Acta Anal (2010) 25:279–284 DOI 10.1007/s12136-010-0098-5
The Incoherence Objection in Moral Theory Eric Wiland
Received: 4 March 2010 / Accepted: 7 June 2010 / Published online: 26 June 2010 # Springer Science+Business Media B.V. 2010
Abstract J.J.C. Smart famously complained that rule utilitarianism is incoherent, and that rule utilitarians are guilty of “rule worship”. Much has been said about whether Smart’s complaint is justified, but I will assume for the sake of argument that Smart was on to something. Instead, I have three other goals. First, I want to show that Smart’s complaint is a specific instance of a more general objection to a moral theory—what I will call the Incoherence Objection. Second, I want to illustrate how the Incoherence Objection can apply both to consequentialist and, surprisingly, some nonconsequentialist theories. Finally, I want to demonstrate at least one way nonconsequentialist theories that make use of rules, principles, and the like can dodge the Incoherence Objection. Keywords Utilitarianism . Deontology . Consequentialism . Rules . Moral theory 1. J.J.C. Smart famously complained that rule utilitarianism is incoherent and that rule utilitarians are guilty of “rule worship” (Smart 1956). He thought it was a confusion to believe both that utility is the only thing that really matters, and that a token act that in fact maximizes utility can be wrong. We feel pressured either to embrace act utilitarianism, or to abandon utilitarianism altogether. Much has been said about whether Smart’s complaint is justified, but I will assume for the sake of argument that Smart was on to something. Instead, I have three other goals. First, I want to show that Smart’s complaint is a specific instance of a more general objection to a moral theory—what I will call the Incoherence Objection. Second, I want to illustrate how the Incoherence Objection can apply both to consequentialist and, surprisingly, nonconsequentialist theories. Finally, I want to demonstrate at least one way some (but not all) nonconsequentialist theories that make use of rules, principles, and the like can dodge the Incoherence Objection. E. Wiland (*) Department of Philosophy, University of Missouri, St. Louis, St. Louis, MO 63121-4400, USA e-mail:
[email protected]
280
E. Wiland
2. How does rule utilitarianism (RU) get into trouble? RU holds that an act is right just in case it is ordered by a rule whose general observance would produce more utility than would be produced by any alternative rule.1 Rules that produce the most utility are the morally relevant rules because RU holds that utility is the only thing that is intrinsically good. Let’s call these utility-maximizing rules UM-rules. Other rules—for example, those that pass Kant’s Formula of Universal Law test—don’t necessarily have the same status in the theory that UM-rules do. Now although a token act ordered by an UM-rule might itself produce at least as much utility as every alternative action, this will not always be the case. From time to time, the act ordered by an UM-rule will produce less utility than some other act available to the agent, as least if RU is not to collapse into act utilitarianism (AU) (Lyons 1965). When this occurs, RU says that it would be wrong to act in whatever way in fact maximizes utility. But this seems absurd. Why think it wrong to produce as much utility as you can, if you think that utility is the only intrinsically good thing, and you are already in the business of valuing rules according to how much utility compliance with them produces? It seems rather arbitrary then to refuse to value actions according to how much utility they produce. RU seems incoherent. Let me specify in a more formal way what I take to be the thrust of Smart’s complaint. The Incoherence Objection says that (IO) A moral theory is incoherent if it says that (1) an act A1 is right if and only if A1 is R-related to the entity E that is best S-related to value V, and yet (2) there can be some alternative act A2 (which is not R-related to an entity E that is itself best S-related to value V) that is itself better S-related to V than A1 is. We can recapture Smart’s particular complaint against RU if we let R = “ordered by”, E = “generally observed rule”, S = “maximizes”, and V = “utility”: (IORU) RU is incoherent because it says that (1) an act A1 is right if and only if A1 is ordered by the generally observed rule that maximizes utility, and yet (2) there can be some alternative act A2 (which is not ordered by a generally observed rule that maximizes utility) that itself maximizes utility, while A1 does not. Presented this way, it is easy to see what the putative problem with RU is. RU supposedly prizes rules simply on the grounds that they are S-related to V, but refuses to treat actions the same way. Nothing can justify this discrepancy unless one is worshipping rules; that is, by ascribing to rules some value independent of their S-relation to V. We can generate versions of IO for other two-level consequentialist theories by tinkering with the values of some of the parameters appearing in IORU. For 1
Of course, there are other ways to specify rule utilitarianism. For present purposes, I don’t think too much hangs on exactly how we spell it out. I should, however, explicitly exclude those versions of rule utilitarianism according to which the value of rules does not derive merely from the value of the utility of general compliance with them; see (Hooker 2000, 99–102). And it also matters that I am considering RU to be a criterion of right action, and not merely a decision procedure.
The Incoherence Objection in Moral Theory
281
example, we get a version of IO that applies to rule-consequentialism simply by letting V = “good consequences”. And we can get a version of IO that applies to practice-consequentialism by also letting E = “the generally observed practice” rather than “the generally observed rule”. To get a version of IO that presents a problem for character consequentialism, we have to tinker with the R-relation as well. Characters don’t order or mandate actions; rather, they—in combination with a set of circumstances—generate them. If we let R = “are generated by”, E = “character”, S = “maximizes”, and V = “good consequences”, we see that character consequentialism is incoherent if there can be some alternative act A2 (not generated by the character that maximizes good consequences) that itself produces better consequences than A1 does. 3. It should now be fairly obvious how analogues of Smart’s objection to rule utilitarianism apply equally to its consequentialist cousins. Next, I will discuss how IO can apply to nonconsequentialist theories as well. So far I have considered only moral theories for which the S-relation is one of maximization. But nonconsequentialist theories relate character traits, principles, practices and the like to values in other ways. For purposes of illustration, I will focus upon a particularly simple version of virtue theory. I am not claiming that anyone in fact holds this version of virtue theory, but I believe that the nonconsequentialist version of the Incoherence Objection can be most easily grasped by considering a theory of this kind. The theory under question (VT) maintains that an act is right if and only if it would be generated by a particular character trait. But, unlike character consequentialism, this theory does not identify the relevant character traits by considering the merits and demerits of their consequences. Rather, it identifies the relevant character traits, i.e. the virtues, as those possessed by individuals who flourish. Flourishing people possess certain character traits, and the actions these character traits generate are the right ones. It is crucial to see that this version of virtue theory is not consequentialist. The virtues are not identified as those character traits that produce flourishing; rather, they are merely the character traits correctly attributable to those who flourish. Similarly, the actions that these virtues generate are not those actions that produce flourishing; rather they are merely the acts generated by the character traits correctly attributable to those who flourish. Neither virtues nor right actions are identified as such by considering their consequences. Even so, the Incoherence Objection applies here too. Right acts are supposedly those generated by the character traits correctly attributable to those who are flourishing, but are not necessarily always the acts that themselves are correctly attributable to those who are flourishing. If flourishing people ever do things out of character, then the act generated by the character trait correctly attributable to a flourishing person will not be the act correctly attributable to a flourishing person. So if we let R = “generated by”, E = “a character trait”, S = “is correctly attributable to those who”, and V = “are flourishing”, we see that (IOVT) VT is incoherent because it says that (1) an act A1 is right if and only if A1 would be generated by the character trait that is correctly attributable to
282
E. Wiland
those who are flourishing, and yet (2) there can be some alternative act A2 (which would not be generated by the character trait that is correctly attributable to those who are flourishing) that is itself correctly attributable to those who are flourishing, while A1 is not. It seems incoherent to say that an act is right just in case it would be generated by the character trait that is correctly attributable to those who flourish, if a flourishing person would in fact act differently. Here is an example that may help. Suppose Joshua is a flourishing person, and he is also a mild man. Yet there is one moment when Joshua acts out of character: seeing some people sell geegaws at his house of worship, he overturns their table of wares in an uncharacteristic act of rage. According to VT, this act is not right, since it is not generated by a character trait correctly attributable to a person who is a flourishing. Yet this act is performed by a person who is indeed flourishing. Hence, VT picks out character traits that are attributable to those who flourish, evaluating actions based upon whether they are generated by these character traits, while refusing to evaluate actions based upon whether the actions themselves are attributable to those who flourish. VT and rule utilitarianism thus have the same sort of problem. It may be objected that those who flourish never act out of character. I don’t wish to debate this at length here, but I do suspect that this objection can be sustained only if we redefine the terms “character trait” and “flourish” in a way to guarantee this result. It seems to me that those who in fact flourish really do act in ways that are out of character sometimes. But if it is insisted that flourishing people never act out of character, then the theory in question devolves to the simpler position that an act is right if and only if a flourishing person would perform it—a theory that makes no mention of character traits or virtues at all! (This is the analogue to the worry that rule utilitarianism might collapse into act utilitarianism). Let me illustrate in a different way how the Incoherence Objection might apply to a nonconsequentialist theory of the right. Consider a version of contractualism (CONT) much like Scanlon’s.2 On this view, an act is right just in case it is permitted by a set of general principles governing our interactions that no one could reasonably reject. Whether actions are right or wrong depend only upon the relation to some other entity, viz. a set of principles. And the feature that determines whether a set of principles is the proper one to be considering is whether the set is reasonably rejectable. It is obvious that CONT is not a consequentialist theory. And yet IO can apply to CONT as well. It surely seems possible for there to be a token act that, considered all by itself, no one could reasonably reject, even if that act were forbidden by a set of general principles no one could reasonably reject. For it can be plausibly argued that there will always be some reasonable actions that slip through any decent codification of appropriate standards of behavior, actions whose reasonableness comes to light only when considered not in the context of general rules, but by themselves. No matter how much we refine the set of principles regulating the way we live with one another, reasonable exceptions to these principles can emerge.
2 Whether Scanlon’s actual view is vulnerable to the Incoherence Objection would require the sort of exegesis I am no position to undertake here. Again, I discuss contractualism only to illustrate my main contention, not to rebut any particular view.
The Incoherence Objection in Moral Theory
283
This possibility does not rest upon the truth of particularism. Particularism, as it is usually understood, is in the first place a view about the metaphysics of morals. My contention instead concerns both the role of novelty in the circumstances we face, and a view about the limits of complexity that a set of general principles can embody. For a set of principles to be not reasonably rejectable, it has to be usable. And for it to be usable, it has to be simple enough for a person to understand. But it is implausible to think that any set of principles simple enough for a person to understand could properly anticipate the reasonableness of each and every token act falling under its domain. This is a point Elijah Millgram has repeatedly argued, and I find his arguments persuasive (Millgram 1997, 2004). I cannot rehearse them all here, so instead I’ll re-use one of his examples. A philosopher reports having been very bored in high school, and so frequently skipped class. The school had a rule that gets you expelled for frequently skipping class, but an exception was made in her case, one to which she owes her subsequent college education and career. The rule governing skipping class was a reasonable rule to have on the books; perhaps any other alternative rule would have been even more objectionable. Still, it was reasonable for the school’s administrators to make an exception to the rule in her unique case. I don’t expect every reader to be persuaded by this sort of example, but that’s not critical to my present goal. Let’s just say that if this sort of case can emerge, then the Incoherence Objection gets its grip on CONT. For if we let R = “permitted by”, E = “a set of principles governing our interactions”, S = “is (of predication)”, and V = “not reasonably rejectable”, we get (IOCONT) CONT is incoherent because it says that (1) an act A1 is right if and only if A1 would be permitted by a set of principles governing our interactions that is not reasonably rejectable, and yet (2) there can be some alternative act A2 (which is not permitted by a set of principles governing our interactions that is not reasonably rejectable) that is itself not reasonably rejectable. Contractualists of this sort prize a value—the value of not being reasonably rejectable—when it is borne by principles, but not when it is borne by actions. CONT seems no less incoherent than RU. Nonconsequentialist theories, then, sometimes characterize right action as action that is appropriately related to some psychological feature (e.g. character trait, attitude), rule (e.g. principle, standard), or social institution (e.g. practice, society) that itself has or is appropriately related to some value. But when this value can also be had or appropriately related to an action itself, the Incoherence Objection rears its head. That is, when actions can have the same sort of value as the item that the theory says is really the source of the value of actions, the Incoherence Objection gets its grip. It may appear that any moral theory that identifies valuable acts in terms of their relations to other valuable things is vulnerable to the charge of incoherence. 4. Next, I want to explore one way nonconsequentialists attracted to two-level theories can nevertheless dodge the Incoherence Objection. The Incoherence Objection gets its grip on a two-level theory when an act not R-related to the E that is S-related to V—when that act—is itself best S-related to V. The easiest way to pre-empt such an outcome is to select the values of S and V so that actions cannot be S-related to V in the first place.
284
E. Wiland
To illustrate, consider a version of divine command theory (DCT). According to this view, an act is right just in case that act is commanded by another being, and the feature that makes a particular being’s commands authoritative is that the being is omnipotent. That is, let R = “commanded by”, E = “a being”, S = “is (of predication)” and, V = “omnipotent”. But DCT—whatever its other merits or demerits—does not fall prey to the Incoherence Objection, since actions aren’t metaphysically the sort of thing that can be omnipotent in the first place. Only agents can be omnipotent, not actions. And so there can’t be an act that is S-related to V, despite not being R-related to the E that is S-related to V. The Incoherence Objection can’t get any traction with DCT. Likewise, consider a moral theory maintaining that an act is right just in case it would be performed by someone with integrity (INT). That is, let R = “would be performed by”, E = “a person”, S = “having”, and V = “integrity”. The rightness of an action, then, depends upon, not its own intrinsic features, but upon whether it is related in the right way to something else of value. Suppose A1 is the act that would be performed by a person with integrity. Could there be some alternative act A2 that itself possesses integrity more than A1? No, for actions are not the sort of thing that can possess (or lack) integrity. A person can be weak-willed, insincere, hypocritical, conflicted, or two-minded, and thus lack integrity. Or a person can be strong-willed, sincere, consistent, and single-minded, and thus have integrity. Persons are the sort of thing that can, metaphysically, have or lack integrity. But actions are the wrong sort of thing metaphysically to have or lack integrity. And so there can’t be an act that is S-related to V, despite not being R-related to the E that is S-related to V. The Incoherence Objection can’t get any traction with INT either. So although the Incoherence Objection threatens moral theories according to which actions can have the sort of value had by the entity to which right actions are appropriately related, it poses no problem for moral theories according to which actions cannot bear this sort of value. Philosophers attracted to theories that understand right action as action that is properly related to something else of value will want to ensure that their theory is not incoherent in the way Smart first identified. Can consequentialists escape the Incoherence Objection in the same way? It seems not. For consequentialism of every stripe prizes something (rules, characters, etc.) in virtue of its superior consequences. And acts also have good and bad consequences. So I don’t see how consequentialists (other than act-consequentialists) can successfully dodge the Incoherence Objection. But at least they have some company. Acknowledgements I thank Thad Metz, Dale Miller, and the audience at a meeting of The International Society for Utilitarian Studies for their help and suggestions.
References Hooker, B. (2000). Ideal Code, Real World. Oxford: Oxford University Press. Lyons, D. (1965). The Forms and Limits of Utilitarianism. Oxford: Oxford University Press. Millgram, E. (1997). Practical Induction. Cambridge, MA: Harvard University Press. Millgram, E. (2004). Does the Categorical Imperative Give Rise to a Contradiction of the Will? The Philosophical Review, 112, 525–560. Smart, J. J. C. (1956). Extreme and Restricted Utilitarianism. Philosophical Quarterly, 6, 344–354.