KATERINA DELIGIORGI | The Scope of Autonomy: Kant and the Morality of Freedom | Oxford University Press 2012


By Adrian Piper

Deligiorgi’s erudite and wide-ranging study The Scope of Autonomy. Kant and the Morality of Freedom situates the core concepts of Kantian ethics—autonomy, freedom, reason, universalizability—squarely within the context of contemporary debates in ethics, philosophy of mind and language, metaethics and moral psychology. The discussion brings the exegetical dialogues among Kant scholars into pertinent lines of communication between Kant and contemporary philosophers who may not even have realized that they were in conversation, or whence came the arguments and theories that preoccupy them. It expands and deepens the shared argumentative resources available to Kant scholars, Kantians, and philosophers in other specialties alike. In this essay, I scrutinize a circumscribed section of the territory revealed and mapped by Deligiorgi’s subtle exploration, and suggest some means by which some of its resources might be thrown into even sharper relief.

The core concept of autonomy Deligiorgi aims to defend is inspired by Kant’s “idea of the will of every rational being as a universally legislating will” (GMS, AA 4:431.20–2; emphasis in text).[1] In this formulation of Kant’s categorical “imperative” (actually he formulates the moral requirement in the indicative mood in this passage), Deligiorgi finds the foundation for a concept of autonomy which ventures beyond the traditional notion of rational self-legislation, to encompass one’s moral relation to others. As she says at the outset,

[t]he moral content of autonomy is given by the notion of the law (nomos) and the demand that one think of oneself in relation to others under a shared law. (p. 4)

[…] [A]utonomy does not describe how an agent stands with respect to herself, her desires or plans of action, but rather how she stands with respect to other agents. Importantly this relation is conceived in law-like terms. (p. 5)

The law-like term that defines the relation of autonomy among agents is Kant’s requirement of universalizability: that we are to act on only those principles of action on which other such autonomous agents could also act under relevantly similar circumstances. The reference to other agents that is implicit in the requirement of universalizability is the textual anchor for Deligiorgi’s interpretation, and grounds the theory of autonomy she develops.

Thus the vision that guides Deligiorgi’s reading contains both intrasubjective and also intersubjective aspects. On the one hand, I determine my own behaviour according to principles of action that I myself formulate to guide them. On the other hand, these principles must be such that other agents not only could act on them, but also, on Deligiorgi’s theory, would be willing to act on them under relevantly similar circumstances. Moreover, the principles I formulate exclude from the scope of her analysis of autonomy “how an agent stands with respect to herself, her desires or plans of action”, in favour of “how she stands with respect to other agents”. Thus Deligiorgi’s reading replaces Kant’s central focus in the Groundwork of the Metaphysics of Morals on the relationship between the principles I formulate and the other aspects of self they are supposed to govern, with a focus on the relationship between the principles I formulate and the others who co-govern with me. This is a different conception of autonomy than that which is to be found in Kant’s Groundwork itself. But although Deligiorgi makes frequent use of that and other texts by Kant, hers is a work of substantive theory construction, not textual exegesis. This strong conception is implicit in her co-legislative guiding vision:

The specific demand of universalizability requires us to imagine stepping in the shoes of a legislator who faces her peers in a constitutional court and who is not in a position to tailor the membership of that court to suit her purposes. (p. 24; cf. also pp. 138, 172)

Universalizability within Deligiorgi’s co-legislative conception does not only require me to envision others doing under similar circumstances what I intend to do. From the fact that others could enact that intention when similarly situated, it does not follow that they would. Deligiorgi’s interpretation of the universalizability requirement also demands that I know enough about the diversity of those others—of their interests, values, motives, plans—to anticipate whether or not they would endorse what I intend to do. This requirement, of actual familiarity and connection with others in the human community sufficient to assess their likely opinion of my proposed action, is one of the central features that differentiate Deligiorgi’s view from Kant’s own in the Groundwork.

So Deligiorgi’s co-legislative conception of the universalizability requirement redistributes more widely the recalcitrance that Kant explicitly builds into our relation to our own inclinations. For her, by contrast, this intrasubjective recalcitrance is not compounded but rather replaced by the recalcitrance with which other agents might react to our inclinations because they have different ones. She proposes the concept of co-legislation “to capture the idea that we think of ourselves as sitting among other legislators who have a legitimate claim to passing judgment on the law we propose to pass” (p. 138). The validity of a principle of action is thus determined not by whether it satisfies rationality criteria that I require of myself; but rather by whether the satisfaction of those criteria itself passes muster among the members of a community of my fellow co-legislators.

In German as in English, it is possible to distinguish between two interpretations of the universality requirement at GMS AA 4:431.20–2, namely the collective versus the distributive interpretations.[2] A principle that satisfies the universality condition collectively applies conjointly to all of the objects—or in this case, subjects—to which it refers. A principle that satisfies it distributively applies to each one of those subjects disjunctively. These two interpretations are logically independent. For example, it might be that everyone collectively could end political corruption, even though any one individual cannot. Or it might happen that any one individual could score a winning goal, even though everyone collectively could not. The difference between these two interpretations has nothing to do with time-indexing: just as the collective interpretation does not require everyone to act simultaneously or in concert, similarly the distributive interpretation does not prohibit this. Everyone collectively might end political corruption if no one ever tolerated any manifestation of it in each of his or her individual lives. And all individuals who were opportunely placed to receive the ball might simultaneously score a winning goal in each of the individual soccer games in which they played.

Deligiorgi understands the concept of universalizability expressed at GMS AA 4:431.20–2 as implying the distributive interpretation of universality (p. 12). This is significant because the concept of universality occurs twice in this passage: first in the concept of “the will of every rational being” (emphasis added) as fixing the conditions under which the principle is formulated; and second in the description of that will as “universally legislating” (emphasis added) as fixing the scope of subjects to which the principle is applied. This yields four possible interpretations of the passage, depending on whether the concept of universality is interpreted collectively or distributively in each case.

Deligiorgi’s across-the-board commitment to the distributive interpretation implies, first, a commitment to the ability of any rational being to independently formulate lawlike principles—rather than to the ability of every rational being collectively to formulate lawlike principles. That is, I (or anyone else) may formulate a universalizable maxim of action regardless of what anyone else does. But from this it does not follow that all rational beings collectively may do so: collective formulation might be blocked by non-cooperation, failures of communication, lack of interpersonal coordination, conflicting or competing interests and the like, even though each individual rational being is perfectly capable of formulating the principle on her or his own. Second, similarly, it implies a commitment to the application of such principles to any rational being distributively—rather than to their application to every rational being collectively. That is, the universalizable maxim I formulate must apply independently to anyone else in the same circumstances, regardless of whether or not everyone else is or could be in those circumstances.

I agree with Deligiorgi that the distributive interpretation of the concept of universality is the correct one for understanding Kant’s various formulations of the categorical “imperative”, not least of all because it briskly disposes of some frivolous purported counterexamples to Kant’s universalizability requirement; for example, that everyone is to be first through the door, or that everyone is to turn up at Times Square tomorrow. However, her across-the-board distributive interpretation, i.e. with respect to both formulation and application, sits uneasily with her co-legislative conception of that requirement. For the co-legislative conception does imply that each rational agent either actually or imaginatively consults the judgement of all the others when formulating a principle of action. That is, it implies that the formulation of a principle of action is actually or imaginatively a collective enterprise among co-legislators, not a distributive one:

Each and every rational being is to occupy the position of universal legislator. To make this rather abstract idea more vivid we might say that the agent is asked to submit her prospective law to a constitutional tribunal whose members all have legislative rights. (p. 17)

This makes the enterprise of formulating a principle much more demanding, for it requires that potential obstacles to collective formulation be not merely removed—as they were by Rawls’s stipulation of the veil of ignorance in the original position, but rather resolved. Rawls’s stipulation had the effect of reducing the collective deliberation of the parties in the original position to the distributive deliberation of any one of its members chosen at random: since all reasoned identically in the absence of knowledge of their own goals and interests, it was irrelevant whose deliberation was at issue. That move is not open to Deligiorgi, for she begins with the premise of individual deliberation but then requires that it satisfy the stronger and more demanding criterion of collective validation. By resisting the individualistic reduction, she ratchets up the conditions that must be met in order to achieve collective formulation. The stakes involved can be illustrated by the complex and laborious processes undertaken by actual constitutional tribunals, for example, by the level of prolonged consultation and negotiation that must have gone into formulating the Preamble of the American Constitution, or Article I of the German Grundgesetz. It is in this sense that the co-legislative conception of autonomy requires of each one of us a certain level of deference to the “other legislators who have a legitimate claim to passing judgment on the law we propose to pass” (p. 138).

Deligiorgi might concede that the formulation of a universalizable maxim is an actually or imaginatively collective, co-legislative enterprise, but nevertheless insist that its application is distributive, as is required in order to dispose of the above counterexamples. This would bring her collective, co-legislative procedure of formulating a universalizable moral maxim into alignment with the intersubjective procedure of formulating a deductive scientific hypothesis on the basis of experimental evidence—an hypothesis which, if successfully formulated, would apply distributively to any phenomenon picked out by its terms and concepts. The next step would be to extend the analogy: just as intersubjective cooperation, consultation and confirmation are central to such a procedure in the laboratory, similarly they are central to the analogous procedure in the courtroom. The third step would be to extend the analogy again, this time to Deligiorgi’s co-legislative conception of moral deliberation as well, for which these three desiderata of collective decision-making in the service of gaining objective knowledge would be just as central.

Thus this analogy would also serve her endorsement of Kant’s argument, at GMS AA 4:431.12–15, that “[t]he ground of all practical legislation lies objectively in the rule and the form of universality, which (according to the first principle) makes the rule capable of being a law (indeed a law of nature) etc.”[3] On Deligiorgi’s reading of this passage, objectivity is achieved when the formulation of a principle satisfies the formal requirement of universality:

[T]he requirement for objectivity is satisfied by the formal requirement of the rule of universality and the basic item that concerns us in this search for objectivity is […] ‘legislation’. (p. 13)

Politically, legislation is a very different enterprise from execution or enforcement, with different procedures, offices of government and personnel. Whereas the former concerns the formulation of positive law, the latter concerns the manner of its implementation, i.e. its application to a citizenry. The requirement of objectivity is satisfied, then, when the formal rule of universality is successfully formulated, rather than when it is applied. Hence the achievement of objectivity on Deligiorgi’s account requires the intersubjective enterprise of co-legislation, i.e. of the collective formulation of the rule of universality. By contrast, the achievement of justice would require the distributive application of that rule to anyone selected by its terms. Henceforth I shall therefore take Deligiorgi to in fact endorse a collective interpretation of the universality requirement with regard to the formulation of a universalizable principle; and a distributive interpretation of that requirement with regard to its application.

Deligiorgi characterizes a principle as objective if it is “an object of knowledge and so capable of being true […] prior to and independent of our choices” (p. 45n.22). For example, I may independently try to formulate the law-like free rider principle that anyone may break the rules when doing so promotes their individual self-interest. This is precisely the kind of principle that her co-legislative conception of autonomy is designed to exclude. It violates her criterion of objectivity on several counts.

First, it cannot be possibly true prior to and independently of our choices, because its content itself refers to and is meant to serve our choices. It presupposes not only that we have choices, but also what kinds of choices we have, namely self-interested ones.

Second, therefore, it cannot be an object of knowledge, but rather justified rational belief at best. It is not a principle that we can know with certainty to be true, because our appraisal of it is not independent of the very same self-interested considerations to which it itself refers. It is too susceptible to the sorts of self-aggrandizing bias that intersubjective replication and confirmation in science is supposed to filter out.

Third, each member of the co-legislative community could anticipate quite a lot of intersubjective recalcitrance in the formulation of such a principle from all of the others, should she or he propose to formulate it. For each might validly worry that each of the others might, like oneself, try to tailor its application to suit her or his own circumstances.

The value of the above analogy between scientific and courtroom procedure for Deligiorgi’s co-legislative conception of autonomy would be that it provides a collective and intersubjective procedure explicitly designed to achieve objective knowledge in Deligiorgi’s sense; a procedure that can be imported and reframed in terms of the specifics—intersubjective cooperation, consultation and confirmation—of Deligiorgi’s actual or imaginative co-legislative process. It is this process that is supposed to exclude the free rider principle.

But that the co-legislative conception of autonomy succeeds in excluding the free-rider principle by underpinning that conception with a collective interpretation of universality in its formulation does not show that it is that intersubjective process of co-legislation that is doing the heavy lifting in getting rid of it. For that process is consistent with the adoption of a principle or hypothesis that we can recognize as false or misguided on independent grounds. The trial and execution of Socrates by his fellow Athenian citizens, or Galileo’s advocacy of the Copernican theory of the revolution of the planets against the combined resistance of his scientific colleagues and the Roman Catholic Church, or the U.S. Supreme Court’s majority decision in 2000 to deny the U.S. Presidency to Al Gore even though he won the popular vote, would be three well-known examples.

But we do not need to search far afield in order to draw the relevant conclusion, namely that intersubjective agreement is neither a necessary nor a sufficient condition of truth; nor, therefore, of objective knowledge. These examples demonstrate that it does not matter for establishing the moral rightness of a principle of action whether or not one’s peers endorse, or could or would collectively endorse that principle; for any such group or community is in the end merely a mob waiting to happen. Hence Deligiorgi needs to establish a basis for the legitimacy of the claim that other legislators have to “pass judgment on the law we propose to pass” (p. 138) that lies beyond the mere procedure of actual or imaginative cooperation, consultation and confirmation with them. She needs to call on a stronger conception of moral objectivity than intersubjective agreement can provide.

Such a basis is implied by the logical structure of a coherent principle of action. It is this structure that the free rider principle violates; and its violation of this structure is what disqualifies it as a candidate for objective truth. For the free rider principle can be neither formulated nor applied universally at all, without self-contradiction—whether collectively or distributively. That is the point of Kant’s argument at GMS AA 4:423–4, and it is a good one. If anyone suitably positioned could break the rules in order to promote their self-interest, the very concept of a rule would be incoherent, hence so would the possibility of breaking it. That is, the free rider principle fails Onora O’Neill’s contradiction in conception test. If it is the inherent self-contradiction of the free rider principle that excludes it as a candidate for objective truth, then the fulcrum of Deligiorgi’s account of objectivity as legislative universality is to be sought not in the co-legislative procedure to which she is committed. Rather it is to be sought in the internal logical consistency of a coherent universalizable principle. Because this criterion of objectivity can be satisfied by any rational agent’s formulation of an action principle, regardless of whether or not any other rational agent does, could or would endorse it, it implies a distributive rather than a collective interpretation of the formulation of such a principle. This conclusion returns us to Deligiorgi’s original, across-the-board commitment to a distributive interpretation of universality with respect to both formulation and application. The price is that the intuitively appealing connection between intersubjective agreement in the co-legislative procedure and moral objectivity must be relinquished.

I think Deligiorgi is right to cash out the notion of moral objectivity in terms of the universalizability of a principle of action. But the objective force of a universal principle is not secured by the range of subjects who formulate it, nor by the range of subjects to whom it applies. Universality of scope is a logical requirement on valid principles. A categorical judgement, whether indicative or imperative, requires of its ascription of predicate to subject that the property in question apply to each and every variable designated as within its scope, on pain of self-contradiction. For Kant, the attributive copula “is” that is used to make that ascription signifies its objectivity (KrV, B142.01–13). Kant’s conception of universality as a requirement of action principles is based on the brand of objectivity expressed in the irrefutable logical force of that requirement—that the concepts constitutive of the principle cannot even function successfully as concepts unless they apply universally to the objects they pick out.

The language in which Kant chooses to couch his claim at GMS AA 4:431.12—15, that “[t]he ground of all practical legislation lies objectively in […] the form of universality” deliberately refers the reader back to his conception of the logical forms of judgement constitutive of the Table of Judgements in the Critique of Pure Reason, as structuring the categories of the understanding—which in turn supply the universally necessary preconditions of any object or event we can experience, including our own deliberation and action. It is thus the satisfaction of these purely logical requirements, without which unified consciousness itself would be impossible, that secures the objective force of an action principle for Kant. For Deligiorgi as well, these are the requirements satisfaction of which can, in fact, be “an object of knowledge” and “true […] prior to and independent of our choices” (p. 45n.22). The rules of logic structure the course and content of our world whether we like it or not. Deligiorgi agrees that this is the way objective knowledge is supposed to work. But of course there are many principles of action that satisfy the requirement of universalizability thus understood which are very far from achieving the status of rightness.

Like Kant, Deligiorgi resists this conclusion. She, too, aspires to a deep and sure connection between the objectivity of an action principle and its moral rightness:

In autonomy-based ethics, the objectivity requirement is met simply by the moral law: the moral law is an interpretation of the concept of ‘right’ in terms of the concept of universalizability. Universalizability is what sustains the claim to objectivity. (p. 41)

I have argued elsewhere that Kant aimed too high; for it is not logically possible to deduce a substantive moral theory from the formal requirements of reason, as central to his metaphysics and epistemology as these requirements may be.[4] Deligiorgi does not make this mistake, for she does not attempt to deduce the concept of right from them. Rather, she asserts the objectivity of the moral law; and asserts its requirement of universalizability as an interpretation of the concept of right. So moral rightness, as expressed in and given form by the moral law, has objective validity; and this moral rightness consists in the universalizability of a principle of action. From these two claims it follows that a universalizable action principle both has objective validity and is morally right. It is this last conjunct of the conclusion that I mean to call into question.

If we understand the idea of universalizability as the universalizable formulation of principle, and that, in turn, in the collective sense required by Deligiorgi’s co-legislative procedure of formulation, we are forced to conclude to the objective validity and moral rightness of principles generated by that procedure. However, the case of Galileo would be a counterexample to the objectivity claim. The case of Socrates would be a counterexample to the moral rightness claim. And the case of Al Gore would be a counterexample to the reliability of the co-legislative procedure itself. If, on the other hand, we understand the idea of universalizability in the weaker, logical sense I have been advocating on Kant’s behalf, then although the objective validity of a universalizable principle may follow from its conformity to the rules of logic, this is no guarantee of its moral rightness. I would recommend this weaker sense to Deligiorgi to underpin her explicit commitment to distributive universality in the formulation of a universalizable principle, even if it means having to rethink her co-legislative account of autonomy. For even if we were to take that account at face value, its implication that our co-legislative formulations of universalizable principle are therefore morally right is problematic. That would be to assert that autonomous action is therefore morally right action. Yet surely any viable account of autonomy must include the freedom to do wrong.


[1] “[…] die Idee des Willens jedes vernünftigen Wesens als eines allgemein gesetzgebenden Willens” (GMS, AA 4:431.20–22).

[2] J. H. Sobel, ‘”Everyone”, Consequences, and Generalization Arguments’, Inquiry 10 (1967): 373–404.

[3] “Es liegt nämlich der Grund aller praktischen Gesetzgebung objektiv in der Regel und der Form der Allgemeinheit, die ein Gesetz (allenfalls Naturgesetz) zu sein fähig macht […]” (GMS, AA 4:431.12–15).

[4] A. M. S. Piper, ‘Kant’s Self-Legislation Procedure Reconsidered’, Kant Studies Online 14 (2012): 203–77.


© 2014, Adrian M. S. Piper.

Adrian Piper obtained her Ph.D. in Philosophy from Harvard University in 1981 under the supervision of John Rawls. She taught philosophy at Georgetown, Harvard, Michigan, Stanford, and UCSD and is now Professor Emeritus and Director of the APRA Foundation Berlin. Adrian’s principal philosophical publications are in metaethics, Kant, and the history of ethics. She has published in numerous journals, such as Journal of Philosophy, Ethics, Philosophical Studies, Southern Journal of Philosophy, Political Theory and American Philosophical Quarterly. Her two-volume study in Kantian metaethics, Rationality and the Structure of the Self, Volume I: The Humean Conception and Rationality and the Structure of the Self, Volume II: A Kantian Conception is available as an open access e-book.