PREFÁCIO
This book brings together two large and distinct areas of enquiry. The first is the nature of good reasons for action, where we ask what it is for something to give us a reason to act in one way rather than another. The second is that part of the theory of action that asks what it is to act for a reason. It would not be quite correct to say that what is being offered here is a combination of value theory and the theory of motivation, for value theory is distinct from the theory of reasons. But there are views on which the distinction between values and reasons becomes pretty slight, and the view presented here is one (but only one) of those.
In broad outline, the first part of the book makes claims about the nature of practical reasons, those reasons that favour one action rather than another. The second part uses those claims to support an unfamiliar account of what it is to act for a reason, and to undermine more standard accounts.
The discussion is conducted at a fairly high level of abstraction, with few of the sort of examples that discuss the doings of Peter, Jane, and John that one often finds in contemporary philosophical writing. This has given me one problem. I speak constantly of the agent’s reasons, of things that are reasons for an agent, and of reasons why the agent acted. As a result, I have been forced to choose between two practices. The first is that of allotting this abstract but omnipresent agent a gender and sticking to it; the second is that of varying the gender from case to case, in hopes of doing so arbitrarily. The second option, when I tried it, seemed to me to be just confusing and artificial, creating two abstract agents in the mind of the reader when it would be better to have only one. So I decided to think of this agent as effectively an abstract representation of myself, and as a result it—now he—is of the male gender throughout. To those whom this irritates, or worse, I apologize. I suppose I am hoping that after about twenty years of variable genders it might be possible for all authors now to write in terms of their own gender, without raising hackles.
Reasons for Action
I. Good Reasons and those for which the Agent Acted
This enquiry begins from a perfectly ordinary and unproblematic distinction, expressed in terms of reasons. When someone does something, there will (normally) be some considerations in the light of which he acted—the reasons for which he did what he did.There are not so many things that we do for no reason at all. Intentional, deliberate, purposeful action is always done for a reason, even if some actions, such as recrossing one’s legs, are not—or not always, anyway. So normally there will be, for each action, the reasons in the light of which the agent did that action, which we can think of as what persuaded him to do it. When we think in terms of reasons in this way,we think of them as motivating.The consideration that motivated the agent was his reason for doing what he did.
We may, of course, know the reasons for which the agent acted without yet knowing whether there was a good reason to do what he did.Was there, for instance, some moral reason for him to do it? Or was there a reason of another sort? Was it at least a sensible thing to do in the circumstances? If it was a sensible thing to do, there will be some reason why this was the case. When we think of such reasons, we think of features that speak in favour of the action (or against it). They are good reasons for doing the action. So they are normative, both in their own nature (they favour action, and they do it more or less strongly) and in their product, since they make actions right or wrong, sensible or unwise.When we think in terms of reasons in this way,we are thinking normatively. We could even call such a reason a normative reason.All such reasons are good reasons, though some will be much better than others. This is because a reason that favours an action must be a good one. A bad reason for doing something, if it is not merely a not very good reason for doing it, can only be no reason at all for doing it; if so, it is not a reason in the sense intended, since it does not favour the relevant action.
The distinction with which I have started is not particularly obscure or complicated. There are these two ways of using the notion of a reason for action, which address different questions. There is the question what were the considerations in the light of which, or despite which, he acted as he did. This issue about his reasons for doing it is a matter of motivation. There is also the question whether there was good reason to act in that way, as we say, any reason for doing it at all, one perhaps that made it sensible in the circumstances, morally required, or in some other way to be recommended, or whether there was more reason not to do it. (There are other reasons besides those that make actions right or sensible; there might be an aesthetic reason in favour of acting gracefully.) This second question raises a normative issue.
Because there are different questions at issue, the answers to them can differ. The consideration in the light of which someone acted as he did need not have been a very good reason; what he did may have been a pretty silly thing to do, with little or nothing actually to be said in its favour. In such a situation we can give his reasons for acting, and in that sense explain his action, but cannot find much to say in favour of what he did. Motivatingreasons are supposed to be the ones that actually made a difference to how he acted; they constitute the light in which he chose to do what he did. Normative reasons are the ones we try to cite in favour of an action, because they are the ones that should show that the action was sensible or right or whatever. So if an agent does something extremely silly, we might say he had a motivating reason for doing what he did but little normative reason. What this means is that we can give a certain sort of explanation for his acting as he did, but we cannot defend it.
If we do speak in this way,of motivating and normative reasons, this should not be taken to suggest that there are two sorts of reason, the sort that motivate and the sort that are good. There are not. There are just two questions that we use the single notion of a reason to answer.When I call a reason ‘motivating’, all that I am doing is issuing a reminder that the focus of our attention is on matters of motivation, for the moment. When I call it ‘normative’, again all that I am doing is stressing that we are currently thinking about whether it is a good reason, one that favours acting in the way proposed.
If an agent does a truly bad or silly action, then, he will surely have acted for some reason; he will, that is,have had a motivating reason for doing what he did. But there will have been little reason to act in that way. In the worst possible case,indeed, we would have to say that there was no reason at all for doing what he did (no good reason, that is). This would happen if the features that the agent took to count in favour of acting in fact do nothing of the sort; this might happen when making complicated financial decisions, for instance, for one can easily get confused in such cases and make a complete mistake about which way a consideration counts. Things are not often quite as bad as this, of course. In the best case, there is some good reason for doing the action, and the reason that motivates the agent coincides with that reason. In another case, there may be a wealth of good reasons present, with only some of these actually doing the motivating,i.e. functioning as the agent’s reason for action. In yet another, there may be perfectly good reasons for doing it, though actually what motivated the agent was something quite different. In all these cases, we might say,there is some defence for the agent acting as he did, or at least for what was done, even if not for the way he did it.
On the other hand, there may be very good reasons favouring an action though these influence the agent not at all, either because he is simply ignorant of them, or because, though perfectly well aware of them, he is lazy, selfish, pig-headed, or whatever.If so, the action, though there are good reasons to do it, willjust not get done. None of those reasons motivated the agent to do this sensible thing. He went off and did something else, without a second thought.
The main awkwardness of what we have so far is that it leaves us saying that an agent can act for a reason (a motivating one) that is no reason (no good reason, that is), or that there was no reason to do what he did, even though he did it for a reason. Some motivating reasons, that is, are not good reasons. But I take this to be no more than a little local difficulty. It seems awkward because the phrase ‘good reason’ is pleonastic, when we are thinking normatively, and this leads us to say that a motivating reason that is no good reason is not a reason either. But we should remember that the notion of a reason is used to answer two distinct questions.The first is why someone acted, and the second is the pleonastic one of whether there were any (good) reasons for so acting. In specifying an agent’s motivating reasons we answer the first question, and in that sense motivating reasons are all reasons. It is only when we have our eyes on the second question that we want to allow that a motivating reason can be no reason at all. In what follows, the phrase ‘good reason’ will be a mark that we are thinking in normative terms.
In laying out the distinction between good reasons and those that motivate, I have spoken as if, when we act, the reasons that motivate us are all on the side of doing what we did. I have given the impression that the considerations in the light of which we act all pull us in the direction of doing the action, with none tugging unsuccessfully in the other direction. But in fact there can be considerations despite which we acted, considerations that would have led us not to do it if they could have done (if, for instance, there had been fewer considerations pulling in the other direction), and so on. I might decide to buy that new book even though it is very expensive. Here we have a consideration pulling against the action that I choose to do, but pulling unsuccessfully. I recognize it as a reason, and am influenced by it, but not enough to stop me from acting. As we might say,’I was moved by her plight,but not enough to do anything about it’. What is more, as well as reasons that pull directly against doing the action, I might recognize reasons for doing each of two things, when I cannot do both at once. This weekend I am motivated both to dig over the vegetable patch and to overhaul my bicycle, but I have not the time to do both. Suppose I decide to deal with the bicycle and leave the garden until later. Are my defeated motivators to be thought of as among the reasons that motivated me? They are not among my reasons for doing what I did. In that sense, then, they are not among the reasons that motivated me to act. Thinking in relation to the action that was done, then, they do not count among the motivating reasons. But still I was influenced by them, and they do figure in my motivational economy. We should not forget this, but all the same I will normally speak as if the reasons that do motivate all pull in the same direction. This should not blind us to the fact that not all such reasons actually succeed in getting one to act.
There is a similar issue about whether all good reasons lie on the same side, or whether there can be good reasons on both sides of the question. But it is easier to take a firm stance on this one. There can be good reasons not to do an action even when there are better reasons to do it. That an action was right does not show that there were no (good) reasons not to do it. Equally,if an action waswrong, this does not mean that there was no reason to do it; it merely means that there was insufficient reason.
2. The Reason Why
We can normally explain an agent’s doing what he did by specifying the reasons in the light of which he acted. But there are other ways of explaining an action—ways that do not involve specifying the agent’s reason (Darwall 1983: 29). For instance, we might say that the reason why he did this was that he had forgotten his promise to her. In so explaining his action, we are not involved in laying out the reasons in the light of which he acted. We are not supposing that the considerations that he took to favour the action included that he had forgotten his promise to her. When we mention the fact that he had forgotten his promise, we mention something of which he was wholly unconscious, and which could not have been among his reasons for acting. This is quite apart from the fact that one’s having forgotten something is not normally able to be taken even by others as a reason in favour of doing what one did, though of course it can be some excuse.
On other occasions we explain an action by appeal to quite different sorts of consideration. What explains why one person yawned may be that someone else yawned just next to them. What explains why he responded so aggressively may be that he is having trouble at home or that he has taken a particular form of medication. What explains why he gave this student a better grade than she deserved is that he was unconsciously influenced by the fact that she always dresses so neatly (or something even less defensible). What explains why so many people buy expensive perfume at Christmas is the barrage of advertising on the television. What explains why he didn’t come to the party is that he is shy. In none of these cases are we specifying considerations in the light of which these things were done. But in all of them we are explaining why they were done. It seems, therefore, as if there is a wide range of things we think of as capable of giving answers to the question ‘Why did he do that?’ These answers range from specifying the things in the light of which the agent chose to do what he did,which we have sometimes called the agent’s reasons for doing what he did, to something that is not a reason at all,really,but rather a cause (as in the explanation that talks about medication).
So we need to keep the notion of a motivating reason separate in our minds from the more general notion of ‘the reason why the agent did what he did’. In what follows, the phrase ‘in the light of will be used to signify the relation between an agent and the reasons for which the agent acted. And I will speak of the agent’s reasons in a similar way.The agent’s reasons are the reasons that motivated him, namely the considerations in the light of which he chose to do what he did. But not all explanations of action specify motivating reasons in this sense.
3. Other Versions of the Motivating-Normative Distinction
I have characterized the distinction between the reasons why we do things and the reasons in favour of doing them in terms of the motivating and the normative. In doing so,I have tried to avoid any suggestion that we are dealing here with two sorts of reason. This is because the main argument of this book is that one and the same reason can be both motivating and normative. A reason for acting can be the reason why one acted. This should not look very surprising, one would have thought—until one notices that most contemporary philosophers take it to be impossible. All this will emerge in Chapter 5.
The terms ‘motivating’ and ‘normative’, though perfectly standard in the literature, are not the ones that everyone uses. Some writers speak of justifying rather than normative reasons; others— sometimes the same ones—speak of explanatory rather than motivating reasons. The distinction intended is,I think, the same all the time. I give here some defence of my preference for the terms I have adopted.
Talking of justifying reasons rather than normative or good ones runs foul of one difficulty that I hope to be able to keep under control. This is that agents may do things that are actually illadvised, but be able to explain why they did these things in such a way as to defend their choice of action, without this showing that what they did was in fact the sensible thing to do. Suppose, for instance, I make a not very sensible choice about what arrangements to make about my pension. And suppose that I can later explain the choice I made by pointing out that there were some crucial facts that I happened quite reasonably to have got wrong, and in this way, as we might put it, exculpate myself.There is a sense of ‘justify’ in which I can be said to have justified doing what I did. But this does not show that the balance of reasons was in favour of the action. It wasn’t. Indeed, the features in the light of which I made my choice turned out all to be a mistake, and so cannot count even as defeated reasons. So I prefer to keep the notion of a reason that justifies separate from that of a normative reason. One might try to talk of a rectifying reason, one that shows the action that it supports to have been right, in preference to a justifying reason, one that defends the action without showing it to have been right. But this is too moralistic to be usable in every case.
Some writers speak of explanatory reasons rather than of motivating ones. My difficulty with this is that there is a sense in which normative reasons are explanatory; they explain whythe action was right, or sensible, or whatever—why it was worth doing. What I have been calling ‘reasons-why’ will also be explanatory without being motivating,too, for I am thinking of motivating reasons only as considerations in the light of which the agent acted, and ‘reasonswhy’ are not like that.
Some writers use phrases like ‘the agent’s normative reasons’ (Smith 1994: 131). These appear to be a sort of hybrid. We might have thought that the reasons that motivated belong to the agent, as it were, and the normative ones are not particularly the agent’s reasons at all; they are just the reasons why the action was to be recommended, the good reasons for doing it. What leads people to speak of the agent’s normative reasons is that commonly (some would say universally) the reasons that motivate the agent are taken by the agent to be good reasons for doing what he did. It is because of this that they motivate the agent at all, we might say. When this happens, the reasons that motivate at least include what is here being called ‘the agent’s normative reasons’. I take it, however, that this phrase really refers to what the agent took to be good reasons, rather than to a special species of good reason, those somehow belonging to the agent. And that is how I will understand it in what follows.
4. Justification and Explanation
We have seen that there are (at least) two different sorts of question that we use the notion of a reason to answer.There are questions about why the agent did what he did, and questions about whether there was good reason to do it; there are reasons why people act, and reasons in favour of acting. Both questions are in a sense requests for explanation.The first sort of question asks for an explanation of the action that was done or is in prospect. The second sort of question asks for an account of whether the action is worth doing,which we give by giving the reasons that made it so, i.e. by explaining what made it so. But there is, it would seem, no one sort of request for explanation that both sorts of question address.
This has, however, been denied. Michael Smith has suggested that there is a broader sense in which, though there are indeed two questions, our answers to them are all aimed at the same thing. They are all attempts to render the relevant action intelligible, though there is more than one way of doing this, for we can render an action intelligible without explaining why it was done. Smith writes:
Motivating and normativereasons do have something in common in virtue of which they both count as reasons. For citing either would allow us to render an agent’s action intelligible.This is essential. For there is an apriori connection between citing an agent’s reasons for acting in a certain way and making her acting intelligible: that is, specifying what there is to be said for acting in the way in question. In virtue of their differences, however, motivating and normative reasons make actions intelligible for quite different reasons. (1994:95)
Smith does not seem to tell us here what exactly the supposed a priori connection amounts to. But I presume that he has in mind the idea that to explain an action is to show how doing that action can have come to seem worthwhile to the agent. The action is explained once we can show that doing it made good sense in the lights of the agent. Now of course it is not required for this that we show that there was indeed good reason to do the action. For if it were, stupid and/or wrong actions would be inexplicable, which they are not. To explain them we show,not that they were sensible, but how it was that doing them appeared sensible to the agent. We lay out, as it were, the favourable light in which the action appeared to the agent.
This means that, to explain the action, we show how doing it can have made sense to the agent.And to do this we lay out the agent’s take on things, and show that the action made sense in that light. So what we show is that, relative to how the agent took things to be, it made good sense to do this action. If things had been as he supposed, this action would indeed have been the right or the sensible thing to do.
There are two questions we can ask, then:
1. Did he do the right thing?
2. Had things been the way he supposed them to be, would what
he did have been right?
The answer to the first question may be no even if the answer to the second is yes. Take a case where he (perhaps reasonably) believes that she would welcome his advances but is too shy to do much to encourage him. He acts in a way that would have been right if things had been as he supposed. But in fact things are not as he supposed, and what he does is not right (though perhaps pardonable, if his original belief was reasonable rather than, say, the result of lustful thinking).
Smith’s remarks were not concerned only with moral reasons, but with practical reasons in general. And indeed we see that the distinction drawn above for the moral case can be drawn more generally, in terms of what there is most reason to do. There are two questions:
1. Did he do what there was most reason to do?
2. Had things been the way he supposed them to be, would his
action have been the one there was most reason to do?
We explain the action by showing that the answer to the second question is yes. We make sense of the agent’s doing what he did, even when there was inadequate reason to do it, by showing that he would have had good reason had things been as he supposed.
What this shows is that to explain an action is to justify it only in a certain sense. It is not to show that it was what there was most reason to do. It is to show that it would have been if the agent’s beliefs had been true. But to show this much is to mount some defence of the agent for so acting. For the agent’s behaviour is pardonable if (roughly) his beliefs were pardonable and had they been true he would have been doing what he ought, or what there was most reason to do. *
* There are, in fact, two notions of a pardonable but false belief that we need to keep separate in our minds.There are beliefs that are unpardonable because they are beliefs that it is morally wrong to have; one should not, perhaps, believe that one is more morally significant than others are. This is an absolute notion of what is and is not pardonable. There are also beliefs that are pardonable given one’s epistemic circumstances—the evidence available, for instance, or the time available for reflection. We could think of this as a relative notion of what is pardonable or unpardonable, since it is determined by considerations of what resources of various sorts were available to the agent.
We can now see what lies behind Smith’s remark that we make acting in a certain way intelligible by specifying what there is to be said for acting in that way.But to do so we have had to take the phrase ‘what there is to be said for’ as meaning rather ‘what there seemed to be said for’. The link between explanation and justification is therefore a link between explaining the action and showing how, had things been as the agent supposed, there would have been most reason to do what he did.
5. Four More Distinctions
In this section I consider four further distinctions The first two of these concern motivation, the last one concerns the nature of good
reasons, and the third concerns the relation between good reasons and motivation.
Humeanism and Anti-Humeanism in the Theory of Motivation
There is a classic position in the theory of motivation that isknown as Humeanism, despite the fact that it bears little resemblance to the views of its supposed progenitor, David Hume. The basic claim made by this theory is that intentional actions are to be explained by reference to the beliefs and desires of the agent. For an intentional action to take place, its agent must have a suitable combination of beliefs and desires. There must be something that the agent wants, an aim or goal which the proposed action subserves in some way.It may do this as a means to an end, or in some other way, perhaps constitutively, as when taking care over my children’s education subserves my aim of being a good parent, not by being a means to that end but by being part of that end. Further, the agent must have suitable beliefs to the effect that the action is likely to subserve that goal. To take a very ordinary case, why is the agent walking towards that cupboard? He is doing this because he wants a glass of beer, and he believes that there is beer in the cupboard. So walking towards the cupboard subserves his aim of having a beer.
It is possible to object to Humeanism on the very simple grounds that we often do things that we do not want to do, either in themselves or as means to some end that we want to pursue or promote. Certainly,the ordinary use of the word ‘desire’ is restricted in a way that would render ridiculous any claim that all action is motivated by desire; its main use nowadays seems to be in sexual contexts. But when philosophers speak of desire, at least in the context of the theory of motivation, they are to some extent constructing a term of art. I may say that I am doing this because it is my duty, though I have no desire to do it whatever. The reply is that I must none the less take a positive attitude to some aspect of what I am doing—a ‘pro-attitude’, in current terms owed to W.D. Ross. And the current philosophical use of the term ‘desire’ takes it to be equivalent to this much broader notion of a pro-attitude. In this sense it becomes much more plausible to say that all action ismotivated by desire.
Humeanism does not content itself with asserting the need for both beliefs and desires if there is to be intentional action. It also claims that the desire is in some way the dominant partner in the belief-desire combination. This point used to be made in terms of a hydraulic metaphor. If an agent is to change from rest to movement, there has to be some motive force that is capable of creating this change. Desire is such a force, being an active, urgent, pushy state. Belief,by contrast, is a static state, a passive, inert, or at least inactive conception of the situation. So the desire plays a role in the belief-desire combination that the belief is quite incapable of playing, and, as we might put it, it is the desire that makes the real difference. Desire can do nothing without belief, of course, since if we cannot attribute to the agent some suitable conception of the situation, we would be unable to explain the eventual action. For there would be no reason for the agent to do one thing rather than another, and so we would be unable to explain what he in fact did. Desire needs belief, then, as belief needs desire. But desire predominates. If the sense in which desire predominates is not yet clear, that is a fair representation of the present state of Humean theory of motivation. Perhaps the nearest Humeans have got to making it clear is to ground talk about the predominance of desire in the distinction between two directions of fit. Belief has one direction of fit and desire has the other. A belief is out to fit the world, and if it fails to fit that world, the fault, as we might put it, lies with the belief rather than with the world. A desire is not shown to need revision if it fails to fit the world.Desire knows that things may not be (indeed probably are not) the way it represents them as to be, and is not to be criticized for this. Desire is out to get the world to fit it. So belief is not out to change things, but to catch them as they are, while desire is out to change things. In this sense, a Humean might say, desire is active in a way that belief is not, and it is in this sense of activity that we can think of the active partner as dominant. *
* I have here suggested that a Humean might appeal to the notion of direction of fit to give sense to the idea that desire is active in a way that belief is not. This might be thought to be an admission that the very notion of a direction of fit can be used to support Humean claims about the dominance of desire in motivation. Worse, indeed, for I will shortly accept the distinction between two directions of fit, and may therefore be thought to commit myself to Humeanism, unwittingly. Not so, however. What I say in the text is that in virtue of its direction of fit, desire can be thought of as active in a way that belief is not. This admission, however, is perfectly compatible with the vigorously anti-Humean position I expound in Ch. 4. If, as I will there argue, desire cannot be part of what motivates, this imposes on us a nonHumean account of what motivates, but does not prevent desire (i.e.motivation) from being active in a way that belief is not. It does nothing to show that belief is not active in some other sense.
Anti-Humean arguments have tended to be of one of two forms. The first takes its cue from Thomas Nagel’s distinction between motivated and motivating desire (Nagel 1970, ch. 5). Humeanism understands desire as leading, and as subserved by belief, and holds this conception in place by trying to think of desire as active in a way that belief is not. An opposing view, then, could argue that, though many actions are indeed to be explained in the Humean way, some are not. For some beliefs are capable themselves of generating the necessary desire, so that the belief is leading, dominant. The belief that the action would be unjust could cause in me a desire to have nothing to do with it, and the belief-desire combination thus formed would be what motivates me to abstain.
It is unclear from Nagel’s own discussion whether this was the picture that he wastrying to put up against the Humean orthodoxy. But let us leave that aside for the moment, and turn to the other form of anti-Humeanism. This is the view that belief alone is capable of motivating action. I argued for this view in the first three chapters of my Moral Reasons (1993). I accepted there (for most cases, at least *) the Humean view that motivation requires a combination of two distinct elements in the agent, but held that both could be beliefs. On my view, a desire is never a necessary part of what motivates. So we have two beliefs which together motivate. One of these is about how things are, and the other is about how they would be if the action were successfully performed. I called this view’pure cognitivism’. (I argue for this position more formally in Chapter 4 below, but here lay out the bare bones of the thing.)
* The reason for this qualification is that it is possible for an action to be motivated by a single state of the agent—one belief rather than two; see Dancy (1993, ch. 2.6)
What has pure cognitivism to say about the role of desire in motivation? Does it deny altogether the need for desire? This would be unwise. For the core of the Humean distinction between belief and desire is captured in contemporary debate by the distinction between two directions of fit, as we have seen. The crucial point here is the availability of an a priori argument that for motivation one needs two distinct states, one with each direction of fit. Intentional action is only comprehensible if one can attribute to the agent some views about how things are and some views about how they are to be should the action be successful. Let us suppose that these ‘views’ are both beliefs. Still, the agent is motivated by those beliefs, and if so, he is in a state whose success conditions consist in things coming to be as,in that state, he conceives of them as to be. Such a state is a desire, not a belief, since the success conditions for a belief consist in things already being as they are there conceived. This is the present form of the original Humean claim that one needs both belief and desire. Pure cognitivism has a choice between denying the need for a state with each direction of fit and allowing the need for a desire as essential to motivation.
The move that I made, in response to this dilemma, was to allow that wherever there is motivation, there will be desire, but to deny that the desire was a part of what motivates. The ‘motivating state of the agent’ itself consists entirely of beliefs, on my account; in fact such a state will normally consist of two beliefs. But where there is motivation, there is desire, since a desire is a state of being motivated. So I allowed the a priori argument that there is a need for two distinct states, one with each direction of fit, if there is to be motivation, but I removed desire entirely from the centre-stage position that it occupies in classical Humeanism. Desire never motivates, on my account. (If one is thirsty, one is motivated by the prospect of drinking.) But there can be no motivation without desire.
Psychologism and Anti-psychologism in the Theory of Motivating Reasons
Pure cognitivism accepts two central features of the Humean picture. The first of these is the asymmetry of belief and desire, i.e. the fact that belief and desire play quite different roles in motivation; this is captured in the distinction between two directions of fit.Pure cognitivism accepts that asymmetry,in accepting that what belief does desire cannot do, and vice versa. But it rejects the characteristic Humean stress on the dominance of desire in the generation of motivation. Desire is not the leading partner in anything, even though desire is a state that is out to change the world to be the way it wants, while belief is merely out to represent the world as it is.
Pure cognitivism also accepts the Humean claim that desire is an independent existence, with its own phenomenology. Desires occur, and they may have a distinctive ‘feel’, at least on occasion (the strong ones, probably). Their occurrence is required for motivation, but this is not surprising, since to desire isjust to be motivated, and being motivated can have a distinctive feel.
There is, however, a further thing that pure cognitivism took unexamined from Humeanism, and this is much more significant than the two mentioned above. This is the idea that the theory of motivating states, as conceived by Humeans, constitutes or is the theory of acting for a reason—the theory of motivating reasons. To put it another way, Humeans presume that the theory of motivation is the theory of motivating reasons, and that this theory is to be written in terms of motivating states of the agent. The question we were debating was whether such reasons could be beliefs— whether a belief alone could be a reason for acting, in particular a moral belief—or whether there needed also to be a desire. If desire dominates belief in motivation, then it would be true to say that our desires are the ‘real’ reasons why we act. If not, beliefs could be reasons too, reasons as real as the desires. But this debate is premissed on something itself hidden but still eminently debatable, namely that our motivating reasons are psychological states of ourselves. I call this view psychologism. We were not arguing about psychologism; we took it for granted, and argued only about which psychological states were involved, and in which combinations.
As soon as one sees this, new possibilities swing into view. Are any motivating reasons psychological states of the agent? Even the most cursory glance at the sorts of reason we actually give, in explaining either our own actions or those of others, reveals that while some certainly seem to be characterized in terms of a psychological state of the agent, others equally certainly do not. We say ‘I am giving you this medicine because I believe that it will reduce your temperature’; but we also say ‘I stayed in because it was raining’, or ‘We are going to sit outside because the sun has come out’. Some reason-givings offer a belief of the agent as a reason, then, in a way which might indeed be understood as a reference to a psychological state of the agent, but some offer something that seems not be a state of the agent at all, but a state of affairs (the sun has come out); and there is a world of difference between these. If the reason that motivates is the thing specified on the right-hand side of the ‘because’ in ‘I did it because p\ many such reasons are apparently not psychological states of the agent. If reasons that motivate are all psychological states of the agent, by contrast, no reasons are properly (i.e. fully) specifiable in the form ‘A acted because p’; the proper form will be ‘A acted because A believed that p\ or something of that sort.
Internalism and Externalism in the Theory of Normative Reasons
Our understanding of this distinction is owed to Bernard Williams. Internalism is a view about normative reasons, and it holds that an agent A only has a good reason to “x” if, were A to know all the relevant facts, and deliberate rationally, A would be motivated to “x”. The idea is that we start from A as he now is, with his existing set of desires, or motivations, but supposedly not yet motivated to “x”; and we ask whether, starting from there, A would become motivated to 0 if he were to know all the relevant facts, and deliberate rationally. If the answer is no, A has no reason as things stand to “x”.* Internalism, then, imposes a necessary condition on any supposed normative reason.
* Even ifA does at present have desires that the action would subserve. Note that Parfit (1997: 100) defines internalism disjunctively. According to him, it is the claim that ‘for it to be true that (R) we have a reason to do something, it must be true that either (D) doing this thing might help to fulfil one of our present intrinsic desires, or (M) if we knew the relevant facts, and deliberated rationally, we would be motivated to do this thing’. Parfit’s disjunctive version is untrue to the position that Williams originally presented. Williams wrote as follows, using ‘5’ to represent the agent’s subjective motivational set, i.e. roughly the agent’s desires: ‘A member of S, D, will not give A a reason for 0-ing if either the existence of D is dependent on false belief, or A’s belief in the relevance of 0-ingto the satisfaction of D is false’ (1980:103). On this account,a desire that does not satisfy the subjunctivecondition (M) is incapable of,as we might put it, ‘sustaining’ a reason; on Parfit’s account, satisfying condition (D) is enough for the sustaining of a reason. So Parfit’s internalist would have to allow that if I do desire to drink what is in this glass, I have at least some reason to drink it, even though I would have no desire to drink it at all if I knew what it really contained. Williams’s internalist need not allow this. This matter begins to make a difference early in Ch. 2.
Calling this suggested necessary condition for having a reason “motivation in condition C”, we could more snappily express internalism as the view that A only has a reason to “x” if A would be motivated to “x” in condition C. Externalism about normative reasons is the claim that it is possible to have a good reason even when one would not be motivated accordingly in condition C.*
* The formulations of internalism and externalism here concern the claim that an agent has a reason—some reason, one might say. For internalist accounts of the condition under which a specific feature F is a reason for A, or gives A a reason, see the beginning of Ch. 2.
Internalism and externalism, so understood, are not views in the theory of motivation. They are views about the relation between good reasons and motivation. Internalism, understood as above, amounts to a motivational constraint on good reasons; something can only be a good reason for me if it is related to what I would want if I deliberated rationally and knew the relevant facts, starting from where I now am. Internalism therefore specifies a necessary condition on normative reasons. We cannot suppose that there are such reasons without specifying which agents they are reasons for, and no such attribution of a reason to an agent is true unless that agent would be suitably motivated in condition C. So we cannot say,for instance, that there is a reason for anyone to be concerned about the needs of the homeless, unless we can show that all agents whatever would be motivated accordingly in condition C. Could all agents be got, starting from where they now are, to be motivated by thoughts about the homeless, if they thought harder and knew more? Somehow I doubt it.
Desire-Based and Value-Based Normative Reasons
This distinction is much harder to characterize, since there has been no authoritative discussion of it.The desire-based view seems to be that all normative reasons are ‘provided’ by desires of the agent, while the value-based view is that some reasons, at least, are provided by (‘grounded in’) values such as the value of achievement, of pleasure, of friendship, or whatever. * As such, these two views seem to be about the metaphysical ground for a reason—about what creates or generates it. They are not about necessary conditions for something to be a good reason for a particular agent—or at least, they are not directly about that. The ‘grounding’ or ‘providing’ relation is different from, stronger than, and much more interesting than the much weaker relation of ‘being a necessary condition for the existence of. This point seems occasionally to escape Derek Parfit’s attention in his masterly typology of theoretical positions in this area, to which I am much indebted and which essentially I am using here (Parfit 1997). He makes the following claims, all within three pages (pp. 128-30):
Internalist theories are desire-based.
According to many Internalists, all reasons are provided by desires.
On Internalist theories, the source of all reasons is something that is not itself normative: it is the fact that we have some desire.
When we have a reason to do something,this reason is not provided by, and does not require, the fact that after Internalist deliberation we would want to do this thing.
Parfit here seems to suggest both that all internalists believe in desire-based reasons, and that most do but some do not. He also suggests both that a desire-based reason is one based on the fact that we have some desire, and that it is one based on the desire itself. Most important, however, is the distinction that he draws in the last quoted sentence above, between a reason being ‘provided by’ some fact about the effect of internalist deliberation and its ‘requiring’ that fact. Talk of ‘requiring’ is talk of a necessary condition.
* There is the intermediate possibility that all reasons are desire-based, without the desires concerned needing to belong to the agents who have the reasons. We might think that every reason for action is eventually grounded in the desires of someone or other, but that that person need not be the agent in the case. Contrarywise, as Lewis Carroll would say,the value-basedview mightgo ‘agent-centred’ and claim that all reasons for A are grounded in what is of value for or to A, rather than in what is of value taut court. This shows that in considering the relation between internalism and the distinction between desire-based and value-based reasons we need to bear two questions in mind. The first is whether we are dealing with a necessary condition for a reason or with the possible grounds for reasons. The second is whether that which stands as condition, or as ground,is to be thought of in an ‘agent-centred’ way or not. The desire-based view, as I have characterized it in the text, concerns grounds, but in an agent-centred way;the value-based view, as characterized, concerns grounds, but in an agent-neutral way.A neutral version of the desire-based view would distance it noticeably from internalism, which is clearly agent-centred.But it is normal to think of the desire-based viewin an agentcentred way,i.e. as groundingreasons for A hi desires of A’s.
Internalism does not need to claim anything more than a necessary condition for something to be a reason if it is to serve Bernard Williams’ original purpose, which was to show that if one cannot be taken from one’s present motivationalstate, understoodroughly as a set of present desires, to a state in which one has an appropriate desire, one has no good reason at all. Internalism can restrict itself to asserting that if our desires were different, we would have different reasons. It need not identify our desires, actual or hypothetical, with our reasons, nor suppose that our reasons are given by, or provided by, our desires; nor need it claim, as Parfit at one point puts it, that our desires are the ‘source of our reasons. *
* The same goes for facts about our desires, rather than those desires themselves.
Internalism of this limited and non-reductive sort * amounts to a constraint on the reasons that we have, and to no more than that. **
* I do not mean by this to suggest that this sort of limited internalism is the only non-reductive form of internalism. One can claim that for A to be motivated in condition C is both necessary and sufficient for A to have a reason, while still stressing that the fact that A has such reason is a distinctive further fact, over and above the facts about actual and potential desires or motivations of A’s. The same applies if one takes internalism as a theory about what gives us our reasons (the grounds for reasons) rather than merely about conditions under whichwehave reasons.
** Reductive forms of internalism are of course different in this respect, but only because, being reductive, they suppose a far tighter relation between (R) and either (D) or (M); this tighter relation derives from the reductiveness of the internalism, not from the internalism itself. (For (R), (D), and (M), see n. 4 above.) This will be true whether reductive internalism comes in the analytic or the nonanalytic form, i.e. whether it claims that the meaning of (R) is given by the disjunction of (D) and (M), or merely that the fact reported by (R) is the same fact as that reported by the disjunction of (D) or (M). For all these distinctions, see Parfit (1997).
Now if we take internalism in this way, Parfit’s claim that ‘Internalist theories are desire-based theories’ becomes a mistake. We can abandon desire-based theories and still stick to this form of internalism.
Parfit writes:
According to normative Internalism:
(A) Some acts really are rational.There are facts about these acts, and their relations to our motivation, which give us reasons to act in these ways. (p. 129)
But our limited internalism does not say this. What it says is rather:
(A*) Some acts really are rational. There are facts about these acts, and their relation to our motivation,without which we would not have reasons to act in those ways.
The characterization of internalism in terms of ‘giving reasons’ introduces a claim that goes beyond A*. There is, then, a possible version of internalism that is weaker than the one that Parfit is expressing here, that allows that reasons are never psychological states of the agent, or ‘facts about his motivation’, but that still insists on a certain relation to desires of the agent if anything is to be a reason for that agent. It requires that any reason mesh in a certain way with the motivation of the agent for whom it is a reason, without demanding that such motivation create, generate, or be the source of the reason. We do not argue against this view by showing that some reasons are not ‘desire-based’. The only objection to it that I can find in Parfit’s paper—and, like many of Parfit’s best objections, it is perfectly simple—is that according to internalism, ‘those who were sufficiently ruthless, or amoral, would have no duties, and … could not be held to be acting wrongly’ (pp. 102-3). To run this objection we have to hold that if someone is acting wrongly, he has at least some reason not to do what he is doing, whatever his actual or potential desires.
6. Looking Back and Looking Forward
This chapter has been largely concerned with laying the ground for what follows, and constructing the terms in which the debate is to take place. I have to apologize for trying my readers’ patience with the plethora of distinctions that it has contained. The excuse, of course, is that what is to come cannot be understood without at least some grasp on those distinctions, and I have, of course, to make it as clear as possible how I understand them. To lay my cards on the table, then, the next two chapters concern the possible grounds for a good practical reason. I will be arguing that such reasons are not based on the agent’s desires (Chapter 2),and that they are not based on the agent’s beliefs either (Chapter 3). In Chapter 4 I turn away from good reasons to the topic of motivation, and argue that pure cognitivism is the soundest form of psychologism in the theory of motivation as standardly conceived. In Chapter 5,however,I argue that we should not accept any form of psychologism anyway. We should attempt to understand the reasons that motivate us as features of the situation rather than as features of ourselves. Chapters 6-7 continue in this vein,buildingup the position I prefer and replying to objections, and the final chapter discusses whether the account of motivatingreasons that I have givenis either itself causal or at least compatible with a causal account.
APPENDIX
The History of the Distinction between Motivating and Normative Reasons
One of the main thrusts of this book will be that those who have recently worked on the distinction between good reasons and those that motivate have managed to mangle it more or less out of all recognition.They have turned a distinction between two contexts in which we use the notion of a reason into a distinction between two sorts of reason, in such a way as to tear apart something that needs to be kept in one piece if intentional action is not to be rendered unintelligible. In supposing that there are two sorts of reason, motivating ones and normative ones, they often claim historical precedent. In this Appendix I argue that those alleged precedents are not worth appealing to. The earliest propounder of the distinction,supposedly, was Francis Hutcheson, who wrote:
When we ask the reason of an action, we sometimes mean ‘What truth shews a quality in the action, exciting the agent to do it?’ Thus, why does a luxurious man pursue wealth? The reason is givenby this truth, ‘Wealth is useful to purchase pleasures’. Sometimes for a reason of actions we shew the truth expressing a quality, engaging our approbation. Thus the reason of hazarding life in just war is that ‘it tends to preserve our honest countrymen, or evidences public spirit’. The former sort of reasons we will call exciting,and the latter justifying. Now we shall find that all exciting reasons presuppose instincts and affections, and the justifying presuppose a moral sense. (1728: 404)
The first thing to say is that if this is really a version of the distinction between normative and motivating reasons, there seems to be no justification for thinking of all ‘exciting reasons’ as given by truths. For an agent may be ‘excited’, i.e. motivated, by a false conception of the situation and the prospects it offers. A man may pursue wealth on the grounds that wealth increases security,little reckoning that the wealthy man is the one who is the target of predators, so that wealth actually diminishes rather than increases security. (I am assuming for the purposes of the example that this is in fact the case.) If the exciting reason is that wealth increases security, Hutcheson is wrong if he supposes that all reasons are given by truths.This matter will become significant later on. I will only sayhere that Hutcheson appears not to be in favour of a rewriting of the supposed exciting reason which would make it the truth that the agent believed that wealth increases security rather than the falsehood that wealth increases security.
It may be, however,that Hutcheson intends both exciting and justifying reasons to come in the category of normative reasons, which do all need to be true—or at least to be ‘shewn’ by truths. If it is true that wealth is useful to purchase pleasures, this may be thought to be a good reason for pursuing wealth, at least for those whose inclinations lie in the direction of the luxurious.Understood in this way, the exciting reasons are just those good reasons for acting that appeal to ‘instincts and affections’ such as a predilection for luxury rather than to our moral sense. Prudential reasons will be exciting reasons, on this account, rather than justifying ones. Hutcheson is not drawing the distinction between normative and motivating reasons at all. He is distinguishing between two sorts of normative reason.
Hutcheson’s distinction was revived in recent times by William Frankena:
It seems to me, at any rate, that we must distinguish two kinds of reasons for action, ‘exciting reasons’ and ‘justifying reasons’, to use Hutcheson’s terms. When A asks ‘Why should I give Smith a ride?’, B may give answers of two different kinds. He may say, ‘Because you promised to’, or he may say, ‘Because, if you do, he will remember you in his will’. In the first case,he offers a justification of the action, in the second a motive for doing it. In other words, A’s ‘Why should I . . . ?’ and ‘Why ought I … ?’ are ambiguous questions … ‘Should’ and ‘ought’ likewise have two meanings (at least) which are prima facie distinct: a moral one and a motivational one. Thus a motive is one kind of reason for action, but not all reasons for action are motives. Perhaps we should distinguish between reasons for acting and reasons for regarding an action as right or justified. It is plausible to identify reasons for acting with motives, i.e. with considerations which will or may move one to action … but it is not plausible to identify motives with reasons for regarding an action as morally right or obligatory. (1958: 44)
Like Hutcheson, Frankena offers here two kinds of good reason for action, i.e. two kinds of normative reason. The first is a moral reason, and the second a prudential one. He is not distinguishing between the reasons for which we act and the reasons in favour of acting. He is distinguishing between two sorts of reason in favour of acting. But Frankena adds to what Hutcheson had to say something quite distinct and idiosyncratic. One sort of reason he calls a motive, and he supposes that when we speak of this kind of reason, the word ‘reason’ has a ‘motivational’ meaning. Moral reasons for acting are to be understood in a quite different way, however, as reasons for regarding an action as morally right. They are not motives, and the word ‘reason’ in the moral context does not have a ‘motivational’ meaning. It concerns reasons for regarding, i.e. for believing something or other, I suppose, rather than for acting. Moral reasons, it turns out, are not really reasons for action at all.
Now this announcement of Frankena’s has appeared from nowhere. What justification could there be for insisting that moral reasons are not ‘motivational’, and that moral reasons are reasons for regarding an action as right rather than reasons for doing it? Frankena just announces that an ethical justification for an action is not what he calls a motive for doing it, as if to give a good moral reason for doing it is not to give one a motive to do it.This can only rest on some antecedent conception of what a motive is.
What is more, one can argue that there is a general distinction to be drawn between the reasons why it is the case that p and any reasons there may be for believing that p. Suppose, for instance, that a respected and wise person (if such can still be found) has come to the conclusion that a certain course of action is morally dubious. This is perhaps some reason to believe that that course of action is dubious. But it is not itself among the reasons why that course is dubious. Those features that are the reasons why the course is dubious, such as, perhaps, that it involves an element of deceit, may also play the role of reasons for believing it to be dubious; but that role is a secondary role, which can be played by other considerations that are not themselves moral reasons. (This point is a general point, not especially a point about moral reasons. That the hedgehogs are hibernating early may be a reason to believe we will have a dry winter, but it is probably not among the reasons why the winter will be dry; the latter will more probably be facts about climatic shift, or whatever.) We should not, then, be tempted by any general analysis of moral reasons as reasons for judgement rather than for action.
Of course we should remember that Frankena’s discussion of Hutcheson’s distinction is produced in the course of an argument about the rights and wrongs of what is called an ‘internalistic’ conception of moral judgement, the conception according to which it is impossible for an agent to make a sincere moral judgement and not to be motivated accordingly. (This is Moral Belief Internalism, in Parfit’s typology, to be kept severely distinct from the Williams-type internalism I have been discussing above; see again Parfit 1997.) Frankena’s argument against this internalism is certainly helped by imposing on the debate the idea that moral reasons are only reasons for regarding an action as right, leaving the question whether to do that action as yet unaddressed. But this imposition is surely contentious in general, and especially so in that context.
It is one thing to announce that moral reasons are not motives, and another to announce that other reasons, such as prudential ones, are motives. The purpose of claiming that other reasons are motives consists, I think, in the underlying idea that being a motive consists in a tripartite relation between the consideration that is the motive, the agent, and the action, while being a moral reason consists only in a bipartite relation between the consideration that is the reason and the action, the relation of making it true that the action is right or justified or wrong or unjustified. A motive involves the agent in a way that a ‘mere’ reason does not. We might suppose that there cannot be motives that nobody has got, since a motive is essentially something belonging to an agent; but there can be moral reasons that nobody has got, since the status of such a reason as a reason does not depend on any relation to an individual agent. Reasons can exist in a sort of void, but motives cannot. I do not recommend this picture at all, but something like it may underlie Frankena’s pronouncements.
I leave this matter here, only noting that to call non-moral reasons motives is to open a door, the door that leads to thinking of desires as motives, or of motives as desires, and hence to thinking of ordinary practical reasons as essentially related to desire. This matter is the topic of the next chapter.
I have argued in this Appendix that the simple distinction between reasons that motivate and normative or good reasons which I am interested in is not to be found in either Hutcheson or Frankena. In fact, the best account of the distinction that I know of is to be found in Baier (1958: 148-62).* Baier’s general picture of the relation between the motivating and the normative (for which he uses the term ‘justifying’) is very congenial to the views I shall go on to propound. In particular, he is careful not to talk of two types of reason, but of two contexts in which the word ‘reason’ is used. These are the context of explanation and the context of deliberation and justification. Baier suggests that in both contexts we specify facts (or, in the context of explanation, maybe just supposed facts; see p. 159) that are taken to be good reasons for acting. In the case of the context of explanation,it will be the agent who supposes this;in the context of justification, the enterprise of justification requires that the facts offered be good reasons. Baier is especially interested in using this distinction to explain what he takes to be the mistaken view that facts can be intrinsically motivating. He explains this by imputing to those who hold it a sort of conflation of the two uses of ‘reason’. To be mentioned as a reason in the context of motivation, a (supposed) fact must indeed motivate.(Things are even worse if one speaks of reasons as motives. Baier writes (p. 157), ‘If reasons are not distinguished from motives, it is obvious that… if they are facts, then the moving power must somehow be lodged in these facts.’)
* I should also cite Joseph Raz’s distinction between guiding and explanatory reasons, discussed in his introduction to Raz (1978)
But to be a good reason, i.e. to be mentioned as a reason in the context of justification, the fact that is the reason need not be motivating anybody at all. Normative reasons, therefore, are not intrinsically motivating.
The fact that the supposed history of the distinction between good reasons and those that motivate turns out not to be quite as generally supposed should not disconcert us. As I see it, the basis of the distinction, when it is properly understood, is not itself either philosophically deep or significant, nor does it need any great acumen to discover or understand, in the ordinary sense of ‘understand’. It is what philosophers have said about the distinction, that is,the castles that they have erected on it, that are philosophically challenging,for very significant matters depend on our philosophical accounts of motivation, of good reasons, and of the relation between them. The construction of a distorted history is just one sign that things are going wrong.
Hutcheson’s distinction amounted to a general distinction between moral and other good reasons, related to a (contentious) distinction between instincts and affections, on the one hand, and a ‘moral sense’, on the other. Frankena’s distinction again started from a general distinction between moral and other good reasons, one which was itself quite inca pable of sustaining the position he then erected upon it. The rights and wrongs of the views of these two writers will not occupy us further. The distinction between moral and other reasons is not the topic of this book, whose general assumption is that there may not be a huge difference between the two.
Be the first to comment on "Jonathan Dancy: Practical Reality (2000)"