Review - Personal Autonomy
New Essays on Personal Autonomy and its Role in Contemporary Moral Philosophy
by James Stacey Taylor (Editor)
Cambridge University Press, 2005
Review by Christian Perring, Ph.D.
Nov 25th 2008 (Volume 12, Issue 48)

Personal Autonomy is a collection of 15 previously unpublished papers by philosophers on the topic of autonomy.  It is divided into three parts.  The first is on theoretical approaches, with seven papers, including work by Michael Bratman, Bernard Berofsky, Alfred Mele, and Laura Waddell Ekstrom.  The second is on moral responsibility, including a paper by Susan Wolf, and the third is the relation of the autonomy debate to other areas of philosophy such as political theory and applied ethics, and has papers by John Christman, Thomas May, Tom Beauchamp, and R.G. Frey. 

The introduction by Taylor sets out the most influential theories about the nature of personal autonomy, with focus on the work of Harry Frankfurt and Gerald Dworkin.  He argues that the hierarchical theories they defend do not work, on the grounds that they do not match our intuitions about some cases.  One central case of that of manipulation of minds, through hypnotism or other means; he claims that even if a person deliberates and endorses a desire, that person is not autonomous with respect to that desire if her thought processes have been manipulated by some third party.  Another example he uses repeatedly is of a man who, in becoming a monk, freely decides to subordinate his will to the abbots.  Taylor claims that such a person effectively makes himself an automaton, and thus gives up autonomy.  Once a monk, none of the person's desires or actions will be autonomous, according to Taylor.  Taylor's arguments rest heavily on these intuitions, yet it is far from clear that they will be widely shared.  This is partly because the examples are sketched with so little detail, and could be filled out in different sorts of directions. 

There are many ways in which our desires can be manipulated.  For example, one could make a person hate pickles by making them violently ill when they eat some pickles, and so they come to associate the taste of pickles with being violently ill.  They may refuse ever to eat a pickle again.  Is their dislike of pickles autonomous?  I doubt that we have a clear pre-theoretical intuition here.  Famous cases of charismatic manipulative people often leave us bemused: when adults join cults, do they do so autonomously?  Did the adult followers of Jim Jones who drank the poisoned drink he told them to drink in Jonestown do so autonomously?  It is hard to give a simple yes or no answer.  We can move to simpler cases that are more idealized and fictional, supposing that our desires and our deliberations about them are controlled by evil scientists who implant devices in our brains or even evil demons who directly control our thoughts.  Then it is easier to agree that in such cases we are not autonomous in having the desires they made us have, even if we want to have them.  However, one is left with the suspicion that since these cases are so unusual, defenders of hierarchical theories of autonomy need only make small modifications to the original theories to protect them from such potential counter-examples.  One might say, for example, that to be autonomous, it is also necessary that one's thought processes are not being directly interfered with by a third party.

The monk example is equally tricky.  If they monk has decided to let the abbots tell him what he should want, and that he will not critically evaluate their commands, and will even quash any countervailing desires he has, and he has made this decision freely, could we not say that his desires are indeed autonomous in an important sense?  They stem from his decision to subvert his will to that of the abbots.  It is far from clear that Taylor's description of the monk as an automaton is accurate.  It may take a great deal of practice to learn how to subvert his will, and he may indeed be very pleased when he at last is successful at it.  We could again change the example to one with more science fiction content, and suppose that the trainee monk gets a brain-implant linked to a computer, and the abbots only have to list what desires he should have at their computer terminal for it to transfer those desires to him.  Even if he has voluntarily agreed to this arrangement, our intuitions here will be much stronger that his desires are not autonomous.  But those defending theories for which this case is a counterexample can rather easily modify their theories to take account of such bizarre possibilities. 

This highlights the difficulty of using such cases as ways to adjudicate among different theories of autonomy.  Our intuitions about cases that happen in real life are often not clear, and fantastic cases are often not very helpful in furthering the rational discussion.

The first paper, by Michael Bratman, provides an overview of his theory of the importance of planning to our understanding of intentional action and autonomous action.  He endorses a hierarchical view of autonomy derived from Harry Frankfurt's theories, but supplements it in order to enable it to counter objections from Gary Watson.  The self-reflection involved in autonomous action will be policies about planning.  For example, these policies will be about which of our desires justify our actions.  Self-governance will be constituted by the guidance by such policies, and when those policies are not in conflict with other personal policies.  These policies will have a "systematic role in constituting and supporting the cross-temporal organization of practical thought and action by way of Lockean ties" (45); this solves the problem of which of a person's desires she identifies with, because the plans are systematically related to each other.  Bratman emphasizes that conative hierarchy is not necessary for self-governance, but it does explain how self-governance is possible.  The paper does valuable work locating his position relative to other theorists, but it is also remarkably vague in its proposals.

The paper by Bernard Berofsky argues that autonomy is possible even if we do not have free will.  It's a slightly strange paper since Berofsky himself is a compatibilist who believes we have free will, so in defending the possibility of autonomy without free will, he is staking out an intellectual position as a conditional.  Nevertheless, his claim is plausible. 

Robert Noggle offers a nicely argued paper on the relation between authenticity and autonomy, considering the problem of how one gains authentic desires.  If we require that if we require that they derive from other authentic desires and beliefs, we will run into a problem of infinite regress.  He wonders how we first become authentic, and his answer is that we do so gradually, though incremental processes in childhood.  He emphasizes the distinction between the initial formation of the self, which does not rob a person of authenticity, and the subsequent manipulation of their core beliefs and desires through a manipulative process such as brainwashing.  Noggle's position here is plausible, although his problem of infinite regress and some of his examples invite a more skeptical view of authenticity, that it is just a way to invoke our disapproval of manipulating people's beliefs, and is very hard to cash out as a concept of its own.  Although we like to say that a brainwashed or radically transformed person is not herself, and is not acting from authentic desires or beliefs, it is not clear that we either have a theory or a consistent set of intuitions to strongly justify such a claim.  That is to say, it is difficult to justify the distinction in authenticity between a person who went through a radical person change autonomously and one who was manipulated.  The real test for our notion of authenticity is whether it can be justified by a well-supported theory. 

Alfred Mele has a theory of autonomy for which he argues at length in his book Autonomous Agents.  His contribution to the volume summarizes some of his main ideas and nuances one of his examples in arguing that we can have autonomy whether take a compatibilist or libertarian view of free will. 

Paul Benson defends what he calls a "weak substantive account" of personal autonomy, in between the procedural accounts of personal autonomy of Diana Meyers and the more strongly substantive account of Natalie Stoljar.  The central cases for this debate arise with the formation of some women's preferences in patriarchal societies, and whether women who identify with their oppressors are really autonomous, or whether their views are a result of their oppression.  Benson argues that personal autonomy requires normative competence, which is different from accepting any particular normative claims.  Normative competence simply involves the ability to recognize and identify normative claims.  Benson's proposal here is quite tentative, since he says it needs further development, but it promises to deliver a middle way between the two other views of personal autonomy he discusses, and thus escaping the weaknesses of both.

Laura Waddell Ekstrom proposes a different theory of autonomy that links it to authenticity, and she argues for a coherentist view of authenticity.  That's to say, a person's preference is authentic, and the action on it autonomous, depending on whether it coheres with the person's other beliefs, values, and desires.  The tricky part of the theory is providing a definition of coherence.  Ekstrom's definition is somewhat technical:

A preference P coheres with the character system of a person at a time if and only if, for any competing preference, it is either more valuable for the person to have P than the competing preference on the basis of her character system at the time, or it is as valuable for her to have the competing preference and a neutralizing attitude, as it is for her to have the competing preference alone, on the basis of her character system at the time.  One preference competes with another preference for a person at a time  just in case it is less valuable for that person to have the preference on the assumption that the object of the other is good than on the assumption that the object of the other is bad, on the basis of her character system at a time. (151)

An odd feature of this definition is that it makes coherence an all-or-nothing concept, which is very different from our ordinary concept of coherence.  A footnote directs the reader to other papers by the author for further elaboration on the meaning of these definitions.  She further says that an act is autonomous if and only if it is "nondeviantly caused by an uncoercively formed, personally authorized preference" (151).  It is easy to get the basic idea, but it is far from clear that these definitions are very helpful, and they give the impression that they don't help much, if the definition of autonomy relies on a prior understanding of what coercion is.  Furthermore, it is hard to judge, except in somewhat unusual cases, when a preference coheres with a character system.    Suppose I prefer to go for a walk than attend a useless committee meeting with people who like to talk too much, because I have been to all the previous committee meetings and I am tired of going, even though I know that the Dean expects me to go.  Does that preference cohere with my character system?   Ekstrom's basic idea is that coherence is related to what one values.  One basic worry is that some values are incommensurable, and so one is not going to be able to judge the coherence of a person's preferences.  Another worry is that this proposal makes no distinction between weakness of will and compulsion.  All weak willed actions are not performed autonomously on this view, but as many have argued, this is an implausible position.  The mere fact that an action performed due to a preference that does not fit with our values, such as being inappropriately rude to a store clerk, does not mean that we were not autonomous.  Ekstrom does not address these worries in this paper, so one would have to look elsewhere for answers to them.

This concern about the relation between weak willed actions and non-autonomous actions is central to Nomy Arpaly's paper, although she frames most of her discussion in different terms.  She argues that theories in philosophy of mind about the true self, identification, and the like tend not to provide a theory of autonomy that is directly related to moral responsibility.  That is to say, theories such as those of Frankfurt or Velleman do not equate autonomous actions with those for which the agents are morally responsible.  On their views, people can act without autonomy and yet still be morally responsible for their actions.  Arpaly nicely excludes a possible loophole in this, for those who wish to draw a distinction between being directly and indirectly morally responsible.  Suppose one loses patience with a friend and snaps at her: on a Frankfurtian theory, one may not identify with the emotion behind that action, yet one can still be morally responsible for one's rudeness.  One could try to explain the link between the theory of autonomy and the moral responsibility by saying that one is indirectly morally responsible for one's rudeness because even though one was not autonomously rude, one should still have made an effort not to be rude, and could have counted to ten and regained one's patience.  Arpaly says that such an approach will not be successful in all or even most cases, because in most cases we have little control over our characters, and often there are no easily available "counting to ten" tools that can get us to get us to act more in accordance with our values.  Thus, Arpaly make a convincing case that these theories of autonomy have little to do with moral responsibility.  She further argues at the end of the paper that while they are interesting, it isn't really clear what those theories are about.  That's to say, once the practical link to moral responsibility is severed, it seems that the different theories of autonomy may just be talking past each other, defining terms in different ways, and thus failing to even disagree with each other.   Arpaly's paper is one of the clearest and most provocative in the collection.

Indeed, so successful and eloquent is Arpaly's paper that other papers suffer in comparison.  The following three papers by Marina Oshana, Michael McKenna and Ishtiyaque Haji all address the relation between autonomy and moral responsibility, with particular relation to the question of free will, determinism, and having alternate possible actions from the one that the agent actually performed.  Oshana argues that one can be autonomous in a deterministic world.  McKenna argues that moral responsibility and autonomous agency are logically independent concepts, and so one can have one without the other.  Haji argues that alternate possibilities are not required for moral responsibility or autonomy.  One has to be particularly enthusiastic about the contemporary debates over free will, determinism and alternate possibilities in order to grapple with their papers.  It may be that there are strong arguments amid the rush of technical definitions, thought experiments concerning bizarre hypothetical scenarios, and copious footnotes (McKenna has 55 footnotes for 22 page paper), but it is not an inspiring way of doing philosophy.  Most of the careful readers of these papers will probably be the philosophers mentioned in the footnotes.

Susan Wolf brings the level of writing up with a paper summarizing the ideas in her 1990 book Freedom Within Reason.  She sets out different views of moral responsibility: the Autonomy View, the Real Self View, and the Reason View.  The Autonomy View equates autonomy with moral responsibility.  The Real Self View distinguishes between different kinds of motivations we have for actions, and labels some of them as authentic.  Those are the ones for which we are morally responsible.  The Reason View holds that one is morally responsible if one has the capacity to rationally appreciate morality and understand what is wrong with wrong actions, and one is not morally responsible if one lacks that ability.  This view assumes a strong moral realism and also takes a Kantian approach to linking moral appreciation with rationality.  She points out the problems with the first two views and the advantages of the latter view.  It is a helpful paper for anyone wanting an introduction to Wolf's theories, which deserve serious consideration. 

The final section addresses the debates about autonomy in political theory and applied ethics.  John Christman points out that there's very little interaction between the philosophical discussion of the nature of autonomy in political theory and that in moral psychology and philosophy of mind.  Christman explores analyses of autonomy from liberalism and sets out some criticisms that have been made, especially from political perfectionists.  Thomas May goes over familiar ground in his paper about the criticisms of the concept of autonomy in bioethics, setting out the views of communitarians and feminists.  The criticism derives from a concern that the standard understanding of autonomy in bioethics is too narrow, and does not appreciate the way an individual is situated in a community.  Tom Beauchamp's paper is especially useful in addressing the connection between metaphysical and moral theories of autonomy, and like Arpaly, he argues for the priority of the moral conceptions over the metaphysical ones.  R.G. Grey considers the relevance of autonomy and nonautonomy for animal rights, and points out that no theory of autonomy will give the result that all humans are autonomous and no non-human animals are autonomous.  This aids his argument for accepting the moral status of non-human animals, showing difficulty in morally distinguishing between seriously impaired humans and other animals. 

In summary, Personal Autonomy is mainly aimed at professional philosophers and graduate students working on autonomy.  Some papers will serve as introductions to well-known positions, and could be used in undergraduate philosophy courses.  Others are much more specialized, and advance existing debates.  The book gives no indication that the philosophical debates about autonomy are likely to achieve a resolution, but they do show how the debates have progressed in the last few decades.  The persistent method of using thought experiments and appealing to intuition that runs through a large number of papers gets surprisingly little scrutiny, although several authors do comment on the unreliability of pre-theoretical intuitions and address what a theory of autonomy is meant to achieve.  If, as Tom Beauchamp points out, our ordinary concept of autonomy is vague, then there's little point in trying to capture our variable intuitions in a philosophical theory: rather, some other grounds need to be given for adopting one or other view.  The main candidate that arises in this collection is a theory of autonomy that will fit with our ethical practices of finding people praiseworthy and blameworthy.  However, we should also note that there's considerable variation in our moral intuitions and the moral theories philosophers advocate, so the link to morality will not settle debates over autonomy.  Yet the book makes a strong case that autonomy is a central concept in our self-understanding, so further philosophical scrutiny of it is worthwhile.

© 2008 Christian Perring

Christian Perring, Associate Professor of Philosophy, Dowling College, New York.