. If this agent is maximally ideal, then the agent must be able to exercise the capacity for desire realization robustly, across a variety of circumstances. But if the agent robustly realizes the desire to believe
, then the agent will realize this desire in circumstances that include those in which the available evidence tells against
. This case shows that exercising the capacity to realize one’s final desires may conf lict with exercising the capacity to know the world. As Smith points out, “An ideal agent thus turns out to be one whose psychology, by its very nature, displays lots of tension and disunity, as a higher score along one dimension comes at the cost of a lower score along another” (2013: 22). What are the choices for a defender of the standard view? one might hold that an ideal agent would maximize either belief acquisition or desire realization, or that there is some composite in which the ideal agent would have the highest aggregate “score” possible—even if this meant having a very dis-unified and incoherent psychology. Smith finds all of these options unappealing, and concludes that we should take one of them only if there is no available way of adding mental states to the ideal agent so as to make that agent’s psychology more coherent. Fortunately, it is very plausible that there are mental states that render the ideal agent’s psychology more coherent. Suppose the agent had a final desire to not now interfere with the exercise of their belief-forming capacities. Provided this desire exceeded the desire to now believe
in strength, the agent would then not face a dilemma. Because the agent’s psychology would be more coherent with this additional desire than without it, we can infer that the desire would be part of the ideal agent’s psychology. What is interesting is that this suggests that the standard story can not only say something about what beliefs and extrinsic desires the ideal agent would have, but can also say something about what final desires the ideal agent would have. In particular, the ideal agent would have what Smith calls “coherence- inducing desires”. Although this shows that one apparent implication of the standard story was mistaken, it does not yet show how the standard story could ground constitutive moral requirements. Smith next considers an agent who finally desires to believe
in the future. This desire sets up the same incoherence in the agent’s future psychology as the analogous desire creates in the agent’s present psychology. Because this makes the agent’s psychology less robustly coherent, Smith concludes that a desire to believe
in the future makes the agent’s psychology less ideal in the present. So, coherence-inducing desires will also include desires to organize the agent’s 12 Ryan W. Davis LEAP 4 (2016) psychology in the maximally coherent way in the future. not only should the agent desire not to interfere with the future exercise of their rational capacities, they should also have desires (now and in the future) to bring it about that the agent possessed these capacities (Smith 2011: 356). As with other constitutivist projects, grounding moral requirements must involve making the shift from a temporally extended concern with the self to a concern with other agents. Smith offers a couple of different considerations for how this move might be made. First, he suggests that if agents are to fully and robustly possess rational capacities now and in the future, they will have to count on other agents to not interfere with their use of these capacities, and also to help them possess these capacities. Part of having an ideal psychology—and recall this is using only the resources from the standard story—is to then have a concern for the rational capacities of others. Smith writes: “[I]f an agent is to robustly and fully exercise the capacity to believe for reasons, then he also has to be able to rely on the non-interference of other rational agents, assuming that there are such agents…[T]his too is grounded in the reasonableness of his supposing that all rational agents, if they are robustly to have and fully exercise their own capacities to believe for reasons, must desire not to interfere with other rational agents exercises of their capacities. For to suppose that rational agents do not extend their concern for non-interference to other rational agents in this way is to imagine that they make an arbitrary distinction between their reliance on themselves and their reliance on others—despite the fact that all of those on whom they must rely, insofar as they exercise their capacity to believe for reasons, have the very same interests in the non-interference of others as they have in themselves” (2011: 357). Smith has another argument for the same generalizing move to other agents. The reasons to want to maintain the functioning of one’s rational capacities remain in place even if the agent undergoes changes during the course of exercising a rational capacity, such that the changes do not preserve the agent’s personal identity (Smith, 2012: 323-327). The move from self to other agents has long been controversial for constitutivists. one potential concern with the account developed here is how it could capture the type of universality that we characteristically regard as characteristic of morality. Why, that is, would an ideal agent want to help and not interfere with all other rational agents, as opposed to merely that subset who happened to be around her, and could affect her capacities?3 3 I’m grateful to a referee for pressing me to think about this question. Which Moral Requirements Does Constitutivisim Support? 13 LEAP 4 (2016) The crucial idea in the account presented here is that, given the similarity between other agents and oneself, it would be arbitrary to hold the relevant desires with respect to oneself and not to others, and it would be similarly arbitrary to desire to help and not interfere with some other agents, but not with other agents.4 Extending the relevant concern to all agents is a matter of being “fully consistent, treating like cases alike” (Smith, 2015: 192).5 In any case, once we grant that ideal agents finally desire to not interfere with and to help other agents, then we can quickly see that they have reasons to do the same. To rehearse: because the concept of an agent is “goodness-fixing”, agents are evaluated as better or worse, depending in part on the extent to which they fulfill their final desires. The concept of a reason can then be analyzed in terms of what is desirable relative to the agent, which in turn is given by the desires of the idealized version of the agent (Smith 2015: 188-189; 2013). These reasons, for Smith, ground the fundamental moral requirements—not interfering and helping. Smith, again: “In virtue of the fact that every agent’s fully rational counterpart has these desires, every agent has the same reasons for action, and these reasons for action, I hereby conjecture, are reasons to do what agents are morally obligated to do. Agents are morally obliged not to interfere with any rational agent’s exercise of his rational capacities, and they are also morally obliged to do what they can to make sure agents have rational capacities to exercise” (2011: 359-360). If Smith’s account succeeds, it would show that moral obligations can be derived using only the resources of a descriptive explanation of agency. This would be a remarkable achievement. More, it would show that the constitutive moral requirements were extensionally very similar to our ordinary moral beliefs. Among other things, it would show that our moral obligations are non-welfarist, agent-relative, and deontological. non- welfarist: because moral reasons concern the presence and exercise of rational capacities, not well-being. Agent-relative: because the desires of the ideal agents, which ground the relevant reasons, are to help and to not 4 Compare a set of agents who are concerned only with the rational capacities of others in their vicinity, and a set of agents who are concerned with all rational agents. Let us assume that it is, in principle, possible that one’s rational capacities could come to depend on the actions of any other rational agent. If this is true, then the first class of agents will not possess their rational capacities as robustly as the second class of agents. Given that robustness is a feature of the ideal agent, those agents who are concerned with the rational capacities of all other agents are more ideal. 5 The issue deserves more attention than I can provide here, but further investigating this question would divert the essay from its intention of setting aside whether constitutivism is correct, and focusing on the content of its resulting requirements. 14 Ryan W. Davis LEAP 4 (2016) interfere, rather than to bring about the maximization of helping among all agents, or the minimization of interference. This also explains why the requirements are also deontological, in the sense that they cannot be reduced to agent-neutral values.6 To borrow david Velleman’s (2009) phrase, it would vindicate at least a “kinda Kantian” normative ethics. 2. ConFLICTInG rEQuIrEMEnTS Suppose we accept the requirements to not interfere and to help. How would these requirements guide our action? Although many obligations we intuitively accept involve some combination of not interfering and helping, accepting these two general requirements recreates the possibility for conf lict that the standard story created for rational requirements. Problem cases will be those in which acting to help counts as interfering, and refraining from interfering counts as failing to help. Individuals may use their rational capacities at a given time in a way that will undermine those capacities in the future. This possibility creates along with it the prospect of tension between the two fundamental moral requirements. For example, consider an individual’s decision to use potentially rational-capacity impairing drugs. If I withhold from interfering with the agent’s exercise of their capacity to realize their desires, then I will be failing to help secure the conditions under which their capacities to realize their desires or to know the world will be effective in the future. Moreover, this tension between a moral concern for the exercise of the capacities, and the capacities themselves, is likely to arise often.7 If the moral requirement to not interfere with people could be compromised anytime they act in ways that undermine their future use of their rational capacities, then Smith’s constitutivism might not deliver a morality as consonant with our intuitions as he might have hoped. As Jessica Flanigan writes: “Sleeping aids, roller coasters, alcohol, standing on one’s head during yoga class, falling in love, and falling out of love can all be seriously incapacitating in their own way, but no one would ever say that 6 They are not deontological in a much stronger sense, in which the deontic facts could not be reduced to any evaluative facts. Cf. Smith (2009). 7 As a referee points out, an ideal agent would have a dominant desire to not impair their rational capacities, and so would not make this choice. I acknowledge as much; it is important that the case I describe here could not arise among ideal agents. Below, I will address this issue by considering conf licts of this sort that could arise among ideal agents, and then extending the concern to non-ideal agents. For now, my aim is just to motivate the case for actual agents who try to live by the helping and non-interfering requirements. Which Moral Requirements Does Constitutivisim Support? 15 LEAP 4 (2016) interference on behalf of the would-be incapacitated is justified” (Flanigan unpublished: 3; cf. Flanigan 2012). Although all of the above cases involve taking some action that undermines one’s rational capacities, we can also imagine conf lict cases in which an agent has a desire to not act in a way that would develop rational capacities. I am told that my capacity to know the world would be much improved if only I would learn some econometrics. However, I have a very strong desire to not spend any summers studying econometrics. Thus I remain at my middling state of being able to know the world, because I privilege my exercise of my capacity to achieve my desires. Again, helping my development of my rational capacity could only come at the price of interfering with its exercise.8 There is no problem with thinking that we have conf licting reasons for action, because the presence of a reason does not imply that there are no countervailing reasons. Whether agents can be subject to conf licting obligations is more controversial. The presence of an obligation typically indicates that the obligated agent might be blamed if they fail to act on the obligation, and they have no excuse (Cf. darwall forthcoming). Perhaps incompatible obligations merely indicate that an agent could be blamed no matter which action is chosen, but this might sit in tension with our ordinary practice of blaming—which supposes that the blamed agent could have acted so as to avoid being blamed.9 nevertheless, many philosophers have developed strategies for allowing inconsistent obligations (Horty 2003; Goble 2009; nair 2014). It is beyond the scope of this essay to address this general matter, so I will set it aside in order to focus on whether there is a special problem with conf licting obligations for Smith’s constitutivist account. To proceed, compare the psychology of an ideal agent who was subject to conf licting obligations with an ideal agent who was not subject to con- flicting obligations. Would one psychology be more coherent than the other? I suspect there are several ways in which an ideal agent’s psychology would be rendered less coherent by conf licting obligations. Suppose the agent intended to comply with all of their obligations. Then the agent would have intentions that were not jointly realizable. Allowing that intention must involve at least the belief that one may do as one intends, the presence of incompatible intentions would imply that the agent had conf licting beliefs, 8 Smith emphasizes that possible ideal agents may know a wide variety of different things, and have a wide variety of different final desires. 9 Bart Streumer (2007) defends the claim that “it cannot be the case that a person ought to perform an action if this person cannot perform the action.” Even if this is not true for “oughts” generally, it may still be true for all-things-considered moral obligations. Cf. Graham (2011: 367-378). 16 Ryan W. Davis LEAP 4 (2016) and was therefore incoherent.10 Alternatively, the agent might intend to comply with only some of their obligations. Partial compliance threatens other kinds of incoherence. In one case, the agent might decide arbitrarily which obligations to fulfill. of course, choosing purely arbitrarily which obligations to fulfill will likely produce diachronically sub-optimal results. Let us grant the possibility of deontic inconsistency. An agent who chooses arbitrarily which obligations to fulfill may well end up satisfying fewer obligations overall, relative to an agent who chooses current actions with an eye toward being able to better fulfill obligations in the future. So an arbitrary selection strategy will not be used by an ideal agent.11 next consider an agent who determines which obligations to fulfill so as to maximize the total number of satisfied obligations. This strategy looks roughly analogous to the “highest aggregate score” strategy mentioned earlier to describe a more basic level of agential functioning. Again, this is intuitively incorrect. Philosophers who accept deontic inconsistency still allow that some obligations are more important than others, a fact obscured by simple aggregation. The constitutivist picture can support this intuition. The highest aggregate score model was previously rejected, since it accepts a bundle of incompatible desires as constituting the ideal psychology. A more ideal psychology would not take such “dysfunction” to be a feature of the ideal (Smith 2012: 314). As we have seen, a more ideal psychology would include dominant, coherence-inducing desires. Likewise, compare the psychology of an agent who simply maximized obligation satisfaction with an agent whose psychology included elements that provided reasons to prioritize some obligations and not others. For considerations analogous to the earlier case, the latter psychology would be more coherent, and so also more ideal. The last option would suggest that an ideal agent would choose which obligations to fulfill on the basis of reasons. Yet, how could there be reasons on the basis of which to make such a choice? To say that I am obligated to do something suggests that I have decisive reason to do it, or at least that I have sufficient reason to do it. If I have sufficient reason to perform either of two incompatible obligations, then on what could I deliberate between them? If there are reasons to deliberate on, then it seems that I may not have sufficient reason to do one of the things I am obligated to do after all. That would deny what I am taking as a conceptual truth about obligation. one possibility here is to think that there are “enticing reasons” to 10 Cf. Bratman (2009). Also on the irrationality of incompatible intentions, see Liberman and Schroeder (2016: 110). 11 This conclusion is consistent with Smith’s rejection of arbitrary discrimination, in other areas—for example, among other agents. Cf. Smith (2011: 357). Which Moral Requirements Does Constitutivisim Support? 17 LEAP 4 (2016) discriminate among obligations, where an enticing reason to perform some action does not undermine the sufficiency of the reasons supporting an alternative.12 Another possibility is to distinguish between an agent’s being obligated, simpliciter, and an agent’s being obligated, all-things- considered.13 In either case, there will be a further question about how an ideal agent would prioritize obligations. To sum up, the obligations to help and to not interfere can come into conf lict. on the constitutivist picture, such conf licts can be characterized in terms of incoherence in the psychology of ideal agent. one way of managing this incoherence is to deny one of the obligations in question; a second is to allow conf licting obligations, but locate some further considerations to establish priority; a third is to locate some further considerations to establish what the ideal agent is all-things-considered obligated to do. For any of the three, further attention to the ideal agent’s psychology is demanded. It is tempting to think that we might appeal to the agent-relativity of the requirements to help and to not interfere in order to explain away any conf lict in obligations (Smith 2011: 361; 2015: 192; 2003). The ideal agent is concerned about that agent’s own compliance with the two requirements, not with maximizing compliance generally. In some well-known cases, an apparent dilemma between competing obligations can be dissolved by appealing to agent-relativity. For example, if an agent is concerned only with her own non-killing of other agents, and not the reduction of killing overall, she might refrain from killing an innocent, even it will bring about that some other killing of an innocent occurs. Agent-relativity can thereby support a distinction between “doing” and “allowing”, which might be thought to bear on dilemmas between helping and not interfering. Although I will not explore this matter in detail, I do not regard this direction as promising. no such solution is likely to be in the offing, because agents have agent-relative reason both to not interfere and to help. The agent has a reason to avoid interfering, but the agent also has an agent-relative reason to bring it about that helping is produced through their own efforts. Agent- relativity cannot offer any traction in choosing between apparently conf licting obligations. Another strategy might be to think about cases of resolving conf lict within a single agent, and then try to generalize this to the case of moral obligations toward another agent. And in fact, it is plausible that the tension 12 A referee provided this suggestion, which I had not previously considered. See, for example, dancy (2004). 13 Thus allowing for different obligations to have different weights, as favored by Liberman and Schroeder (2016). 18 Ryan W. Davis LEAP 4 (2016) between helping and not interfering would arise within a single agent. Consider again the dangerous drug case. An ideal agent might have any first-order desire, and so might desire to take capacity-damaging drugs. However, the ideal agent would also have a coherence-inducing desire to avoid interfering with the agent’s future use of belief-forming and desire- realizing capacities. To achieve coherence, the latter desire would have to be dominant, and so the ideal agent would never have a dominant desire to take the drugs in the first place. This shows that in interactions between ideal agents, one agent will never have to consider whether to help or not interfere with another agent in this kind of case, since the ideal organization of the patient’s psychology will prevent the conf lict from arising. notice, however, that this conclusion does nothing to help the ideal agent who has to interact with non-ideal agents. There is no assurance that a non-ideal agent will have the relevant coherence inducing dominant desires, and so there is no assurance that non-ideal agents will not act in ways that threaten their rational capacities. Thus, if an ideal agent is interacting with a non-ideal patient, the patient may well decide to take a capacity-impairing drug. In this case, the ideal agent will be forced to prioritize either helping or not interfering. If the ideal agent helps (by interfering so as to stop the non-ideal agent from taking the drug), the ideal agent will be failing to comply with the obligation to not interfere. Likewise, helping can only be achieved through interference. Although the tension might not happen to arise for residents of the Kingdom of Ends, the actual world seldom affords such morally propitious conditions (Cf. Korsgaard 1996; Schapiro 2003). There is no reason this kind of case could not arise within Smith’s constitutivist system. Although Smith formulates the view initially within a community of ideal agents interacting with each other, he explicitly allows that ideal agents have reason to abide the moral requirements with respect to non-ideal as well as ideal fellow agents (Smith 2015: 191; 2012: 329). For simplicity it may help to begin with a conf lict case that could arise within even ideal agents. Smith poses a helpful case: an ideal agent is suffering from an incurable, degenerative disease, but it so happens that forming the false belief that one is getting better actually does delay the progress of the disease, thereby preserving the patient’s deliberative capacities in the future. The patient has a drug that, if taken, will cause the formation of the helpful false belief. Because the agent is ideal, there is a coherence-inducing desire to not interfere with one’s rational capacities in the present, which provides a reason against taking the drug. Likewise, there is another coherence-inducing desire to help one’s rational capacities in the future, which counts in favor of taking the drug. Because the conf lict is between two coherence-inducing desires, it won’t do to say that the Which Moral Requirements Does Constitutivisim Support? 19 LEAP 4 (2016) coherence-inducing desire is dominant. Smith recommends resolving this conf lict “in a principled way, specifically by reference to the relative strengths that these desires have to have vis-à-vis each other simply in virtue of being the desires of an ideal agent” (Smith 2012: 319). As I understand it, the relative strengths that the desires “have to have” are fixed by facts about what would maximize the agent’s satisfaction of final desires, and knowing the world, given the agent’s circumstances in the present and in the future.14 In other words, the agent would have a relatively stronger desire to take the drug if doing so would overall promote the agent’s final desires’ satisfaction and knowledge of the world better than not taking the drug. This way of resolving the case is principled in that it appeals to the agent’s success qua agent across time, rather than to our own intuitions about whether or not taking the drug is rational. What remains to be shown, I suggest, is how to carry out analogous reasoning between two different ideal agents, and then between an ideal agent and a non-ideal agent. Both the aspirations of Smith’s constitutivism, as well as our everyday moral situation, call for extending the theory to cases like these. 3. PrIorITIZInG rEQuIrEMEnTS recall that so far, the constitutivist strategy has given us the following principles. Non-interference: It is impermissible to interfere with any rational agent’s exercise of his capacities. Help: It is morally required to do what one can to make sure that agents have rational capacities to exercise. The last section canvassed the constitutivist view to look for additional resources for resolving conf licts among these principles. The following are relevant. First, agents with more coherent psychologies are, ceteris paribus, more ideal than those with less coherent psychologies. Second, if an ideal agent must decide between either not helping or interfering with their future self, the agent will act so as to maintain the ideality of her future self’s psychology. Third, ideal agents will act to maximize their knowledge of the world and satisfaction of final desires. now we can deploy a similar strategy in the two-person case. If one 14 I am persuaded of this interpretation by a referee. I am not confident that the ideal agent would be one who maximally achieves intrinsic desire satisfaction and knowledge of the world, since this standard sounds similar to the (rejected) “highest aggregate score” criterion, discussed above. All the same, it is a better interpretation than denying there are any facts fixing the desires’ relative strengths. 20 Ryan W. Davis LEAP 4 (2016) ideal agent must choose between helping and not interfering with another ideal agent, the agent should act so as to maximize the patient’s knowing the world, and fulfilling final desires. Further, the acting agent should act to bring about that the patient’s psychology is as ideal as possible, and so, as coherent as possible. So we can add a further principle: Coherence: If one must either not help or interfere with another agent, one should do whichever would be supported by the most coherent rendering of that agent’s psychology. Imagine one ideal agent must choose between helping or not interfering with a second ideal agent. Suppose an agent [A] is deciding whether to take a helpful drug that would cause A to form a false belief. Some other agent [B] must choose between interfering with A’s choice and not interfering. If A is an ideal agent, it is—according to the last section—possible that A will choose either option. Let us suppose, given the facts about A’s circumstances, that A chooses not to take the drug. In this case, it would not make sense for A to also want B to interfere with A’s choice. The reasons for A’s wanting B to interfere with A’s choice would also, by hypothesis, count in favor of A’s not making the choice that A made in the first place. So if A were to then prefer that B interfere with A’s choice, A’s psycholog y would not be ideally coherent. So, A has most reason to want B to not interfere with A’s choice in the case of conf lict. next, suppose that B were committed already to some combination of helping and not interfering with A. If B had these commitments, it would not be coherent for B to then make decisions about whether to help or not interfere that disregarded what A had most reason to want. If B were to do that, then B would be both committed to acting in ways that were sensitive to A’s reasons to want A’s rational capacities to be helped and not interfered with, but insensitive to A’s reasons about how helping and not interfering should be prioritized. This combination of responsiveness and non- responsiveness would, I think, impose a tension within B’s psychology. So in the case of two ideal agents, we can infer how one would prioritize helping and not interfering with respect to the other. The acting agent would honor the priorities of the agent in the role of patient, whatever those priorities might be. This inference can be further refined. Given variation in the circumstances, A may decide to take the drug that will interfere with A’s capacities in the present, but will do so in a way that helps A’s future capacities. or, given other circumstances, A may decide against taking the drug. So A may prioritize either helping or not interfering with respect to future A. However, given A has an ideal psychology, B’s response will be to Which Moral Requirements Does Constitutivisim Support? 21 LEAP 4 (2016) not interfere with A. If A interferes with A’s future self, or if A decides to help A’s future self, B will not interfere in either case. A question arises when we consider the case of an ideal agent confronting a choice of whether to prioritize helping or not interfering with respect to a non-ideal agent. recall from above that ideal agents may well encounter other agents who are non-ideal in a variety of ways. If A is a non-ideal agent, then A may opt to not take the helpful drug, notwithstanding that under the circumstances, A would do better as an agent if A did take the drug. A’s psychology is not coherent. But if it were coherent, then A would want to take the drug. In this case, it seems that B, an ideal agent, should interfere with A to bring about that A takes the drug. Although this may be correct about the case of the helpful drug, cases more enriched with realistic detail may reveal relevant, complicating considerations. Consider again the case of my refusal to study econometrics. An onlooker, persuaded by my social scientifically inclined friends, decides that it would be good for my rational capacities to enroll me in a remedial summer economics class without my consent. Although enrolling me would—by hypothesis—help cultivate my capacity to form correct beliefs about the world, it would also interfere with my autonomy. Which should the observer privilege? According to Coherence, my friend should investigate the relative strengths of my desires as part of an investigation of which value would better (that is, more coherently) resolve the internal tensions within my psychology. By hypothesis, if I cannot be persuaded to the take the class by non-interfering methods, then it is probably unlikely that the class will improve my capacities enough to justify the trade-offs with my end-setting and desire-satisfying capacities. I will just resent the infringement on my liberty, lack interest to study effectively, and so on. In other words, the same features of my psychology that make it non-ideal may render the helpful action ineffectual. An ideal agent would also be responsive to this non-ideality, and so would have reason to defer to my refusal, non-ideal though it might be. Maybe this case seems too easy. recall the would-be recreational drug user. This person has an end of using drugs that might damage their rational capacities in the future. Should an onlooker interfere in the drug user’s life for the good of their future rational capacities? This case may seem more challenging—and indeed, a number of philosophers side with the preservation of rational capacities, even by way of state coercion if necessary (Freeman 1999; de Marneffe 2003; but compare Koppelman 2006). While I allow that this outcome cannot be ruled out according to Coherence, there is some reason to privilege non-interference. To see why, imagine the best scenario for the would-be paternalist: the drug user 22 Ryan W. Davis LEAP 4 (2016) judges that he should use the drug, but this judgment is—in coherentist terms—mistaken. We can add detail for the convenience of seeing the case at higher resolution. Suppose the drug user wants to go to school and study a technical subject, he admires people who resist temptation, he wants to set an example of “clean” living for his younger siblings, and he wants to preserve his cherished memories of his youth. A dominant desire to take the drug does not cohere with these other attitudes. My suggestion is that even if the drug user is in this sense mistaken, it does not follow that the would-be paternalist should interfere. This is because the drug user’s own judgment must now be included as a member of the elements of his psychology, and that judgment tells in favor of taking the drug. Even if the drug user’s original judgment was mistaken, its mental genealogy does not make it any less a part of the drug user’s psychology. Further, the drug user is likely to make additional plans based on that judgment, forming intentions and policies that cohere with it. After adopting these plans, the drug user’s other attitudes may naturally shift in ways that cohere better with the judgment in favor of taking the drug (Cf. Velleman 2006; 2008). The preceding claim is about human psychology, but in principle my suggestion does not rely on any psychological conjecture. It only requires that an ideal agent’s judgments about what to do will affect the content of their subsequent attitudes such that those attitudes will tend to cohere with the judgment. This alone, I think, is enough to at least tip the scales in the direction of deference to the agent’s choice. If this is right, then Coherence supports another principle for deciding how to reconcile the original two. Deference: one should not, ceteris paribus, interfere with an agent for the sake of promoting their rational capacities. Again, the ceteris paribus clause makes Deference defeasible. Here it will help to distinguish between the agent’s local and global coherence.15 The drug user’s judgment may create a series of attitudes that cohere with that judgment, but these attitudes will likely include only a part of the total set of his desires and beliefs. However, the drug user’s taking the drug may contribute to thwarting the agent’s completely unrelated desires, and may undermine the correctness of unrelated beliefs. Thus, the drug user’s judgment in favor of taking the drug may be locally coherent, but—with respect to the total set of the agent’s attitudes—globally incoherent. The extent to which Deference is generalizable depends on how significantly considerations of local coherence impinge on an agent’s global coherence. In the case of the drug user, local coherence of attitudes that fit with the drug user’s plan are outweighed by its global incoherence. In other cases, 15 I’m grateful to an anonymous referee for this suggestion. Which Moral Requirements Does Constitutivisim Support? 23 LEAP 4 (2016) the opposite may be true. Imagine an agent who irrationally decides to embark on a career to which he is not well suited. This decision is non- ideal; it will, in expectation, realize the agent’s ex ante desires less well than other career options. However, the agent not only strongly desires to embark on this career, but thereby adopts a whole series of related desires, beliefs, and plans. The agent might form other final desires to develop the skills necessary for the career, may form plans to receive training for the career, and might intend to move to different parts of the country to facilitate the career. As the set of relevantly connected attitudes expands, it becomes more likely that the local coherence with the agent’s initial judgment will affect what is globally coherent for that agent. If this line of reasoning is correct, we should expect that Deference will apply more often in cases in which an agent has formed many desires and beliefs around a given judgment, and less often in cases where a judgment is at odds with an agent’s other desires and plans. To a considerable extent, this fits with our intuitions about when we ought to respect a person’s sub- optimal choices. Choices that are more central to a person’s beliefs, which ref lect “deep commitments” or “personal integrity”, are plausible candidates for respect, whereas choices less connected to other attitudes are correspondingly more plausible candidates for paternalism (Williams 1973). In fact (although I cannot pursue this conjecture here), the constitutivist program followed here might provide one way of explaining the normative significance of such locutions. on this view, a choice would be “deeper” or more associated with an agent’s “integrity” if it impinges to a greater degree than other choices on an agent’s global coherence. 4. THE PrIorITY oF non-InTErFErEnCE So far I have tried to show that constitutivism, at least in the form presented here, tips the scales of moral obligation slightly in favor of deference to individual choice. It tends toward what rawls called the “priority of liberty” (rawls 1999: 214-220). I regard shoring up this presumption as sufficient to satisfy the original aim of this paper. nevertheless, in this section I hope to find support for a stronger version of the thesis. recall that Smith’s argument moves from the premise that agents must desire to not interfere and to help to the conclusion that they are morally required to not interfere and to help. The argument is something like this: 1. Ideal agents have dominant desires to help and to not interfere. 2. If ideal agents have dominant desires to Φ, then their real-world counterparts have decisive reason to Φ. 24 Ryan W. Davis LEAP 4 (2016) 3. obligations are grounded in decisive reasons. 4. So, agents are obligated to help and to not interfere. There are several questions we could ask about how the desires to help and to not interfere could ground moral obligations. First, one might ask how morality got into the story at all. For none of the previous desires that Smith considered, including coherence-inducing dominant desires, did he infer that their corresponding reasons were moral reasons. I take it that Smith simply infers the moral character of the reasons to not interfere and to help from their extensional similarity to our ordinary judgments about the content of moral reasons. He writes: “The striking similarity of these acts to those that we ordinarily take to be morally required is, the Constitutivist insists, manifest. The only reasonable conclusion to draw is that every agent isn’t just rationally required to help and not interfere, but that, at the most fundamental level, every agent is morally required to help and not interfere as well” (Smith 2013: 26). Smith may not have much at stake in whether this inference to moral requirements holds. In a passage cited earlier he describes it as his “conjecture”, and here he recommends it as a kind of obviously reasonable conclusion. Granting that we have moral reasons to not interfere and to help, I am less sure that it follows that these are requirements. I am also less sure that this inference follows from Smith’s constitutivist account. To begin with the former, we ordinarily accept that we have many moral reasons that we are not required to act on, even in the absence of strong opposing reasons. If we take for granted a basic moral category of supererogation (or even something like imperfect duties), then there are likely many moral reasons that do not yield a requirement to perform any particular action (driver 1992; darwall 2006; Wolf 2009; Harman 2016). Moreover, it would be strange if we were morally required to act in ways that promoted the development or acquisition of others’ rational capacities. It does not fit with our intuitions that we have obligations to ensure that other people (at least, other adults) go to class, or refrain from taking drugs, or avoid falling in love—notwithstanding that these all correspond to ways of ensuring various capacities for knowing the world. These concerns form part of a larger worry, which is that a set of moral requirements to “help” would ask more from us than a commonsense mo- rality supposes. There are—to put it mildly—many people in the world whose rational capacities are not fully and robustly realized (Caplan 2007). doing what we could to help them would likely require living very Which Moral Requirements Does Constitutivisim Support? 25 LEAP 4 (2016) differently than we now live, but this is at odds with our current practice of moral praise and blame. We do not resent people who fail to dedicate them- selves to helping in the same way that we resent those who stand us up for lunch. This echoes the standard “overdemandingness” worry prevalent in the moral philosophical literature (railton 2003; Herman 2001; Sin 2010; noggle 2009; Igneski 2006; Jamieson 2005). But within the constitutivist framework underwriting this discussion, we can more precisely frame why the worry poses a theoretical problem. The issue is not merely that the demands are intuitively too demanding. rather, the issue is that such demands would predictably disorder an agent’s psychology. If we were to dedicate ourselves to helping (in Smith’s technical sense), it would likely take so much time as to compromise our pursuit of our other final desires and cultivation of rational capacities. Perhaps if we had significantly re- stricted sets of final desires, or final desires that happened to cohere with a rigorous program of helping, then they would not conf lict with an obliga- tion to help. recall, however, that an ideal agent can have a great variety of final desires. It is not plausible to assume that such incoherence-creating conf licts could be avoided. nor will it help to insist that the ideal agent’s final desire to help will be a dominant (coherence-inducing) desire. As noted above, there will likely be many candidates for helping, requiring a kind of triage in deciding where to help. Choices must also be made about how much to trade off helping others with other dominant-desire support- ed ends, including not interfering with one’s future self, and helping one’s future self. In short, treating the moral reasons to help as requirement-grounding creates much possible incoherence in an agent’s psychology. But was there a good theoretical basis for treating reasons in this way to begin with? Consider again the single agent whose idealized psychology happens to finally desire to believe
. That desire conf licted with another desire that the ideal agent turned out to have—a desire to not interfere with their capacity for belief. Imagine leaving it an open question, for any given case of such conf lict, which desire happened to be stronger. If that question had been left unsettled, there might have been cases in which the desire to believe
prevailed, such as when the importance of believing on the evidence seemed relatively low. That state of affairs would have f louted a rational requirement on belief, which is that beliefs must still be apportioned to the evidence even when the content of the belief is unimportant (Kelly 2002). revisiting this case shows how to locate the emergence of the rational requirement. Here, the requirement on belief is not given by any comparison of the strength of the desires that bear on how to believe. Instead, the 26 Ryan W. Davis LEAP 4 (2016) theory produced a model that extensionally resembled the rational norms on belief by positing an additional mental state—a coherence-inducing desire—and then ensuring that this desire would always be dominant. In the remainder of this section, I will brief ly sketch how conf licts among dominant desires might be managed within the ideal psychology. To be ideal, an agent must satisfy as many of their dominant desires as possible. However not all dominant desires can ground requirements. note that incoherence only arises with respect to helping, but not with respect to not interfering. Any agent may maximally satisfy the requirement to not interfere with other agents’ exercises of their rational capacities. All you need to do is nothing at all. Some philosophers have tried to deny that it is possible to avoid interfering with other agents (Pogge 2002). I will not argue against this view here, but I do not agree (Cf. risse 2005). While I will not try to specify what counts as non-interference here, I am sympathetic to the hypothesis that a great many human endeavors can succeed at not interfering in the relevant sense.16 If not interfering is uniformly possible in a way that helping is not, then not interfering can always be required, while helping cannot. With that distinction in hand, we can say something about the traditional asymmetry between negative and positive duties. Very generally, negative duties (which forbid actions) seem morally more stringent than positive duties (which require actions) (for example, Foot 1977; Thomson 2008). The constitutivist account helps to explain the difference. Because helping requires action while not interfering does not require action, the asymmetry between the moral status of helping (sometimes required) and the moral status of not interfering (always required) fits the asymmetry between negative and positive duties. now the question is: How can we add a mental state to the psychology of an ideal agent in order to fix the terms of when helping is a moral requirement? We could try to say that the ideal agent would maximize helping overall, or maximize instances of that agent’s own helping actions. These would also threaten incoherence, given that they would predictably interfere with the agent’s dominant desires to develop rational capacities and not interfere with the agent’s own exercises of those capacities. But this fact may give a clue to discerning when helping others could be required. Perhaps if an act of helping would not conf lict with any of the agent’s dominant desires with respect to the agent’s future self or with respect to other agents, then it could also be promoted to the status of a requirement (Ebels-duggan 2009). The ideal agent might have some 16 See ripstein (2009) on the difference between interfering with a person, and changing the circumstances of their choice. Which Moral Requirements Does Constitutivisim Support? 27 LEAP 4 (2016) additional mental state that facilitates this coherence. For example, an ideal agent might be required to perform those helping actions that he had promised or otherwise committed to perform. Beyond this, the ideal agent would intend to act on some combination of desires to help others, and desires to help and exercise the agent’s own capacities. How could we ensure that a requirement to help some given agent would not conf lict with any of an ideal agent’s other dominant desire supported ends? There may be a variety of ways to achieve this result, but one suggestion is to expect that dominant desires to not interfere will be especially weighty, relative to dominant desires to help. Although the details of how such a weighting might be developed will have to be left aside for now, the general contour of this idea fits with many first order intuitions, as well as widely accepted theoretical commitments. For example, it conforms with an intuition mentioned earlier: the fact that another could be helped by our action is generally not sufficient to require our action. It also fits with the diversely motivated theoretical commitment that there is “a clear sense in which [morality’s] fundamental prohibitions (its ‘thou shalt nots’) are more strict than its fundamental exhortations (its ‘thou shalts.’)” (Graham 2011: 377). For now, all I want to suggest is that “helping” and “not interfering” can both be correct principles, provided that we see the limits of the requirements they together create. 5. ConCLuSIon This essay takes constitutivism about morality for granted. Suppose that moral requirements are grounded in what is constitutive of agency. What would that tell us about the content of moral requirements? Michael Smith answers that it would reveal that helping ensure that other agents have rational capacities, and not interfering with the exercise of those capacities, are the fundamental moral requirements. The problem is that these requirements can conf lict. That news is not too bad, though, because Smith’s entire constitutivist project is worked out in terms of resolving conf licts in an agent’s psychology. using similar strategies, this paper has argued that the potential for conf lict can be solved. The weaker thesis of this paper is that there is reason to defer to the agent’s choice in deciding between helping and not-interfering, and so we should have what rawls called a “presumption of liberty.” The stronger thesis is that not- interfering is always required, but helping is only required sometimes. one interesting upshot of these claims is that, if correct, they can contribute to explaining other aspects of our moral practice, such as the asymmetry between doing and allowing. Another interesting upshot is that it will turn 28 Ryan W. Davis LEAP 4 (2016) out that Immanuel Kant—at least on one reading—was right about how we are obligated to other persons.17 Put in a mundane way, my essay has tried to make one modification to one existing version of constitutivism. But put in a more dramatic way, the proposal of this essay shares the aspiration of constitutivist theories since their start—to vindicate the truth of Enlightenment liberalism. If the amendment offered here is right (along with, I suppose, all of the foregoing theory as well), then we are rationally required to treat the liberty of persons as sacred.18 BIBLIoGrAPHY Alm, d., 2011: “defending Fundamental requirements of Practical reason: A Constitutivist Framework”, Journal of Philosophical Research 36: 77-102. Bertea, S., 2013: “Constitutivism and normativity: A Qualified defense”, Philosophical Explorations 16: 81-95. Bratman, M., 2009: “Intention, Belief, and Practical rationality”, in Reasons for Action, eds. d. Sobel and S. Wall, Cambridge: Cambridge university Press. Caplan, B., 2007: The Myth of the Rational Voter: Why Democracies Choose Bad Policies, Princeton: Princeton university Press. Cohen, A.J., 2008: “Existentialist Voluntarism as a Source of normativity”, Philosophical Papers 37: 89-129. dancy, J., 2004: “Enticing reasons”, in Reason and Value: Themes from the Moral Philosophy of Joseph Raz, eds. r.J. Wallace, P. Pettit, S. Scheffler, and M. Smith, oxford: oxford university Press. darwall, S., 2006: The Second-Person Standpoint: Morality, Respect, and Accountability, Cambridge: Harvard university Press. —Forthcoming: “Morality, Blame, and Internal reasons”, in Does Anything Really Matter?: Parfit on Objectivity, ed. P. Singer, oxford: oxford university Press. davidson, d., 1963: “Actions, reasons, and Causes”, in Essays on Action and Events, oxford: oxford university Press, 1980. 17 on Karen Stohr’s (2010) reading of imperfect duties, “it appears that it is up to me to choose the occasions on which I will be beneficent.” If Kant thought interference was always wrong; that helping was sometimes required; but that no arbitrary case of helping was sometimes required; and that what determined whether a given case of helping was required was up to the helping agent—then Kant was right about everything. 18 I am grateful for discussions of earlier versions of this paper with the graduate fellows at the Edmund J. Safra Center for Ethics, and in the graduate workshop in moral and political philosophy at Harvard university. I would also like to express thanks to Eric Beerbohm, Selim Berker, Jessica Flanigan, Katie Gallagher, Christine Korsgaard, Mark Hanin, Kelly Patterson, Thomas Scanlon, Michael Smith, and two anonymous referees for Law, Ethics and Philosophy. Which Moral Requirements Does Constitutivisim Support? 29 LEAP 4 (2016) de Marneffe, P., 2003: “Against the Legalization of Heroin”, Criminal Justice Ethics 22: 38-44. Ebels-duggan, K., 2009: “Moral Community: Escaping the Ethical State of nature”, Philosopher’s Imprint 9: 1-19. driver, d. 1992: “The Suberogatory”, Australasian Journal of Philosophy 70: 286- 295. Ferrero, L., 2009: “Constitutivism and the Inescapability of Agency”, in Oxford Studies in Metaethics, ed., russ Shafer-Landau, IV: 303-334. Flanigan, J., 2012: “Three Arguments Against Prescription drug requirements”, The Journal of Medical Ethics 38: 579-586. —unpublished: “When deontological Moral obligations Conflict”, ms. Foot, F., 1977: “The Problem of Abortion and the doctrine of double Effect”, in Virtues and Vices and Other Essays in Moral Philosophy, oxford: oxford university Press, 2002. Freeman, S., 1999: “Liberalism, Inalienability, and the rights of drug use”, in Drugs and the Limits of Liberalism, ed. P. de Greiff, 110-130, Ithaca: Cornell university Press. Graham, P.A., 2011: “‘ought’ and Ability”, Philosophical Review 120: 337-382. Goble, L., 2009: “normative Conflicts and the Logic of ‘ought’”, Nous 43: 450-489. Harman, E., 2016: “Morally Permissible Moral Mistakes,” Ethics 126: 366-393. Herman, B., 2001: “The Scope of Moral requirement”, Philosophy & Public Affairs 30: 227-256. Horty, J., 2003: “reasoning with Moral Conflicts”, Nous 37: 557-605. Hume, d., 1740: A Treatise of Human Nature, oxford: Clarendon Press, 1968. Igneski, V., 2006: “Perfect and Imperfect duties of Aid”, Social Theory and Practice 32: 439-466. Jamieson, d., 2005: “duties to the distant: Aid, Assistance, and Intervention in the developing World”, Journal of Ethics 9:151-170. Katsafanas, P., 2011: “deriving Ethics from Action: A nietzschean Version of Constitutivism”, Philosophy and Phenomenological Research 83: 620-660. Kelly, T., 2002: “The rationality of Belief and Some other Propositional Attitudes”, Philosophical Studies 110: 163-196. Koppelman, A., 2006: “drug Policy and the Liberal Self”, Northwestern University Law Review 100: 279-293. Korsgaard, C., 1996: “Two distinctions in Goodness”, in Creating the Kingdom of Ends, 249-274, Cambridge: Cambridge university Press. —“The right to Lie: Kant on dealing with Evil”, in Creating the Kingdom of Ends, 133-158, Cambridge: Cambridge university Press. —2009: Self-Constitution: Agency, Identity, and Integrity, oxford: oxford university Press. Langton, r., 2007: “objective and unconditioned Value”, Philosophical Review 116: 157-185. Liberman, A., and M. Schroeder, 2016: “Commitment: Worth the Weight”, in Weighing Reasons, 104-122, eds. E. Lord and B. Maguire, oxford: oxford university Press. nair, S., 2014: “Consequences of reasoning with Conflicting obligations”, Mind 123: 753-790. 30 Ryan W. Davis LEAP 4 (2016) noggle, r. 2009: “Give Till it Hurts? Beneficence, Imperfect duties, and a Moderate response to the Aid Question”, Journal of Social Philosophy 40: 1-16. Pogge, T., 2002: World Poverty and Human Rights, Cambridge: Cambridge university Press. railton, P., 2003: “Alienation, Consequentialism, and the demands of Morality”, reprinted in Facts, Values and Norms, Cambridge: Cambridge university Press. rawls, J., 1996: Political Liberalism, new York: Columbia university Press. — 1999: A Theory of Justice, revised edition, new York: Columbia university Press. ripstein, A., 2009: Force and Freedom: Kant’s Legal and Political Philosophy, Cambridge: Harvard university Press. risse, M., 2005: “How does the Global order Harm the Poor?” Philosophy & Public Affairs 33: 349-376. Schapiro, T., 2003: “Compliance, Complicity, and the nature of nonideal Conditions”, The Journal of Philosophy 100: 329-355. Shiffrin, S., 2011: “A Thinker-Based Approach to Freedom of Speech”, Constitutional Commentary 27: 283-308. Sin, W., 2010: “Trivial Sacrifices, Great demands”, Journal of Moral Philosophy 7:3- 15. Smith, M., 1994: The Moral Problem, oxford: Wiley-Blackwell. —1996: “Search for the Source”, Philosophical Quarterly 49: 384-394. —2003: “neutral and relative Value after Moore”, Ethics 113: 576-598. —2009: “Two Kinds of Consequentialism”, Philosophical Issues 19: 257-272. —2011: “deontological Moral obligations and non-Welfarist Agent-relative Values”, Ratio X XIV: 351-363. —2012: “Agents and Patients, or: What We Learn about reasons for Action by reflecting on our Choices in Process-of-Thought Cases”, Proceedings of the Aristotelian Society CXII: 309-330. —2013: “A Constitutivist Theory of reasons: Its Promise and Parts”, Law, Ethics and Philosophy 1: 9-30. —“The Magic of Constitutivism”, American Philosophical Quarterly 52: 187-200. Stohr, K., 2011: “Kantian Beneficence and the Problem of obligatory Aid”, Journal of Moral Philosophy 8: 45-67. Streumer, B., 2007: “reasons and Impossibility”, Philosophical Studies 136: 351-384. Thomson, J.J., 2008: “Turning the Trolley”, Philosophy & Public Affairs 36: 359-374. Thomson, J.J., 2008: Normativity, Chicago: open Court Press. Velleman, d., 2006: Self to Self, Cambridge: Cambridge university Press. —2008: “A Theory of Value”, Ethics 118: 410-436. —2009: How We Get Along, Cambridge: Cambridge university Press. Walden, K., 2012: “Laws of nature, Laws of Freedom, and the Social Construction of Morality”, in russ Shafer-Landau (ed.) Oxford Studies in Metaethics VII: 37-79. Williams, B., 1973: “A Critique of utilitarianism”, in Consequentialism: For and Against, 77-150, Cambridge: Cambridge university Press. —1981: “Internal and External reasons”, in Moral Luck, 101-113, Cambridge: Cambridge university Press. —1995: “Internal reasons and the obscurity of Blame”, in Making Sense of Humanity, 35-45, Cambridge: Cambridge university Press. Which Moral Requirements Does Constitutivisim Support? 31 LEAP 4 (2016) Wolf, S., 2009: “Moral obligation and Social Commands”, in Metaphysics and the Good: Themes from the Philosophy of Robert Merrihew Adams, 343-367, eds. S. newlands and L.M. Jorgensen, oxford: oxford university Press.