About the Author
Read an Excerpt
Cambridge University Press
0521809878 - Ethics and the A Priori - Selected Essays on Moral Psychology and Meta-Ethics - by Michael Smith
When we act, we act for reasons. It is easy to hear this as a truism or platitude. "Surely," it might be said, "what makes an action an action is the fact that it is something that someone does for a reason!" (Davidson 1963).
But in fact the claim that when we act, we act for reasons, is ambiguous. When interpreted in one way it is indeed a truism - all actions are things that people, or more generally animals, do for reasons - but, when interpreted in the other, it is no truism at all. Though some acts are done for reasons in this alternative sense, it isn't the case that all acts are done for reasons. Some people act because there is reason not to do what they do (Stocker 1979).
The claim that the term "reason" is ambiguous is, of course, familiar in the philosophical literature (Woods 1972; Smith 1987). On the one hand, talk of reasons is much the same as talk of causes. When we talk of reasons for action we thus sometimes have in mind the psychological states that teleologically and causally explain behaviour. This is the use of the word "reason" that is in play when I say that my reason for (say) tapping away on the keys of my laptop is that I want to write an introduction to my collection of essays and believe that something I can do - namely, tap away on the keys to my laptop - will lead to that outcome. It is also the use that is in play when we say that the cat's reason for meowing at the door in the morning is that she's hungry and wants some food. In earlier work I have called these "motivating reasons." Motivating reasons are psychological states that teleologically, and perhaps causally, explain behaviour: they are constituted (or so I say) by pairs of desires and means-end beliefs. On the other hand, however, when we talk of reasons for action we sometimes have in mind something completely different: considerations that (allegedly) justify. This is the use of the word "reason" that is in play when I say that my reason for tapping away on the keys to my laptop is that collections of essays require an introduction. A cat simply isn't sophisticated enough to have reasons in this sense. In earlier work I have called these "normative reasons." Normative reasons are propositions whose truth would justify acting in a certain way: they are (roughly speaking) facts about the desirability of so acting (or so I say).
If a distinction of this kind is along the right lines then the ambiguity in the claim that, when we act, we act for reasons, becomes readily intelligible. For what makes an action an action is the fact that there are certain motivating reasons that teleologically, and perhaps causally, explain an agent's doing something: actions are actions in virtue of their distinctive psychological pedigree. In this sense, the claim that when we act, we act for reasons, is indeed a truism. But it is a substantive truth about any particular action - an achievement on behalf of the agent of that action, an achievement that requires not just conceptual sophistication but also, perhaps, the possibility of conscious control - that there are normative reasons for doing what is done. For it is an achievement to act on a consideration that does, in fact, justify what is done. In this sense, the claim that when we act, we act for reasons, is no truism at all. It is a compliment that at most some of us deserve some of the time. Some of us are even so perverse as to be motivated to act in ways that we believe to be dysjustified, to use Michael Stocker's term (Stocker 2004).
The fact that there are sometimes, but not always, normative reasons for doing what we do, but always motivating reasons, raises several important questions about the nature of motivating and normative reasons and about the relationship between them. The task of the essays in the first part of this collection is to raise and answer some of these important questions.
Part One Moral Psychology
The first question is what, exactly, makes it the case that there is a normative reason to act in one way as opposed to another. In other words, what is the truth-maker of the claim that a certain consideration provides a justification for what an agent does?
This question is taken up in Chapter 1, "Internal Reasons." In Bernard Williams's famous paper "Internal and External Reasons," he argues, in effect, that what makes it the case that there is a normative reason to act in a certain way is the fact that so acting accords with certain of the agent's idealised desires, where the idealisation in question is a matter of the agent's desires conforming to certain principles of reason, and where the agent in question has the capacity to grasp this fact (Williams 1980; see also Pettit and Smith forthcoming). Though this is an intuitively attractive account of normative reasons - it explains, for example, why cats aren't capable of having or acting on normative reasons - once we remember that such idealised desires are supposed to make it the case that we have normative reason to act in a certain way in circumstances that may well be less-than-ideal, we see that there are two very different models of such idealised desires. Moreover we also see that these models aren't on a par in terms of plausibility.
Put colloquially, the first of these models amounts to the idea that what we have normative reason to do is to follow the advice of our idealised selves. Thus, according to the advice model, what I have normative reason to do in my less-than-ideal circumstances is a matter of what my idealised self, in his idealised circumstances, would desire me to do in my less-than-ideal circumstances. The second model amounts to the quite different idea that what we have normative reason to do is to follow the example set by our idealised selves. Thus, according to the example model, what I have normative reason to do in my less-than-ideal circumstances is a matter of what my idealised self wants himself to do in his ideal circumstances. One of the main aims of "Internal Reasons" is to show how implausible the example model is, and to show how its implausibility points us in the direction of the advice model.
A good deal of time is also spent revisiting Williams's own account of what he calls "internal" reasons. I ask whether his account of internal reasons provides us with a sound basis for an advice model of normative reasons. I argue that, duly amended and supplemented, it does. One main way in which it requires amendment is in the assumptions it makes about the transformative powers of the idealisation process. According to Williams, the desires any particular subject ends up having after we idealise her desires need not be the same as the desires other subjects end up with after we idealise their desires. But I argue that, in order to suppose that the objects of our idealised desires are capable of providing justifications, we cannot go along with Williams on this score. We must assume that subjects would all converge in the desires they have when we idealise, otherwise facts about the objects of our idealised desires would not be facts of the right kind to provide justifications. They would be too arbitrary. (Note that it is consistent with what I've just said that though this is how we conceive of normative reasons, there may well be no normative reasons, as there may be no objects of desire upon which subjects would converge if we were to idealise their desires. This issue is taken up again in various of the remaining essays, but especially in Chapter 12 and Chapter 14.)
Towards the end of "Internal Reasons" I argue that, when Williams's internal conception of reasons - that is, his conception of normative reasons in terms of idealised desires - is augmented and supplemented in the way I suggest, then we come to see one of its great attractions. For, I argue, this account enables us to see how and why our beliefs about our normative reasons are capable of both causing and rationalising our having corresponding desires, that is, our having corresponding motivating reasons. However when I wrote "Internal Reasons" I was still unclear in my own mind how exactly the account of normative reasons in terms of idealised desires enables us to do this. (I now think that the explanation offered in the paragraph that ends with footnote 9 is completely wrong, for example.) In subsequent work I revisited this explanation. I finally settled on a formulation in terms of coherence.
An idealised set of desires is, as I said above, a set of desires that conforms to all of the principles of reason that govern them. Let's say that when we have such a desire set that desire set is, inter alia, maximally coherent. In these terms, what allows us to explain the connection between our beliefs about our normative reasons, on the one hand, and our motivating reasons, on the other, is that coherence itself would seem to require that, if we believe that we would want that we ɸ in certain circumstances C if we had a maximally coherent desire set, then we desire that we ɸ in C. In other words, the coherence of our psychology is itself enhanced when we have desires that accord with our beliefs about what we would desire if we had a maximally coherent desire set. What explains the transition from the belief to the desire is thus none other than the capacity we have, as rational creatures, to have a coherent psychology. In a phrase, our being rational does all of the explaining.
This claim, which is a crucial premise in many of the arguments provided in the essays in this collection - indeed, as will become clear, both the claim and the argument for it (such as it is) is repeated over and over in many of the essays - has received a good deal of attention in its own right. I explicitly respond to two of these discussions in this collection. Chapter 2, "The Incoherence Argument," is an attempt to respond to Russ Schafer-Landau's criticisms of the argument (Schafer-Landau 1999). Since many people object to the Incoherence Argument because they have an interpretation much like Schafer-Landau's, and since such objections turn on a misinterpretation of my intentions, my hope is that this paper will help clarify what is, and what is not, being asserted in the premises of the Incoherence Argument. (The other explicit discussion of the Incoherence Argument appears in Chapter 13, "In Defence of the Moral Problem: A Reply to Brink, Copp, and Sayre-McCord," in which I discuss Geoff Sayre-McCord's criticisms [Sayre-McCord 1997].)
The real significance of the fact that we can explain the transition from having beliefs about our normative reasons to having motivating reasons in terms of the exercise of our rational capacities emerges when we see how this enables us to explain both the nature of various forms of practical irrationality and our ability to respond to the fact that we are practically irrational in these various ways. Chapter 3, "Philosophy and Commonsense: The Case of Weakness of Will" (co-authored with Jeanette Kennett) argues for the superiority of the resultant account of weakness of will by comparing it to Donald Davidson's famous account. Our preferred account of weakness of will is in terms of a subject's having, but failing to exercise, her capacity to have motivating reasons that accord with her beliefs about what she has normative reason to do: in other words, it amounts to her having, but failing to exercise, her capacity to have coherent pairings of belief and desire. We suggest that this gives us the required contrast between weakness and compulsion.
Chapter 4, "Frog and Toad Lose Control" (also co-authored with Jeanette Kennett), argues for the superiority of the resultant account of self-control. The very idea of self-control looks, after all, to be quite puzzling. If all action is motivated by desire, then when we act in a non-self-controlled way we must be acting on a desire, a desire that, in some sense, we shouldn't be acting on. But what exactly does this "should" mean? Furthermore, if all action is motivated by desire, and if an exercise of self-control is needed, then how is that exercise of self-control so much as possible? Wouldn't the exercise of self-control require the presence of a desire that, by hypothesis, we do not have? In "Frog and Toad Lose Control," Kennett and I offer a unified solution to these two puzzles.
The solution turns on two distinctions. First, Kennett and I distinguish between the exercise of synchronic, as opposed to diachronic, self-control. As we show, the puzzle only arises in cases of the exercise of synchronic self-control: exercises of diachronic self-control are unproblematic. Second, we distinguish between the distinct causal roles played, in the genesis of action, by desire, on the one hand, and by beliefs about our normative reasons, on the other. Exercises of synchronic self-control are required because rationality demands that our desires align themselves with our beliefs about what we have normative reason to do: the "should" mentioned above is thus the "should" of rational coherence. However, though our desires should so align themselves, they may fail to do so. It is in such cases that exercises of synchronic self-control are needed. But what makes such exercises possible is that the fact that they are one and all non-actional: that is to say, what causes the realignment of our desires with our beliefs about our reasons is our havings of various thoughts, our engaging in various imaginative exercises, and the like, where these are all in turn explained by the fact that it is rational for us to think such thoughts, to engage in such imaginings, and the like. The relevant causal factor is thus the tendency or capacity we have to move from an incoherent overall psychological state into a more coherent overall psychological state, not a desire and means-end belief pair.
Chapter 5, "A Theory of Freedom and Responsibility," uses the accounts of self-control and a capacity to do otherwise developed in Essays Three and Four to build a comprehensive theory of freedom and responsibility. Here, it seems to me, we see the real pay-off of asking the sorts of questions we asked in the previous essays: in order to build a comprehensive theory of freedom and responsibility we have no choice but to build on the foundation laid by a plausible theory of both normative and motivating reasons and the relations between them. In many respects, the theory of freedom developed here is similar to David Lewis's theory of freedom (Lewis 1981). There are, however, some striking dissimilarities. In particular, the theory offered here assumes that the capacity to act freely consists in two quite distinct capacities: the capacity to match our desires with our beliefs about what we have reason to do, and the capacity to match our beliefs about what we have reason to do with the facts about what we have reason to do. The latter is a crucial component. It explains why freedom is not a power of arbitrary significance. For this reason I argue that the theory of freedom and responsibility developed here brings out the crucial flaw that lies at the core of Harry Frankfurt's account of freedom of the will (Frankfurt 1971).
In the three papers just discussed - "Philosophy and Commonsense," "Frog and Toad Lose Control," and "A Theory of Freedom and Responsibility" - free use is made of the idea that we may have a capacity to make our psychology more coherent and yet fail to exercise this capacity. However, as Gary Watson pointed out some years ago in his seminal essay "Skepticism about Weakness of Will," the idea of an unexercised capacity is much more difficult to make sense of than it might initially appear (Watson 1977). Chapter 6, "Rational Capacities," attempts to provide the needed explication of this idea. I should perhaps say that the problem addressed in this paper - articulating the sense of "could" required for freedom of the will - seems to me to be the most difficult discussed in the essays in this collection. Though I would be amazed if the proposal is entirely successful - if it is then the problem of free will and determinism is solved! - my firm conviction is that a solution to this difficult problem must be found in some such proposal. My hope is that the proposal will stimulate profitable discussion and development of related proposals.
In the final three papers in the first part I return to the issue of motivating reasons. Chapter 7, "Humeans, Anti-Humeans, and Motivation," is a response to Philip Pettit's published reply to "The Humean Theory of Motivation" (Smith 1987; Pettit 1987; note that Smith 1987 is not reprinted in this collection). The aim is to make it clear why two lines of response to my view that motivating reasons are constituted by desires and means-end beliefs are not to the point. According to the first, I overlook the possibility that though motivating states are one and all constituted by desires, since the desires in question are beliefs - they are evaluative beliefs - it follows that our motivating states are one and all constituted by such beliefs as well. I point out that this is merely terminologically different from an objection I consider and rebut in "The Humean Theory of Motivation," which is that motivation requires the presence of a kind of psychological state that is both belief-like and desire-like but identical with neither. The second line of response is that I fail to answer the crucial question whether the desires that motivate actions, though distinct from beliefs, are none the less caused and rationalised by beliefs. But though this is a crucial question - indeed, it is the question that animates the essays that appear earlier in this collection - it is distinct from the question about the nature of our motivating states. One issue is whether we should accept a Humean theory of motivation. The argument in favour of this is that an explanation in terms of motivating reasons is a teleological explanation, from which it follows that motivating states must be constituted by desires and means-end beliefs, where belief and desire are distinct existences. Another quite distinct issue is whether we should accept Hume's own account of the rational status of desire. I argue that though we should accept a Humean theory of motivation, it is moot whether we should accept Hume's own account of the rational status of desire. (As the earlier essays make clear, I think we should in fact reject Hume's own account of the rational status of desire.)
In Chapter 8, "Humeanism, Psychologism, and the Normative Story," I respond to Jonathan Dancy's views about the nature of motivating reasons (Dancy 2000). If we must think that psychological states figure in the explanation of action then, according to Dancy, we should suppose that these psychological states are beliefs rather than desire-belief pairs. But in fact he thinks that we have no business supposing that psychological states typically figure in the explanation of action at all. For though it is indeed a truism that actions are explained by reasons, he argues that psychological states are only rarely, if ever, reasons. He thus prefers what he calls the "normative story," a story which contents itself with explaining actions by laying out the considerations in the light of which the agent acted as he did. But while I find myself agreeing with Dancy's premises, I do not find his conclusion convincing. I explain why.
Chapter 9, "The Possibility of Philosophy of Action," was conceived as a sequel to "The Humean Theory of Motivation." The paper addresses various challenges to the standard account of the explanation of intentional action in terms of desire and means-end belief, challenges that didn't occur to me when I wrote "The Humean Theory of Motivation." I begin by suggesting that the attraction of the standard account lies in the way in which it allows us to unify a vast array of otherwise diverse types of action explanation. I illustrate this with an explanation of action by ignorance. When we explain an action by ignorance, I say, we do not displace, but rather presuppose the availability of an explanation in terms of desire and means-end belief. With this illustration in mind I go on to consider a range of other challenges to the standard account of the explanation of action: Rosalind Hursthouse's challenge based on the possibility of what she calls "arational" actions (Hursthouse 1991); Michael Stocker's challenge based on the idea that some explanations of action are non-teleological (Stocker 1981); Mark Platts's challenge based on the idea that our evaluative beliefs can sometimes explain our actions all by themselves (Platts 1981); a voluntarist challenge based on the possibility of explaining actions by the exercise of self-control; and a challenge from Jonathan Dancy based on the idea that reasons can themselves sometimes explain actions all by themselves (Dancy 1994).
Part Two Meta-Ethics
In the second part of the collection the focus turns from general issues concerning the explanation of action to more specific issues in meta-ethics. As becomes clear, however, these more specific issues in meta-ethics seem to me to be continuous with those that arise regarding action-explanation.
In the essays described previously I defend the view that facts about what agents have reason to do are best understood as facts about what they would ideally want themselves to do. But, as even a casual glance at the meta-ethical literature makes clear, we find a very similar idea in meta-ethics. Indeed, Roderick Firth goes so far as to suggest that something like this idea is defended by all of the classic moralists (Firth 1952). According to the dispositional theory of moral value, for example, facts about moral values are facts about idealised desires. But if this is right then perhaps we can simply collapse the two stories into one: facts about what we have reason to do are simply facts about what it would be good or desirable for us to do where these, in turn, are facts about what we ideally desire ourselves to do. Moral values are a sub-class of the values: facts about what we morally ought to do are a sub-class of the facts about what we have reason to do. The essays in part two defend this conception of moral facts and, as well, locate that defence in the context of broader issues in meta-ethics.
Chapter 10, "Moral Realism," is an extended statement and defence of the version of moral realism that I myself favour. According to this version of moral realism, moral facts reduce to idealised psychological facts, facts which in turn constitute reasons for action. My aim in writing "Moral Realism" was to write something reasonably accessible to those not familiar with the various moves that are standardly made in the vast meta-ethical literature. A wide range of material is therefore covered, including an explanation of the difference between moral realism, nihilism, and expressivism; an explanation of why moral realism becomes truistic on certain minimalist conceptions of truth; an account of the problem that moral realism, so construed, faces; an account of what a moral realist who is a naturalist would say about moral facts; an account of the various standard objections to this kind of naturalistic moral realism; an account of what a non-naturalist would say about the nature of moral facts; an account of the various objections to this non-naturalistic kind of moral realism; two replies that a naturalistic moral realist might give to the objections made earlier, one of which commits the realist to internalism - that is, to the idea that there is an internal or necessary connection between moral judgement and the will - and the other of which commits the realist to externalism; a reason for preferring the version of moral realism that commits the realist to internalism (this is the version that reduces moral facts to a certain sort of idealised psychological fact); a discussion of the relativistic and non-relativistic versions of this kind of moral realism (this is related to my objections to William's view at the end of Chapter 1); and, finally, an argument in favour of the non-relativistic version.
Chapter 11, "Does the Evaluative Supervene on the Natural?," is a discussion of an important challenge to an orthodoxy in meta-ethics. Virtually everyone writing in meta-ethics takes if for granted that evaluative facts supervene on natural facts: no two worlds can be naturalistic duplicates and yet differ in evaluative terms. One notable exception to this orthodoxy is James Griffin (Griffin 1992). I begin the essay by clarifying the claim that the evaluative supervenes on the natural. This proves to be a much more difficult task than we might have thought it would be. I then consider, and ultimately reject, Griffin's various reasons for being skeptical about the supervenience thesis. I trace the attraction of the supervenience thesis to a fact about ordinary moral discourse, namely, the fact that it is always appropriate to ask what makes a moral claim true and that what we require by way of a response is an answer in terms of certain natural features.
Chapter 12, "Objectivity and Moral Realism: On the Significance of the Phenomenology of Moral Experience," is about a well-known exchange between John Mackie and John McDowell. Mackie famously argued for an error theory: in his view, though we have moral beliefs, these beliefs are one and all mistaken (Mackie 1977). John McDowell's reply to Mackie's argument is that it merely underscores the fact that Mackie has a mistaken conception of moral qualities (McDowell 1985). To be sure, McDowell insists, if moral qualities had to be like primary qualities - like being round, being extended, and the like - then, just as Mackie says, it would be impossible to make sense of the internal connection between moral judgement and the will. However, according to McDowell, this simply shows that we should conceive of moral qualities as being more like secondary qualities, rather than primary qualities: more like being red than being round. Once we adopt the secondary quality conception of moral qualities, a conception that he thinks is well supported by the phenomenology of moral experience, McDowell claims that Mackie's error theory becomes a non-starter: moral qualities, so conceived, are both out there in objects and internally connected to the wills of moral agents.
In "Objectivity and Moral Realism," I consider McDowell's argument at some length. I argue that even if we do adopt a secondary quality model of moral qualities, the error theory still looms large. For even if moral qualities are just dispositions to elicit appropriate desires in us under suitable conditions, if, as Mackie seems to think, these conditions would have to be conditions in which, in virtue of our perfect rationality, we all come to have the same desires in response to the same facts - here we once again appeal to the conception of facts about what we morally ought to do as facts about our normative reasons for action, as these are analysed in the essays in the first part of the collection - then skepticism about the possibility of this kind of rationally underwritten convergence in our desires would still provide us with grounds for skepticism about the existence of such dispositions. In other words, to the extent that the phenomenology of moral experience provides us with support for a certain kind of rationalism, which it does indeed seem to do, Mackie's error theory remains an option that needs to be ruled out. McDowell's objection thus misses its mark.
© Cambridge University Press
Table of ContentsIntroduction; Part I. Moral Psychology: 1. Internal reasons; 2. The incoherence argument: reply to Schafer-Landau; 3. Philosophy and commonsense: the case of weakness of will; 4. Frog and toad lose control; 5. A theory of freedom and responsibility; 6. Rational capacities; 7(i) On Humeans, anti-humeans and motivation: a reply to Pettit; 7(ii) Humeanism, psychologism, and the normative story; 8. The possibility of philosophy of action; Part II. Meta-Ethics: 9. Moral realism; 10. Objectivity and moral realism: on the significance of the phenomenology of moral experience; 11. In defence of The Moral Problem: a reply to Brink, Copp and Sayre-McCord; 12. Exploring the implications of the dispositional theory of value; 13. Does the evaluative supervene on the natural?; 14. Internalism's wheel; 15. Evaluation, uncertainty, and motivation; 16. Ethics and the a priori: a modern parable.