Book Review : Michael Smith, the Moral Problem

[Je publie ici un devoir d'anglais réalisé dans le cadre de mon master de philosophie l'an dernier : il s'agit d'une revue de "The Moral Problem" de Michael Smith. Dans la partie "discussion", je défend l'idée (inspirée par Smith) que les raisons normatives, impliquées dans les jugements moraux, portent sur des désirs de second ordre. L'objectivité des jugements moraux proviendrait de l'abstraction inhérente aux désirs de second ordre. Le texte est assez dense : le devoir était limité en taille...]

In `the moral problem', Smith discusses what he thinks to be the central problem in meta-ethics. He begins the first chapter by contrasting normative ethics, which seeks to establish some moral facts in specific situations, with meta-ethics, which concerns the very status of moral facts: do they exist, are they objective and how do they affect our actions? Smith is interested in what he believes to be the central problem of meta-ethics: the apparent incompatibility between the objectivity and practicality of moral facts. Moral facts seem objective facts which we can believe in, and at the same time, they seem to imply a motivation to act. However both these statements do not fit with a humean psychology, which makes a clear distinction between beliefs and desires, and following which motivation implies the existence of corresponding desires.

Smith expresses this problem in the form of the following three, apparently incompatible statements:

• Moral judgements express a subject's beliefs about an objective matter of fact about what is right.
• If someone judges that it is right that she does something, then, ceteris paribus, she is motivated to do it.
• An agent is motivated to act in a certain way just in case she has an appropriate desire and a means-end belief, where belief and desire are distinct.

Thus formalised, the trillemma allows for a classification of the different theories in meta-ethics. Expressivists reject (1): moral judgements are not beliefs, but expressions of desires. Externalists reject (2): there is no connection between moral judgements and motivation. Anti-humeans reject (3): there is no clear distinction between beliefs and desires. However, none of these accounts are convincing (or so Smith will argue).

Chapters 2 to 4 aim at arguing against each of these positions. In chapter 5 and 6, Smith presents his own position, which consists in an adaptation of humean psychology following which not only beliefs, but also desires are subject to rational criticism. Smith's view is based on a distinction between normative and motivating reasons: the latter are involved in action, while the former are involved in changing our desires.

I will briefly summarize each chapter, then discuss Smith's solution in more details.

Summary of the book

The second chapter is directed toward the expressivist challenge, and in particular the objection to the objectivity of moral facts formulated by Ayer. On the one hand, Ayer argues, if moral facts can be naturalised, for example, if they are reducible to approbation by an agent or by the population in general, or to what is pleasant or useful, then they are not necessarily good or bad, because natural facts are a priori neutral. On the other hand, if they are not natural, then they cannot be verified by experience. Ayer concludes that moral judgements are mere expressions of one's desires.

Ayer's objection against non-naturalism is based on the doctrine of positivism, which is now generally rejected, but nevertheless Smith believes that Ayer's argument can still be defended on the basis of the a priori supervenience of moral facts on natural facts. However he rejects Ayer's objection against naturalism. There can exist a synthetic, a posteriori definition or structural account of what is good based on natural facts and on platitudes about morality, in the same way that there can exist a definition of colours, although no natural fact (such as the reflectance of surfaces) is in itself `colourful'. Smith stresses that no such definition exists at the moment, but he hopes that the analysis he will provide in the following chapters will enlighten this point. His main argument for defending the objectivity of moral judgments is that we usually engage rational discussions about moral judgments as if they were matters of facts rather than mere subjective viewpoints.

The third chapter is directed toward the externalist challenge. Externalists, such as Brink, emphasise that amoral people can make good moral judgements, yet still fail to be motivated to act morally. According to the externalists, we cannot deny that these persons are rational. It follows that moral facts are not intrinsically prescriptive. Smith argues that on the contrary, amoral people are not rational and fail to make proper moral judgements: their failure to act in accordance with moral principles shows that they don't master morality.

Furthermore, externalists are unable to explain why normal people are motivated. One could say that normal people are motivated to do the good, whatever they think the good is, or to deny that our motivation to do the good is rational. For example, Foot claims that moral facts are akin to etiquette: they are institutionalised prescriptions. However, generally, our motivations are not de dicto but de re: we don't aim to do the good in general, but we aim to do what we think is good in particular cases and we generally expect that people follow moral principles in virtue of their rationality, not because we think they accepted these principles as a condition for belonging to a specific community, otherwise there would be no point in engaging in moral discussions.

The fourth chapter is directed toward the anti-humean challenge. Anti-humeans hold that a moral belief can be a source of motivation by itself, or that beliefs and desires are not clearly distinct, that there are `besires'. However Smith argues that desires and beliefs have a distinct direction of fit: a belief aim to fit the world, while a desire aim to change the world. Therefore they are incompatible.

According to Smith, one of the reason why some authors fail to see that humean psychology is correct is that they think of desires as phenomenological entities, which raises an objection: we do not always act according to our feelings. Smith argues that desires are not necessarily phenomenological. They are propositional and dispositional. Following a dispositional account of desires, no important difference can be found between desiring and having a goal, and humean psychology is obviously true, for being motivated is \emph{inter alia} having a goal.

In the fifth and sixth chapters, Smith intents to explain how he departs from humean psychology and develop his own view. For Hume, desires, contrarily to beliefs, are not subject to rational criticism. However both intentionality and deliberation over moral values are valid explanations to one's actions and some examples, such as that of an addicted person who does not value her addiction, show that valuing and desiring are distinct, although both relevant to motivation.

Some authors attempt to reduce values to particular modes of desiring, or to second order desires (desire to desire). The problem is that then values do not seem to imply rationality, although they are subject to rational deliberation. They seem to imply that the acceptance of values depends on the agent. However then values are not normative any more because there cannot be any genuine disagreement on them. It seems that values have a social component, akin to shared desires.

Smith proposes his own view of normative reasons: they are what our rational self counterpart would desire that we do. Someone who is rational will take into account these reasons, as if they were advices, and change her desires accordingly. The constraint of rationality implies that there should be a convergence on normative reasons among different persons. However Smiths admits that this definition of normative reasons is purely conceptual, and that nothing proves that normative reasons substantively exist, i.e. that such a convergence is indeed possible. However looking at history can enjoin us to think that moral progress exists (we abolished slavery) and that rightness could finally be found by debating.

Discussion

Smith's account of the `moral problem' is very useful, in that he offers a clarification of the debates in the form of a trillemma. His criticisms of each position appear to be carefully argued. However Smith himself admits that the objectivity of reasons is merely supported by the fact that we engage in moral debates as if there existed objective normative reasons to be found, which does not mean that there are. Moreover his definition of normative reasons is conceptual. It does not really throw light on what exactly `right' and `wrong' mean, but only tells us that it is the object of rational deliberation.

In the final analysis, the solution he provides appears to be a bit frustrating: it is merely an account of what normativity must be given the constraints of objectivity, practicality and humean psychology, but not a positive account of the meaning of rightness.

Another problem is that following Smith's definition, normative beliefs are still based on some desires: the desires hypothetically shared by rational agents. Maybe it is not clear how we should distinguish between the desires which pertain a specific situation, and which are the objects of rational deliberations, and the desires which drive these deliberations. One could wish to integrate these desires as objects of deliberation as well, but then what motivates us to deliberate? Following this view, rational deliberation could become empty, or merely a set of rational constraints, such as consistency, and only our shared desires would explain that a convergence occurs between rational agents. Normativity would be the result of a compromise, but not an objective matter of fact after all.

I think we can find a solution to these problems. The case of slavery is an interesting example. We abolished slavery in part on the basis of the fact that black people are no different from white people (or not in a relevant way): there is no rational reason to treat differently people who are basically the same. Of course, one could support slavery for both white and black people, but this could have the consequence that she would become a slave herself under certain circumstances. More generally speaking, promoting principles which rest on contingencies is not rational insofar as such principles could be turned against you.

In other words, rational discussion can indeed lead to the discovery of objective values shared by all rational agents if we take for granted that rational agents will try to find universal principles to guide their actions, and that in virtue of the constraints of sociability, these principles could be turned against them. These assumptions are quite well supported. Beliefs always have the particularity of being somehow universal: my belief that sugar dissolves in water supposedly apply to any piece of sugar. Generally speaking, beliefs are universal principles for guiding our actions efficiently, and so should be moral beliefs. The constraints of sociability are also well supported: in the final analysis, we all depend on each other, and a lot of our behaviours are based on communication, imitation and mutual agreement. We can suppose that moral beliefs are well suited for the kind of situations described by the prisoner's dilemma (where cooperation is desirable, but only at the condition that other people cooperate): the right would be a general way to reach an optimal situation in a society by mean of cooperation, regardless of specific desires.

Note that this view also provides a solution to the problem of supererogatory actions. If the desirability of cooperation is subordinated to other people's willing to cooperate, and if, in a complex society, cooperation is a matter of degree, then one's duty is to reach the average level of cooperation in the society one belongs. We should not expect people to do more, for cooperating more than the average in a prisoner's game is risky and thus not totally rational, but people should be grateful of such behaviour because it tends to increase the average level of cooperation in the society: it is supererogatory.

The problem is that this view slightly departs from Smith's solution to the moral problem. In this account, moral beliefs are merely mean-ends beliefs: they are what we believe is the best way to cope with sociability issues (including empathy and animosity). However then our motivation to do the good is derivative, not \emph{de re}, which, as Smith observes, does not seem to fit with the phenomenology of morality (although some studies show that we are more willing to act morally when we are observed and when we see others behave morally, which suggests that social optimality is indeed an important motivation). This problem can be overcame by combining the present view with some of Smith's insights.

The main import of Smith's solution is to provide a solution to the trillemma by accepting that beliefs and desires are related after all, in that normative beliefs can make an agent change her desires (this could be related to what Mc Dowell calls a `conversion'). One can observe that grown-up human beings do not only act according to purely natural desires (such as eating). They also have cultural, educated desires. These educated desires are probably based on more profound, innate desires, but it is not obvious that they are completely reducible to them and behavioural studies have shown that we can be conditioned in our dispositional behaviour. Smith provides a connection between beliefs and desires through rational deliberation which accounts for this fact, and which can be interpreted as self-conditioning.

Interestingly, the view that agents can change their desires could be combined with an hedonist conception of desires. Let us define a desire as something which fullfilment anticipation procures some pleasure to a subject. This definition does not contradict the fact that desires can be shaped during our maturation: obviously, having pleasure in solving mathematical problems is not natural. Changing one's desires could be seen as training oneself to have pleasure in such or such circumstances. Following this observation, altruistic people are not to be blamed because they take pleasure at what they are doing: their pleasure expectation might be a cause of their current behaviour, but what is admirable is not their current behaviour, but the fact that they shaped their desires in accordance to what is right. Similarly, someone who lacks motivation to do something good can still be blamed for not having shaped her desires in the past, although she has a good reason not to do the action in the present.

In other words, pleasure and suffering are not the central aim of morality, but instruments of and guides to morality, whose aim is social optimality. Blaming or prising someone is expressing a judgement on the moral intelligence of that person: someone is blamed for not having acquired a sufficient knowledge of the principles which promote social optimality and changed her desires accordingly.

This account is apparently circular, because being motivated to change one's desires is also required, which requires another desire. As we observed, it is not clear how to distinguish between the desires which pertain a specific situation and the desires which drive and motivate these deliberations, and thus it is not clear why moral principles are objective rather than relative to specific desires.

However we have good reasons to think that desires whose satisfaction involve changing other desires are abstract, higher-order desires and that they are less constrained by specific situations or specific traits of character, such as having such or such desire, which, precisely, we are ready to change, and therefore they are more prone to be shared by different agents independently of their individual characteristics, and to apply to a wider range of situations. If this is the case, then these higher order desires could be typically subject to sociability constraints rather than individual ones.

Let us take, for example, the higher-order desire not to be generally treated on the basis of contingencies. Such a desire does not depend on particular individual traits or desires, and is merely constrained by sociability. In particular, it can be considered a fact that not treating others on the basis of contingencies in general is a mean not to be treated by others on the basis of contingencies. Here, `in general' means that one should be disposed not to treat others on the basis of contingencies. Following a dispositional account of desires, we get the following moral principle: one should desire not to treat others on the basis of contingencies. If one is rational, one should believe that this principle is true.

In summary, the fact that higher-order desires require changes in our desires stems from their abstractness and their possible application to a wide range of situations, from which follows that their practical satisfaction involve dispositions rather than discrete actions. In this view, moral beliefs are mean-ends beliefs to achieve our higher-order abstract desires. They are objective because higher-order desires are more prone to be shared by all agents in virtue of their abstractness from specific first-order desires, and they imply motivation ceteris paribus because they generate or change the corresponding first-order desires.

It follows from these considerations that if normative reasons are based on moral facts and shared abstract desires, then normative reasons are merely informed higher-order motivating reasons applied to specific cases:

$p$ has a normative reason to $\phi$ just in case $p$, suitably informed, would have a motivating reason to desire to $\psi$ and $\phi$ is a mean to $\psi$.

For example, John has a normative reason not to hire a slave just in case John, suitably informed, would have a motivating reason to be generally disposed to treat people with respect, and if not hiring a slave is a mean to treat people with respect. A failure to be rational can stem from different issues: either John fails to know that being disposed to treat people with respect is a mean to be treated with respect in a social context, or he fails to know that not hiring a slave is a mean to treat someone with respect. The former issue concerns the general validity of moral principles, while the latter concerns its applicability to a particular case. Generally speaking, moral discussions can concern one or the other (John could argue either that he has no motivating reason to be disposed to treat others with respect, or that slavery is compatible with respect).

A third possibility is that John struggles with his desires and fail to change them. Perhaps his first order desire to enslave someone for his personal comfort is stronger than his higher-order desire to be treated with respect. This case is similar to the case of an addicted person who fails to get rid of her addiction, although she desires it. This case is problematic: is John rational, but his addiction to personal comfort constitutes an impediment to his behaving morally? Or is John irrational because he overlooks the long-term social benefits of treating others with respect?

A final possibility is that John does not desire to be treated with respect at all (or not very strongly). This case is problematic too: is John irrational in virtue of not having a commonly-shared abstract desire (perhaps in virtue of even more abstract normative reasons? But we might face an infinite regress here), or is he rational but does not really belong to the community of `normal' human beings? It might be the case that, as Foot has argued, normative reasons can sometimes be relative to our belonging to specific communities. If normative reasons cohabit at different levels, from individuals or small groups to humanity at large, this could lead to complex situations where they are in conflict, which would explain why moral values are at the same time diverse and subject to disagreement, why they lead to dilemma in their application to specific cases, and why nevertheless we often consider them to be objective.

Conclusion

As we have seen, `The Moral Problem' has many qualities. First, it clarifies the debate of meta-ethic by providing a clear formulation, from which the different positions can be classified. Second, it provides important arguments against each solutions: expressivism, externalism and anti-humean accounts. Finally, it provides a solution to the moral problem. It distinguishes between normative reasons and motivating reasons and provides a definition of normativity under the constraints of objectivity and practicability, and explains the practicality of normative reasons in terms of changing one's desires.

Smith does not provide a straightforward definition of rightness and his account is conceptual rather than substantive, as he admits himself. Moreover, the role of abstract, idealised desires in normativity, and how these are related to other desires, deserves clarification. However we have seen that such a clarification is possible in terms of second-order desires, which are abstracted from individual constraints, and therefore subject to sociability constraints only. Then possibly, the aim of morality is to reach a social optimum. We have seen also that shaping one's desire to achieve a higher-order desire can be interpreted as self conditioning, and how this relates to pleasure and suffering.