer to the second question about the action/motivation distinction: the question of its moral significance. The action/motivation distinction can be drawn along the lines he proposes, but it does not align with the obligation/creditworthiness distinction as neatly as he claims.

2. Action and manner

Turning now to the distinction between action and manner, it is Audi’s treatment of the first question – what criterion is there for making the distinction? – that invites resistance. This time, the question cannot be answered by simply adapting his criterion for distinguishing action from motive. Actions, we saw, can be distinguished from motives by appealing to the ends towards which the agent’s exercise of control makes their behaviour an intentional action: this works, since motives are not part of the content of those ends. However, this will not help to distinguish action from manner. Manner can be part of the content of those ends. Agential control can be exercised not just towards helping but also towards helping gently – that can be the whole point of helping, in some circumstances. You might try saying: what makes helping the action is simply that helping is an act type that can be performed in more than one manner. However, that does not work: it is equally true of helping gently. Acts of that type can be performed with solicitous gentleness or with other forms of gentleness.

In light of this, a deflationary picture of the distinction between action and manner looks attractive. The non-adverbial names we have for act types sometimes specify a particular manner – ‘berate’, ‘soothe’, ‘scoff’ or ‘cheer’ – and sometimes do not – ‘keep your promise’, ‘tell the truth’, ‘help’ or ‘harm’. Manner-involving act types can then be subject to finer-grained further specification through adverbial modification: you could berate someone vehemently or not. There is no canonical way of distinguishing action from manner. Some names for act types specify a manner, and others do not; through progressively more fine-grained adverbial modification, we can specify a manner of acting more narrowly; and when we do, we are identifying a progressively more specific act type. Whenever A-ing is an act type, A-ing M-ly gives us a further act type, and it is more a matter of linguistic than moral significance whether or not our language happens to have a nonadverbial term to capture any given manner-specific act type.

In Chapter 3, Audi opposes this deflationary picture, maintaining that ‘it is at best unwise to regard adverbial modifiers as simply indicating a more “fine-grained” act-type’ (46). An argument he emphasizes is that to identify an act type is to attribute a property (berating, say) to the agent, whereas to identify a manner of acting is to attribute a property (say, vehemence) to the act token. However, that does not seem quite right. We get problems in cases where a single token action instantiates two very different act types. Take this case: in a single movement, a guest at a party turns away from Cedric to greet Delia. This could be a deliberate snub to Cedric as well as a welcome to Delia, and it could be an instance of both ignoring abruptly and greeting eagerly. We could say if we liked that the one act token has two different manner properties: it is both abrupt and eager. However, clearly, it is as an instance of ignoring that it is abrupt and as an instance of greeting that it is eager. The most accurate thing to say is that the manner of an action is a property of the tokening of an act type, and the manner qualifies the type in a way that generates a more specific act type that is tokened (here, both an abrupt ignoring and an eager greeting).

So it is not easy to see a convincing objection to the deflationary picture. An ascription of manner is always relative to a given act type, and it qualifies the type so as to generate a more specific act type. Furthermore, it is not clear that manner ascriptions have any kind of uniform moral significance. There are many act types (e.g. keeping a promise) with respect to which the manner in which the type is tokened can, when intentional, affect the ‘merit’ of the act token but not its permissibility. However, sometimes manner does affect permissibility. As Audi rightly emphasizes, there are obligations to treat others respectfully: the manner in which you criticize someone can turn your action into an insult that wrongs them. Moreover, the manner of action can affect whether one has discharged an obligation of concern, not respect. Cold conscientiousness is not sufficient to fulfill the obligations of parenthood.

This raises further interesting questions. Sometimes, behavioural manner is significant for obligation, sometimes it is significant for creditworthiness, and sometimes neither. When and why is that so? As far as I can see, the explanation for this has to come from elsewhere. We need a further theory of moral obligation that explains which act types (and therefore, which manner-involving act types) are obligatory and why. We also need a theory of the communicative content of action that explains why some forms of expressive disrespect (such as the case of discrimination described earlier or simple personal insults) are impermissible and are covered by what we can morally demand from each other, while others detract from the creditworthiness of action but are not strictly owed to each other.

Audi is right that we should not overlook the moral importance of the manner of action, but it can have different sorts of importance: there is no canonical way of distinguishing between what one does and the manner in which one does it, and that distinction itself is not morally significant.

3. Reasons and factivity

In the philosophy of practical reason, normative reasons are standardly thought of as facts. It is the fact that you will enjoy it that is a reason for you to go to the concert, and the fact that you were kindly helped that is a reason to express your gratitude. This leaves open a debate about whether the facts in question should be identified with true propositions (Setiya 2014) or states of the world (Dancy 2000: 114), but that is an intramural debate between two refinements of a now-standard view.

Audi challenges this orthodoxy in Chapter 10. He points out that ordinary reason reports often do not cite facts, but instead take an infinitival form (‘to save electricity’ or ‘to avoid spreading the virus’). This way of speaking is not confined to giving the motivating reasons with which we answer the question, ‘Why did you do that?’ It can also be used to give a normative reason in answering the question, ‘Why should I do that?’ – as addressed either to an informed adviser or to oneself in deliberating about what to do.

A stronger and a weaker thesis are available here. The weaker thesis is an equivalence thesis: propositional and infinitival reason reports are intersubstitutable. Audi argues for a stronger thesis: we should reject the ‘factivity view’ on which normative reasons are facts, because it is too narrow (221). The kind of case he emphasizes is one where an agent acts to achieve something that does not come about. The sun is shining in Sue’s face, so you close the blinds to prevent this, but the blinds are too thin so your action does not succeed. When you are asked, ‘What was your reason for closing the blinds?’ you can answer, ‘To prevent Sue from squinting’, thereby citing your reason for doing so, and your action’s being unsuccessful does not make it appropriate for an observer to say afterwards that you had no reason to do what you did. Your action was rational, so surely there was a normative reason for doing it, and not merely a motivating reason why you did it (220–21). However, that normative reason is not a fact.

This line of argument becomes hard to sustain, though, when we reflect on two important roles that normative reasons play in practical thought that apparently require thinking of reasons as facts. First, there is their role in prospective deliberation, which involves trying to figure out what good reasons there really are and then assess their relative strengths in coming to an overall conclusion about what you should do. When you try to identify what good reasons there are, you are trying to ascertain what is true: the object of your enquiry is directed outwards, towards ascertaining what state the world is in, not inwards, towards ascertaining what state your own mind is in. When, having identified competing reasons, you are trying to assess their respective strengths, it is hard to see how your thought can proceed without treating reasons as facts that stand in support of relations of different strengths to action. How can you weigh infinitives against each other?

Another role that normative reasons play is to serve as the content of advice. If I have previously tried closing the blinds in a similar situation and discovered that it does not work, I can advise you that what appears to you to count in favour of closing the blinds does not in fact count in favour of doing so, because what you take to be true is actually false. Using ‘ought’ in one way – the ‘ought’ of advice – I can express this by advising you that you ought not to close them.

In Audi’s example, the observer who comes along afterwards is doing a third thing: neither deliberating nor offering advice, but rather making an after-the-fact performance evaluation, assessing whether what you did in closing the blinds was rational. Judgements of rationality evaluate an agent’s responses to the world given the information available to them. When you are innocently ignorant, that does not make you irrational. However, your ignorance is ignorance about the considerations that really do count in favour of your action: that is why it makes sense to be concerned to rectify your ignorance when you find out about it. It is hard to see how to avoid thinking of those considerations – the ones that a rational agent can be innocently ignorant about – factively.

A further point is relevant to the blind-closing example. One plausible view is that the expectational properties of action can generate normative reasons. If I go for a wild joyride and injure no one, it remains true that what I did carried a risk of injuring someone, and that fact was a normative reason against doing what I did. A parallel claim can be made about your closing the blinds. Although your action did not turn out to prevent Sue from squinting, what you did increased the expectation of preventing Sue from squinting, and that fact was still a normative reason for acting as you did.

So there are two replies to Audi’s use of this kind of example to argue that normative reasons are not facts. To secure that conclusion, this needs to be a case in which there is a normative reason for acting, but no reason-constituting fact. However, first, it is plausible that there is a reason-constituting fact, and second, even if there was not, the retrospective evaluation of your action as rational should not lead us to infer that there really was a normative reason counting in favour of what you have done – the sort of reason that is the proper target of deliberation and advice.

4. Consequencism and consequentialism

Finally, we can consider the treatment of consequentialism and deontology in the book’s final chapter. The best elements of both, Audi argues, can be retained and combined in a view that he calls ‘consequencism’. What this view endorses in consequentialism is the idea that what we morally ought to do depends on ‘the value of what we stand to bring about’ (276), but it retains the deontological idea that ‘some act-types … are by their very nature wrong-making’ (269). The way it achieves this synthesis is by attaching moral disvalue to the performance of actions of certain types – lying, killing, promise-breaking and so on – and thereby allowing that, although the rightness of an action depends on the goodness of the state of affairs it produces, the moral value of the action itself is an important contributor to the value of that state of affairs, along with its further downstream causal consequences.

In filling out this view, Audi develops a value theory in which experiences are the bearers of basic intrinsic value, but items of many other kinds are recognized as ‘inherently’ valuable: persons, works of art, states of the natural world, just distributions, actions, policies and so on. These various items can be valuable non-relationally, in themselves, but their value is secondary to the intrinsic value of experience. Something’s inherent value is ‘its power to yield intrinsic value in an experience that is appropriately responsive to the right kind of focus on its intrinsic properties’ (242). A fine work of art’s inherent value is its power to elicit the intrinsically valuable experience of appreciating it. Actions of morally significant types have (positive or negative) inherent moral value: this value is inherited by the states of affairs that result when actions of those types are performed (since the state of affairs contains the action) and in this way contributes to determining the rightness or wrongness of the token actions. This gives us a value theory anchored in the attractive and aversive qualities of experience; those qualities metaphysically ground the intrinsic goodness or badness of experiences (232–33); items of other kinds acquire value through their power to furnish the contents of those experiences; states of affairs thereby come to possess intrinsic, inherent or instrumental value; reasons for action derive their normative authority from the value of the states of affairs realized by the action (227–28, 285); and the moral rightness or wrongness of an action is determined by the balance of value-dependent reasons for and against it (271).

Audi’s view can be located by contrasting it with other, more thoroughly non-consequentialist theories that also make the good prior to the right. Some forms of virtue ethics do this. If you say that what makes right actions right is that they are the actions that a virtuous person would perform, then you are making the goodness of persons prior to the rightness of actions. More broadly, there is what Raz (1999: 22) calls the ‘classical view’ that attributes goodness (as Audi does) to things of many different kinds – not just states of affairs and persons, but institutions, artworks, motives, parts of the natural environment, processes, and so on – sees the different kinds of goodness that these things have as corresponding to the different kinds of positive responses they fittingly evoke (promoting, respecting, honouring, protecting, appreciating, sustaining, nurturing), and then takes the right action to consist in making those fitting responses. Elizabeth Anderson’s ‘rational attitude theory’ is a prominent recent example of a theory with that value-first, non-consequentialist structure (1993: ch. 2).

However, what makes those views non-consequentialist is that they deny that the value of states of affairs alone determines the rightness of action. In this, they differ from Audi’s view. His ‘consequencism’ does take the rightness of action to be solely determined by the value of states of affairs, while insisting that the moral quality of the action itself is an important contributor to the value of a state of affairs, and therefore to the rightness of the action, alongside the nonmoral value of its causal consequences.

On its own, that would not be enough to prevent Audi’s view from being a form of standard consequentialism. Standard consequentialism holds that the rightness of an action is determined by its relationship to an impersonal ranking of states of affairs (in the simplest forms of act consequentialism, the relationship of producing a higher ranked state of affairs than any alternative). This allows for an axiology that assigns great disvalue to certain act types, which contribute negatively to the value of the state of affairs that will obtain if they are performed.

Audi’s view departs from standard consequentialism in his denial that acting morally rightly is ‘a matter of enhancing value impersonally conceived’ (272; emphasis added). This allows him to deliver deontological verdicts in cases such as lying to prevent two lies, torturing to prevent two tortures and so on – cases in which, no matter how much disvalue you attach to an act of a certain morally bad type, performing an act of that type could produce a state of affairs containing less of that disvalue, impersonally speaking. The moral badness of my action can still provide sufficient reasons for me not to perform the action, even though I would thereby be reducing the incidence of those actions.

Given this, the natural way to classify Audi’s moral theory is as a form of what is sometimes called relative-value consequentialism. It is a view of the same type as Douglas Portmore’s (2011) ‘commonsense consequentialism’, making the rightness of an action dependent on the value of the resulting state of affairs, while adding that those value assignments to states of affairs is relativized to the agent’s own perspective. However, this means that it faces the following sort of objection.²

Take a case in which I have some sort of inducement to torture Edwin, and I can see that doing so would be wrong. Now consider the question: what makes it right for me not to torture Edwin? A proponent of the classical view can appeal to the value of Edwin himself, in virtue of which it is fitting that I respond to him with various forms of respect, which include not torturing him. Relative-value consequentialism gives a different answer: it appeals instead to the value-relative-to-me of the state of affairs in which I do not torture Edwin. It says, it is because the world’s not containing my action of torture is good-relative-to-me that it is right for me not to torture Edwin. However, the worry is that that makes the determinants of rightness too self-regarding. It says that what matters morally, in a situation like this, is whether the world turns out to contain my performing a certain kind of action. It is more plausible to say that torturing Edwin is wrong because it is a failure to respect Edwin than that it is wrong because it makes the world bad (relative to me) by putting my torture into it.

Audi’s theory of inherent value gives him a reply to this line of objection. His view does attribute inherent value to persons. So he can say that the agent-relative badness of the state of affairs in which I torture Edwin derives, in turn, from the inherent value of Edwin. That is an attractive feature of his view. However, his consequencism still commits him to say that the reason I have not tortured Edwin derives from the comparative personal value of the two possible states of affairs in which I do or do not torture Edwin, and that does not seem to be why torturing Edwin is wrong. Consequencism’s explanation travels in the wrong direction. My torturing Edwin would be a bad feature of the state of affairs that contains it because that action is a failure to respond to the reasons there are not to torture Edwin. Consequencism says instead that the badness of the state of affairs containing the action is itself the source of the reason not to perform it: this gets things the wrong way around.

Those are some points on which central theses in Of Moral Conduct invite resistance. Nevertheless, this is a strength of the book, not a weakness. On these, as on many of the other topics that the book cohesively addresses, it has distinctive and thought-provoking theses to advance, provides a stimulus to further thought, opens new paths in moral theory and challenges other philosophers who address those topics to aspire to the same level of authority, breadth, perceptiveness and skill.

Footnotes

References