Deontology, Rationality, and Agent-Centered Restrictions

Brandon Hogan, University of Pittsburgh

I. Introduction

Deontological ethical theories appear to be inconsistent on their own terms because they include agent-centered restrictions. Samuel Scheffler defines an agent-centered restriction as “a restriction which it is at least sometimes impermissible to violate in circumstances where a violation would serve to minimize total overall violations of the very same restriction, and would have no other morally relevant consequences.” ¹ It is claimed that such restrictions create an inconsistency in deontological theories because they forbid the performance of some action that would bring about a state of affairs that the deontologist should prefer, namely, a state of affairs in which more people abide by the restriction rather than less. For example, most deontological theories contain an agent- centered restriction on killing innocent persons, even in circumstances where one violation of the restriction would prevent the killing of five innocent people. As Scheffler rightly argues, one who wishes to defend deontological theories that contain such restrictions should care to address this apparent inconsistency. ²

In what follows, I evaluate the nature of the claim that agent-centered restrictions render deontology inconsistent and address three seemingly promising responses available to the deontologist. ³ The first response is inspired by Kant’s essay “On a Supposed Right to Lie Because of Philanthropic Concerns.” The latter two responses appeal to the importance of personal moral integrity and the moral worth of actions, respectively. I conclude that neither response will allow the deontologist to refute the charge of inconsistency. I take the failure of the latter two responses to reveal that the charge of inconsistency, despite appearances, is both a serious and deep one. This is so because the failure reveals that the deontologist’s endorsement of agent-centered restrictions appears inconsistent even when one takes into account the importance that most deontologists place on personal moral integrity and the moral worth of actions.

II. Are Agent-Centered Restrictions Really Irrational?

Scheffler argues that agent-centered restrictions appear problematic because of the seeming irrationality of forbidding violations of such restrictions even in circumstances in which a violation would reduce the total number of violations. ⁴ It seems that one who thinks that the killing of innocent persons is morally objectionable should also (on pain of irrationality) hold that it is at least sometimes permissible to kill in order to reduce the total number of killings. It is not clear, however, how one should understand the contemplated relationship between the violation of an agent-centered restriction and the decrease in the total number of violations of that restriction. I contend that we can either understand the relationship in strict causal terms or in terms of motivation or influence. Under either interpretation, agent-centered restrictions do not seem to require any genuinely irrational behavior. ⁵

I will use the following example to aid this discussion. Imagine that B tells A (a deontologist) that unless he (A) kills C, B will kill C and five other innocent persons. In this situation, A can either kill C and hopefully prevent B from killing six innocent persons or A can adhere to the agent-centered restriction against killing, which will most likely result in B violating that same restriction six times.

Let’s first think about what follows if we take the relationship between A’s violation of the agent-centered restriction and the decrease in total violations (that is, B’s not killing C and five other innocent persons) as one of cause and effect. Without delving too deeply into the metaphysics of causation, it seems that we can safely say that if action φ was necessary for effect E to occur, then φ was at least one of the causes of E. Consequently, had action φ not been taken, effect E would not have occurred. And it is irrational for one to desire E, yet forbid the performance of all actions that are necessary for E to obtain. ⁶

Those who endorse agent-centered restrictions obviously presuppose that there are agents. Here I will not explicate the concept of agency, but will simply assume the controversial position that agents are at least capable of freely making decisions and acting on reasons. On this assumption, agents are capable of causing things to happen and are not themselves caused. Thus, if we take B to be an agent, A’s actions cannot cause B to do anything. A can only take actions that will influence B’s choice of action. Thus, as a general matter, the relationship between the violation of an agent-centered restriction and the decrease in total violations cannot be one of cause and effect.

We can see that this is the case by noticing that if A’s action does cause B to refrain from killing, B cannot be considered an agent and thus his inaction cannot count as a reduction in the total number of violations of an agent-centered restriction. Likewise, if B is not an agent, his “killing” C and five other innocent persons cannot count as an increase in the total number of violations of agent-centered restrictions. If the relationship between the violation of a restriction and the decrease in total violations is taken to be causal, we run into conceptual difficulties in even attempting to formulate the problem Scheffler raises for the deontologist. The deontologist’s endorsement of agent-centered restrictions cannot be irrational on this picture because there could be no circumstance in which one’s violation of a restriction could decrease the total number of violations of the very same restriction.

As mentioned above, we can also understand the relationship between the violation of an agent-centered restriction and the decrease in total violations as one of motivation or influence. If this is the case, A’s actions can only influence B’s decision not to kill. B himself is free to decide whether or not to kill independent of what A decides to do.

On this picture, agent-centered restrictions could be seen as irrational in that they require adherence even in circumstances in which it is most likely that such adherence would serve to increase the total number of violations. However, if we view the irrationality in this way, it appears to be less of a problem for deontology. Scheffler’s charge was taken to be serious because it at first seemed to commit the deontologist to a certain type of irrationality. How, it was asked, could one rationally prefer a certain state of affairs and forbid the performance of an action that is necessary for that state of affairs to obtain? But if the connection between the violation of the restriction and the desirable state of affairs is only one of probability, then it does not appear irrational to both forbid the violation and desire that the desired state of affairs obtain. In short, it seems rational to uphold an absolute rule against killing and desire that others not take one’s adherence to this rule in any particular circumstance as a reason to kill (even in situations in which it is likely that others will do so).

Thus, regardless of how we understand the contemplated relationship between the violation of an agent-centered restriction and the state of affairs in which the total number of violations is decreased, the restrictions do not appear to be genuinely irrational. Either the relationship must be understood as one that partially destroys agency or as one under which the adherence to agent- centered restrictions is consistent with the assumed desires of the deontologist. There is, however, a weaker version of Scheffler’s critique that should also be troubling to the deontologist.

III. The Real Trouble with Agent-Centered Restrictions

I said that we can only understand the relationship between one’s violation of an agent- centered restriction and the actions of another as one of motivation or influence. In which case, one’s violation can only influence and cannot cause the actions of another. While it is not genuinely irrational to forbid the violation of a restriction in circumstances in which such a violation would make it highly probable that others will refrain from violating the restriction, such a stance does appear at least rationally troubling at first glance. It does not seem that a deontologist can easily justify his strict adherence to agent-centered restrictions in such circumstances. How, it may be asked, can one defend agent-centered restrictions in cases in which the most probable result of adhering to the restriction would be the occurrence of further violations of the very same restriction? I next turn to a response to this question inspired by Kant.

IV. Kant on the Rationality of Agent-Centered Restrictions

In the essay “On a Supposed Right to Lie Because of Philanthropic Concerns,” Kant offers an interesting defense of the agent-centered restriction against lying. While in this essay Kant is only concerned with the agent-centered restriction against lying, his position can be understood as a defense of all agent-centered restrictions. Kant argues that even in circumstances in which one supposes that violating an agent-centered restriction will bring about a favorable result, one should resist because (1) one can never know what will result from one’s violation of an agent-centered restriction, and (2) persons are morally responsible for the immediate negative results brought about by their violations of agent-centered restrictions. ⁷ For Kant persons are responsible in this way even if the negative effects of their actions were completely unforeseeable. ⁸ Thus, Kant presents both an epistemic and normative defense of agent-centered restrictions. I will assess each in turn, though I admit that the two defenses cannot be sharply distinguished.

Kant illustrates the epistemic component of his argument with an interesting example. He imagines a situation in which one lies to a murderer in order to throw him off the trail of his intended victim. Kant tells us that while it may be reasonable to believe that this lie will serve its purpose, the lie could lead the murderer to take actions that ultimately lead to his finding and murdering his intended victim. ⁹ Kant imagines as intuition in which the victim moves away from his hiding place in the liar’s house only to encounter the murderer who is in the back of the house because he was told that his intended victim was not hiding in the house. ⁹ For Kant, since this or any other distant or unforeseen possibility cannot be ruled out, one cannot be certain about the consequences of one’s actions.

Kant’s analysis seems correct. One can never be absolutely certain that one’s actions will have their intended consequences. But Kant’s point here seems to obscure the issue at stake. If we understand the problem created by agent-centered restrictions as I have explained it above, one who takes it to be rational to sometimes violate an agent-centered restriction does not need to claim to know for certain that his violation will bring about a particular result. The claim is that a violation will most likely bring about a certain result, and that it would be irrational to act against this high probability. Thus, Kant’s warning should be of little consequence as directed to one who contemplates violating an agent-centered restriction on the grounds of rationality. What Kant needs to show is that as a general matter, violations of agent-centered restrictions are more likely to bring about bad results than adherences to those restrictions. Since Kant does not do this, the epistemic portion of his defense of agent-centered restrictions appears misguided.

In addition to the above argument from the ignorance of result, Kant argues that one should never violate an agent-centered restriction because one is morally responsible for any negative results brought about by one’s violation of an agent-centered restriction. Note that if we are to take Kant’s argument as an argument for the rationality of agent-centered restrictions, we cannot take the relationship he contemplates between the lie and the murder to be one of cause and effect. Otherwise, as argued above, there is no issue of irrationality because in this circumstance Kant’s “murderer” can no longer be regarded as an agent.

Consequently, we must understand Kant’s argument as follows. One should not violate an agent-centered restriction even when one thinks that doing so will reduce the total number of such violations because agents are morally responsible for all outcomes made more likely by their violations of duty. ¹⁰ Interestingly, Kant does not seem to present any arguments in favor of adopting of this picture of responsibility. In fact, he seems to think it obvious that responsibility attaches to the liar in his example. But this is far from obvious for several reasons. Specifically, Kant’s formulation suffers because he ignores the concepts of intent and foresight, which appear to be central to our concept of responsibility as applied to the consequences of one’s actions.

In lieu of a more nuanced definition, I will simply understand intent as follows: A intends a certain result if A acts with the conscious desire that his actions produce that result. The concept of intent allows us to distinguish between what a person does and what simply happens as a result of his actions. The concept of intent also allows us to determine the degree to which one is responsible for one’s actions. For example, say A lights a candle which, unbeknownst to A, falls over igniting a curtain which burns down B’s house. In this circumstance it would be inappropriate to say that A “burned down the house” without further explanation. Here A does not burn down the house, but the house burns down as a result of A’s lighting of the candle. On the other hand, if A lit the candle with the desire that it ignite the curtain and burn down the house (or in order to burn down the house) it would be appropriate to say that A burned down the house. In the second scenario, burning down the house is something that A does. The difference between these two scenarios is A’s intent.

In the first scenario A would only be held partially responsible, if responsible at all, for the house’s burning down. And in the second scenario, A would be held fully responsible for burning down the house. This is so because we look to intent to measure A’s mental culpability and, thus, blame worthiness. ¹¹ Kant makes no mention of intent and there foreign or a concept that is central to our notion of causal responsibility.

Kant’s understanding of causal responsibility is problematic not only because he ignores the applicability of the concept of intent, but also because he ignores the relevance of the concept of foresight. In this context we can understand foresight as the knowledge of the probable consequences of one’s action. To use the above example, in the first scenario A does not foresee the consequences of his actions because he is not aware of the likelihood that his lit candle will cause a house fire. However, such consequences were foreseeable. A could have known that his lit candle was likely to cause a fire if not monitored. We tend to hold persons more responsible for consequences that were foreseeable, even if not foreseen, because in certain circumstances persons should have foreseen the likely consequences of their actions and acted accordingly. ¹²

In Kant’s example, it is clear that the liar does not intend to bring about the victim’s death. And it is not clear that the liar could have or should have foreseen the consequences of his lie. As such,it is not clear that the liar maybe “just accused as the cause of [thevictim’s] death.” ⁹ Infact, such an accusation seems morally problematic, especially if we take the murderer to be a rational agent as well.

To summarize, Kant argues that one should never violate an agent-centered restriction, even in situations in which a violation would most likely result in a reduction of the total number of violations because (1) one can never know what will result from one’s violation of an agent-centered restriction, and (2) persons are morally responsible for the immediate negative results brought about by their violations of agent-centered restrictions. I argued that (1) does not resolve the inconsistency created by agent-centered restrictions because the opponent of agent-centered restrictions need not suppose that one can know for certain the consequence of one’s actions in order to generate the inconsistency. The opponent of agent-centered restrictions only needs to claim that one can know the probable consequences of one’s actions. I also argued that (2) is morally problematic because it is based on a view of causal responsibility that neglects the morally salient concepts of intent and foresight.

V. Moral Integrity

Though Kant’s solution to the problem of agent-centered restrictions is unsatisfactory, the deontologist may attempt to resolve the problem by arguing that the restrictions are needed in order to uphold the virtue of personal moral integrity. We can understand moral integrity as the degree to which a person consistently endorses and adheres to a particular moral code. One who is consistently honest can be said to have more integrity than one who is honest only when he thinks that being honest will allow him to derive some personal benefit. Also, integrity comes in degrees. One can have more or less integrity depending on the degree to which one complies with the dictates of a particular moral code. If the deontologist takes integrity to be a moral virtue valued more than any decrease in the total number of violations of agent-centered restrictions and obeying agent-centered restrictions allows agents to retain this virtue, then no inconsistency is created by the presence of agent-centered restrictions in deontological ethical theories.

The deontologist would likely value his own integrity more than any decrease in the total number of violations of agent-centered restrictions because, for the deontologist, the agent-centered restrictions themselves have value partially because they are self-legislated. For the deontologist, the agent-centered restrictions constitute the rules that he has chosen to adhere to. On this picture, since moral rules are self-legislated, one should value one’s moral integrity because the degree to which one is morally integrated is the degree to which one lives as one thinks one should live. While the deontologist cares that others adhere to the moral rules he takes to be correct, he seems to have more reason to care that he adheres to these rules because he alone must live with himself. ⁹

Additionally, adhering to agent-centered restrictions will allow the deontological agent to maintain a certain type of moral integrity. ¹³ Agent-centered restrictions admit of no exceptions, and, as such, one cannot help but be morally consistent over time if one adheres to these restrictions. By contrast, an ethical theory which contains no agent-centered restrictions may permit or even require that one commit acts that would compromise one’s moral integrity. For example, a utilitarian would require that one steal in circumstances in which doing so would increase overall happiness, even if one did not wish to steal and had never stolen anything in one’s life. Thus, agent-centered restrictions allow the deontologist to maintain a certain type of integrity.

This appeal to integrity, however, will not allow the deontologist to resolve the inconsistency created by agent-centered restrictions. This is so because there may be cases in which an agent’s violation of an agent-centered restriction would make it most likely that that same agent would refrain from such violations in the future. In which case, the agent can increase his own integrity by violating an agent-centered restriction.

An example may help to make this point clear. Imagine that A believes that stealing is morally wrong, but knows that by stealing a decent dress shirt now, in order to look presentable for his upcoming job interview, he will make it much more likely that he will not have to steal in the future in order to survive. In this case A’s life will likely be more morally integrated, by his own lights, if he steals now. Thus, even if we take moral integrity to be an important virtue, it still appears inconsistent to endorse a moral theory which contains agent-centered restrictions.

VI. Moral Worth

The deontologist may also attempt to respond to the charge of inconsistency by arguing that the state of affairs that would result from a violation of an agent-centered restriction would lack moral worth and would, therefore, be undesirable from a moral point of view. If this is the case, the deontologist’s restrictions are not irrational because they do not prevent the obtaining of any state of affairs that the deontologist values.

From the moral perspective of the deontologist, only actions performed for the right reasons are morally praiseworthy. ¹⁴ From this perspective, one receives no moral credit for doing the right thing out of selfishness or for some other non-moral reason. Additionally, it seems that this calculus should apply to cases in which one refrains from doing the wrong thing. That is, from the perspective of deontology, one should receive no moral credit for refraining from committing a morally objectionable act for the wrong reasons (for example, out of fear of punishment). I assume that for the deontologist, a state of affairs in which more moral credit is awarded is preferable to one in which less moral credit is awarded. Thus, for the deontologist, there is no moral reason to prefer a state of affairs in which one commits a morally blameworthy act to one in which one refrains from committing the same act, but for the wrong reasons. ¹⁵ I realize that this last claim seems odd, but hopefully it will seem less so after some explanation.

In order to understand why a deontologist may hold such a position, one must understand why the deontologist denies moral credit to those who do the right thing for the wrong reasons. Credit is denied in such circumstances partly because the deontologist places great value on person’s motives. The deontologist recognizes that in cases in which persons do the right thing for the wrong reasons, had those bad reasons not been present, the person in question would not have done the right thing. For example, suppose A donates money to Oxfam just so he can brag to his friends about his generous donation. The deontologist would not give A moral credit for this donation because had A not had friends to impress, he would not have made the donation. In fact, it does not seem that the deontologist should regard A as being in any better a position morally than someone who did not donate but would have had he had friends he wanted to impress. In both circumstances the person fails to act on moral reasons. The actions that the deontologist deems praiseworthy are those that would be performed regardless of circumstance.

It seems also that the deontologist would want to extend this reasoning about moral credit to inaction as well. That is, the deontologists should not give moral credit to those who refrain from committing morally objectionable acts for the wrong reasons. As in the case of actions, it seems that the deontologist should not take the person who refrains from doing wrong for the wrong reasons to be in any better a moral position than the person who actually does wrong. Had circumstances been different, those persons who refrain from doing wrong for the wrong reasons would likely have committed the morally objectionable acts that they currently avoid. Thus, from a strictly deontological moral point of view, there seems to be no reason to prefer a state of affairs in which one commits a morally blameworthy act to one in which one refrains from committing the same act, but for the wrong reasons. ¹⁶

An example used above will help to show how this conclusion bears on the charge of irrationality leveled against agent-centered restrictions. Recall that B tells A (a deontologist) that unless he (A) kills C, B will kill C and five other innocent persons. Let’s expand the example and assume that B takes himself to be playing a sick game in which he attempts to coerce others to kill. Assume that if A were to kill C, B would keep his promise and not kill the five other persons. However, he would do this only because he cares about the success of his game. B knows that if he does not make good on his promise to A, he will be unable to coerce the others into killing at some time in the future.

It was argued that it would be irrational for A to forbid himself from killing C because, even if he takes killing to be wrong, he should desire the state of affairs in which C and the other innocent persons are not killed. But if what I have said about the moral perspective of deontology is correct, A has no moral reason to prefer the state of affairs in which B refrains from killing only because he cares about the success of his game. In this situation, B is just as worse off morally as he would have been had he actually killed C and the other innocent persons.

In fact, A does seem to have a moral reason to adhere to the agent-centered restriction and not kill C. In this circumstance, A’s inaction would have moral worth because he adheres to the restriction for the right reasons. On the other hand, if A were to kill C, preventing B from killing, the entire state of affairs would be less preferable than the alternative from the deontological perspective. This is so because in this latter circumstance A would have violated a restriction and B would be in no better a position morally. As such, the deontologist does not behave irrationally in adhering to the agent-centered restriction in this circumstance.

However, in order for this argument to establish that the deontologist’s adherence to an agent-centered restriction is never irrational we must assume that in all possible cases in which a violation of an agent-centered restriction would most likely minimize the total number of violations, the resulting state of affairs would lack moral worth. This, however, is not the case. There are circumstances in which violating a restriction would most likely bring about a morally valuable state of affairs in which the total number of violations is reduced.

To see that there are situations in which violating a restriction would most likely bring about a morally valuable state of affairs in which the number of overall violations is reduced, consider the following example. Imagine that A, in order to prompt B to recognize that stealing is wrong, steals $100 from B. Assuming that B is the type of person who is best instructed by means of example, the most likely result of A’s theft will be a reduction in the number of overall thefts (at least those committed by B), and B would refrain from stealing because he recognizes stealing to be wrong. There seem to be many possible situations of this type, that is, in which the violation of an agent- centered restriction would influence an agent in a way that would make him more likely to act on moral reasons in the future.

We can also apply this line of reasoning to the situation in which the violation of an agent- centered restriction would make it most likely that the agent himself would refrain from such violations in the future. Recall the person who contemplates stealing a dress shirt to wear to his job interview. Assume that A takes it to be his duty to avoid stealing, but knows that if he does not get a job soon, he will most likely place himself in a situation in which he will not be able to resist the temptation to steal. In this scenario, A’s stealing would most likely bring about a morally valuable state of affairs—one in which the total number of thefts is reduced and he refrains from stealing for the right reason. Thus, the deontologist’s theory seems inconsistent on its own terms even when we take into account the deontologist’s views on moral worth.

VII. Conclusion

We have seen that Kant’s likely defense of agent-centered restrictions is untenable as well as those defenses which appeal to moral integrity and moral worth. In the introductory section, I said that the failure of these latter two responses reveals that the charge of inconsistency against the deontologist is both a serious and deep one. For those sympathetic to deontological theories, the charge of inconsistency can at times appear crude and misplaced. The crudest versions of this charge seem to simply ignore the morally salient concepts of moral integrity and moral worth. As such, the charge of inconsistency can at times seem naïve and even lacking in moral depth. But, as I hope to have shown, even if those who object to deontology on the basis of rationality do take into account the importance that most deontologists place on the concepts of moral integrity and moral worth, their charge of inconsistency maintains its force. One who wishes to defend deontology must show either that despite appearances, agent-centered restrictions are not irrational or that deontological theories are preferable even though they may require that persons act irrationally.

Bibliography

Darwall, Stephen. “Agent-Centered Restrictions from the Inside Out.” Philosophical Studies 50 (1986): 219-319.

Kant, Immanuel. Grounding for the Metaphysics of Morals: With on a Supposed Right to Lie Because of Philanthropic Concerns. Trans. James W. Ellington. Indianapolis, IN: Hackett, 1993.

Korsgaard, Christine M. “The Right to Lie: Kant on Dealing with Evil.” Philosophy and Public Affairs 15.4 (1986): 325-49.

LaFave, Wayne. Principles of Criminal Law. St. Paul, MN: West Group, 2003.
Scheffler, Samuel. “Agent-Centered Restrictions, Rationality and the Virtues.” Mind94 (1985):

409-19.

Smart, J. J. C. and Bernard Williams. Utilitarianism: For and Against. Cambridge: Cambridge UP, 1973.

Samuel Scheffler, “Agent-Centered Restrictions, Rationality and the Virtues,” Mind 94 (1985): 243 (my emphasis).
Scheffler 243.
Throughout the paper, I refer to “the deontologist.” The deontologist I have in mind is largely Kantian in orientation. However, I realize that not all deontologists are Kantians.
Scheffler 244.
Roughly, I take one to behave in a genuinely irrational manner if one believes that one’s performance of action φ is necessary to bring about state of affairs E and desires that E obtain, yet does not perform action φ (given, of course, that performing φ is practically possible).
Here I just assume that in the contemplated circumstance A’s action is necessary for, and thus a cause of, B’s not killing C and five other innocent persons.
Immanuel Kant, Grounding for the Metaphysics of Morals: With on a Supposed Right to Lie Because of Philanthropic Concerns, trans. James W. Ellington (Indianapolis, IN: Hackett, 1993) 64-65.
Christine M. Korsgaard, “The Right to Lie: Kant on Dealing with Evil,” Philosophy and Public Affairs 15.4 (1986): 326.
Kant 64-65.
I admit it is not clear how we should understand the use of probability in this argument. On one view, it seems that by lying to the murderer, the person in Kant’s example makes it less likely that the intended victim will be killed. But, if we restrict our scope to the facts of the circumstance at hand, we can regard the lie as making it more likely that the intended victim will be killed. We can just hold as constant that the victim will move away from his hiding place in the house. In which case, the lie makes it more likely that the murder will find his victim.
I realize that some philosophers have argued that the concept of intent should play no role in our judgments of responsibility. According to utilitarian J. J. C. Smart, we should only blame persons in situations in which blaming would increase overall happiness. See J. J. C. Smart and Bernard Williams, Utilitarianism: For and Against (Cambridge: Cambridge UP, 1973) 49-50. While this principle may be correct, the practice Smart encourages would not be our practice of moral blame. You cannot “blame” someone who you do not take to be responsible or blameworthy. True, one could yell and point fingers at anyone one wishes, but these actions would not constitute “blame” unless the one thought the receiving party was guilty.
The concepts of foresight and intent are closely connected. Persons are not usually held responsible for the unforeseeable consequences of their actions partly because such consequences could not have been intended. Additionally, we tend to assume that people intended the highly foreseeable consequences of their actions. In fact, in American criminal law, intent is often derived from forseeability. That is, persons will be taken to have intended the most foreseeable consequences of their action. See Wayne LaFave, Principles of Criminal Law (St. Paul, MN: West Group, 2003) 164-65.
I say that agent-centered restrictions allow the deontologist to maintain a certain type of integrity because the utilitarian does not clearly lack moral integrity. The utilitarian consistently does what will increase total happiness.
Kant 397-99. Kant believed that an agent’s action has moral worth only when that action is (1) in accordance with duty and (2) motivated by the agent’s desire to act in accordance with duty. Of those actions done out of habit or inclination, Kant writes: “…however dutiful or amiable it [the action] may be [it] has nevertheless not true moral worth.”
I realize that it may not be the case that all deontologists hold this position, but I do think it is a plausible position for the deontologist to hold given his understanding of moral worth.
However, I think the deontologist would admit that there are practical reasons to prefer one state of affairs to the other. We can simply understand a practical reason as one having to do with the consequences of one’s actions and not one grounded strictly in some conception of duty.