No Unconditional Oughts?

Long ago (and far away?), Roger Bissell claimed, following some statements to be found in Rand's writings, that there were no unconditional shoulds. (I hope he won't mind if, in what follows, I refer to this as a claim about unconditional oughts, which seems to mean the same and strikes me as more euphonious.)

Now, it can immediately be agreed that this is true in some sense. There are no oughts that are not conditional upon anything whatever. There are not, for example, any oughts that apply to the behavior of rocks. (If I say, I dropped it, so it ought to fall, that's a different sense of ought, one that expresses an expectation about rather than a prescription for its behavior.) This, I take it, is not what he had in mind or what he meant to be denying (at least, not all he meant to be denying).

Along another dimension, I take it he did not mean (just) to be limiting the applicability of oughts to beings for which something can be good (or bad) or, therefore, to be saying that there are no oughts that are not conditional upon being such that something can be good (or bad) for it. Though we can say that some things are good for cockroaches or cockroach-kind, we do not (seriously or in the same sense that we apply oughts to human behavior) say that there is anything that cockroaches ought to do. (Besides, they seem to flourish well enough without attending to any advice we might proffer.)

Now, though I've taken the trouble to point these out, I don't think there's any danger that Roger is confusing his claims about unconditional oughts with either of these senses in which oughts are conditional. (Rand may be responsible for some confusion here in that she tended to blur the distinction between the beings for which something is good and the ones who face or may face moral oughts by saying such things as that a fish ought to live in water.) In the effort to achieve some clarity on this, I shall try to point to one more thing upon which oughts are conditional and then say how (I take it) Roger is claiming that oughts are conditional upon something beyond that.

Recognizing oughts, acting on them, understanding them to be addressed to oneself or others is conditional upon being what I shall call a reflective chooser. This may be the same as or part of what Rand meant (or should have meant) by calling people rational. But I think putting the matter in terms of reflective choosers (and reflective choice, etc.) helps to focus on the right issues without distractions derived from other connotations of rationality. A reflective chooser is one who not only makes choices or who acts voluntarily -- in the ordinary sense of the words, that includes many non-human animals -- but who is first, capable of thinking about and reflecting upon the choices she makes and second, capable of revising or altering future choices on the basis of the results of that reflective consideration. That phrase, on the basis of, is crucial. It's not enough that behavior changes because that can happen just through training. A dog may learn not to jump on guests -- by being punished for doing so or rewarded for not doing so -- but presumably does not do so because it has chosen not jumping (in general) over jumping. It has just (unreflectively) acquired a new habit. A reflective chooser is one who has the capacity to make her choices the object of (other) choices. And, insofar as a chooser is reflective, no choice is (in principle) exempt, though of course not all can be subjected to reflective scrutiny at once. (If any choices areexempt, if, say, there are irresistable drives that directly control behavior and thus are not open to modification even of where, when or how they are acted upon -- they are also beyond moral assessment.) To expand this slightly further to questions about how choosing is related to desire (or aversion or other motivations), I'd say that for a reflective chooser a desire is not simply and without qualification a reason for action but rather something that may be taken as a reason for action. Part of reflection's capacity to range over choices is that reflection can range over motivations as well. A reflective chooser can ask whether wanting something is a good (or good enough) reason for action.

This I take to be a general condition for the applicability of oughts. No ought applies, in the relevant sense, to a being who is not a reflective chooser. Here, we are getting within range of the main issue. Some people, myself for example and possibly some others on this list, think that being a reflective chooser is all that is needed for some oughts to apply. (Of course, there's no need to assume, and I do not, that all obligations or oughts flow from [nothing but] being a reflective chooser. There are further conditions for some.) Roger, it appears -- and I suppose some here will agree with him thinks that there is a further condition, beyond being a reflective chooser, that must be met for any ought to apply. He (I assume) would agree with me that being a reflective chooser is a necessary condition for the applicability of oughts. I think that, for some oughts, being a reflective chooser is also a sufficient condition, and I think he would disagree on that. We draw the lower bound or minimum condition for the application of oughts in different places.

So, what further condition does he -- and, on his reading, Rand have in mind? It appears that an obligation or some goal (or something else?) that implies the obligation must itself be chosen:

>A should or an ought is an obligation. And if there's one thing
>that Ayn Rand taught me about morality, it is that there is no such
>thing as an unchosen obligation. Unchosen obligations are the
>hallmark of intrinsicist theories of morality. Shoulds, rational
>or not, must be chosen by me in order to be my shoulds.
>Anyone else's good intentions notwithstanding, they may not
>choose my shoulds for me! If someone wants to tell me what he
>thinks I should do, he must justify it as a logical implication of
>some obligation I have chosen.

Now, this is not as clear as one would like. What, in particular, is choosing an obligation? Could I, for example, have an obligation never to step on cracks in a sidewalk by choosing to have it? Could I do all the (behavioral) things that are ordinarily involved in making a promise but choose not to have an obligation to keep it? Further confusion looms when he says Shoulds, rational or not, must be chosen by me in order to be my shoulds. Does Roger think there is a class of oughts, the ones that are not rational, that I can nevertheless choose (and therefore have)? Or should I be rational? Does that, too, depend on some prior choice? Then, there is the curious phrase that others may not choose my shoulds for me. That certainly sounds as if it is saying that others have an obligation not to. Is that obligation one that they can choose to have or not?

Roger comes closest to dealing with this cluster of questions with his example of a chosen obligation:

>If I want to be healthy, I must
>have proper nutrition. I have expressed my desire to be healthy.
>Therefore, I have logically obligated myself to pursuing proper
>nutrition. Q.E.D.

But matters are not clear even here. In the first place -- obviously! -- his example appears to make no reference to a choice at all, but rather to something he wants or desires. Since wants and desires are not typically chosen -- though it may be that something chosen results in one's having particular wants and desires -- it's not completely clear how this is relevant.

If someone can have an obligation just because of a desire which is not itself chosen, then Roger is mistaken in saying there are no unchosen obligations. Perhaps he would not mind this and could be content with the thesis that there are no obligations that are not conditional upon something (whether it be a choice or desire) -- something more than just being a reflective chooser. Now, I think two things can be said here. First, there are good reasons for resisting the assimilation of choices to desires, and second, even if we allow desires (in the place they need to occupy in this sort of argument), the problem that I will be concerned to point out can be exhibited in the same way. Since the issue I'm concerned about does not depend upon whether it is choices or desires or both upon which it is said that obligations are conditional, I will stick to talking about choices.

I think we can make sense of what Roger has in mind, if not of what he says, if we take him to be talking about instrumental or means-end reasoning.

[Before we go on, let me point to a slight additional complication. In what follows, I shall assume that the musts or obligations that occur are musts or obligations, other things being equal. They are things that we have reason to do or have, provided that there are no countervailing considerations. If there ever are or can be countervailing considerations, that will provide a different sense in which an obligation is conditional. But that sense is not the same as the one in which Roger is concerned to claim that all obligations are conditional, because it will appeal not just to choices (or desires) but to other obligations. To say that there is a (decisive) consideration against doing x that one otherwise had reason to do (i.e., should do) is to say that one ought not, all things considered, to do x. One can't make the case that all oughts are conditional if what they turn out to be conditional upon is other oughts.]

With these preliminaries out of the way, I think the argument would go something like this:

1. If health is my goal, I must have proper nutrition.
2. Health is my goal (in some sense that implies that I have endorsed or chosen pursuing it, not just that I want it).

3. Therefore, I must have proper nutrition. (1, 2)

This can be developed further with a subsidiary argument like:

4. In order to have proper nutrition, I must pursue it.
5. I must have proper nutrition. (3)

6. Therefore, I must pursue it. (4, 5)

These arguments are certainly valid and seem to be conditional upon having the goal asserted in the second premise. If I did not have that goal, all that could be validly inferred from the underived premises (1 and 4) would be:

6a. Therefore, if health is my goal, then I must pursue proper nutrition.

I want to note several things about this argument as I've reconstructed it. First, I think that it exhibits the sense in which Roger means to be claiming that obligations are all conditional. To fill out an argument of this form, you need to insert a premise (like 2) stating that something is a goal or has been chosen. Second, however, it only illustrates his sense rather than makes his case because the fact that a premise like 2 is needed does not show that one could not have an obligation to have some goal (that isn't conditional upon having some other goal or making some other choice) or that there are no other forms of argument that lead to obligations. Third, note that his actual conclusion, that he is obligated to pursue proper nutrition, does not appear in the reconstructed argument. Let's try to remedy that by rephrasing as follows:

1. If health is my goal, I must have proper nutrition.
2. Health is my goal (in some sense...).
3. Therefore, I must have proper nutrition. (1, 2)
4. In order to have proper nutrition, I must pursue it.

6b. Therefore, I am obligated to pursue it. (3, 4)

Plainly, that is not valid unless 'I am obligated to' means the same as or can be derived from 'I must have'. Let us assume that it does mean the same and rephrase once more:

1a. If health is my goal, I am obligated to have proper nutrition.
2. Health is my goal (in some sense...).
3a. Therefore, I am obligated to have proper nutrition. (1a, 2)
4. In order to have proper nutrition, I must pursue it.

6b. Therefore, I am obligated to pursue it. (3a, 4)

Once again, we've got a valid argument, this time with an obligation in the right place in the conclusion. But notice that we've also got an obligation in the first premise. Where is that supposed to have come from? Is there something from which it is derived, that does not itself use just the same kind of premise that already includes some notion of obligation? Let's try, generalizing a bit:

7. My goal is x.
8. I can't get x without doing or having y.

9. Therefore, I'm obligated to do or have y.

Clearly, that's not valid without the addition of a premise (or inference rule) of the form, 'If x is one's goal, and doing or having y is necessary to getting x, then one is obligated to do or have y.' That premise really is necessary to validly infer the conclusion because, without it, we could fill in:

8a. If x is one's goal, and doing or having y is necessary to getting x, then one is not obligated to do or have y

or

8b. If x is one's goal, and doing or having y is necessary to getting x, then one is obligated not to do or have y,

or

8c. If x is one's goal, and doing or having y is necessary to getting x, then one is obligated to kill oneself,

or

8d. If x is one's goal, and doing or having y is necessary to getting x, then one has no obligations with respect to anything.

And so on. But this necessary premise is precisely the kind of thing we we're trying to get rid of! It looks as though we can't do it.

And that is what I do think. We can't do without some general principle -- call it the instrumental principle -- that says that we have reason to take or must take or are obligated to take (hedged, as always, with a ceteris paribus clause) the means to our ends. Moreover, this principle is not and cannot be conditional on any particular goals or ends that we have because, without it, no goal or end could give rise to any obligation. The truth of the instrumental principle is needed for there to be any such thing as reason-giving force in means-end reasoning and therefore cannot itself be derived from means-ends reasoning.

Now, I do not believe that the instrumental principle is the only case in which the nothing-but-conditional-obligations model breaks down, but even a single case is enough to show that the model does break down. If there are no unconditional obligations -- for example, to take the means to one's ends -- there are no conditional obligations, either.

Rob

P.S. I have discussed some of these issues at greater length in Can Instrumental Reasoning Stand Alone?

P.P.S. Fruitful ground for consideration of other ways in which the no-unconditional-obligations model breaks down can be found in considering some of these questions:

What ought to be done if obligations derived from different goals are in conflict?
Is there any obligation to be rational?
Are there any goals that it is irrational to have?
Are there any goals that it is irrational not to have?

---
Rob Bass


A great many people think they are thinking when they are merely rearranging their prejudices.

-- William James

Let me clear up a couple of points first before addressing Allen Costell's post. First, I didn't use the term duty, as he did in titling his reply, mainly because it leaves a bad taste in Objectivists' mouths and might distract from other things I had to say. A second reason for avoiding it is that duty is more commonly associated with specifically moral obligations, while I did not mean to be so limiting the terms I used. As I was using 'ought', 'obligation', etc., I'm just referring to what people have reasons to do or act upon. (I'm not claiming there's a sharp distinction between moral and other obligations, but I also don't want to overlook differences: Compare [based on an example from Darwall] "if you want to commit a particularly bloody murder, you ought to use a chainsaw" and "if you want to commit a particularly bloody murder, you ought to reconsider.")

Allen complains, with respect to one of my examples, that

>Rand's position is not a deduction from premises, but an induction from
>reality. So as far as I see, expressing it in the form of a syllogism is
>unnatural and will inherently misrepresent her view.

This strikes me as unclear. If the claim is that Rand's view in general is an induction from reality, that surely does not preclude using deductive arguments within the general view. So understood, the objection just doesn't come to grips with anything I said.

Now, I don't think that's what Allen meant, but he doesn't really make it clear what else he had in mind. I will try to suggest a different interpretation later (of which I'm also unsure), but doing that requires going back to where my last piece ended. There, I had posited a principle, which I called the instrumental principle, to the effect that we have reason to take or should take or are obligated to take the means to our ends (where have reason to take [etc.] is understood to refer to what we have reason to do, ceteris paribus).

My claim was that unless the instrumental principle is true, means-end reasoning has no rational or reason-giving force. Moreover, the instrumental principle could not itself depend upon any particular goal or end we have, because unless we already have reason to take the means to our ends, the presence of some end which cannot be achieved unless the instrumental principle is accepted would not give us reason to accept the instrumental principle. (Thats a bit too simple, as I shall indicate below.) So, what status does Allen think the instrumental principle has?

It seems to me that there are exactly four general possibilities:

1)The instrumental principle is true: we have reason to take the means to our ends.

2)At the other extreme, it could be that the instrumental principle is not true but also not false. In some way, it is ill-formed and so does not succeed in being either true or false. (An example: Suppose, for the moment anyhow, that the truth or falsehood of claims regarding fictional characters depends on the intentions of the author. If Rand never considered the question or formed any intention about it, then the claim that John Galt had a mole in his left armpit would be neither true nor false, since the only thing that could make it either true or false, a particular authorial intention, would not have been present.)

3)The instrumental principle is false in general: we do not (ever) have reason to take the means to our ends, at least not because they are the means to our ends. (There might be other reasons for action which result in our taking the means to our ends in some cases, but the fact that something is the means to one of our ends will not play a role in the justification for the action.)

4)The instrumental principle holds for some ends but not others: There is some subset of the ends that we could have which are acceptable as starting points for means-end reasoning while others are not. The simplest possibility of this sort might be described as dividing a range of possible ends into those that are good (and therefore acceptable bases for means-end reasoning) and those that are bad (and therefore not acceptable starting points). But there are more complicated orderings as well, perhaps including continuous gradations from more to less acceptable starting points. Ignoring these differences, let's apply to all accounts of this type the tag, graded instrumentalism. (This class of cases is what I had in mind above in saying that it was too simple to say that, unless the instrumental principle was true, there could be no reason-giving force in means-end reasoning. If some version of graded instrumentalism were true, then instrumental reasoning starting from acceptable ends would have reason-giving force.)

I think it's clear that neither Allen nor Rand would accept the second or third possibilities. Means-end reasoning has to have some reason-giving force for their accounts to get off the ground. Graded instrumentalism, I think, can also be dismissed (as an expression of their views). It assumes that there is some grading of possible ends that has to precede the (correct) application of means-end reasoning, but obviously our reason to employ that grading (if we have any) cannot be conditional upon any ends that we have chosen.

That leaves only the instrumental principle, and my guess is that Allen would agree that it's true. (If I'm mistaken in this, I'd appreciate it if he would say either which of the three possibilities I ruled out he accepts instead or what possibility there is that I didn't include at all.)

But if he does accept the truth of the instrumental principle, he's accepting that there is reason to (or that we have an obligation to or that we ought to) take the means to our ends -- and once again, we're face to face with the question how this obligation is related to the particular ends that we choose.

Now, as I argued previously -- I don't see that Allen has addressed this at all -- it can't be that this obligation is conditional upon some choice, because unless the instrumental principle (or some version of graded instrumentalism, which above I argued Allen would have to reject) is true, no goal could give rise to any obligation including the one expressed in the instrumental principle.

Is there any way out? Well, there's one worth exploring (and, so far as I can see, only one). I suspect it may be what Allen has in mind. Briefly, the proposal would be that the instrumental principle is a generalization derived from particular cases.

For example, in a particular case, I select a goal -- say, drinking a beer. I see that I can't drink a beer unless I get one from the refrigerator. So I conclude that I ought to get one from the refrigerator. Or, to use Allen's example, I decide to type a certain sentence. I see that typing that sentence requires that I press certain keys. So I conclude that I ought to press those keys. And so on.

In each case, I find that, because of some goal (choice) and certain facts, namely, that the goal will not be achieved unless I act in a certain way (causality), there is something particular that I ought to do. (There are complications this picture doesn't take into account such as the possibility that there may be multiple ways to achieve the same goal, but those don't have a bearing on the current argument, so I'll omit them.) Eventually, after having many such experiences and observing them in others, I realize that all these cases have certain abstract features in common, that they are all cases of taking means to ends. So, as a generalization, I formulate the instrumental principle: one ought to take the means to one's ends. (This may be, and I'm speculating that it is, what Allen meant by saying that Rand's position is inductive.)

Note that if this is the approach Allen has in mind, then, in order for it to work, it has to be possible to identify an obligation to take the means to one's end in the single case and without relying upon or presupposing the instrumental principle (or any other unconditional principle of obligation). If it's not true that I can identify an obligation to take the means to a chosen end in the single case -- and, for that matter, in all the single cases that I'm in a position to check -- then treating the instrumental principle as an inductive generalization from single cases fails.

Now, in my earlier post, in talking about single cases, I (once -- not with respect to all the arguments I considered, though Allen doesn't mention any of the others) used x and y to stand for the goal and the means to it respectively. My reason for doing so was that, by using variables, I could talk about what all cases of the relation of means to ends have in common. However, Allen objected that with its abstract brevity, it fails to include choice and causality. I think this was a mistake on his part. I did not omit choice and causality; I omitted the words choice and causality in the particular passage he quoted. I had elsewhere said that by saying that something is someone's goal, I meant to imply that he had chosen or endorsed pursuing it. And, as for causality, I had used as one premise 'I can't get x without doing or having y' -- which is one way, though one that would need to be filled out further, of expressing a relation in which doing or having y is causally necessary for getting x. Since I didn't fail to include choice and causality, including the words in the appropriate places in my argument (as I did in the schematic description above) will make no difference, and so, I'm happy to do so.

Let's return to considering a single case. Remember that we have to be able to identify an obligation to take particular means to a particular end in single cases (without presupposing the instrumental principle) or we won't be able to view the instrumental principle as a generalization from single cases. Suppose that I choose to lose weight (by following a certain diet). Suppose further that I will not lose weight if I indulge in between-meal snacks. That's saying that there is a causal relation between avoiding snacks and achieving the goal, specifically that avoiding snacks is causally necessary to achieving the goal. So we have a chosen end, losing weight, and a causally necessary means to that end, avoiding snacks. How, without appealing to the instrumental principle or something like it, do we get the conclusion that I ought to avoid snacks (or even that I ought to avoid snacks or else give up the goal of losing weight)? The short answer is that we do not. Nothing in the statement of the claim about my choice or the causal necessity claim implies, either by itself or in conjunction with the other, that I ought to avoid snacks.

I think Allen obscures this fact (I don't mean that he is doing so deliberately) by using, following Rand, language that refers to what must be true or must be done. For example:

>So Rand's view should be expressed as: If I choose to attain x, then I
>must do y. And why must I do y? Because that's the nature of reality.
>It's what's required -- required by the natures of the entities involved --
>to attain the goal. If you choose to attain a goal, then you must do the
>necessary actions to attain it.

To say that something must be the case or must be done is to say that alternatives are ruled out or absent. But 'must' is used in multiple contexts and with reference to differing sets of relevant alternatives. Consider these:

We can speak about what logically must be the case, where we mean that there are no consistent alternatives.

We can speak about what morally must be done, where we mean that there are no morally acceptable alternatives.

We can speak about what people rationally must accept, where we mean that all the alternatives are irrational.

We can speak about what causally must be the case, where we mean that there is no alternative consistent with other facts and the relevant laws of nature.

We can speak about what people rationally must do, where we mean that all alternative actions are rationally defective.

Accordingly, if you're trying to construct a non-fallacious argument, it's important to be sure that you're using 'must' in the same way and with respect to the same set of alternatives throughout. Otherwise, you may introduce a 'must' in your premises referring to one set of alternatives but interpret a 'must' in your conclusion as referring to a different set of alternatives.

Is Allen using 'must' in the same way throughout? Let's fill in my example of a diet in his statement of Rand's position:

If I choose to lose weight, then I must avoid snacks. And why must I avoid snacks? Because that's the nature of reality. Avoiding snacks is what's required -- required by the natures of the entities involved -- to lose weight. If I choose to lose weight, then I must do the necessary actions (avoiding snacks) to end up losing weight.

Does the first 'must' signal a causal claim? If so, it signals a false causal claim. It is not true that choosing to lose weight is a causally sufficient condition for avoiding snacks. So, it is better to read it as an answer to the question whether I ought to avoid snacks. Then, the question, 'why must I avoid snacks?' will be asking why I ought to avoid snacks, why no other alternative is rationally open. The answer Allen gives is a causal claim to the effect that I must avoid snacks if I am actually to lose weight.

Here, I hope the problem is beginning to be apparent. The question asked is about why there is no rational alternative to avoiding snacks, but the answer given -- though phrased in the same 'must' language -- is about why there is no causal alternative that results in losing weight. But that isn't an answer -- not to the question asked unless some further assumption is made to the effect that one ought to lose weight (at least if that's what one has chosen) or, to bring us back to our starting point, that one ought to take the causally necessary means to one's chosen ends.

Allen appears to be offering an answer because he's not using 'must' in the same way throughout. If he were, we'd get either the unenlightening claim that avoiding snacks is a causally necessary condition of losing weight because it is a causally necessary condition of losing weight or the equally unenlightening claim that one who has chosen to lose weight ought to avoid snacks because one who has chosen to lose weight ought to avoid snacks.

What this means is that he can't do what had to be done in order to treat the instrumental principle as an inductive generalization. He cannot, that is, identify an obligation to take the causally necessary means to a particular chosen end in single cases without assuming something more general -- like the instrumental principle.

This has been a lengthy discussion, but really it is just an application of a fairly simple point, though one Objectivists seem to have trouble seeing: Substantive ought-statements do not follow from any is-statements whatsoever (that do not involve or presuppose substantive ought-statements). Using is-statements about what is chosen or what is causally related to what doesn't change that.


Rob
---
Rob Bass
rhbass@gmail.com


A great many people think they are thinking when they are merely rearranging their prejudices.

William James