1 The future actions as circumstances or components of actions

Transcription

1 Pluralism about Reasons and Agent-Units in Consequentialism: Why Actualism and Possibilism are not Incompatible Kaoru Ando (Kobe University) ISUS Yokohama National University 0 Preliminaries Consider the following familiar case *1 : On the first of January, a graduate student asks her supervisor, Professor Procrastinate for comments on a paper she is planning to read at a job interview. If he accepts the task and comment on the paper by the end of February, the paper will be much improved and the student will have a highly successful interview (suppose this is the best outcome). If he declines the task, the student will revise the paper for herself and have a neither good nor bad interview (suppose this is worse than the former). On the other hand, if he accepted the task, he would though he surely could comment actually fail to comment on the paper and ruin the career of his student (suppose this is the worst that can happen). What ought Procrastinate to do? Ought he to accept the task? In this paper, I take for granted that consequentialism is, at least arguendo, the right moral theory though I believe the problem arises equally for non-consequentialists. If you are consequntialists, however, which should you choose? As you all know, there are two main camps: possibilists and actualists. The former answer with unanimous Yes, and the latter answer univocally with No. The possibilists deem what could happen or, perhaps precisely, what you could make happen if you accepted the proposal as relevant and what would happen as irrelevant, while the actualists do exactly contrariwise. Then, which is the right theory of our moral obligation? It seems, however, dubitable that either is the right answer. It is not because there is none, but because both are right answers, or so I will argue. 1 The future actions as circumstances or components of actions Suppose that there is another person say, Procrastinate s paternalistic secretary and she determines whether to accept or decline the task instead of him. Now we will not wonder whether she should accept or decline the task and surely favour the latter. The difference is the fact that the one making a *1 This example is mixed up from those of [Goldman 1978, pp ] and [Jackson 1987, p. 110]. 1

2 decision is different from the one who will (fail to) comment on the paper. When we evaluate the secretary s act (accept it/decline it), we consider the act of Procrastinate (comment/fail to comment) to be relevant as a circumstance of the former. The actualists say Decline it! in both cases. They deem the future action of Procrastinate to be a circumstance of his present action in the first case as well as in the second, while the possibilists suppose the fact that the future actions are his is morally relevant not as a circumstance but as an integral part of larger acts of his. Hence the disagreement. The background intuition of the possibilism is that the future actions are components of your diachronically extended acts (accept the task + comment / accept the task + neglect / decline the task + do other things) and we shouldn t isolate them from the whole acts in evaluating them. This might be interpreted as a form of the seemingly plausible deontic principle: Deontic Principle: (D) If you ought to do a + b, then you ought to do a Nevertheless, (D) is highly problematic. Now that the deontic status of a is determined by that of a+b, when you would like to know the deontic status of some act, you must examine beforehand whether there is a larger act which contains it as a component and is obligatory. As living a whole life may be itself a maximal act and has no larger whole, and as consequentialism obligates the best possible act call this the Consequentialist Principle it may be the case that you ought to do φ when you do φ in the life which is the best of all the possible lives which is open to you at the moment of your decision *2. However, as you would surely often fail to what you could and should do, if you tried to comply with this prescription, you might make no significant - or even worse as in the example - outcome. This shows that (D) can easily lead to possibilism and it may have a disastrous consequences when adopted by a non-ideal agent. In light of this, (D) doesn t seem to be a plausible principle, but I am sure the possibilists are willing to bite the bullet and that possibilism has its merit. One thing should be noted here. I believe most of the actualists will accept you ought to do a + b *3. It is a subtle but important point that the actualists will surely accept If you do a, you ought to do b, too, but this must be distinguished from You ought to do a and b. However, it seems the actualists must reject not only (D) but also any other principle which derives an obligation to do a from an obligation to do a + b. 2 Consequentialism and Alternative-Sets Consequentialism provides a criterion which determines the deontic statuses of actions. In short, it says that an action is right/obligatory/required if and only if it is the best of all the possible alternatives *4. *2 This view is closely similar to the extreme possibilism the so-called world utilitarianism of Fred Feldman. *3 See [Jackson and Pargetter 1986] and [Goble 1993]. The actualists can accept (D) if only they admit the possibility that you ought not to do u and you ought to do u + u and, therefore, you ought to do u at once, which means if they admit the possibility of a genuine moral dilemma, they can also be possibilists at once. I will not enter into details here, but I will take up a synchronic analogue of this manoeuvre in 5.2. *4 If there are more than one best alternatives, it is permissible to do either of them but, prohibited to do neither. I will ignore this complication in this paper. 2

3 Under consequentialism, the deontic status of any given act cannot be determined without reference to its alternatives. Consequently, if the actualists want to attribute obligatoriness to incompatible actions, those actions must belong to different alternative-sets. To put it in the other way around, insofar as they belong to different alternative-sets, you may be able to suppose that you ought not to do a and that you ought to do a + b at the same time with no incoherence. If we can divide all the actions of yours into the sets that doesn t overlap one another notice that a and a + b are different actions consequentialism itself doesn t prevent us from supposing so. Of course, there are deontic principles like (D) that bridge between different alternative-sets and make the actualist position difficult, but why should we accept that they hold across the alternative-sets? It might be the case that (D) holds within an alternative set like {a, a + b, a + c} *5, but this doesn t commit us to hold it across different alternative sets like {a, d} and {a + b, a + c, d + e}. 3 Units of Agency: What Alternative-sets Represent You might object this manoeuver is too ad hoc, but I will argue that the actualists have good reasons to reject applying (D) to this case. First, when the actualists think Procrastinate s action accept the task & comment is obligatory, the alternative-sets taken into consideration are {accept the task & comment, accept the task & neglect, decline the task & do other things}. They are all taken to be his actions at the moment of making a decision. However, when considering the action decline the task to be obligatory, they suppose the alternativeset concerned to be {accept the task, decline the task}. The possibilists have been rightly pointing out that the opponents deem the future actions comment and neglect to be not integrated parts of his actions but just the circumstances of them. The actualists deny that Procrastinate is an agent of the future actions at the moment of making a decision *6. Actually there must be two distinct agents here. As alternative-sets must be merged if they represent the choice-situtations of the same agents at the same time, the actualist must assume {accept the task, decline the task} and {accept the task & comment, accept the task & neglect, decline the task & do other things} represent the choice-situations of two distinct agents. I think this view is well-motivated. In evaluating the deontic status (e.g. obligatoriness) of the actions, they think about the different agents and their actions between two evaluations. In the former, they take a relevant agent s alternative sets to be {accept the task & comment, accept the task & neglect, decline the task & do other things} and the agent to be a temporally extended one, while in the latter, they take an agent in question to be temporally truncated and his alternative-set to be {accept the task, decline the task}. In short, they take the future self to be another agent who might be cooperative or uncooperative with the present self. *5 Whether any alternative set of this form where an action and its proper subaction occur simultaneously can be an adequate object of consequentialistic evaluation is a very difficult problem, which I cannot but ignore in this paper. *6 You might contend that this is the very point where the actualists make a mistake and that Procrastinate does have an ability to do u + u at the moment in question. For this view, see [Oddie 1993]. I don t think the actualists are wrong, for there are actually two distinct agents. The January temporal part of Procrastinate surely lacks an ability to do u and thereby u + u, whilst Procrastinate as the temporally extended whole has an ability to do u + u at the moment of doing u. 3

4 You may have a deja-vu here because this point was made by the arch-actualist Frank Jackson in his [Jackson 1987]. He contends that the January temporal part of Procrastinate ought to decline the task while the temporally extended Procrastinate ought to accept it and comment. The question What ought Procrastinate to do? is ambiguous because this can be read as What ought the January temporal part of Procrastinate to do? and also as What is the January part of the extended programme that the temporally extended Procrastinate ought to follow? [ibid. p. 110, n. 13] *7. It should be noted here that this manoeuvre attributes the two alternative sets in question to two distinct agent-units the January part Procrastinate and the whole Procrastinate and separated them so that the bridging principle like (D) may be invalid. Now that decline the task is the former s duty and accept the task & comment the latter s, either agent has no conflicting duties. However, it surely seems dubitable that there is no bridging principle. It sounds highly implausible that there is no essential relation between the temporal part of the whole Procrastinate s duty and the duty of the temporal part of the whole Procrastinate. What s the relation? Doesn t it undermine the plausibility of this way out? 4 Agent-Unit Pluralism about Moral Obligations Let p denote the January part Procrastinate and p, the whole Procrastinate. Now, if you simply assume a bridging principle between the obligations of these two agents like If p ought to do a + b, p ought to do a., the actualists must admit that p has a genuine moral dilemma. This is an unwelcome result. There should be some bridging principle between them, but how could there be? I propose that we should relativise in some way other than those of the typical moral relativism or agent-relativism moral obligations to the agents, that is, the subjects of the alternative-sets which consequentialism refer to in order to identify the moral obligations in question. As consequentialism attributes obligatoriness to the action accept the task and comment with reference to the alternativesets whose subject is p, any derivative obligations - e.g. p s obligation to accept the task has their obligatoriness only with reference to p. Then, even if p has a moral p -obligation to accept the task because of p s moral p -obligation to accept the task and comment, p s moral p-obligation to decline the task doesn t conflict with it since they are relativised to p and p respectively. The relativised (CP) and this bridging principle can be formulated as follows: Relativised Consequentialist Principle: (CP-R) S has a moral S-obligation to do the best action in S s alternative-set. Bridging Principle: (B) If S has a moral S-obligation, any part of S has a moral S-obligation to do its part in S s obligation. *7 It might be objected that a mere temporal part of a person cannot be an agent, because the notion of the agents require that it should be temporally extended. I disagree. Imagine that a person u was physically duplicated from you with SF-like technologies. However, just after that moment, some technological issue happened and u was killed. I think it is evident u was an agent at the moment though his life is very short. If so, as u s agency is an intrinsic property of the temporal part of u at the moment with which u is identical, it follows that a temporal part can be intrinsically an agent. 4

5 I think this proposal is well-motivated. While consequentialism is a monistic moral theory, we must admit some deontic principle that derives further duties from the original duties the consequentialist principle recognise. However, this can easily endanger the monism of the consequentialism since there are now more than one moral principles. As the bridging deontic principle mustn t conflict with the sovereign consequentialist principle, the derived duties must be strictly distinguished and separated from the original duties so as not to conflict with them. Since a p-obligation of the agent q is an original duty if and only if p and q are identical, if p and q are distinct, this obligation is merely derivative and must be incapable of conflicting with any other original duty. This should lead to the kind of relativism I will call it agent-unit pluralism mentioned above. We can generalise (B) to (B*), which can derive further duties from derivative duties *8 : Generalised Bridging Principle: (B*) If S has a moral S -obligation, any part of S has a moral S-S -obligation to do its part in S s obligation. I am not sure the derivation from the original duties should be many-tiered and would like to be neutral on whether to adopt (B*) instead of (B), at least in this paper. 5 Group Morality and Group Agency The above-mentioned principle (B) is attractive because it seems valid also in the context of group morality. In this section, I will consider some problems about it and vindicate it. 5.1 Group Morality and (B) The following principle seems plausible, at least at first sight: Group Principle: (G) If a group G has a moral obligation, any member has a moral obligation to do her part in G s obligation. However, there is a well-known problem about this kind of bridging principle. Imagine a good Samaritan situation. Two persons, p and q, come across a man who has been left half dead. If both of them sacrifice their individual interests considerably, they can save his life. If not, the man will die anyway. Evidently, the group let G denote it that consists of p and q has a moral duty to help the man. Yet if either is uncooperative, the other has no moral obligation to sacrifice her interest to no avail. Most consequentialists will agree to this judgement. Nevertheless, the above-mentioned principle (G) bridging between the deontic statuses of the actions of the group and its members is conflicting with the consequentialist principle and thereby contaminating the monistic hygiene of consequentialism, but it seems implausible that there is no principle connecting them. *8 For any u, u -u duties can be reduced to u -duties. 5

6 One seemingly attractive alternative is this: Jackson s (P) If a group has a moral obligation, and it is in fact discharged, then each member has a moral obligation to do her individual constituent act. This is Jackson s view [Jackson 1987, p. 107], but I think this has a fault that I personally believe to be fatal. (P) can be plausible only when factualism about consequentialism the opposite position is probabilism is adopted *9. Under probabilism, if p and q are sufficiently unlikely to sacrifice their interests respectively, they have no moral obligation to help the dying man because the act has only a low-level expected utility owing to the other s uncooperativeness even if they still have a group moral obligation as a group. So (P) fails * 10. Even if the group duty is fortunately discharged, it is nothing more than a matter of sheer luck. Although I wouldn t like to go into details here, I believe factualism necessarily deprives consequentialism of the action-guiding-ness, which is an indispensible feature of any moral theory and that it is implausible. Anyway, I would like to set aside those matters and ask those who endorse factualism to assume, arguendo, that (P) is not unproblematic. On the other hand, (B) is more attractive than (P) in that it keeps its plausibility whether fatulalism is tenable or not. Both p and q have G-obligation to perform self-sacrifice respectively, while they haven t p-obligation nor q-obligation to do so. Precisely, each has p(q)-obligation to refrain from futile selfsacrifice. They have individual moral obligations (and thereby moral reasons), which are a G-obligation to sacrifice herself and a p(q)-obligation not to do so and are not conflicting with each other. 5.2 Willingness Requirement The fact that (B) is plausible as a group morality principle is important, for a certain point of (B) rejection of the apparently plausible assumption that S in (B) must be an agent so that its part may have a derivative obligation can be elucidated clearly in the context of group morality. First, in the above-mentioned Samaritan case, is G an agent? The group action save the life of the man through cooperation is surely conceivable and appropriately liable to moral evaluation even when its members p and q are uncooperative, but then it is dubitable that G has a group agency. I think this gives a reason to suppose that it is unnecessary for G to be actually a group agent in order that the deontic evaluation may be attributed to the acts of G. There is no problem insofar as G is capable of being an agent. Consequently, according to (B), even when G is not an agent and hence, its members are uncooperative each member has her moral G-obligation, whether she herself is cooperative or not * 11. Christopher Woodard has a closely similar view about this problem [Woodard 2008]. He thinks that each member has a pattern-based roughly, rule-consequentialistic reason to do her part in the group act even though the others are uncooperative, whereas she may have an act-based act-conseqnentialitic reason not to do so at the same time. As I did above, he explicitly rejects the popular assumption that *9 About factualism/probabilism, see [Carlson 1995, pp. 20-4]. *10 See [Oddie 1996, pp ] for this point. *11 In the diachronical settings, this means that the whole Procrastinate himself needn t be an agent at all. 6

7 there cannot be pattern-based reasons when members are uncooperative, which he dubs Willingness Requirement (WR). I fully agree to this rejection of (WR) and follow his position that several units of agency may be relevant to single deliberative problem (ibid. p. 110). However, there is a subtle but important difference between his view and mine. He seems to be of the opinion these act-based reasons and pattern-based reasons conflict with each other in moral deliberation (ibid. pp ), but I don t think so. I suppose that they are not conflicting but talking past because they are different kinds of moral reasons. My view is more deeply pluralistic or relativistic, if you like than Woodard s. This is partly because the Consequentialist Principle as I understand it provides primarily not a criterion of reasons for actions but that of the rightness and obligatoriness. Moral obligations cannot conflict there cannot be more than one conflicting right actions unless you allow the possibility of genuine moral dilemmas, whereas moral reasons can easily conflict without raising any genuine dilemma. However, I suppose those reasons cannot conflict, at least qua moral reasons, because they are originated from the unconflicting moral obligations identified by the Consequentialist Principle. 5.3 Units Proliferation Problem There is a reason though indecisive to prefer mine to Woodard s. Think of a group g including you and let n be a number of the members of g. Then, you have 2 u 1 units of agency including you from combinatorial mathematics and you have 2 u 1 kinds of pattern-based reasons because we reject the (WR). The act-based act-consequentialistic reasons correspond to the singleton unit comprised of just you, and the typical pattern-based rule-consequentialistic reasons correspond to g. However, there are 2 u units and each must constitute its own kind of pattern-based reasons * 12. According to Woodard, the pattern-based reasons are conflicting with one another in our practical deliberation and pull it in various directions respectively. This is highly counter-intuitive because it is inadequate as a description of the phenomenology of our practical deliberation * 13. It is an almost self-evident fact that we don t see such many 2 u 1 and n can be a number of the whole population of the globe kinds of reasons in deliberating. Woodard s theory suffers from this problem I will call it the Units Proliferation Problem precisely because he rejected the Willingness Requirement, which drastically reduces the number of reasons because it deems almost all of those numerous agent-units to be irrelevant * 14. *12 You might object that this worry is utterly insubstantial. I think not. For instance, Michael Ridge s variable-rate ruleutilitarianism does take these middle units into consideration and its theoretical appeal consists mainly in this point (cf. Ridge 2006). When the effects of the individual actions on utility accumulate super-linearly, all the units of perfect compliance will be rather significant. *13 You might object that this isn t a problem at all. If consequentialism is primarily a theory of the criterion of right-making property of actions, the fact that it doesn t reflect the structure of moral phenomenology is irrelevant to the tenability of it. It is true this objection has some force, but I think it fails ultimately. For even if you endorse indirect consequentialism, the subjective decision-procedure must reflect the objective criterion of consequentialism. If there are such vastly many pattern-based reasons, the act-based reason will surely be imperceptible in the whole bunch of reasons, and optimal subjective-decision-procedure will not reflect it significantly. Then, the act-utilitarian intuition that Woodard tries to preserve will be lost considerably anyway. *14 Of course, Woodard carefully considered the problems of the proliferation of pattern-based reasons arising from the rejection of (WR) in his marvellous work (cf. ibid. pp. 90-9), but I couldn t find the arguments against this sort of proliferation. 7

8 You might contend that I will surely suffer the same defect as I also reject the (WR) to vindicate (B), but my view is that there are 2 u kinds of moral reasons but that they are almost all irrelevant in practical reasoning. I think that moral reasons are not practically relevant intrinsically this means that I am endorsing a sort of moral anti-rationalism * 15 and that what makes them good reasons, that is, the reasons which it is practically irrational to ignore is a matter external to morality. To put it briefly, it depends solely on which agent-unit you identify yourself with * 16. Only those reasons are your reasons. In general, we don t identify ourselves with more than a few agent-units (e.g. the present temporal part, the whole person, family, nation, humanity), and therefore, almost all of those moral reasons are not ours * 17. I think my view is preferrable in that it can avoid the Units Proliferation Problem safely and keep (B) intact. I don t deny that various kinds of moral reasons can conflict in our deliberation. If you are identifying yourself with more than one units of agency at once, you may have a conflicting identities and desires, and it is natural that a conflict occurs in your practical deliberation. However, it occurs not because morality has any conflict in it it doesn t but because you have conflicting desires that cannot be fulfilled simultaneously. If there should be a conflict, it would be a matter of your practical identity, not of morality itself. 6 Conclusion: Back to the Main Problem Now, we can go back to the main problem and draw some analogical conclusions from the arguments in the previous section. Consider the Professor Procrastinate (p ) case again. Let p 1 denote a January part and p 2,... be the temporally succeeding parts of p respectively. As we have observed, (B) yield the Proliferation Problem, but we can avoid it and consequently, we can also plausibly hold (B) in the diachronic settings with the same manoeuvre. If we focus on the alternative-set of p 1 as a particular unit of agency, the consequentialist principle tells us that p 1 has a moral p 1 -obligation (and therefore, a p 1 -reason) to decline the task because p 2,... are uncooperative. At the same time, it has a moral p -obligation (and a p -reason) to accept it because of (B) and the p s p -obligation to accept it and comment, even though the succeeding parts p 2,... are uncooperative with p 1. True, those procrastinating parts p 2,... would actually fail to carry out the task, but it is utterly irrelevant to whether p 1 has a moral p -obligation to accept it. After all, the actualists are right in that p 1 has a moral p 1 -obligation to decline the task and no other *15 I find David Copp s skeptical argument about the unity of Practical Reason fairly convincing [cf. Copp 2007]. I don t think any kind of reasons are intrinsically good reasons simpliciter in the first place, hence my moral anti-rationalism. *16 About the problem of the practical identity and group agency, see [List and Pettit 2011, ch. 9]. *17 I think the willingness of the other members in the agent-unit in question affects our practical deliberation exactly at this point in two ways. First, it is difficult for us to identify ourselves as a member of a group whose members are uncooperative. In general, the less a group has agency, the more difficult it is for us to deem it to be an object of practical identification (I think this is a background intuition of the WR). This is ultimately a matter of psychological facts about us. Second, the actors outside the group I am identified with are deemed to be a circumstance of the actions of the group, and the (un)willingness of those outer actors are morally relevant just as the (un)willingness of everyone other than myself are morally relevant under act-consequentialism. To put it the other way around, the extent of the persons that I deem to be a circumstance is conversely just the extent of my self-identification. The more you fear the recklessness problem, the smaller the extent. 8

9 p 1 -obligations, whilst the possibilists are also right in that p 1 has a moral p -obligation to accept the task whether or not the succeeding temporal counterparts p 2,... (and therefore, p ) would fail to fulfill the task. Both of them correctly identify the moral obligations of p 1 respectively, and they are compatible because those moral reasons they provide are not conflicting at all with each other in the realm of morality though perhaps so in the realm of the practical reason of p 1. Bibliography [Carlson 1995] Erik Carlson, Consequentialism Reconsidered, Kluwer, [Copp 2007] David Copp, The Ring of Gyges: Overridingness and the Unity of Reason, in D. Copp, Morality in a Natural World: Selected Essays in Metaethics, Cambridge, [Feldman 1986] Fred Feldman, Doing the Best We Can, Reidel, [Goble 1993] Lou Goble, The Logic of Obligation, Better and Worse, Philosophical Studies, 70: (1993) [Goldman 1978] Holly Goldman (Smith), Doing the Best One Can, in Alvin Goldman and Kim Jaegwon eds. Values and Morals, Reidel Publishing Company, [Jackson 1987] Frank Jackson, Group Morality in P. Pettit, R. Sylvan, and J. Norman eds., Metaphysics and Morality, Blackwell, [Jackson and Pargetter 1986] Frank Jackson and Robert Pargetter, Oughts, Options and Actualism, The Philosophical Review, 95: (1986) [List and Pettit 2011] Christian List and Philip Pettit, Group Agency, Oxford, [Oddie 1993] Graham Oddie, Act and Maxim: Value Discrepancy and Two Theories of Power, Philosophy and Phenomenological Research, 53: (1993) [Oddie 1996] Graham Oddie, The Consequences of Action, in J. Copeland ed. Logic and Reality, Clarendon Press, [Ridge 2006] Michael Ridge, Introducing Variable-Rate Rule-Utilitarianism, The Philosophical Quaterly, 56: , (2006) [Smith 1976] Holly Smith (Goldman), Dated Rightness and Moral Imperfection, The Philosophical Review, 85: (1976) [Woodard 2008] Christopher Woodard, Reasons, Patterns, and Cooperation, Routledge, [Zimmerman 2008] Michael J. Zimmerman, Living with Uncertainty, Cambridge,