The Journal of Value Inquiry

, Volume 44, Issue 2, pp 199–207

Agency and Responsibility


    • Department of PhilosophyOakland University

DOI: 10.1007/s10790-010-9211-7

Cite this article as:
McDonald, F.J. J Value Inquiry (2010) 44: 199. doi:10.1007/s10790-010-9211-7

According to Christine Korsgaard, Kantian hypothetical and categorical imperative principles are constitutive principles of agency. By acting in a way that is guided by such imperatives, an individual makes herself into an agent. On her theory, there is an inextricable link between the nature of agency and the practical issue of why we should be rational and moral. The benefits of such an account would be great. In Korsgaard’s view, an account that bases morality on the nature of agency is the basis for a refutation of any kind of moral skepticism, providing an indubitable and objective foundation for morality. This may seem too good to be true, and it is. Korsgaard would only be able to succeed at offering a foundation for morality at the cost of an account of agency that is too restrictive. Korsgaard does not present a coherent account of irrational or immoral agency, and the inability to offer an account of such agency implies an inability to offer a proper account of responsibility. Korsgaard’s view shares a fundamental flaw with Immanuel Kant’s account of morality in the Groundwork of the Metaphysics of Morals. Korsgaard cannot give a full, adequate account of individual responsibility. In light of the failure of Kant’s and Korsgaard’s accounts, Kantians need to provide a better, more comprehensive characterization of agency. Presenting a proper account of agency requires a rejection of a central tenet of traditional Kantian meta-ethics, but rejection of the central tenet does not require a full rejection of Kantianism.

1 Korsgaard on Agency

In order to be clear on how Korsgaard’s account is supposed to account for morality, we need to be clear on what she means by agency. Unfortunately, she does not offer a single, precise definition of this key notion. Korsgaard has defined an agent as an active person: “A person is both active and passive, both an agent and a subject of experiences.”1 This definition could be called the active person definition of agency. Korsgaard goes onto say, in the same essay just cited, that “we may regard ourselves as agents, as the thinkers of our thoughts and the originators of our actions.”2 That characterization goes beyond the active person definition. If being an agent requires being the originators of our own actions, then it is not sufficient to just act: we must act in a way that is self-originating. This could be termed the self-originating active person definition.

To make full sense of this, we need to be clear on the notion of self-origination. Korsgaard’s mention of self-origination suggests a kind of libertarian account of freedom of the will, the sort of account offered by Roderick Chisholm, on which the agent or the self is an Aristotelian unmoved mover.3 Perhaps Korsgaard’s notion of agency, of self-originating action, might best be understood in terms of agent-causation: we are agents only insofar as we are active selves, specifically active selves that are not determined to act in the way we act by prior events. However, this metaphysical, theoretical notion of agent-causation is not what Korsgaard has in mind. Korsgaard invokes a Kantian distinction between theoretical and practical points of view, to suggest that what is relevant to the nature of agency is not the theoretical, metaphysical fact of whether or not our choices are uncaused causes, but instead the purported fact that, in order to make a choice at all, a person must regard herself as an unmoved mover. Korsgaard writes: “We must view ourselves [as agents, as free, as responsible] when we occupy the standpoint of practical reason—that is, when we are deciding what to do. This follows from the fact that we must regard ourselves as the causes—the first causes—of the things that we will.”4 To properly characterize the self-originating active person conception of agency offered by Korsgaard, we may say that to be an agent is to be an individual who must see herself as an uncaused cause from the practical point of view.

This characterization of agency, drawing as it does on the distinction between practical and theoretical standpoints, is far from clear. How we should understand the notion of causation, of first causation, in a non-theoretical fashion is never clearly spelled out by Korsgaard. Perhaps her characterization suggests that what is required is a positive account of self-origination, of how we supposedly determine action without prior determination from a practical point of view. In The Sources of Normativity, Korsgaard reiterates the self-originating active person conception of agency: “Minimally, then, I am not the mere location of a causally effective desire but rather am the agent who acts on the desires.”5 Korsgaard claims that this is achieved through consistency: “It is because of this that if I endorse acting a certain way now, I must at the same time endorse acting the same way on every relevantly similar occasion…. For if all my decisions were particular and anomalous, there would be no identifiable difference between my acting and an assortment of first-order impulses being causally effective in or through my body.”6 In the same work, she claims that consistency is required for an act to be an act. She says: “This claim to generality, to universality, is essential to an act’s being an act of the will.”7 Hence we have a third conception of agency at work in Korsgaard’s discussion, the consistency conception of agency.

The consistency conception of agency draws on Korsgaard’s reading of the categorical imperative, a reading that gives emphasis to the formula of the universal law. She writes: “The categorical imperative, as represented by the formula of the Universal Law, tells us only to act on a maxim which we could will to be a law. And this, according to Kant, is the law of a free will.”8 This is a fairly accurate representation of Kant’s formulation of the categorical imperative, “act only in accord with that maxim through which you can at the same time will that it become a universal law.9 The way in which Korsgaard goes onto spell out the categorical imperative as she interprets it is a serious departure from Kant. Korsgaard claims that the only “constraint on our choice” presented by the categorical imperative is that our choice “has the form of the law. And nothing determines what the law must be. All that it has to be is a law.10 This cannot be what Kant has in mind.

For Kant, for a maxim to pass the categorical imperative test, it must be capable of being a universal law, and it must be a universal law someone could will to be a universal law. As Kant notes, in his discussion of indifference to a person in need, it is possible to make it a universal law to ignore the needs of others, “if such a way of thinking were to become a universal law the human race could admittedly well subsist…. But although it is possible that a universal law of nature could very well subsist in accord with such a maxim, it is still impossible to will that such a principle hold everywhere as a law of nature.”11 It is this test of what a person is able to will to be a universal law that in Kant’s view would make it wrong to ignore someone who is in need. We would not want to live in a world where others would ignore us in similar circumstances. Hence, according to Kant, it is impossible to genuinely will ignoring others in need into universal law. Setting aside the issue in Kant exegesis, the account Korsgaard offers of the categorical imperative helps to clarify the consistency conception of agency. It also helps to explain Korsgaard’s view that the categorical imperative is a constitutive principle of agency. In order to act, a person must act in a way that is guided by universal laws, which are consistent principles of actions. Insofar as a person does not act in such a consistent fashion, she is not an agent.

There are two more points worth noting about Korsgaard’s account of agency. Korsgaard alternates between making claims about the constitution of agency, and the constitution of the self. It seems clear, then, that Korsgaard’s account of agency is intended as an account of the nature not only of agency but also of the self. Also, Korsgaard thinks that being an agent, or being a self, is in some crucial way necessary for human beings. She claims that it is impossible not to be an agent or a self. This is important for her justification of morality and her response to moral skepticism. If morality and rationality are based on the nature of being an agent or self, and a person cannot help but be an agent, then morality and rationality have an indubitable foundation in human nature.

Korsgaard claims, as we have seen, that “if all my decisions were particular and anomalous, there would be no identifiable difference between my acting and an assortment of first-order impulses being causally effective in or through my body.12 To claim this is to assume that there cannot be a way to be both self-controlled and acting, but not acting in a way that would require consistency over time. The assumption is mistaken. It leaves out an alternative of self-controlled action for particular reasons in particular circumstances. Let us suppose that Mary is in a situation where, by cheating on her tax return, she would be able to pay for an expensive medical treatment for her child. She wants to do so on one occasion only. Fearing the possibility of being caught for being a tax cheat, she decides to cheat now, but not to cheat on her taxes whenever she is in difficult circumstances. On Korsgaard’s account, we would have to consider Mary a sort of wanton, lacking self-control, ruled by the particular impulse on which she acts. This seems a misreading of the facts. We may ask why it cannot be the case that Mary is fully in control of herself when she makes her decision. Korsgaard might claim that this case presents an impossible circumstance. Mary has made the decision to cheat on her taxes just once, but does not will the universal principle that a person should always cheat on her taxes when in need. Insofar as Mary is not willing a universal principle, she is not, on Korsgaard’s account, acting at all. Mary wants to cheat on her taxes just once, while not wanting to cheat again the next time she is in a similar predicament. As there is no consistent universal principle behind her action, Korsgaard would suggest that Mary’s action is simply impossible. Insofar as she does not will a universal principle, Mary is not, by Korsgaard’s lights, an agent. It is far from obvious that Korsgaard’s denial of Mary’s agency would be correct. Only on a narrow, technical definition of the notion of agency would it be the case that such an action is impossible. If it were the case that one had to be able to will a consistent universal principle each time we act, then a vast number of human actions would be ruled out as possibilities by this definition. Observation of human behavior reveals that such actions are not just possible but actual and common. One troubling aspect of such a narrow definition of agency is that it would rule out the possibility of irrational or immoral acts, insofar as such agency is not based on universal principles of rationality and morality. On Korsgaard’s account, the only way to be an agent is to act on rational principles, including the categorical imperative. If this is so, it is difficult to see how immorality or irrationality is even possible.

2 Whither Irrationality and Immorality?

Plainly there are irrational and immoral people in the world. Whatever our characterization of immorality and irrationality might be, we can regard the person who throughout her life fails to do what is in her own self-interest to be irrational, and we can consider the parent who abuses her children to be an immoral person. There is no need for any further argument for this claim, since it is an easily observed fact that stands in no need of a defense. Given that immorality and irrationality exist, a constraint on any adequate account of moral agency is for the account to allow for the possibility of immorality and irrationality. It must be the case that we can allow not only for the existence of successful attempts at rationality and morality but for the possibility of failure as well. Korsgaard’s account, on the face of it, does not meet this failure constraint. Principles of rationality and morality, such as the categorical imperative, are, on her view, constitutive principles of agency or the self. In order for a human being to be an agent, she must act in accord with the categorical imperative principle. However, there are clearly individuals in the world who fail to act in accord with the categorical imperative.

This circumstance presents a dilemma for Korsgaard. Either individuals who fail to act in accord with the categorical imperative are immoral agents, or they are not agents at all. In order to take takes the first horn of the dilemma, Korsgaard would have to reject her own account. Insofar as an agent is an agent, she must constitute herself by acting in a way that is guided by the categorical imperative. However, an immoral individual, on Korsgaard’s account, would be an individual who is not guided by the categorical imperative. On her account, immoral agents would have to act on the categorical imperative, insofar as they are agents, and not act on the categorical imperative, insofar as they are immoral. This is clearly incoherent. The second horn of the dilemma presents the difficulty that if immoral individuals are not agents at all, then there cannot be immorality. Korsgaard indicates that she may take this horn of the dilemma in her most recent book, Self-Constitution, writing that “the laws of practical reason govern our actions because if we don’t follow them, we aren’t acting.”13 Insofar as an individual is not governed by the laws of practical reasoning, she is not an agent. It would only make sense to attribute the property of being immoral to an agent. However, if being immoral, failing to act on the categorical imperative, is tantamount to not being an agent, then there cannot be immorality, because there cannot be immoral agents.

When it comes to Mary’s tax evasion, we might consider Mary to be performing an immoral act. In her circumstances, she wants to make an exception for herself to the tax laws. This could be seen as a violation of the categorical imperative, because Mary might be willing to make an exception for herself without willing it to be a universal law that others be allowed to cheat on their taxes in difficult circumstances. Based on Korsgaard’s definition of agency, we might deny that Mary is an agent at all, given that there is no universal principle behind her action. It would follow from this that it would not be possible to attribute the immorality of the action of cheating on her taxes to Mary, insofar as she is not an agent. Hence Korsgaard’s narrow definition of agency precludes the attribution of immorality.

In her recent essay “Self-Constitution of the Ethics of Plato and Kant,” Korsgaard raises this concern about her account. As Korsgaard notes, her neo-Kantian account shares this problem with Kant’s own account in the Groundwork of the Metaphysics of Morals: “For a well-known problem in the Groundwork is that Kant appears to say that only autonomous action, that is action governed by the categorical imperative, is really free action.”14 It is not entirely clear why Korsgaard claims that Kant appears to say this. The account of free action offered by Kant in the Groundwork of the Metaphysics of Morals clearly has this implication. For Kant, causation requires a law, and there can only be two kinds of laws, laws of nature and the moral law. The moral law, Kant claims, would be the law that governs the free, noumenal will. This account has just the same problematic implication of not meeting the failure constraint as Korsgaard’s account. If individuals are only responsible for free actions, and the only way for an action to be free is if it under the moral law, then there cannot be genuine immorality at all. On Kant’s account, the only possible actions a person can perform freely, the only actions for which a person can be responsible, are moral actions.

Korsgaard clearly sees that this is an implication of her account and Kant’s account as she writes, “So it looks at first as if nothing exactly counts as a bad action.15 Surprisingly, Korsgaard claims that this problem is not a concern at all but “our main reason for embracing” this account of the self.16 Korsgaard addresses the concern by weakening her account of what is required for agency. An immoral or irrational person is a person who is making a failed attempt to act on principles of rationality such as the categorical imperative. Korsgaard claims that “even the most venal and shoddy person must try to perform a good action, for the simple reason that there is no other way to try to perform an action.”17 This revision contradicts her claim that the categorical imperative is a constitutive principle of agency. On her revised view, it is trying to act on the categorical imperative that is constitutive of agency, not acting on the categorical imperative. In Self-Constitution, Korsgaard states this view by claiming that the categorical imperative is among the “constitutive principles of action, principles to which we necessarily are trying to conform insofar as we are acting at all.”18 On this account, we cannot act without trying to act in conformity with moral principles. What occurs in the case of a bad action, what Korsgaard calls a defective action, is that the attempt to act in conformity with moral principles fails. “The kind of practical deliberation that results in bad action is not a different activity from the kind of practical deliberation that results in good action. It is the same action, badly done.”19 Morally wrong action is a failed attempt to act on the same principles that lead to morally right action.

Even this revised account does not meet the failure constraint. It is reasonable to think that people should be given just as much credit for trying to do what is right as for succeeding in doing what is right. The celebrated baseball player Roberto Clemente died in a airplane crash while trying to deliver aid packages to the victims of an earthquake in Nicaragua. He was clearly trying to do something good, and is perhaps given even more credit for the tragic circumstances of his death. Given that Clemente tried, but failed, to do something good, he deserves moral praise. Korsgaard, as a Kantian, is committed to the view that a person ought to be judged not based on the consequences of actions but on the intention behind the action. Hence, Korsgaard should think Clemente praiseworthy. As a result, Korsgaard is hoist by her own Kantian petard. If the intent to do what is right is the standard for judging an individual, then we ought to consider all persons who try to conform to the categorical imperative as good persons. The class of good persons would not only include moral saints like Clemente, but every person, including a shoddy or venal person, insofar as every agent is making an attempt to conform to the categorical imperative.

This is not what we would ordinarily say about a venal or shoddy person. However, if, as Korsgaard suggests, venal and shoddy people are always trying to do what is right, then it is not clear that we should consider them venal or shoddy as much as unfortunate. If the aim of all people is to act in accord with the categorical imperative, and the categorical imperative is the fundamental principle of morality, then everyone is aiming to be moral. As a result, on an intention-based morality, everyone is morally good. This revised version of the account, like the original account, violates the failure constraint by not allowing for the possibility of immorality.

Furthermore, the suggestion that all agents must be trying to act in a way that is guided by the categorical imperative, is an ad hoc claim that is given no support by Korsgaard. We may ask why we should believe that venal and shoddy people are really trying to act in accord with the categorical imperative. When a greedy businessman embezzles a pension fund for his own personal gain, why should we think that the businessman has tried, but failed, to be moral? This naïve, early Socratic account of moral psychology, on which all persons must be trying to be rational and good, does not do justice to the facts.

3 Explaining Agency

In order for an account of moral agency to be adequate, it must allow for the possibility of moral and immoral agency, and rational and irrational agency. Failure to allow for these possibilities is failure to offer an account of agency that can properly characterize an irrational or immoral agent as responsible for irrationality or immorality. Furthermore, as we have seen in the case of Mary, who wants to cheat on her taxes once and only once, the consistency conception of agency offered by Korsgaard is too narrow for another reason. It would have us treat anyone who acts for a particular reason on a particular occasion as a non-agent, lacking self-control. The quality of self-control, however, is not plausibly identical to the quality of acting in a fully consistent fashion. Persons can be self-controlled without ideal consistency.

A better account of agency would allow that inconsistent, irrational, and immoral agency is still agency. Better yet, it would be an account that fits a generally naturalistic picture of the world, and does not require the postulation of any kind of ontological oddity such as the Kantian noumenal will. Progress has been made on matters of this kind in the compatibilist literature on freedom of the will. It is a lacuna in Korsgaard’s writing that she does not relate her accounts of agency to the current literature on freedom of the will, writing: “Agency is almost as mysterious as freedom of the will” in her most recent essay collection.20 The great advantage of compatibilist accounts is that they demystify freedom and agency. Moreover, compatibilists offer an account of freedom that does not violate the failure constraint.

In “Freedom of the Will and the Concept of a Person,” Harry Frankfurt offers such an account of agency.21 Frankfurt’s account can allow for agency that is consistent or inconsistent, rational or irrational, and moral or immoral. Frankfurt characterizes freedom in terms of a particular kind of desire, which he considers essential to being a person. To act in a way that is free is to act on a desire that a person desires to desire. Having free will is a matter of having second-order desires. For instance, if Frank wants to write a book, and Frank wants to have this want to write a book, then he may write the book of his own free will. Alternatively, if Mitch is compulsive hand-washer, and while Mitch wants to wash his hands, he does not want to have this want, then when Mitch washes his hands he is not acting of his own free will.

Unlike Korsgaard, Frankfurt can allow for desires that are particular but self-controlled. If Mary both wants to cheat on her taxes just once, and wants to want to cheat on her taxes, then she can be self-controlled, freely acting, even if her action is not based on a desire to act in this fashion on every similar occasion. Frankfurt’s account also allows for blameworthy instances of irrationality and immorality. If a parent wants her child to be able to go to college, but she does not save any of her money to provide her child with the education, we can say that the parent is irrational. If she desires to spend her money, not save it, and desires to have this desire, then we can rightly say that she has made a free, irrational choice. If a business executive who embezzles a pension fund is not suffering from an uncontrolled compulsion, but instead both desires to embezzle the fund for his own well-being and desires to have this desire, Frankfurt’s account, unlike Korsgaard’s, allows for the possibility of claiming that the businessman makes a choice that is both free and immoral.

If Kantians accept an account of agency of the sort presented by Frankfurt, based in second-order desires, it seems that they must do so at a significant cost. It is difficult to see how such an account of agency could serve as a foundation for morality. If it possible for an agent to be rational as well as irrational and moral as well as immoral, then agency cannot serve as any sort of guarantee of morality or rationality. The hope of Kant and Korsgaard that anyone who is an agent or a self must be moral is empty.

There are good, more general meta-ethical reasons to think that the grounding of morality in agency could never succeed. If “ought” cannot be derived from “is,” then any sort of metaphysical account of the nature of the world, or the nature of persons, does not suffice to answer the questions of what one ought to do and why. Korsgaard’s attempt to base morality in agency is based on a similar mistake that is made by the sort of moral realist accounts she rightly opposes. Much as moral realists attempt to base morality on the metaphysical structure of reality but fail, seeking a ground for what we ought to do in the nature of what exists, Korsgaard attempts to ground morality in the metaphysical structure of the self. The accounts differ, but the mistake is the same.

The search for a metaphysical foundation of morality is a nonstarter. This is no cause for Kantian moralists to be concerned. The search for a justificatory foundation of morality is similar to the epistemic search for a justificatory foundation of knowledge. While there have been many philosophers who have realized that foundationalism about knowledge is a failure, there seems to be a failure to recognize that the search for a metaphysical foundation for morality is similarly flawed as an attempt to find a justification based on something that is not a justification. The Kantian categorical imperative principle, that we should not act on maxims that would lead us to treat persons as mere means but always at the same time treat persons as ends in themselves, does not need a metaphysical defense. The only way to justify such a principle is through moral thinking by considering whether the principle fits well with other considered, moral views.22


Christine M. Korsgaard, “Personal Identity and the Unity of Agency: A Kantian Response to Parfit” in Creating the Kingdom of Ends (Cambridge, England: Cambridge University Press, 1996), p. 363.


Ibid., p. 377.


See Roderick M. Chisholm, “Human Freedom and the Self,” The Lindley Lecture (1964).


“Personal Identity and the Unity of Agency: A Kantian Response to Parfit,” p. 378.


Christine M. Korsgaard, The Sources of Normativity (Cambridge, England: Cambridge University Press, 1996), p. 228.




Ibid., p. 232.


Ibid., p. 98.


Immanuel Kant, Groundwork of The Metaphysics of Morals in Practical Philosophy, ed. and trans. Mary Gregor (Cambridge, England: Cambridge University Press, 1996), p. 73.


Korsgaard, The Sources of Normativity, p. 98.


Kant, Groundwork of The Metaphysics of Morals, p. 75.


Korsgaard, The Sources of Normativity, p. 228.


Christine M. Korsgaard, Self-Constitution (Oxford: Oxford University Press, 2009), p. 32.


Christine M. Korsgaard, “Self-Constitution in the Ethics of Plato and Kant” in The Constitution of Agency (Oxford: Oxford University Press, 2008), p. 110.


Ibid., p. 111.


Ibid., p. 112.


Ibid., p. 113.


Christine M. Korsgaard, Self-Constitution, p. xii.


Ibid., p. 132.


Christine M. Korsgaard, “Introduction” in The Constitution of Agency (Oxford: Oxford University Press, 2008), p. 10.


See Harry Frankfurt, “Freedom of the Will and the Concept of a Person,” the Journal of Philosophy, vol. 68, no. 1 (1971).


I would like to thank the University Research Committee of Oakland University for a Faculty Research Fellowship that supported my work. I am grateful to Radu Neculau, Christopher Tindale, Phillip Rose, Jeff Noonan, and an audience at the University of Windsor for their comments and to Rosemary Twomey, David Pereplyotchik, David Enoch, and Keota Fields for discussions of Christine Korsgaard’s philosophy. I am especially grateful to two anonymous referees and Thomas Magnell, the Editor-in-Chief of the Journal of Value Inquiry, for their careful critiques and help on this article.


Copyright information

© Springer Science+Business Media B.V. 2010