Can Instrumental Reasoning Stand Alone?

Robert Bass

I. Introduction

There is something appealing about ordinary instrumental or means-end reasoning. One begins with a want, a goal or a desire and considers available options as means to its satisfaction or achievement. If, among the available options, one is the best or only way to satisfy the desire or achieve the goal, one has a reason to select it. If two or more options both seem to lead to the goal, they may still differ in other ways, e.g., in the probability with which they lead to the goal – in which case (if that was the only difference) one would have reason to choose the option which led to the goal with higher probability.

To consider things in the simplest form possible, consider a being with only a single desire. Suppose that this being wants nothing but to break a street-lamp. Even in so simple a case, we can begin to say what he ought to do. Any number of things may be effective. If he has no other goals – not even going unapprehended so that he can do it again with some other street-lamp – he may use a rifle, a pistol, throw rocks at it, climb the lamp-post to bash it with his fist, etc. But we can say that there are some things that, in terms of his goal, he ought not to do, for example, that he ought not to try breaking it (because he won't succeed) by throwing feathers at it, one by one.

It looks as though, even in this deliberately simplified case, means-end reasoning, combined with some knowledge of the world, is enough to tell us something about what he ought to do. This is not, to be sure, a moral 'ought,' but we seem to have generated a normative conclusion, an ought-judgment of a modest sort, without appealing to any mysterious non-natural properties or other entities or processes difficult to fit into a naturalistic world-view.

Of course, with a richer set of goals and motivations, things become more complicated. It may even be, some have thought, that if one allows for sets of goals sufficiently rich to be typical of human beings, one can get a moral theory out of instrumental reasoning.¹ It is tempting to think or to hope that instrumental reasoning can be autonomous, that either we can understand all practical reasoning in terms of instrumental reasoning or that, at least, we don't need to understand any other forms of practical reasoning, if there are any, in order to understand instrumental reasoning.

However, in a recent, closely-argued paper, "The Normativity of Instrumental Reasoning,"² Christine Korsgaard has challenged this:

Most philosophers think it is both uncontroversial and unproblematic that practical reason requires us to take the means to our ends. If doing a certain action is necessary for or even just promotes a person's aims, the person obviously has at least a prima facie reason to do it.... But philosophers have, for the most part, been silent on the question of the normative foundation of this requirement. The interesting question, almost everyone agrees, is whether practical reason requires anything more of us than this....
My arguments ... have [an] implication which I will be concerned to bring out in the course of the paper, namely, that the instrumental principle cannot stand alone. Unless there are normative principles directing us to the adoption of certain ends, there can be no requirement to take the means to our ends. The familiar view that the instrumental principle is the only requirement of practical reason is therefore incoherent.³

Whether or not one aspires to an account of practical reason which will, finally and in the last analysis, leave the instrumental principle as "the only requirement," this seems an unsettling claim. It suggests that we cannot simply get on with the mundane, but reasonably well-understood, business of instrumental reasoning without settling apparently more difficult questions as to what ends we are rationally required to adopt or endorse.

I think that, in the end, Korsgaard is not quite saying this. She is not saying that we need to solve other problems in order to be justified in the instrumental reasoning in which we engage. Rather, she is taking it for granted that we are justified in instrumental reasoning (or, at least, are not always unjustified) while arguing that the justification must appeal to other principles or requirements of practical reason.

This is still a challenging claim and one that has a considerable bearing on what may be possible in ethical theory.⁴ What I wish to do is examine her argument. I shall claim that her argument is substantially correct, but that the conclusion she draws (and is entitled to) is somewhat less dramatic than it appears. The second section will be devoted to an examination of the central features of her argument, while the third will address the question as to what thesis she has actually established and, since I agree that she has established that instrumental reason cannot be autonomous, offer a proposal that at least partially reconciles her argument with the position of those who would like to view instrumental reasoning as autonomous.

II. Why Instrumental Reason Cannot Stand Alone

In developing her argument, Korsgaard addresses alternative positions, of those she terms "empiricists" and of those she terms "dogmatic rationalists," in considerable detail. The core of her argument, though, can be set out without the detail.

We can begin by specifyingwhat Korsgaard means by "the instrumental principle." Briefly, the instrumental principle just claims that a person has reason to take or adopt means to her ends. In the limiting case, where only a single end is in question and only a single option qualifies as a means to that end, the person ought to take that means (to achieve her end).

Her central argument can be put in the form of a dilemma: If we are justified in relying on the instrumental principle, then instrumental reasoning must be capable both of guiding us and of motivating us. However, in the absence of some other normative principle, instrumental reasoning can perform one of these functions only if it fails to perform the other.

So, what do these two requirements come to? Let's begin with a consideration of the sense in which the instrumental principle must be able to guide us. According to Korsgaard,

[L]et us say that a rational agent is one who is motivated by what I will call the rational necessity of doing something, say, of taking the means to an end, and who acts accordingly. Such an agent is guided by reason, and in particular, guided by what reason presents as necessary. A comparison will help illustrate the point. If all women are mortal, and I am a woman, then it necessarily follows that I am mortal. That is logical necessity. But if I believe that all women are mortal, and I believe that I am a woman, then I ought to conclude that I am mortal. The necessity embodied in that use of "ought" is rational necessity. If I am guided by reason, then I will conclude that I am mortal. But of course it is not logically necessary that I should accept this conclusion, for if it were, it would be impossible that I should fail to accept it. And it is perfectly possible for someone to fail to accept the logical implications of her own beliefs, even when these are pointed out to her. A rational believer is guided by reason in the determination of her beliefs. A rational agent would be guided by reason in the choice of her actions. (pp. 8f.)

Though there may be difficulties in understanding how this kind of guidance is possible, the crucial point is that it must be guidance. In particular, if I am to be guided by instrumental reasoning (or by any other rational principle), then it must be possible that I should fail to be guided by the relevant principle. If I cannot fail to be guided, I am not being guided; rather, I am (perhaps) being caused to conform to the principle.⁵

Now consider the other requirement, that the instrumental principle must be able to motivate us. This may be presented as relying upon a thesis about the internality of reasons, the claim that for something to be a reason for a given person, it must be possible for it to motivate her.⁶ The central claim here, which I take to be correct, is that something cannot be a reason for me to act in one way rather than another if it is not something that I could (in some appropriate sense) have a motivation to act upon. Stating the requirement with real precision is not easy, but also, for present purposes, not necessary. For it will turn out that, if the guidance requirement is met and if there is no other normative principle involved, that any plausible version of the internalism requirement will be violated.

So, if the instrumental principle is to be a requirement of practical reason, it must both guide and motivate. Why does Korsgaard think this is a problem? Consider first the kind of answer that Korsgaard attributes to empiricists – by which she means those who follow Hume in thinking that reason does not have a role in determining our ends. "On an empiricist view, to be practically rational is to be caused to act in a certain way – specifically, to have motives which are caused by the recognition of certain truths which are made relevant to action by one's pre-existing motives." (p. 5) The instrumental principle says that an agent has reason to take means to her ends. Well, what are an agent's ends? If reason has no role in determining ends, then the empiricist has to take the agent's desires to be her ends. But this is problematic. If the desires are enough, when combined with the agent's knowledge and/or beliefs to cause her to take means to satisfy them, then the motivational problem is solved – there's no problem saying what motivates enacting the means to the satisfaction of desire – but the instrumental principle is providing no guidance, because, ex hypothesi, the desire is sufficient to cause the action. Hence, there is no possibility that she would fail to act in accordance with the instrumental principle – which possibility is, as we saw above, essential if the principle is to guide.

Suppose on the other hand, that the agent's desires are not sufficient (in conjunction with her beliefs) to cause her to enact means to their satisfaction.⁷ Then, it might seem that the instrumental principle could tell her to take the means to satisfy desire. But this will not do, either.

[I]f you hold that the instrumental principle is the only principle of practical rationality, you cannot also hold that desiring something is a reason for pursuing it. The principle: 'take as your end that which you desire' is neither the instrumental principle itself nor an application of it. If the instrumental principle is the only principle of practical reason, then to say that something is your end is not to say that you have reason to pursue it, but at most to say that you are going to pursue it.

In the case that we're considering here, where desire is not sufficient to cause one to enact means to its satisfaction, the person has no reason that the empiricist is in a position to recognize to enact those means unless there is some further normative principle than just the instrumental principle.

It might be suspected that this argument turns upon a trick. Why, the empiricist might ask, must the instrumental principle be formulated so that there is a gap between end and desire that stands in need of being bridged by some further normative principle? Why not simply formulate the principle to refer to desire in the first place? – "One has reason to take means to the satisfaction of desire." However, this doesn't really escape the dilemma: If desire (together with belief) is sufficient to cause one to take those means, one will – but will not be able to fail to do so and, hence will not be guided by the instrumental principle, while if desire is insufficient, the principle will fail to motivate.

So, the general problem with the empiricist approach to instrumental reasoning is that insofar as it succeeds in addressing the motivational issue, by providing a connection to desire, it fails to treat the instrumental principle as normative or guiding. But if the empiricist view does not work, Korsgaard does not think that the rationalist alternative fares any better. The rationalist's problem is a mirror image of the empiricist's: he can understand and account for guidance, but cannot explain how we can be motivated to conform to the instrumental principle.

The empiricist's problem came from the assumption that all ends ultimately derived from desire, there was nowhere else for ends to come from and nothing that could be meant by the claim that we have an end but that we are caused to pursue it. The rationalist is willing to deny that and admit that there may be rational principles which (somehow) may or should constrain or direct desire.

So, the rationalist has no general problem with how the instrumental principle can guide us. It tells us to take the means to our ends – which we may or may not do. His problem is the different one of explaining the connection between the instrumental principle and our motivations.

In order to explain that connection, the rationalist has to say something about the way in which having something as an end is or may be distinct from desiring it. According to Korsgaard, "the instrumental principle can be rescued only if we take 'my end' to be something other than 'what I actually, just now, desire.' One possibility is to distinguish desire from volition, and to say that your end is what you will, not merely what you want."^{8, 9}(p. 23)

The problem at this point, of course, is that any number of action-guiding principles might be suggested, including the counter-instrumental principle that one should always avoid the means that lead to one's end. Even given that the rationalist doesn't insist that requirements of practical reason all derive directly or indirectly from desire, why act on the instrumental principle rather than some other inconsistent principle?¹⁰ In short, what is the status of the instrumental principle?

Typically, according to Korsgaard, the rationalist will say that the instrumental principle is somehow a logical truth. The hypothetical "imperative [the instrumental principle] derives the concept of willing the means from the concept of willing the end, with the aid of some synthetic proposition telling us what the means are." (p. 24) The question is: What kind of logical principle is this? If it says something like, "one always takes the means to one's end," (and if there is some way of identifying one's end apart from desire or, more broadly, watching what the person does), then it is not any kind of logical principle; it is simply false. If, on the other hand, the instrumental principle says something like, "one must take the means to one's end," we can ask what the role of the "must" is. If it is not just going to collapse into causal necessity, it will have to mean something like what Kant formulated (and called analytic¹¹ ): "Whoever wills the end, wills (so far as reason has decisive influence on his actions) also the means that are indispensably necessary to his actions and that lie within his power." (quoted on p. 24, emphasis added)

The Kantian formulation of the instrumental principle may provide a respectably logical or analytic principle for the rationalist to use, but how, Korsgaard asks, can a principle like that motivate? In her words, there is "a problem about how the analytic proposition is supposed to make it possible for the agent to combine willing the end with knowing the means to arrive at a rational requirement of willing the means." (p. 28) Somehow, the instrumental principle is going to have to do this, because, otherwise, one could trivially comply with the principle by claiming that refusal to employ means either indicated that something was not (ever) an end or that, in the light of further knowledge, one had revised one's ends. How, though, can the instrumental principle require one to will the means?

Suppose, for example, that I have an end and do not will the means. I find the means repulsive. On the version of the principle on which it is analytic,¹² it says that I will the means insofar as I am rational. How is that supposed to get a grip on me? Can't I just reply that I'm not or, at least in this case, don't care to be rational? It looks like I need a further normative principle to tell me to be rational.¹³

III. Toward a Richer Normative Theory

Summarizing, it appears that whether we take desire to be the ultimate source of all ends or whether we allow for some other source, we will have to admit that some other normative principle is needed (or other principles are needed), not just in addition to the instrumental principle but needed in order for it to be intelligible that we can be both guided and motivated by the instrumental principle.

Further, Korsgaard's argument makes it clear that it will not be easy simply to tack on some additional principle to one's prior theory of practical reason and solve the problems she has raised. The arguments she has offered and the problems that she has highlighted are likely to have a bearing on any simple modification of either an empiricist or rationalist theory. Her own proposal is that both should be replaced with a Kantian theory where the principles of practical reason, including the instrumental principle, are laws that we give to ourselves. If that is correct, then we can draw the distinction between will and desire that the rationalist employs to explain how the instrumental principle can guide without unanswerably raising the question how we can be motivated to conform to it.

She may be correct in this, but within the scope of this paper, I could not begin to adequately assess the correctness (or otherwise) of that claim. Much less, if the claim were judged mistaken, could I propose an alternative. However, I think that something useful may be said with respect to two issues. First, the conclusion that she reaches at the end of her paper appears less dramatic than what she had claimed to be arguing for at the beginning. Second, if we assume that there is some way, whether by means of Korsgaard's Kantian proposal or not, to have an instrumental principle that can both guide and motivate, then, given the fact that her actual conclusion is not so dramatic as the earlier announcement appeared, there may be room for a normative principle that amounts to a defense of the near-autonomy of instrumental reasoning.

The divergence between what Korsgaard initially appears to claim that her argument will show and what she ends up claiming is easy to establish. Near the beginning of the paper, there is a programmatic remark, already quoted above, with respect to what she expects to show: "Unless there are normative principles directing us to the adoption of certain ends, there can be no requirement to take the means to our ends." (p. 7) This certainly seems to say that certain specific ends will turn out to be required (if the instrumental principle is a requirement of practical reason). The friends of instrumental reasoning, especially its empiricist friends, will be quick to detect allusions to the categorical imperative or to values subsisting in the world of Forms.

However, near the end of the paper, Korsgaard is making what is plainly a more restricted claim: "I claimed before that what my argument showed was that hypothetical imperatives cannot exist without categorical ones, or anyway without principles which direct us to the pursuit of certain ends, or anyway without something which gives normative status to our ends." (p. 40) She does, it is true, offer some considerations in favor of saying that we have to attach normative weight to some of our specific ends:

If I am to will an end, to be and to remain committed to it even in the face of desires that would distract and weaknesses that would dissuade me, it looks as if I must have something to say to myself about why I am doing that – something better, moreover, than the fact that this is what I wanted yesterday. It looks as if the end is one that has to be good, in some sense that goes beyond the locally desirable. I have to be able to make sense to myself of effort and deprivation and frustration ... (p. 40)

But, she continues,

I do not have an argument that shows that this [committing oneself without judging anything to be good] is impossible. I suppose that through some heroic existentialist act, one might just take one's will at a certain moment to be normative, and commit oneself forever to the end selected at that moment, without thinking that the end is in any way good ... (p. 40)

Plainly, though Korsgaard is not claiming (any longer?) to have a proof that we must accept "normative principles directing us to the adoption of certain ends" (p. 7), she doesn't find the alternative she has described to be very attractive.¹⁴ (Nor do I.) Part of the reason that she finds no credible alternative to the view that rational principles "direct us to certain ends" is, I think, that she has simply overlooked a possibility.

To understand the possibility that she has overlooked, it must be clear that I take her, in referring to rational principles that direct us to particular ends, to mean that the rational principles would themselves pick out the ends for us, independently of or prior to desire or any desire-based argument. Thus, such a principle might direct us (immediately and without further ado) to value health. A principle that directed us to value health either because we directly desired it or because of the way that it fit in with and promoted the satisfaction of (or avoided the frustration of) other desires that we have would not, in this sense, be a rational principle "that directs us to certain ends."

I think, though I am not sure, that we can do without rational principles in this sense. And, if I am correct, we can do so without either rejecting the instrumental principle or finding ourselves in the position of making (or not making) ungrounded existentialist choices. I propose the following normative principle:¹⁵

(P) Desire is prima facie legitimate.

Almost every word of that sentence stands in need of further explanation. We can begin with "desire." I understand the term, as it is used in P, quite broadly. I mean it simply to cover any motivation whatever that may shape an action. I am not here concerned to distinguish what is desired from what is wished, or what is the product of whimsy or habit or preference. "Desire" is meant to cover them all.

The sense in which I wish to use "prima facie" is perhaps more familiar. I mean only that whatever is authorized or supported by this principle, whatever it declares legitimate, is so only in the absence of countervailing reasons. If there are countervailing reasons of some sort, then a desire may not be legitimate.

Last, I take "legitimate" in a very weak sense. I take it to mean "not-wrong to act upon" or, more precisely, "not wrong to take as a reason for action." If a desire is legitimate, then it is not wrong to take it as a reason for action or, ipso facto, to act upon it. If it is prima facie legitimate, it is not wrong to take it as a reason for action unless there is some further argument or reason against it. No stronger claim than that is being made.

I think that the principle, P, is both plausible and correct, but I do not claim to have a proof for it. (I am not trying to deduce an 'ought' from an 'is.') However, I think it can be given dialectical support in two ways.

First, something can be said in response to various attempts to deny it. Except for forestalling a possible misunderstanding, I shall only outline those attempts and claim (but not argue) that they are implausible.

How might P be denied? Obviously, its denial would be the claim that it is not true that desire is prima facie legitimate. However, there is at least one way that P can be misunderstood. The claim that desire is legitimate might be understood to operate with some conception of the goodness or rightness of action which can be denied or, at least, stands in need of defense. In fact, this objection comes from not noting how weak the legitimacy claim is: "Rightness" is not claimed; P only denies wrongness. Even someone who holds that there is no such property as rightness that can attach to some actions or reasons for action as distinguished from others and who, consequently, believes that there is no contrasting property of wrongness that can be attached to actions or reasons, can still – and, given his position, should – agree that it is not wrong to treat desire as a reason for action.

How else might P be rejected? It seems to me that, broadly speaking, there are only two ways, and both will require their proponents to make strong claims. One of these we can call the Original Sin Thesis. P might be rejected on the basis that all our desires are in some way wrong or corrupt. Or, P could be rejected on the basis of what we can call the Lower Nature Thesis. A subset of our actual desires and motivations is declared unworthy to guide action. It seems to me that both of these are, in the first place, implausible, and in the second, involve large commitments that are themselves difficult to defend.¹⁶

Second, something can be said about what it makes possible – namely, just about everything that the partisans of instrumental reasoning have hoped for. Since any preference or desire or motivation may be taken as a reason for action, then anything that can be derived or argued for as a condition on preferences will turn out to be defensible if all desire is counted as legitimate. At least, such results will be defensible to those who care about whether their preferences clash or are harmonious – who, of course, were the only ones to whom decision-theoretic arguments could be offered as advice, anyhow. Those who do not care or perceive possible frustration of preferences as an issue, will be unmoved – just as they were before.

On the whole, though, it appears that we can keep what was appealing about instrumental reasoning and, though we have to introduce some normative principles in addition to the instrumental principle itself, we do not appear to have to introduce one "that directs us to certain ends." If that is really necessary, it requires further argument than Korsgaard has offered in this paper.

Comments? I'd love to hear.