Should I Stop Smoking?

paper presented at the conference Quotidian Ethics, University of Capetown, August 11-12, 1999

Christian Munthe
Dept. of Philosophy
Göteborg University

1. The Problem
A few weeks ago I firmly decided to stop smoking. There are many good prudential reasons for this decision: the health risks, the economic cost, the regular morning-cough, the smell in my home, the troubles I have to go through in order to satisfy my addiction in an increasingly non-smoking society, the uneasy experience of being controlled by the habit and the accompanying loss of dignity. And then there are the health risks, of course. Slow death in lung cancer is hardly appealing and neither are the other various troubles that without doubt will hit me within the next twenty years. Add to these all the benevolent reasons which could be mobilised in support of the same thesis (dangers for people in my surroundings, the smell of my clothes and my body and other inconveniences to them caused by my disgusting habit), and the case for my obligation to stop smoking seems overwhelming, no matter what set of basic ethical norms and values is applied.
 There is one problem, however ? a problem I share with many other smokers: I have a weak character. I could, if I flexed the muscles of my will-power to their extreme, probably endure the problems of abstinence. However, in past attempts to 'kick the habit' I have never in fact done so and this is due to my strong tendency to yield to the temptation of the moment. But, since I in fact have such a lousy character, may I not as well yield immediately? Would the few weeks of pointless struggle really amount to anything more than sheer masochism: agony leading up to nothing? Might I not just as well have another cigarette immediately, thereby avoiding a non-negliable amount of pointless suffering? Intuitively, the correct answer to these questions seems to be affirmative. I should, therefore, have another cigarette. But since the same reasoning applies to any single cigarette I may smoke in the future, and since to stop smoking involves abstaining from each such cigarette, this line of reasoning seems to lead back to the negation of the thesis above: I should not stop smoking!
 This, then, is the problem I want to address: Can the just outlined paradox be resolved, and, if so, should I really try to stop smoking in light of the facts about my weak character? It is a real problem, I should know, because the story I have told is about myself and it is true. But how should we reflect on it? I will approach it from two different viewpoints: first, from the perspective of normative ethics and, secondly, from the perspective of decision theory. My conclusion will be somewhat perplexing: on the one hand, although partly faulty, the reasoning sketched above is mostly correct, but, on the other, people like me (people with a weak character, that is) should in general avoid to think along such (albeit correct) lines.

2. Projects and Future Mistakes
The problem created by my weak character regarding smoking is an example of a more general problem in ethics, discussed under the heading of future mistakes.(1)  It surfaces due to the fact that many of our actions are made up of other actions. For example, to call 911 when wittnessing a crime is made up of a temporal series of 'smaller' actions of the type 'find a telephone', 'insert some change', 'punch in the numbers 9, 1 and 1', 'wait for an answer', 'report the event' and so forth. This means that it is possible for a person to start performing the action 'call 911' without completing it ? after having performed any of the early parts of the action, a person may deviate and perform another action than the one he has to perform in order to call 911. The action 'call 911' may, in other words be seen as a project ? something that takes time to complete and which may be begun without being completed. This, of course, is a trivial observation when seen from the point of view of the unreflected doings of our daily lives. However, from the point of view of ethical analysis, this fact creates problems which are both complex and subtle. The problem of future mistakes ? i.e. how to take into account the anticiptated deviation from an initiated desirable project ? is one of these.
 In a more formal manner, the problem may be illustrated such: Assume that it is desirable that I perform the project A, and that this project is made up of two actions, a and b which have to be performed in temporal series in order for A to have been performed. Instead of A, I could perform two other projects, B and C, which are both clearly less desirable than A. At the same time, if a choice would have to be made between B and C alone, B is clearly preferable. Assume further that the project B is made up of neither a nor b, (but of x and y) while C is made up of a and another action, z. Finally, assume that it is very probable that if I perform a, I will not perform b  but z ? i.e. if I start the project A, I will in fact complete the project C. The diagram below illustrates this situation by having the broken lines represent actions that I will probably not perform (even if the action leading up to it have been performed) and filled out lines represent actions that I will probably perform (given a decision to perform them, or that the action leading up to them have been performed).

 One way of analysing this situation would be to seek guidance from deontic logic, in particular the so-called principle of distribution (POD), which says that if I ought to perform a combination of several actions, I ought to perform each of these actions.(2)  Since I should perform A, it would then follow that I should perform a and that I should perform b, and, ergo, that I should perform a. By the same token, I should abstain from the next cigarette just because I should stop smoking. POD, then would resolve the paradox and solve my problem.
 However, this line of reasoning is only as plausible as POD, and, in cases like these, POD seems to be clearly false. Let, for example, A be a successful revolution which frees a people from brutal opression, while C is a pointless bloodshed which only strengthens the power of the dictator.(3)  The message of POD that the high probability of C being the end result if A is initiated should be disregarded is a logic of calousness rather than ethics. A sound analysis of the problem of future mistakes cannot just ignore such mistakes, but must in some manner deal with them.
 A very influential suggestion for doing exactly this has been suggested by Lars Bergström. His basic idea is that we cannot just look at the available alternatives in a lump ? we have to group them together in different alternative-sets. A simple way of doing this is to sort the alternatives according to the time-span during which they would be performed, so that (in the example) we would get the sets 1:[a, x], 2:[b, z, y], and 3:[A, B, C].(4)
 However, Bergström claims, we cannot rest content with this move, for then we risk having ethics providing us with conflicting prescriptions (recommendations to perform actions which cannot be jointly performed).(5)  For example, looking at set 1, x should clearly be performed (since a would lead up to z being performed). However, looking at set 2, b should clearly berformed (since that would complete A). But if the first recommendation is acted upon, the second recommendation cannot be acted upon, since, if x is performed, this will unable me to perform b later. Moreover, if we look at set 3 we already know that A should be performed. However, we have also seen that it would be wrong to start performing A (since, in the choice between a and x, I should perform x).
 In order to avoid these kinds of conflicting prescriptions, Bergström's proposed strategy is to select some alternative-sets as being relevant from an ethical point of view, so that several alternative-sets cannot have such relevance if application of a moral principle to them would result in conflicting norms.(6)  Consequently, alternative-sets which are not relevant in this way are ethically irrelevant, i.e. alternatives belonging to such sets cannot be ascribed moral status by any ehical theory. Applied to the case of smoking, this means that I have to make a choice between asking the question 'should I have another cigarette?' and the question 'should I stop smoking?'. For, the alternative-set containing the actions concerned by the one of these questions cannot be relevant unless the alternative-set containing the actions concerned by the other question is irrelevant.
 Such a selection of relevant alternative-sets can be made in numerous ways.(7)  However, I will neglect going into these details, since I believe Bergström's basic idea to be mistaken. I have two reasons for this claim.
 First, most specifications of the idea of relevant alternative-sets has to entail the view that some perfectly sensible moral questions are empty or mistaken.(8)  To illustrate, in my example, set 1 and 3 cannot both be relevant (since, if they were, we would get conflicting norms). In effect, among the actions of these two sets only those belonging to one of the sets can have any moral status. If 1 is the relvant set, x is morally right and a is morally wrong, while A, B and C are neither right nor wrong. If 3 is the relevant set, the reverse holds. This means that, if 1 is the relevant set and I ask whether or not A should be performed, I make some kind of mistake. Similarily, if the project to stop smoking belongs to a relevant set, the question whether or not I should have another cigaret is mistaken. It is like asking for the length of beauty ? I am searching for a type of fact which is nowehere to be found. But, clearly, this is a highly unsatisfactory feature of a normative ethical theory. The job of such a theory is to make sense of and provide answers to our moral questions as far as possible. A theory which fails on this account is unfruitful in the same manner as a scientific theory which fails to explain the facts it has been constructed to explain. Therefore, unless some further evidence is presented in favour of such a theory, this is good reason for rejecting it.(9)
 Now, it might look as if Bergström's suggestion can in fact be supported by such further evidence. After all, is it not needed in order to avoid conflicting prescriptions? However, and this is my second reason for rejecting Bergström's idea, this is in fact no evidence at all. Conflicting prescriptions of the kind illustrated above would be a problem if they were all seen as the same type of moral statements. In that case we would end up with prescriptions like 'take three steps to the door and open it, and do not take another step towards the door' ? i.e. recommendations which no one is capable of acting on and which therefore would violate the dictum that 'ought' implies 'can'. However, this is the result of looking at all alternative actions open to an agent as having a certain moral status regardless of whether or not they are in fact performed, and it is not at all necessary to see things like that. Instead, we may say that actually performed actions indeed have a certain moral status (they are right or wrong), while performable but unperformed alternative actions would have had some moral status if they had been performed. In this way, we avoid all risks of unacceptable cases of conflicting obligations, since the actions concerned cannot all be performed (if they could there would be no conflict).(10)  For example, if I perform a, I act wrongly, and if I had performed B, I would have been acting wrongly, too. However, since B is in fact not performed, it is not wrong. Thus, we avoid the unacceptable prescription 'a is wrong and B is wrong', and the problems created by the idea of relevant alternative-sets may therefore be side-stepped entirely.
 Moving back, now, to my initial problem regarding smoking, the following transpires: If I have another cigarette, this is indeed what I should do ? and this can be extended to any additional cigarette I will in fact have further on as well. However, we cannot infer from this that it would have been wrong of me to stop smoking. On the contrary, since it would be far better if I stopped smoking than if I did not, if I were to stop smoking, I would indeed be doing what I should do. Reversely, if I stop smoking I do what I should, but this does not contradict the claim that I would have done what I should if I, at some point, have had another cigarette.
 This analysis of the problem thus exposes one serious fault in the reasoning sketched at the outset: my weak character does not give rise to any paradox in ethics. Although the project to stop smoking is made up of my repeated abstaining from another cigarette, the fact that I have (or would have) good reasons to have any such additional cigarette does not show that I lack good reasons to stop smoking. On the contrary, if I stop smoking, this is what I should do! The fact that, if I have another cigarette, this is what I should do (since I am likely to have one anyway some time in the future) does not cancel the fact that, if I had instead stopped smoking, I would have been doing the right thing.

3. Free Will?
So far, I have argued that the fact that I have a weak character, and therefore have good reasons for having another cigarette, does not show that I should not stop smoking. However, the reasoning above rests on the crucial assumption that my weak character does not make me incapable to stop smoking. But may it not be suspected that my weak character or my knowledge about it actually makes it impossible for me to stop smoking? Does not my past experiences of my weak character makes me not really believe in the feasibility of me giving up smoking? And does not this in fact unable me to go through with such a project? If so, to stop smoking can hardly be what I should do.
 I do not believe this to be the case, however. First, having a weak character does not by itself imply incapability. There is a crucial difference between people suffering from compulsive behaviour (such as cleptomaniacs) and people of weak character. A person of the former kind lacks resources to prevent himself from exhibiting the compulsive behaviour (in a situation where he has an opportunity of exhibiting it). However, a person of weak character does have such resources. The problem is rather that such a person have a very difficult time actually using them. For this reason, he is liable not to use them ? and thus liable not to refrain from exhibiting the behaviour. This does not mean that he is incapable of doing so, however, only that it is difficult for him.
 Secondly, it is perfectly possible even to acknowledge the fact that the probabilities are not in our favour when embarking on some project, but still embark on it in the full resolution to complete it (against all odds). Moreover, althought the odds are not in our favour, we may very well succeed in completing the project. Of course, according to the reasoning presented at the outset, embarking on a project in light of such low chances of completing it might be the wrong thing to do (although the project is a desirable one). However, this does not in any way show that we cannot do such things.
 It seems to be a tempting thought to doubt our ability to perform actions which we know ourselves to be likely not to complete if we begin performing them.(11)  However, I believe that this doubt arises out of a simple confusion of two distinct factors. First we have my knowledge about my character, and then we have the line of reasoning built on this knowledge sketched out above. The fact that I have the former does not imply that I must think according to the latter. I may choose to ignore the information of my weak character entirely, or I may acknowledge it in a different way. The reasoning above may still be valid, of course, but that does not force me to think according to it.

4. How Should I Reason?
So far, I have concluded that neither my weak character, nor my knowledge of it (or the validity of the reasoning based on this knowledge sketched out in section 2) makes me incapable of giving up smoking. I can do it! Moreover, according to the argument in section 2, if I were to do it, I would be doing what I should do. However, we have also seen that, if I were to have another cigarette, I would equally be doing what I should do. Since this holds for any additional cigarette I may have in the future, I may completely fail to even try to stop smoking but still be doing what I should. I have argued that there is nothing paradoxical in this conclusion as long as we ask the ethical question about what I should do.
 However, although saving my free will and avoiding a nasty paradox, all this may seem to be of little help to me ? facing as I am the real problem of deciding what I should do. The conclusions above advise me that, whatever I do, I will be doing what I should. But I want to know how I should decide. I want guidance on how to reason when facing decisions like these. This brings me into the area of decision theory, and here the perspective is rather different than when addressing from an objective viewpoint the issue of what I should do. Now, the perspective is the subjective perspective of a decision maker and the question is what aspects should be taken into account in what way by such a decision maker. Should he, for example, take into account the (correct) answer to the question of what I should do and the reasoning leading up to it described above? I will claim that, in general, he should not.
 The question here concerns whether or not, when taking decisions, it is in practice a wise general strategy to take a pragmatic and strategic stance towards one's own expected future moral mistakes. In particular, is this a wise general strategy for a person who suffers from a weak character? For those who prefer that, the question I am asking could be interpreted as a question regarding virtue: How, the question could be rephrased, should a person be disposed to reason when it comes to decisions about desirable projects which he knows himself to be liable not to complete? My answer to this question is that, in general, he should not employ the (albeit correct) pattern of reasoning sketched out above, and I will back this up with three arguments.
 First, if you know yourself to have a weak character and are disposed to take a pragmatic and strategic stance to your own expected future mistakes caused by this weakness of character, this will block the access to many beneficial projects. If you know that you are likely not to complete such a project, and if (as is usually the case) it involves some inconvenience on the way to completion, you will repeatedly end up with the conclusion that, although you should perform the project, you should avoid trying to perform it (since you will most likely fail). This is exactly what is exemplified by my reasoning above regarding smoking: I should stop smoking, but I should not try to do it (but instead have another cigarette).
 To be sure, this argument has its limits. I am not claiming that we should never take a pragmatic and strategic stance to our own weak characters and embark on any beneficial project, no matter the chance of completing it. On the contrary, we should take such a stance when sufficiently substantial values are at stake. If the completion of a project involves benefits which only slightly exceed the harm that would be caused by my likely failiure to complete it (if I try to perform it), I should be disposed to take this into account. However, in many cases where weakness of character becomes an issue, the picture is reversed. In many cases, very much can be accomplished by the successful completion of a project, while the harm caused by a failiure to complete it (compared to if the agent had never started to perform it) would be comparably small. This, I conjecture, is clearly the case with the project of giving up smoking.
 Secondly, if you are generally disposed to reason according to the lines sketched out in section 2, you will in many cases become a victim of what may be called paralysis. In cases where the carrying out of some project is desirable, but where you are likely to fail to complete it (if you try to carry it out), the line of reasoning you are disposed to follow will lead you to the conclusion that you should carry out the project, but that you should avoid trying to carry it out. Above, I argued that it is possible to uphold this conclusion without theoretical paradox. However, from a practical point of view, there is surely a paradox. From the subjective perspective of a decision maker, the result is a set of recommendations which (although not logically contradictory) in practice pull in completely opposite directions regarding what he is to do. His focusing on one part of the set will pull him towards carrying out the project, while his focusing on the other part will pull him away from trying to carry it out. In the end, therefore, unless other forces of his mind is at work, he will remain locked into continued indecision or paralysis. Not because there is anything beneficial to be found in such paralysis according to the pattern of reasoning he is disposed to employ, but simply because this pattern prevents him from making up his mind. In the case of smoking, it is easy to see that such an outcome is worse than alternative options. Not only will he continue smoking, but he will suffer additionally from an endless inner struggle of trying to decide. Moreover, his preoccupation with this struggle will prevent him from addressing other areas of his life which may need his attention.
 Partly, then, paralysis is equivalent to a preservation of status quo. This leads to my third argument, which is that the disposition to reason in a way which leads to paralysis isconservative with regard to a person's character. If I reason according to this pattern and ? as I have claimed that I will ? end up in paralysis, this means that I must give up any ambition to improve my weak character. A dynamic view of a person's character, according to which it may be improved (for example, by overcoming it), is completely neglected for the benefit of a view of character, according to which a person is condemned to pragmatically adapting his decisions to his own character. This, then, is my third argument: A disposition to reason in such a way unreasonably disfavours a dynamic view of character and, thus, for no good reason whatsoever blocks the opportunity of people to attack their own shortcomings and thereby maybe obvercome them. For example, it blocks my opportunity to improve my character by giving up smoking against all odds.

5. Conclusion
I have defended two claims. First, the line of reasoning sketched at the outset is indeed correct, apart from the fact that it does not involve any theoretical paradox. If I were to stop smoking, this would be what I should do, and if I were to have another cigarette this would be what I should do (but since I cannot do both of these things, they cannot both be what I should do).
 Secondly, people suffering from a weak character should not in general be disposed to reason along such (albeit correct) lines. We should focus on embarking on and complete desirable projects, unless a likely future mistake would lead to substantial harm (compared to if we had never initiated the project). I should, therefore, not in my practical thinking separate the issue of whether or not I should stop smoking from the issue of whether or not I should have another cigarette, but focus on completing the obviously very desirable project of giving up smoking. Although there would be nothing wrong in having another cigarette, I should be disposed to reason in a way which makes me mobilise my will-power in trying to stop smoking and, hopefully, overcome my weak character.

  (1)This problem has traditionally been discussed in a consequentialist context. However, this paper should make it evident that the problem can surface, no matter what ethical framework is employed.
  (2)This seems to be the solution favoured by Fred Feldman (1975).
  (3)Both projects have the same beginning, a, which here can be seen as, for example, initiating a popular uprising.
  (4)Bergström 1966.
  (5)Bergström 1977.
  (6)Bergström 1976.
  (7)The above mentioned texts by Bergström all present one such way (in particular, see Bergström 1976). Other ways have been defended and/or presented in Carlsson 1995, Feldman 1975, Goldman 1978, Jackson & Pargetter 1986; Rabinowicz 1978, and Sobel 1976.
  (8)One exception is Howard Sobel's suggestion (Reported in Munthe 1996, and worked out in a decision theoretical context in Sobel 1983 and Sobel 1994) to distribute the moral status of actions belonging to relevant alternative-sets to those actions belonging to irrelevant sets with which they can be jointly performed. For example, if set 3 is relevant, A is right, and all actions belonging to set 1 and 2 with which A can be jointly performed are right, too. However, this idea can be objected to on the same grounds as POD, since it would entail that a is right, although it most likely would lead to a moral disaster.
  (9)I have developed this argument in Munthe 1992, chapter 6, and Munthe 1996.
  (10)Elsewhere (Munthe 1992, chapter 6, and 1996), I have linked this idea to Torbjörn Tännsjö's suggestion that only 'concrete' actions (i.e. actions individuated by their location in space-time) should be ascribed moral properties (see Tännsjö 1978, 1985 and 1998). However, as shown by Carlson (1995, chapter 8), this particular version of the idea is open to the objection that 'concrete' actions cannot be the objects of choice. For this reason, my proposal is that we should content ourselves with the sugestion that only performed actions can have moral properties, without specifying whether these are 'concrete' or not.
  (11)In any case, this is the impression I have got when discussing the issue with many people.

Bergström, Lars 1966, The Alternatives and Consequences of
Actions, Stockholm: Almqvist & Wiksell International, 1966.
 ? 1976, "On the Formulation and Application of Utilitarianism",
Noûs 1976, 10:121-44.
 ? 1977, "Utilitarianism and Future Mistakes", Theoria 1977, 43:84-
Carlson, Erik 1995, Consequentialism Reconsidered, Dordrecht
1995: Kluwer Academic Publishers.
Feldman, Fred 1975, "World Utilitarianism", in Lehrer, K (ed.) 1975, Analysis and Metaphysics, Dordrecht 1975: Kluwer.
Goldman, Holly S 1978, "Doing the Best One Can", in Goldman &
Kim (eds) 1978, Values and Morals, Dordrecht 1978: Kluwer.
Jackson, Frank & Pargetter, Robert 1986, "Ought, Options and
Actualism", The Philosophical Review 1986, XCV:233-55.
Munthe, Christian 1992, Livets slut i livets början. En studie i
abortetik (The End of Life in Life's Beginning. A Study on the Morality of Abortion), Karlshamn 1992: Thales.
 ? 1996, "The Argument from Transfer", Bioethics 1996, 10:27-42.
Rabinowicz, Wlodzimierz 1978, "Utilitarianism and Conflicting
Obligations", Theoria 1978, 44:19-24.
Sobel, Howard 1976, "Utilitarianism and Past and Future
Mistakes", Noûs 1976, 10:195-219.
 ? 1983, "Expected Utilities and Rational Actions and Choices",
Theoria 1983, 49:159-183.
 ? 1994,Taking Chances: Essays on Rational Choice, Cambridge
1994: Cambridge Uiversity Press.
Tännsjö, Torbjörn 1978, "The Morality of Abstract Entities",
Theoria 1978, 44:1-18.
 ? 1985, "Moral Conflict and Moral Realism", Journal of
Philosophy 1985, 82:113-17.
 ? 1998, Hedonistic Utilitarianism, Edinburgh 1998: Edinburgh
University Press.

Back to the bibliography