1
The Humean Theory of Motivation Rejected [1]
G. F. Schueler
University of New Mexico
Hume famously wrote that “[r]eason is, and ought only to be the slave of the passions, and can never pretend to any other office than to serve and obey them.”[2] Though defenders of the contemporary account of rational motivation that follows Hume’s view would not put this quite so strongly, they still hold that desires, or what Hume called ‘passions’, broadly understood, are necessary features of motivation. In this paper I will argue that this view is mistaken.
Hume proposed an account of rational motivation, but it is important to recall at the outset that the term “rational” has both normative and descriptive senses. The normative sense of ‘rational’ implies that there is some (unspecified) standard and of course that entails the possibility that not everyone lives up to that standard, whatever it is, at least not all the time. In this normative sense ‘rational’ contrasts with ‘irrational’ (or maybe ‘less rational’). But there is also a descriptive sense of ‘rational’ in which it seems to be an important fact about humans that, even when they don’t live up to the standards of rationality, they are still rational in a way that other things in nature are not; they do things for reasons. In this descriptive sense ‘rational’ contrasts with ‘non rational’. It is in this descriptive sense that the Humean Theory is a theory of rational motivation.
Other objects in nature beside human beings can be correctly said to ‘do’ things. Storms flood cities. Trees sprout buds in the spring. But to say that humans are rational in the descriptive sense is to say that some of the things we do can be explained in terms of our reasons for doing them, however good or bad those reasons are. That is, it is to say that these things can be explained with a form of explanation – an explanation in terms of the agent’s reasons – that contrasts sharply with the explanations we give for the things that storms and trees do. When a person performs some action, it always makes sense to ask what her reasons were for doing whatever she did. This is not so for storms or trees.
Most of the things we do, and do for reasons, are not preceded by any explicit or conscious process of trying to figure out what to do, i.e. by practical deliberation. But of course sometimes we do explicitly deliberate, and since it only makes sense to think of rational beings as deliberating at all, one way to focus the difference between explanations of actions in terms of rational motivation and explanations which do not involve appeal to the agents’ reasons is to think about cases where such deliberation does take place and is acted on. Practical deliberation is the process of reasoning about, trying to figure out, whether to do something and even when someone doesn’t consciously deliberate before acting it makes sense to understand her reasons for doing what she did as the things that would have come into her deliberation had she actually deliberated and then acted on the basis of that deliberation. The normative questions here are questions about what sorts of things should be considerations in some instance of deliberation, and how much they should be counted as ‘weighing’. The descriptive questions are questions about what considerations actually did move someone on some specific occasion to perform some action. The Humean Theory of Motivation is a theory about how that sort of motivation works, that is, about how agents’ reasons explain their actions. Of course it should apply whether or not conscious deliberation takes place.
Someone might be tempted to say that Hume’s actual account of motivation wasn’t really an account of ‘rational’ motivation because, of course, Hume did not think ‘reason’ came into motivation at all, as the quotation above about reason being the slave of the passions dramatically illustrates. But for the purposes of this paper, this is just a terminological point. Hume certainly intended his account to apply to the sort of intentional actions that humans engage in but storms and trees do not; that is, he was discussing ‘rational motivation’ in the descriptive sense.
It is just that Hume also tried to get some argumentative mileage out of his terminology by packaging it together with a substantive account of how ‘reason’ works. He held the contentious view that human reason deals only with ‘relations of ideas’ and ‘matters of fact’, that is, roughly with logical or mathematical beliefs and with factual and causal ones, because these are the only sorts of things that can be true or false. That is not just a terminological point though, since he also claimed that discovering or figuring out some mathematical relation or causal connection won’t by itself move anyone to do anything at all. The question for him is then what more is needed to move someone to act and he held that it is a special, motivational state, which he called a ‘passion’. So even discovering for example that the building is on fire, which is of course a matter of fact, will only move someone to get out of the building if she has a desire not to be burned. Lacking such a desire (or some analogous one) the discovery that the building is on fire will not move her to act any more than will any other discovery about which she cares nothing, according to this view, such as that there are an odd number of chairs in the room. And given Hume’s view of the scope of reason, as applying only to things such as beliefs that can be either true or false, he thought that passions, that is desires and other such motivational states, could not be either supported by or opposed by reason. Like hunger or thirst, passions can perhaps be explained causally, but that is essentially the whole story[3]. Either we have them or we don’t and that is it.
This is a substantive and contentious conclusion that goes well beyond terminological issues about where to apply terms such as ‘rational’ and ‘reason’. Contemporary advocates of the Humean Theory of Motivation agree with Hume in holding that a desire or some analogous motivational state is always needed to move anyone to act – that is the heart of this ‘theory’ - but they don’t always think of desires as utterly outside the scope of reason. (I’ll say a bit more about this below) They agree that rational agents do things that can be explained in terms of their reasons. That is central to being a rational agent. Humeans though disagree with non-Humeans about the place of desires in explaining actions,[4] that is, they disagree over whether an agent’s reason must always involve a desire of some sort.
Both Humeans and non-Humeans are trying to answer Davidson’s question: “What is the relation between a reason and an action when the reason explains the action by giving the agent’s reason for doing what he did?”[5] The disagreement is over how to understand what an agent’s reasons essentially are, that is, over what things or sorts of things the phrase ‘the agent’s reasons’ refers to or, more generally, what the form or structure of explanations of actions in terms of agents’ reasons really is. Humeans hold that a desire (or other such motivating state) must always be part of an agent’s reason for doing whatever she does, or at least must figure essentially into any such explanation. Non-Humeans deny this.
In this paper I will argue that the latter group is correct. My argument focuses on practical deliberation and has two parts. I will discuss two different problems that arise for the Humean Theory and suggest that while taken individually each problem appears to have a solution, for each problem the solution Humeans offer precludes solving the other problem. I will suggest that to see these difficulties we must take seriously the thought that we can only understand an agent’s reasons for her action by looking at her actual or possible practical deliberation. So let’s look at the first problem.
1.
The Humean Theory is well summed up by Michael Smith, following Davidson, in a principle Smith calls ‘P1’:
R at t constitutes a motivating reason of agent A to phi iff there is some psy such that R at t consists of an appropriately related desire of A to psy and a belief that were she to phi she would psy.[6]
Davidson puts his version of what looks like the same claim in a principle he calls ‘C1’:
R is a primary reason why an agent performed the action A under the description d only if R consists of a pro attitude of the agent towards actions with a certain property, and a belief of the agent that A, under the description d, has that property.[7]
I will assume here that by “a motivating reason of agent A to phi” Smith is referring to the same thing Davidson calls “a primary reason why [the] agent performed the action”, i.e. to what in the more ordinary terminology I used above (and the terminology of Davidson’s original question) would be called the agent’s reason (or one of the agent’s reasons) for doing what she did. Beyond this terminological difference, though, these two principles differ in some other ways. Davidson speaks of a “pro attitude”, Smith of a “desire”. Davidson’s principle gives only a necessary condition for a reason. Smith says an agent has a motivating reason “[if and only if]” she meets the conditions he describes. But it is a third difference on which I want to focus. Smith says the desire in question must be “appropriately related” to the belief, a phrase which has no analogue in Davidson’s version of this principle.[8]
Smith explains what he means by this phrase in the sentence following his statement of the principle. He says, “To say that the desire and belief must be ‘appropriately related’ is merely to acknowledge that in order for a desire and belief to constitute a motivating reason the agent must, as it were, put the relevant desire and belief together.”[9] This seems to me to be an important claim, one worth examining.
We can see what is at issue, I think, if we shift briefly from ‘practical’ reasoning, where we are thinking about what some agent does, to so called ‘theoretical’ reasoning. Anyone who has ever taught (or, for that matter, taken) a logic course will be able to testify that it is possible for someone to fully believe premises which in fact entail some conclusion without realizing that this conclusion is entailed and hence without actually drawing this conclusion, that is, coming to believe it. This is easy enough to see if one thinks of complex premises in difficult arguments, but actually, human reasoning ability being what it is, the phenomenon is ubiquitous. Books of ‘brain teasers’ and ‘logic puzzles’ are full of examples.
So in the case of theoretical reasoning, if we want to explain why someone believes some conclusion, it is not enough just to cite beliefs she already has which entail the conclusion in question. Entailment is merely a relation between or among the propositions she believes. It tells us nothing about the genesis of the beliefs themselves. If we want to explain why she believes a conclusion entailed by premises she believes we must also add that she has ‘put together’, in Smith’s phrase, the various premises, that is, we must add that she has noticed or figured out that this conclusion is entailed by what she already believes. Otherwise we leave open the possibility that she acquired her belief in the conclusion in question in some other, perhaps completely irrational, way.
It is tempting to put this point by saying that she must also believe that the premises she accepts entail the conclusion here but that is really not right either. It is not enough, or even strictly relevant, that she realizes, that is, holds the belief that the premises she accepts entail the conclusion in question, since she might still fail to put that belief together with the (other) premises in such a way as to come to believe the conclusion.[10] So, on pain of regress, ‘noticing’ or ‘putting-together’ must be understood as a sort of activity rather than as just another belief of the same sort as her beliefs in the premises. But this activity is still a cognitive activity, one that can go wrong. We do, after all, sometimes reason fallaciously, i.e. we sometimes draw conclusions from premises that don’t entail or even support them. In Smith’s terminology this is to say that we sometimes ‘put together’ things in ways we shouldn’t.
To return to the two principles quoted above, this is what the phrase ‘appropriately related’ in Smith’s principle adds to Davidson’s claim. (Of course, in Davidson’s defense, if we are only making claims about the necessary conditions of explaining someone’s belief in some conclusion in terms of her reasons, as Davidson is in C1, then this ‘putting-together’ point need not come up. It could be a necessary condition of such an explanation being correct that the person doing the reasoning hold the relevant premises as beliefs even if it is alsoa necessary condition that she ‘put them together’.) In theoretical reasoning we can explain the agent’s coming to believe some conclusion on the basis of beliefs she has that support (or seem to her to support) this conclusion only if we add that she has ‘put together’ these beliefs in such a way as to actually draw the conclusion. Similarly for practical reasoning, Smith is claiming, it is not enough, in explaining an action, that the agent merely has the relevant desire and belief (even if that is, as Davidson says, a necessary condition). She must also put these together. Without this extra claim the Humean Theory is open to counter examples since someone might have both a desire and related belief but not act on them because she failed to ‘put them together’.
Suppose I want to get to campus and know perfectly well that the bus that stops at my corner goes right there. (I just this morning explained it to someone perhaps.) Still, when my car doesn’t start and I am frantically trying to make it to class on time, I may not put these two things together. I may have gotten so habituated to driving to campus that it never occurs to me to consider taking the bus to get there even though, if I just stopped and thought about it for a moment, I would realize that I could do that. If, in these circumstances, I do in fact get on the bus, this particular desire-belief pair (my desire to get to campus and my belief that this bus will take me there) will not be what explain my action, even though I do in fact have both this desire and this belief. Some other explanation will have to be the correct one. Perhaps I decided to take the bus to my sister’s office, which is only a short walk off the bus route, to see if I could borrow her car.
If this putting-together point about explanations of actions in terms of the agent’s reasons is correct, it is an important, in fact essential, feature of the Humean Theory since it is needed to block counter examples of the sort just described. At the same time, however, it raises a serious question for the Humean, desire-belief account of action explanations, including the one Smith himself proposes. Again, it may be easiest to see the problem by comparing theoretical reasoning. If we are explaining why I believe q by citing my belief that p and my belief that if p then q, then, as just explained, we also have to add, or at least implicitly assume, that I ‘put these two things together’ since otherwise the explanation won’t work. I could hold these two beliefs and yet never draw the conclusion that q. ‘Putting these two things together’ is reasoning, a term which refers to this mental activity (of drawing a conclusion), not to, e.g., consciously rehearsing the relevant sentences to myself or the like, which is neither necessary nor sufficient for reasoning. But as I have just argued this reasoning, that is, this extra activity beyond merely believing the premises, is essential to the explanation. Without it my belief that q won’t be explained even if in fact I believe some premises that entail q.
So Smith’s putting-together point is really just the thought that desire-belief explanations of actions rely for their explanatory force on the fact that the agent whose action is being explained is engaging in some practical reasoning, whether or not this reasoning (the ‘putting together’) is conscious or explicit, which presumably it usually is not. Colin McGinn, in his discussion of one specific account of practical reasoning, the practical syllogism, is I think expressing the same thought when he writes,
‘Whenever someone acts for a reason we can assume some such reasoning [as is represented in the practical syllogism] to have occurred. We can thus say that an action is a bodily movement issuing from such practical reasoning as is codified in the practical syllogism.’[11]
Though I don’t want to commit myself to the thought that the practical syllogism is the correct account of practical reasoning, this doesn’t matter for the point I am making here.
The practical syllogism looks like the ideal way of understanding practical reasoning if one is an advocate of the desire-belief account of action explanation, that is, the Humean Theory of Motivation. It has a place for representing the desire (or pro attitude) in the major premise, a place for representing the associated, instrumental belief in the minor premise, and a place for representing the action in the conclusion, which is the judgment on which one acts, presumably a judgment to the effect that this is the best thing to do or that this is what one should do or will do. As Robert Audi puts it,