The nature of autonomy


The nature of autonomy

Gerald Dworkin*

Department of Social Science and Humanities, University of California, Davis, CA, USA

Gerald Dworkin is a professor of moral, political and legal philosophy. He is currently Distinguished Professor of Philosophy at the University of California, Davis. Dworkin’s main areas of research include the nature and justification of autonomy, paternalism in the criminal law and the issue of which acts may legitimately be criminalised by the state.


In both theoretical and applied contexts, the concept of autonomy has assumed increasing importance in recent normative philosophical discussion. Given various problems to be clarified or resolved, the author characterises the concept by first setting out conditions of adequacy. The author then links the notion of autonomy to the identification and critical reflection of an agent upon his or her first-order motivations. It is only when a person identifies with the influences that motivate him or her, assimilates them to himself or herself, that he or she is autonomous. In addition, this process of identification must itself meet certain procedural constraints.

Keywords: individual autonomy; first and second order motivation; moral philosophy

Citation: NordSTEP 2015, 1: 28479 -

Copyright: © 2015 Gerald Dworkin. This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 International License, allowing third parties to copy and redistribute the material in any medium or format and to remix, transform, and build upon the material for any purpose, even commercially, provided the original work is properly cited and states its license.

Published: 3 July 2015

*Correspondence to: Gerald Dworkin, Department of Social Science and Humanities, University of California, Davis, CA, USA, Email:


The concept of autonomy has assumed increasing importance in contemporary moral and political philosophy. Philosophers such as John Rawls, Thomas Scanlon, Robert P. Wolff and Ronald Dworkin have employed the concept to define and illuminate issues such as the characterisation of principles of justice, the limits of free speech and the nature of the liberal state.

In the most recent formulation of the foundations of his theory of justice, Rawls makes clear – what was implicit in his book – that a certain ideal of the person is the cornerstone of his moral edifice. A central feature of that idea is the notion of autonomy.

[T]he main idea of Kantian constructivism … is to establish a connection between the first principles of justice and the conception of moral persons as free and equal …. [T]he requisite connection is provided by a procedure of construction in which rationally autonomous agents subject to reasonable constraints agree to public principles of justice. (Rawls, 1980, p. 554)

Scanlon’s defense of a Millian principle of free speech relies also on a view of what powers autonomous persons would grant to the state.

I will defend the Millian principle by showing it to be a consequence of the view that the powers of a state are limited to those that citizens could recognize while still regarding themselves as equal, autonomous, rational agents. (Scanlon 1972, p. 215)

Ronald Dworkin, in his article on Liberalism, does not use the word ‘autonomy’, but in discussing the idea of treating people as equals, he is arguing for equal respect for the autonomy of citizens.

According to Dworkin, the liberal theory of equality supposes that political decisions must be, so far as is possible, independent of any particular conception of the good life, or of what gives value to life. Since the citizens of a society differ in their conceptions, the government does not treat them as equals if it prefers one conception to another, either because the officials believe that one is intrinsically superior or because it is held by the more numerous or more powerful group (Dworkin, 1978, p. 127).

Wolff’s essay, In Defense of Anarchism, is devoted to the task of demonstrating that a citizen cannot retain his or her autonomy and at the same time be under any obligation to obey the commands of the state simply because they are the commands of the state.

The autonomous … man may do what another tells him, but not because he has been told to do it … by accepting as final the commands of the others, he forfeits his autonomy … a promise to abide by the will of the majority creates an obligation, but it does so precisely by giving up one’s autonomy. (Wolff, 1970, p. 41)

Bruce Ackerman, in his Social Justice in the Liberal State, speaks of

respect for the autonomy of persons as one of the four main highways to the liberal state. […] It is, in short, not necessary for autonomy to be the only good thing; it suffices for it to be the best thing that there is. (Ackerman, 1980, pp. 368–369)

It is clear that either as interpretations of the idea of liberty and equality, or as additions to them, the notion of autonomy plays a central role in current normative philosophical work. It is also apparent that, unlike the concepts of liberty and equality, it has not received careful and comprehensive philosophical examination.

Proceeding simultaneously, and as far as I can tell, relatively independently, the idea of autonomy has emerged as a central notion in the area of applied moral philosophy, particularly in the biomedical context. All discussions of the nature of informed consent and its rationale refer to patient (or subject) autonomy. Conflicts between autonomy and paternalism occur in cases involving civil commitment, lying to patients, refusals of life-saving treatment, suicide intervention and patient care.

Whether or not this is the same concept that appears in the more theoretical discussions remains to be seen, but we have some reason to believe that philosophical scrutiny will be of more than just theoretical interest.

One more warning by way of introduction. It would be unwise to assume that different authors are all referring to the same thing when they use the term ‘autonomy’. By way of illustration, consider the following brief catalogue of uses of the term in moral and political philosophy.

[T]he law in thus implementing its basic commitment to man’s autonomy, his freedom to and his freedom from, acknowledge(s) how complex man is. (Goldstein, 1978, p. 252)
To regard himself as autonomous in the sense I have in mind, a person must see himself as sovereign in deciding what to believe and in weighing competing reasons for action. (Scanlon, 1972, p. 215)
As Kant argued, moral autonomy is a combination of freedom and responsibility; it is a submission to laws that one has made for oneself. The autonomous man, insofar as he is autonomous, is not subject to the will of another. (Wolff, 1970, p. 14)
(Children) finally pass to the level of autonomy when they appreciate that rules are alterable, that they can be criticized and should be accepted or rejected on a basis of reciprocity and fairness. The emergence of rational reflection about rules … central to the Kantian conception of autonomy, is the main feature of the final level of moral development. (Peters, 1972, p. 130)
I am autonomous if I rule me, and no one else rules. (Feinberg, 1980, p. 161)
Human beings are commonly spoken of as autonomous creatures. We have suggested that their autonomy consists in their ability to choose whether to think in a certain way insofar as thinking is acting; in their freedom from obligation within certain spheres of life; and in their moral individuality. (Downie & Telfer, 1971, p. 301)
A person is “autonomous” to the degree that what he thinks and does cannot be explained without reference to his own activity of mind. (Dearden, 1972, p. 453)
[A]cting autonomously is acting from principles that we would consent to as free and equal rational beings. (Rawls, 1971, p. 516)
I, and I alone, am ultimately responsible for the decisions I make, and am in that sense autonomous. (Lucas, 1966)

It is apparent that, although not used just as a synonym for qualities that are usually approved of, ‘autonomy’ is used in an exceedingly broad fashion. It is used sometimes as an equivalent of liberty (positive or negative in Berlin’s terminology), sometimes as equivalent to self-rule or sovereignty, sometimes as identical with freedom of the will. It is equated with dignity, integrity, individuality, independence, responsibility and self-knowledge. It is identified with qualities of self-assertion, critical reflection, freedom from obligation, absence of external causation and knowledge of one’s own interests. It is even equated by some economists with the impossibility of interpersonal comparisons. It is related to actions, beliefs, reasons for acting, rules, the will of other persons, thoughts and principles. About the only features held constant from one author to another are that autonomy is a feature of persons and that it is a desirable quality to have.

It is very unlikely that there is a core meaning which underlies all these various uses of the term. Autonomy is a term of art and will not repay an Austinian investigation of its ordinary uses. It will be necessary to construct a concept given various theoretical purposes and some constraints from normal usage.


 I shall begin by discussing the nature of autonomy. Given the various problems that may be clarified or resolved with the aid of a concept of autonomy, how may we most usefully characterise the concept? I use the vague term ‘characterise’ rather than ‘define’ or ‘analyse’ because I do not think it possible with any moderately complex philosophical concept to specify necessary and sufficient conditions without draining the concept of the very complexity that enables it to perform its theoretical role. Autonomy is a term of art introduced by a theorist in an attempt to make sense of a tangled net of intuitions, conceptual and empirical issues and normative claims. What one needs, therefore, is a study of how the term is connected with other notions, what role it plays in justifying various normative claims, how the notion is supposed to ground ascriptions of value, and so on – in short, a theory.

A theory, however, requires conditions of adequacy; constraints we impose antecedently on any satisfactory development of the concept. In the absence of some theoretical, empirical or normative limits, we have no way of arguing for or against any proposed explication. To say this is not to deny the possibility we may end up some distance from our starting point. The difficulties we encounter may best be resolved by adding or dropping items from the initial set of constraints. But without some limits to run up against, we are too free to make progress.

I propose the following criteria for a satisfactory theory of autonomy:

Logical consistency

The concept should be neither internally inconsistent nor inconsistent (logically) with other concepts we know to be consistent. So, for example, if the idea of an uncaused cause was inconsistent and autonomy required the existence of such a cause, it would fail to satisfy this criterion.

Empirical possibility

There should be no empirically grounded or theoretically derived knowledge which makes it impossible or extremely unlikely that anybody ever has been, or could be, autonomous. Thus, a theory which required as a condition of autonomy that an individual’s values not be influenced by his or her parents, peers or culture would violate this condition. It is important to note that this condition is not designed to beg the question (in the long run) against those, such as Skinner, who deny the possibility of autonomy. I am attempting to construct a notion of autonomy that is empirically possible. I may fail. This might be due to my limitations. If enough people fail, the best explanation may be that Skinner is correct. Or he may be correct about certain explications and not others. It would then be important to determine whether the ones that are not possible are the ones that are significant for moral and political questions.

We see here how the constraints operate as a system. It would not be legitimate to reject a proposed explication of autonomy on the grounds that we know that nobody is autonomous in that sense if that sense were the very one people have appealed to when deriving normative claims.

Value conditions

It should be explicable on the basis of the theory why, at the least, people have thought that being autonomous was a desirable state of affairs. A strong constraint would require that the theory show why autonomy is not merely thought to be a good, but why it is a good. A still stronger constraint would require that the theory show why, as Kant claimed, autonomy is the supreme good. Because I do not intend my theory as an explication of Kant’s views, and because it is plausible to suppose that there are competing values which may, on occasion, outweigh that of autonomy, I do not adopt the strongest constraint.

As an additional constraint, I suggest that the theory not imply a logical incompatibility with other significant values, that is, that the autonomous person not be ruled out on conceptual grounds from manifesting other virtues or acting justly.

Ideological neutrality

I intend by this a rather weak constraint. The concept should be one that has value for very different ideological outlooks. Thus, it should not be the case that only individualistic ideologies can value autonomy. This is compatible with the claim that various ideologies may differ greatly on the weight to be attached to the value of autonomy, the trade-offs that are reasonable, whether the value be intrinsic, instrumental and so forth.

Normative relevance

The theory should make intelligible the philosophical uses of the concept. One should see why it is plausible to use the concept to ground a principle protecting freedom of speech, or why Rawls uses the idea of autonomous persons as part of a contractual argument for certain principles of distributive justice. One may also use the theory in a critical fashion to argue that a theory which argues from a notion of autonomy to the denial of legitimate state authority has gone wrong because it uses too strong a notion of autonomy.

Judgmental relevance

The final constraint is that the explication of the concept be in general accord with particular judgments we make about autonomy. These judgments may be conceptual: for example, one may believe that autonomy is not an all-or-nothing concept but a matter of more or less. The judgments may be normative, for example, that autonomy is that value against which paternalism offends. The judgments may be empirical, for example, that the only way to promote autonomy in adults is to allow them as children a considerable and increasing degree of autonomy.

I do not believe, however, that one can set out in advance a ‘privileged’ set of judgments which must be preserved. If the judgments do not hang together, then any of them may have to be hanged separately.

These are the criteria. It is possible that no concept of autonomy satisfies them all. Just as Kenneth Arrow discovered there was no social welfare function satisfying certain plausible constraints, we may find there is no concept satisfying ours. That itself would be an interesting discovery and would raise the question of whether we ought to drop or weaken some of the constraints or, perhaps, abandon the idea of autonomy.

What is more likely is that there is no single conception of autonomy but that we have one concept and many conceptions of autonomy – to make use of a distinction first introduced by H.L.A. Hart and developed by Rawls. The concept is an abstract notion that specifies in very general terms the role the concept plays. Thus, a certain idea of persons as self-determining is shared by very different philosophical positions. Josiah Royce speaks of a person as a life led according to a plan. Marxists speak of man as the creature who makes himself; existentialists of a being whose being is always in question; Kantians of persons making law for themselves. At a very abstract level, I believe they share the same concept of autonomy. But when it comes to specifying more concretely what principles justify interference with autonomy, what is the nature of the ‘self’ which does the choosing, what the connections between autonomy and dependence on others are, then there will be different and conflicting views on these matters. This filling out of an abstract concept with different content is what is meant by different conceptions of the same concept.

I intend to present a view that provides specifications for most of these questions, but since I believe that the value of a particular conception is always relative to a set of problems and questions, I want first to indicate the range of issues that I see as relevant to the conception I shall develop.

Autonomy functions as a moral, political and social ideal. In all three cases, there is value attached to how things are viewed through the reasons, values and desires of the individual and how those elements are shaped and formed.

As a political ideal, autonomy is used as a basis to argue against the design and functioning of political institutions that attempt to impose a set of ends, values and attitudes upon the citizens of a society. This imposition might be based on a theological view, secular visions of a good society, or the importance of achieving excellence along some dimension of human achievement. In each case, the argument favouring such imposition is made independently of the value of the institutions as viewed by each citizen. Those favouring autonomy urge that the process of justification of political institutions must be acceptable to each citizen and must appeal to considerations that are recognised to be valid by all the members of the society.

In particular, then, autonomy is used to oppose perfectionist or paternalistic views. It is also related to what Ronald Dworkin refers to as the notion of equal respect. A government is required to treat its citizens neutrally, in the sense that it cannot favour the interests of some over others. This idea is used by Dworkin to argue for the existence of various rights.

Conceptions of autonomy are also used, by Wolff and others, to argue for the illegitimacy of obedience to authority. The emphasis in this argument is on the individual making up his or her mind about the merits of legal restrictions. This use of autonomy seems much closer in content to the ideal of moral autonomy. As a moral notion – shared by philosophers as divergent as Kant, Kierkegaard, Nietzsche, Royce, Hare and Popper – the argument is about the necessity or desirability of individuals choosing or willing or accepting their own moral code. We are all responsible for developing and criticising our moral principles, and individual conscience must take precedence over authority and tradition. I am not defending this line of reasoning, but it is certainly a body of thought which makes use of the notion of autonomy and has a corresponding set of problems connected with responsibility, integrity and the will. A theory of autonomy must throw some light on these problems, even if it does not accept (all of) the proposed solutions.

Finally, we have a set of issues concerning the ways in which the non-political institutions of a society affect the values, attitudes and beliefs of the members of the society. Our dispositions, attitudes, values and wants are affected by the economic institutions, the mass media, the force of public opinion, the social class and so forth. To a large extent, these institutions are not chosen by us; we simply find ourselves faced with them. From Humboldt, Mill and DeTocqueville to Marcuse and Reismann, social theorists have worried about how individuals can develop their own conception of the good life in the face of such factors, and how we can distinguish between legitimate and illegitimate ways of influencing the minds of the members of society.

While Marxists have been most vocal in raising the issues of ‘false consciousness’ and ‘true versus false needs’, it is important to see that the question is one which a wide range of social theorists must address. For it is a reasonable feature of any good society that it is self-sustaining in the sense that people who grow up in such a society will acquire a respect for and commitment to the principles which justify and regulate its existence. It is very unlikely that the development of such dispositions is something over which individuals have much control or choice. Socialisation into the norms and values of the society will have taken place at a very young age. It looks, then, as if we can only distinguish between institutions on the basis of what they convey, their content, and not on the basis that they influence people at a stage when they cannot be critical about such matters. It looks, therefore, as if autonomy in the acquisition of principles and values is impossible.

In all three areas – moral, political and social – we find that there is a notion of the self which is to be respected, left unmanipulated and which is, in certain ways, independent and self-determining. But we also find certain tensions and paradoxes. If the notion of self-determination is given a very strong definition – the unchosen chooser, the uninfluenced influencer – then it seems as if autonomy is impossible. We know that all individuals have a history. They develop socially and psychologically in a given environment with a set of biological endowments. They mature slowly and are, therefore, heavily influenced by parents, peers and culture. How, then, can we talk of self-determination?

Again, there seems to be a conflict between self-determination and notions of correctness and objectivity. If we are to make reasonable choices, then we must be governed by canons of reasoning, norms of conduct, standards of excellence that are not themselves the products of our choices. We have acquired them at least partly as the result of others’ advice, for example, teaching – or, perhaps, by some innate coding. In any case, we cannot have determined these for ourselves.

Finally, there is a tension between autonomy as a purely formal notion (where what one decides for oneself can have any particular content), and autonomy as a substantive notion (where only certain decisions count as retaining autonomy whereas others count as forfeiting it). So the person who decides to do what his or her community, guru or comrades tells him or her to do cannot on the latter view count as autonomous. Autonomy then seems in conflict with emotional ties to others, with commitments to causes, with authority, tradition, expertise, leadership and so forth.

What I shall try to do now is introduce a conception of autonomy that satisfies the criteria set out at the beginning and that is (1) relevant to the moral, political and social issues mentioned above; (2) possible to achieve; and (3) able to avoid the difficulties and problems just enumerated.

The central idea that underlies the concept of autonomy is indicated by the etymology of the term: autos (self) and nomos (rule or law). The term was first applied to the Greek city state. A city had autonomia when its citizens made their own laws, as opposed to being under the control of some conquering power.

There is then a natural extension to persons as being autonomous when their decisions and actions are their own; when they are self-determining. The impetus for this extension occurs first when questions of following one’s conscience are raised by religious thinkers. Aquinas, Luther and Calvin placed great stress on the individual acting in accordance with reason as shaped and perceived by the person. This idea is then taken up by the Renaissance humanists. Pico della Mirandola expresses the idea clearly in his ‘Oration on the Dignity of Man’. God says to Adam:

We have given thee, Adam, no fixed seat, no form of thy very own, no gift peculiarly thine, that … thou mayest … possess as thine own the seat, the form, the gift which thou thyself shalt desire … thou wilt fix the limits of thy nature for thyself … thou … art the molder and the maker of thyself. (Kristeller, 1947, pp. 100–101)

The same concept is presented by Berlin under the heading of ‘positive liberty’:

I wish to be an instrument of my own, not other men’s acts of will. I wish to be a subject, not an object … deciding, not being decided for, self-directed and not acted upon by external nature or by other men as if I were a thing, or an animal, or a slave incapable of playing a human role, that is, of conceiving goals and policies of my own and realizing them. (Berlin, 1969, p. 131)

But this abstract concept only can be understood as particular specifications are made of the notions of ‘self’, ‘my own’, ‘internal’ and so forth. Is it the noumenal self of Kant, or the historical self of Marx? Which mode of determination (choice, decision, invention or consent) is singled out? At what level is autonomy centred – individual decision, rule, values or motivation? Is autonomy a global or a local concept? Is it predicated of relatively long stretches of an individual’s life or relatively brief ones?

Let me begin by considering the relationship between the liberty or freedom of an individual and his or her autonomy. Are these two distinct notions? Are they linked, perhaps, in hierarchical fashion so that, say, interference with liberty is always interference with autonomy, but not vice-versa? Are they, perhaps, merely synonymous?

Suppose we think of liberty as being, roughly, the ability of a person to do what he or she wants, to have (significant) options that are not closed or made less eligible by the actions of other agents. Then the typical ways of interfering with the liberty of an agent (coercion and force) seem to also interfere with his or her autonomy (thought of, for the moment, as a power of self-determination). If we force a Jehovah’s Witness to have a blood transfusion, this not only is a direct interference with his or her liberty but also a violation of his or her ability to determine for himself or herself what kinds of medical treatment are acceptable to him or her. Patient autonomy is the ability of patients to decide on courses of treatment, choose particular physicians and so forth.

But autonomy cannot be identical to liberty for, when we deceive a patient, we are also interfering with his or her autonomy. Deception is not a way of restricting liberty. The person who, to use Locke’s example, is put into a cell and convinced that all the doors are locked (when, in fact, one is left unlocked) is free to leave the cell. But because he or she cannot – given his or her information – avail himself or herself of this opportunity, his or her ability to do what he or she wishes is limited. Self-determination can be limited in other ways than by interferences with liberty.

Both coercion and deception infringe upon the voluntary character of the agent’s actions. In both cases, a person will feel used, will see himself or herself as an instrument of another’s will. His or her actions, although in one sense his or hers because he or she did them, are in another sense attributable to another. It is because of this that such infringements may excuse or (partially) relieve a person of responsibility for what he or she has done. The normal links between action and character are broken when action is involuntary.

Why, then, should we not restrict our categories to those of freedom, ignorance and voluntariness? Why do we need a separate notion of autonomy? One reason is because not every interference with the voluntary character of one’s action interferes with a person’s ability to choose his or her mode of life. If, as is natural, we focus only on cases where the person wishes to be free from interference, resents having his or her liberty interfered with, we miss an important dimension of a person’s actions.

Consider the classic case of Odysseus. Not wanting to be lured onto the rocks by the sirens, he commands his men to tie him to the mast and refuse all later orders he will give to be set free. He wants to have his freedom limited so that he can survive. Although his behaviour at the time he hears the sirens may not be voluntary – he struggles against his bonds and orders his men to free him – there is another dimension of his conduct that must be understood. He has a preference about his preferences, a desire not to have or to act upon various desires. He views the desire to move his ship closer to the sirens as something that is no part of him, but alien to him. In limiting his liberty, in accordance with his wishes, we promote, not hinder, his efforts to define the contours of his life.

To consider only the promotion or hindrance of first-order desires – which is what we focus upon in considering the voluntariness of action – is to ignore a crucial feature of persons, their ability to reflect upon and adopt attitudes towards their first-order desires, wishes and intentions.

It is characteristic of persons, and seems to be a distinctively human ability, that they are able to engage in this kind of activity. One may not just desire to smoke but also desire that one not have that desire. I may not just be motivated by jealousy or anger but may also desire that my motivations be different (or the same).

A person may identify with the influences that motivate him or her, assimilate them to himself or herself, view himself or herself as the kind of person who wishes to be moved in particular ways. Or, he or she may resent being motivated in certain ways, be alienated from those influences, prefer to be the kind of person who is motivated in different ways. In an earlier essay, I suggested that it was a necessary condition for being autonomous that a person’s second-order identifications be congruent with his or her first-order motivations (Dworkin, 1976). This condition, which I called ‘authenticity’, was to be necessary but not sufficient for being autonomous.

I now believe that this is mistaken. It is not the identification or lack of identification that is crucial to being autonomous, but the capacity to raise the question of whether I will identify with or reject the reasons for which I now act. There are a number of considerations that tell against my earlier view.

First, autonomy seems intuitively to be a global rather than a local concept. It is a feature that evaluates a whole way of living one’s life and can only be assessed over extended portions of a person’s life, whereas identification is something that may be pinpointed over short periods of time. We can think of a person who today identifies with, say, his or her addiction, but tomorrow feels it as alien and who continues to shift back and forth at frequent intervals. Does he or she shift back and forth from autonomy to non-autonomy?

Second, identification does not seem to be what is put in question by obvious interferences with autonomy. The person who is kept ignorant, lobotomised, or manipulated in various ways (all obvious interferences with autonomy) is not having his or her identifications interfered with, but rather his or her capacity or ability either to make or reject such identifications.

Third, there seems to be an implication of the position that is counterintuitive. Suppose that there is a conflict between one’s second-order and first-order desires. Say one is envious but does not want to be an envious person. One way of becoming autonomous is by ceasing to be motivated by envy. But another way, on the view being considered here, is to change one’s objections to envy, to change one’s second-order preferences.

Now there may be certain limits on the ways this can be done that are spelled out in the other necessary condition which I elaborated: that of procedural independence. So, for example, it would not do to have oneself hypnotised into identifying with one’s envious motivations. But even if the procedures used were ‘legitimate’, there seems to be something wrong with the idea that one becomes more autonomous by changing one’s higher-order preferences.

Fourth, this view breaks the link between the idea of autonomy and the ability to make certain desires effective in our actions. In this view, the drug addict who desires to be motivated by his or her addiction, and yet who cannot change his or her behaviour, is autonomous because his or her actions express his or her view of what influences he or she wants to be motivating him or her. This seems too passive a view. Autonomy should have some relationship to the ability of individuals, not only to scrutinise critically their first-order motivations but also to change them if they so desire. Obviously, the requirement cannot be as strong as the notion that ‘at will’ a person can change his or her first-order preferences. Indeed, there are certain sorts of inabilities of this nature that are perfectly compatible with autonomy. A person who cannot affect his or her desires to act justly or compassionately is not thought by that fact alone to be non-autonomous. Perhaps, there is still the idea that if justice were not a virtue or that if, in a given case, hardness and not compassion were required, the agent could adjust his or her desires. Susan Wolf (1980) has suggested the requirement that a person ‘could have done otherwise if there had been good and sufficient reason’ (p. 159).

The idea of autonomy is not merely an evaluative or reflective notion, but includes as well some ability both to alter one’s preferences and to make them effective in one’s actions and, indeed, to make them effective because one has reflected upon them and adopted them as one’s own.

It is important both to guard against certain intellectualist conceptions of autonomy as well as to be candid about the ways in which people may differ in their actual exercise of autonomy. The first error would be to suppose that my views imply that only certain types or classes of people can be autonomous. If we think of the process of reflection and identification as being a conscious, fully articulated and explicit process, then it will appear that it is mainly professors of philosophy who exercise autonomy and that those who are less educated, or who are by nature or upbringing less reflective, are not, or not as fully, autonomous individuals. But a farmer living in an isolated rural community, with minimal education, may without being aware of it be conducting his or her life in ways which indicate that he or she has shaped and moulded his or her life according to reflective procedures. This will be shown not by what he or she says about his or her thoughts, but in what he or she tries to change in his or her life, what he or she criticises about others, the satisfaction he or she manifests (or fails to) in his or her work, family and community.

It may be true, however, that there is empirical and theoretical evidence that certain personality types, social classes or cultures are more (or less) likely to exercise their capacity to be autonomous. I do not suppose that the actual exercise of this capacity is less subject to empirical determination than, say, the virtue of courage. To the extent that this is borne out by the evidence, we must be on guard against the tendency to attribute greater value to characteristics which are more likely to be found in 20th-century intellectuals than in other groups or cultures.

To return to our original question of the relation between autonomy and liberty, I would claim that the two are distinct notions, but related in both contingent and non-contingent ways. Normally, persons wish to act freely. So, interfering with a person’s liberty also interferes with the ways in which he or she wants to be motivated, the kind of person he or she wants to be and, hence, with his or her autonomy. But a person who wishes to be restricted in various ways, whether by the discipline of the monastery, regimentation of the army, or even by coercion, is not, on that account alone, less autonomous. Further, I would argue that the condition of being a chooser (where one’s choices are not defined by the threats of another) is not just contingently linked to being an autonomous person but must be the standard case from which exceptions are seen as precisely that – exceptions. Liberty, power and control over important aspects of one’s life are not the same as autonomy but are necessary conditions for individuals to develop their own aims and interests and to make their values effective in the living of their lives.

Second-order reflection cannot be the whole story of autonomy. For those reflections, the choice of the kind of person one wants to become may be influenced by other persons or circumstances in such a fashion that we do not view those evaluations as being the person’s own. In ‘Autonomy and Behavior Control’, I called this a failure of procedural independence.

Spelling out the conditions of procedural independence involves distinguishing those ways of influencing people’s reflective and critical faculties which subvert them from those which promote and improve them. It involves distinguishing those influences such as hypnotic suggestion, manipulation, coercive persuasion, subliminal influence and so forth, and doing so in a non-ad hoc fashion. Philosophers interested in the relationships between education and indoctrination, advertising and consumer behaviour, and behaviour control have explored these matters in some detail, but with no finality.

Finally,1 I wish to consider two objections that can be (and have been) raised to my views. The first is an objection to introducing the level of second-order reflection at all. The second is why should we stop at the second level and is an infinite regress not threatened.

The first objection says that we can accomplish all we need to by confining our attention to people’s first-order motivation. After all, in my own view of the significance of procedural independence, we have to find a way to make principled distinctions among different ways of influencing our critical reflections, so why not do this directly at the first level. We can distinguish coerced from free acts, manipulated from authentic desires and so forth. My reply is that I think we fail to capture something important about human agents if we make our distinctions solely at the first level. We need to distinguish not only between the person who is coerced and the person who acts, say, to obtain pleasure, but also between two agents who are coerced. One resents being motivated in this fashion, would not choose to enter situations in which threats are present. The other welcomes being motivated in this fashion chooses (even pays) to be threatened. A similar contrast holds between two patients, one of whom is deceived by his or her doctor against his or her will and the other who has requested that his or her doctor lie to him or her if cancer is ever diagnosed. Our normative and conceptual theories would be deficient if the distinction between levels were not drawn.

The second objection is twofold. First, what is particularly significant about the second level? Might we not have preferences about our second-order preferences? Could I not regret the fact that I welcome the fact that I am not sufficiently generous in my actions? I accept this claim, at least in principle. As a theory about the presence or absence of certain psychological states, empirical evidence is relevant. It appears that for some agents, and some motivations, there is higher-order reflection. If so, then autonomy will be thought of as the highest-order approval and integration. As a matter of contingent fact, human beings either do not, or perhaps cannot, carry on such iteration at great length.

The second part of this objection concerns the acts of critical reflection themselves. Either these acts are themselves autonomous (in which case we have to go to a higher-order reflection to determine this, and since this process can be repeated an infinite regress threatens) or they are not autonomous, in which case why is a first- order motivation evaluated by a non-autonomous process itself autonomous. My response to this objection is that I am not trying to analyse the notion of autonomous acts, but of what it means to be an autonomous person, to have a certain capacity and exercise it. I do claim that the process of reflection ought to be subject to the requirements of procedural independence, but if a person’s reflections have not been manipulated, coerced and so forth, and if the person does have the requisite identification, then they are, in my view, autonomous. There is no conceptual necessity for raising the question of whether the values, preferences at the second order would themselves be valued or preferred at a higher level, although in particular cases the agent might engage in such higher-order reflection.

Putting the various pieces together, autonomy is conceived of as a second-order capacity of persons to reflect critically upon their first-order preferences, desires, wishes and so forth and the capacity to accept or attempt to change these in light of higher-order preferences and values. By exercising such a capacity, persons define their nature, give meaning and coherence to their lives, and take responsibility for the kind of person they are.


Ackerman, B. (1980). Social justice in the liberal state. New Haven, CT: Yale University Press.

Berlin, I. (1969). Four essays on liberty. Oxford: Oxford University Press.

Dearden, R.F. (1972). Autonomy and education. In R.F. Dearden (Ed.), Education and the development of reason (pp. 448–465). London: Routledge and Kegan Paul.

Downie, R.S., & Telfer, E. (1971). Autonomy. Philosophy, 15, 293–301.

Dworkin, G. (1976). Autonomy and behavior control. Hastings Center Report, 6, 23–28.

Dworkin, R. (1978). Liberalism. In S. Hampshire (Ed.), Public and private morality (pp. 113–143). Cambridge: Cambridge University Press.

Feinberg, J. (1980). The idea of a free man. Education and the Development of Reason, 161, 122–155.

Friedman, M. (1986). Autonomy and the split-level self. Southern Journal of Philosophy, 24(1), 19–35.

Goldstein, J. (1978). On being adult and being an adult in secular law. In E.H. Erikson (Ed.), Adulthood (pp. 19–40). New York, NY: W. W. Norton.

Kristeller, P.O. (1947). The philosophy of man in the Italian renaissance. Italica, 24, 93–112.

Lucas, J.L. (1966). Principles of politics. Oxford: Oxford University Press.

Peters, R.S. (1972). Freedom and the development of the free man. In R.F. Dearden (Ed.), Education and the development of reason (pp. 11–31). London: Routledge and Kegan Paul.

Rawls, J. (1971). A theory of justice. Cambridge, MA: Harvard University Press.

Rawls, J. (1980). Construction and objectivity. Journal of Philosophy, 78, 554–572.

Scanlon, T. (1972). A theory of freedom of expression. Philosophy and Public Affairs, 1(Winter), 204–226.

Thalberg, I. (1978). Hierarchical analyses of unfree action. Canadian Journal of Philosophy, 8, 211–226.

Wolf, S. (1980). Asymmetric freedom. Journal of Philosophy, 77, 151–166.

Wolff, R.-P. (1970). In defense of anarchism. New York, NY: Harper & Row.


The article is an unchanged republishing of Gerald Dworkin’s (1988) ‘The nature of autonomy’ in The theory and practice of autonomy (Cambridge, Cambridge University Press), pp. 3–20, with the kind permission from Cambridge University Press.

1I am indebted to an unpublished manuscript, ‘Autonomy and External Influence’, of John Christman for his ideas on these points. For further discussion, see Thalberg (1978) and Friedman (1986).

About The Author

Gerald Dworkin
Department of Social Science and Humanities, University of California, Davis, CA, USA
United States

Gerald Dworkin is a professor of moral, political and legal philosophy. He is currently Distinguished Professor of Philosophy at the University of California, Davis. Dworkin’s main areas of research include the nature and justification of autonomy, paternalism in the criminal law and the issue of which acts may legitimately be criminalised by the state.

Article Metrics

Metrics Loading ...

Metrics powered by PLOS ALM

Related Content