Can Reasons be Self-Undermining?

Abstract

The characterization of objective, normative reasons to φ as facts (or truths) that count in favor of φ-ing is widely accepted. But are there any further conditions that considerations which count in favor of φ-ing must meet, in order to count as a reason to φ? In this brief paper, I consider and reject one such condition, recently proposed by Caspar Hare.

The characterization of objective, normative reasons to φ as facts (or truths) that count in favor of φ-ing is widely accepted.¹ But are there any further conditions that considerations which count in favor of φ-ing must meet, in order to be a reason to φ? Caspar Hare (2011) recently proposed one such condition: reasons cannot be self-undermining. He relies on this restriction on facts (or truths) which are eligible to be reasons when arguing that there is a distinction between two different ways of thinking about what we objectively ought to do: the ‘ought of omniscient desire’ and the ‘ought of most reason’: some facts (or truths) are reasons for desiring that an agent perform an action, but they fail to be reasons for those agents to perform the actions in question. The distinction between two ways of thinking about the objective ‘ought’ is in turn put to use in arguing for solutions to some outstanding problems in normative ethics.² In this brief paper, I consider and reject Hare’s proposed restriction on what can be a reason.

Hare offers the following, somewhat intricate example to illustrate his proposals. Imagine that you can spin a wheel of fortune; if you spin and get a black, you get nothing; if you spin and get a red, you get $100; if you turn the offer down, you get $20. Nothing but the prize you will receive is relevant to your decision. What do you have most reason to do, accept or decline? Provided we are inquiring into your objective, normative reasons to act, the answer to this question is plausibly thought to depend on what will happen, were you to accept and spin the wheel. If you will get a black if you spin the wheel, you have most reason to turn the offer down, as $20 dollars is better than nothing; if instead you will get a red if you spin, you have most reason to accept it, as $100 is better than $20.

According to Hare, this impression is mistaken: even if ‘if you spin, you will get a black’ is already true prior to acting, and you know that it is true, you still have most reason to accept, because the seemingly decisive consideration ‘if you spin, you will get a black,’ true though it is, is not a good reason for you to decline the offer. This consideration fails to be a reason for you to take the offer because this consideration is self-undermining, and reasons are not self-undermining, in the following sense:

Reasons are not Self-Undermining (RSU): It cannot be that c is a reason for you to φ and yet, if you φ, then c will not be true (2011: 196).

How does RSU bear on Hare’s example? The crucial point is that the claim ‘if you spin, you will get a black’ is true only if it is also true that you will in fact spin the wheel; if you do not spin the wheel, then there is no fact of the matter about what would have happened if you had spun it. This claim is only true if you in fact spin because whether you will get a black or a red if you spin depends on exactly how you spin the wheel (the exact force you apply to the wheel in doing so, etc.). There is only a fact of the matter about the exact way in which you spin the wheel if you in fact spin, and so only then can there be a fact of the matter about whether you will get a black or a red if you spin.³ If you do not spin the wheel, there is no fact of the matter about what would have happened had you spun; this is just a different way of saying that if you decline the offer, ‘if you spin, you will get a black’ will not be true.

If, as Hare suggests, reasons are not self-undermining, then ‘if you spin, you will get a black’ thus fails to be a reason for you to decline the offer; once this (seemingly decisive) consideration against acceptance is excluded, the balance of reasons favors acceptance. It is, though, a reason for an omniscient onlooker to desire that you decline the offer; this is how the two ways of thinking about the objective ‘ought’ come apart. I do not think the argument is successful, because reasons can be self-undermining. In fact, many excellent reasons are self-undermining in the sense defined. Here are three examples:

• (1) Alice needs help. That Alice needs help is a reason for Bernie to help her. If Bernie helps Alice, then it will not be true that Alice needs help.

• (2) Charlene is hungry. That Charlene is hungry is a reason for her to eat the sandwich in front of her. If Charlene eats the sandwich in front of her, then it will not be true that she is hungry.

• (3) David is in pain. That David is in pain is a reason for him to take a painkiller. If David takes a painkiller, then it will not be true that he is in pain.

In all three cases, the consideration cited as a reason is self-undermining in precisely the way that RSU claims they cannot be if they are in fact a reason, yet these considerations all seem excellent reasons for performing the actions in question. In defense of RSU, Hare could either deny that the considerations cited as reasons in (1)–(3) are genuine reasons to act in certain ways, or he could restrict RSU’s application, and hold that while for some values for c (such as the ones cited in (1)–(3)), c can indeed be a reason despite being self-undermining, but that for some other values for c (such as the one cited in the wheel of fortune example), this is not the case.⁴ A closer look reveals that neither of these options is promising.

Denying that the considerations cited in (1)–(3) are reasons to act in the indicated ways is not without cost, as this denial seems to conflict with intuitive assessments about what is and what is not a good reason: on the face of it, these considerations are excellent reasons to perform the acts in question. More importantly, however, is that doing so is unwarranted given the motivation Hare offers for RSU, which appeals to the role that practical reasons must be capable of playing in good practical reasoning⁵:

Considerations are eligible to be reasons only if they can play a role in good practical reasoning, only if they are the kind of considerations that can move a good practical reasoner to act (2011: 196).

The considerations cited as reasons in (1)–(3) are, I take it, the kind of considerations that can—and would—move good practical reasoners to act in the indicated ways: Bernie would surely be reasoning well if he is moved to help Alice on the basis of ‘Alice needs help’, and the same can be said about Charlene and David. If anything, these agents would be bad practical reasoners if they were not moved to act by the considerations cited in (1)–(3). Therefore, these considerations do count as reasons. Given that these considerations are both reasons and self-undermining, RSU in its current formulation is false: reasons can be self-undermining.

Can a restricted version of RSU be defended? Restricting may look like a good move at first sight; after all, the considerations cited in (1)–(3) appear to be different in kind from the consideration that Hare wants to rule out in the wheel of fortune case. Perhaps, then, only some of our reasons cannot be self-undermining. The main problem facing a restricted version of RSU is that the result may turn out to not do any work. Here is why. If being self-undermining does not entail that a consideration is not a reason, then while there may well be some identifiable, non-empty set of considerations bearing (favorably or unfavorably) on certain actions, the elements of which are both (a) self-undermining and (b) not reasons to perform the actions that they bear on, what ground is there for holding that it is the fact that they are self-undermining which explains why they are not reasons?

Call whatever property it is that the elements of this set share which is not shared by self-undermining considerations that are reasons ‘X.’ If (as I have suggested) being self-undermining does not disqualify considerations from being a reason across the board, then it is either instantiating both the property of being self-undermining and the property of being X that explains why these considerations are not reasons, or some it is other property the elements of the set share which does the explanatory work; call this latter property ‘Y.’ If there is some property that we can fill in for Y, a restricted version of RSU comes out as a theoretically idle wheel. If the former, we need some story explaining why the combination of being self-undermining and being X disqualifies considerations from being reasons, a story on which it does not turn out that their being X explains why these considerations are not reasons, for in that case, the restricted version of RSU is also an idle wheel.

Is there a property that fits this bill? I do not have an argument that establishes that there cannot be such a property, but these brief remarks indicate what a successful defense of a restricted version of RSU would need to achieve. To my mind, it is unlikely that, given the rather stringent conditions for success just sketched, such a defense is forthcoming. I therefore conclude that for now at least, we can reasonably reject RSU altogether, and maintain that all kinds of reasons can be self-undermining.