. 2
( 13)


a different source and as invested in a different dimension of human
existence. As a corollary, each pinpoints a different set of contributions
to autonomy as well as a different set of threats to autonomy (apart from
coercive threats).
The unitary self is the independent, self-monitoring, self-controlling
self that has been pivotal to autonomy theory. As the seat of rational-
ity and thus rational deliberation and choice, the self-as-unitary is often
viewed as the ground for free will and responsibility. Indeed, the self-
as-unitary and the autonomous self are so closely identi¬ed that they
almost seem indistinguishable.3 To be rationally re¬‚ective and free to
carry out one™s rationally reached decisions is to be autonomous, on
many accounts. I submit, though, that intelligence and good sense should
not be equated with reason, and that if reason is kept distinct from
these broader desiderata, rationalism is not without its perils for au-
tonomy. A zealous commitment to reasoned decision-making can leave
the individual inhibited, rigid, unspontaneous, and shallow “ in a word,
The social self is the socialized or enculturated self. This conception of
the self underscores people™s assimilation of social norms and mastery
of appropriate ways to act and interact, as well as their assimilation of cul-
turally transmitted values, attitudes, and interpretive frameworks through
Diana Tietjens Meyers

which they perceive and negotiate social relations. Internalized, this mate-
rial contributes to the individual™s identity, and thus the identity of the
self-as-social is invested in a community and its cultural heritage. While
it is obvious that individuals cannot create their own value systems and
styles of conduct ex nihilo and that individuality is parasitic on socialization
and enculturation, it is also clear that these normalizing processes pose
a danger to autonomy. When individuals have little opportunity to ex-
plore alternative value systems and social practices, and when dominant
values and practices are rigorously enforced, socialization and encultur-
ation function as indoctrination, which precludes critical re¬‚ection on
the values and desires that shape one™s choices.
The relational self is the interpersonally bonded self. As relational selves
with lasting emotional attachments to others, people share in one an-
other™s joys and sorrows, give and receive care, and generally pro¬t from
the many rewards and cope with the many aggravations of friendship and
family membership. These relationships are sources of identity, for peo-
ple become committed to their psychocorporeal and to others whom they
care about, and these commitments become integral to their psychocor-
poreal economies. Thus, the self-as-relational is invested in a circle of
family and friends, and the lives of individuals are incalculably enriched
by these ties. Yet, these ties also threaten autonomy, for responding to
others™ needs and ful¬lling one™s responsibilities to them can become so
consuming that the individual is deprived of any opportunity to pursue
personal goals and projects.
The divided self is the psychodynamic self. Split between consciousness
and self-awareness, on the one hand, and elusive unconscious affect and
desire, on the other, the self-as-divided is characterized by inner depth,
complexity, and enigma. The ¬‚uid but distinctive psychocorporeal econ-
omy of the self-as-divided is manifest in a unique “ indeed, a vibrantly
individualized “ personality. In an important respect, the value we place
on autonomy pays tribute to this conception of the self, for autonomy en-
ables people to express their individuality in the way they choose to live.
Yet the self as divided alerts us to another peril for autonomy “ namely,
unconscious drive and repressed desire. In pathologies such as obsessive
compulsive disorder, these forces take over the individual™s agency. But
the peril is not limited to this extreme. To the extent that individuals are
oblivious to unconscious materials, their self-knowledge is incomplete
and possibly distorted, and to the extent that their choices and actions
are shaped by these obscure forces, individuals lack control over their
Decentralizing Autonomy: Five Faces of Selfhood 31

Outside of legal theory, the embodied self is often overlooked in discus-
sions of autonomy. This is rather surprising since embodiment is nec-
essary for taking action or partaking in sensuous pleasure. Moreover,
people are deeply invested in their body image “ their sense of what they
look like and what their physical capabilities are. Consequently, attacks
on their bodily integrity can be traumatic. Because the attributes of the
embodied self are central to individual identity and agency, U.S. law gen-
erally treats the embodied self as sacrosanct.4 Still, the self-as-embodied
deserves sustained attention from autonomy theorists, for health, phys-
ical pro¬ciencies, and vitality expand the scope of autonomy, whereas
illness, frailty, and disability put autonomy in jeopardy.5
Laid out this way, it seems obvious that each of these conceptions cap-
tures a signi¬cant dimension of selfhood “ of what it™s like and what
it means to be a human subject. Before proceeding, though, I would
like to comment brie¬‚y on the terminology I have introduced here. I
have referred to the self-as-unitary, as-social, as-relational, as-divided, and
as-embodied. In what follows, I shall use these expressions interchange-
ably with the more idiomatic expressions “ “the unitary self,” “the social
self,” “the relational self,” “the divided self,” and “the embodied self.”
However, I wish to stress that the latter expressions, though familiar, are
also misleading, for they seem to reify these different selves. They make
it seem that each person is somehow an aggregate of ¬ve selves. The
implausibility of this claim then leads to the supposition that one must
decide which kind of self one really is and somehow subsume the other
four phenomenal selves within that conception. I think that this move is
wrongheaded, and I like the “self-as” terminology because it de¬‚ects this
reductionist proclivity.6 What I mean to convey by the “self-as” terminol-
ogy is that each of these conceptions represents a focus of attention, a di-
mension of subjective life, and a way of framing a self-understanding or a
project. Accordingly, these expressions should not be viewed as mirroring
ontology, but rather as labeling phenomenological and epistemological
It is important to note as well that in the social-scienti¬c and philo-
sophical literature, these ¬ve conceptions of the self are not as discrete
as I have made them seem for the purposes of my argument in this chap-
ter. In psychoanalytic object relations theory, for example, the relational
self is also divided and social. Likewise, philosophical accounts of the au-
tonomous self often recognize that the self is divided while positing the
unitary self as a regulative ideal, and feminist ethics of care often focus
on the relational self while presuming its enculturation and rationality.
Diana Tietjens Meyers

In my concluding remarks, I shall revisit the issue of the interconnections
between these conceptions of the self.

II Authentic Attributes and Decentralized Self-Direction
My aim in this section is to discredit the assumption that autonomous
agency is inseparable from the reasoning skills of the self-as-unitary. To
that end, I shall sketch two predicaments, one pertinent to the self-as-
relational and another pertinent to the self-as-embodied. In both, I would
maintain, I enacted authentic attributes, but it is doubtful that I acted au-
tonomously. In these two cases, what raises doubts about my autonomy
is the fact that I did not decide to act as I did because I would be en-
acting my authentic attributes. Indeed, I experienced no introspectable
decision-making process at all. Nevertheless, instead of concluding that
my autonomy was compromised, I shall suggest that the self-as-relational
and the self-as-embodied sometimes function as agents of autonomous
self-discovery, self-de¬nition, and self-direction. In the interest of conci-
sion, I shall leave it to readers to imagine examples of how the self-as-social
and the self-as-divided might also enact authentic attributes, but in later
sections I shall develop reasons to think that these dimensions of the
self can function in ways parallel to those of the self-as-relational and the
My ¬rst case focuses on the self-as-relational. A few years ago, I learned
that I had developed a metabolic condition that requires me to restrict my
diet. Alas, I love good food. There are almost no foods I don™t enjoy, and
I™ve always enjoyed the conviviality of eating with friends. Consequently,
adhering to this diet is not easy for me, but by and large I™ve managed to.
I™ve come to realize, though, that doing so has been a complex relational
achievement. My husband and some of my close friends have patiently
listened to my gripes, and they™ve even abstained from ordering or serving
forbidden dishes when I™m around. Not only does their compassion and
sympathetic self-restraint reduce my exposure to temptation, but also
their willingness to adapt helps me overcome my resistance to adapting.
This sort of support, valuable as it is, has received considerable attention
from autonomy theorists.8
Instead, I would like to spotlight a quite different relational mechanism
of control. I have discovered that when I am with people who know of
my condition but who don™t refrain from indulging in pleasures I must
forego, I seldom succumb to temptation. The mere knowledge that there
would be witnesses to my delinquency curbs my appetite. I don™t ask
Decentralizing Autonomy: Five Faces of Selfhood 33

people to encourage me to stick to my diet, and no one ever has. Yet
their knowing presence prevents me from violating my diet.
In informing associates about my situation, do I delegate responsibil-
ity? Do I make these individuals into enforcers of my values? It might seem
that I exercise self-control because I created a social network of knowl-
edge that suppresses self-destructive behavior. But in an important respect
this construal is inaccurate, for it exaggerates the role of my rational will.
When I ¬rst told people about my condition, my intentions were differ-
ent. I informed good friends because one tells good friends important
news, and I informed people who invited me to dinner parties because I
wanted to avoid awkward situations. In time, however, as I encountered
these con¬dants in other contexts, I realized that these individuals were
preempting my occasional renegade impulses.9 Without realizing what I
was doing, and unbeknownst to these individuals, I had transferred some
of my agentic powers to our relationships. Now that I know this trick,
I can deliberately recruit acquaintances into this scheme, and to do so
would seem a straightforward case of making them unwitting extensions
of my autonomous will. But I submit that before I anticipated the full
consequences of my telling people about my predicament, a relational
conative capacity simply materialized as an unintended consequence of
my disclosures, and this conative capacity enabled my self-as-relational to
autonomously refuse harmful delectables.
My other case concerns the self-as-embodied. Several years ago, I was
hiking by myself on Mt. Rainier. While descending a vast, steep, hard-
packed snow¬eld quite high on the mountain, I slipped and fell twice,
and I broke a wrist each time. There was no one else around. After
picking myself up from the ¬rst fall, I thought for a moment about
whether I should go back up to Camp Muir, a base camp used for
summiting where I had encountered a few people and that was much
closer than the ranger station at Paradise, thousands of vertical feet be-
low. I quickly decided that returning made no sense, and continued
down the mountain. At that point, my body took over in two respects.
Without ever pausing to ¬gure anything out, I took measures to pro-
tect myself from further injury, for example, sitting and using my legs
to propel myself down especially steep places. Also the wonder-drug
adrenaline kept me energized, pain-free, and fear-free throughout the
ordeal. My body improvised quite ingeniously, and proceeded with ex-
traordinary determination and alacrity in the face of considerable dan-
ger. So I consider myself lucky to have had a clever, courageous self-as-
Diana Tietjens Meyers

Here I must ask for your indulgence and beg you not to dismiss this
seeming category mistake out of hand. Apart from re¬‚exes, we don™t have
good ways of talking about situations in which one™s body assumes control
and acts on one™s behalf, as it were.10 Ordinarily, we think of adrenaline™s
psychoactive properties as analogous to those of Prozac “ the former
functions as an anti-cowardice drug, just as the latter functions as an anti-
depressant. It seems natural, then, to say that this hormone temporarily
made me “ the agentic consciousness “ courageous. Yet putting things
this way seems false in my case, for at the time I did not feel infused with
courage. In fact, it was not until later when friends asked me if I had
been afraid that it occurred to me that I could have been afraid or that
anyone might consider what I™d done courageous. Nevertheless, my body
was certainly doing exactly what a courageous body would do.
It would not be odd either to say that it was lucky that my body was
strong and vigorous enough for me to be able to extricate myself from
the mess I had gotten myself into. Although true, this observation fails to
capture my body™s expeditiousness in managing my descent. Of course
it does not strain credulity to say that one™s body has acquired certain
skills “ for example, the ability to swim “ and that these ingrained ca-
pabilities can take over and ensure survival “ for example, in a boating
accident. I would stress, though, that learned mountaineering skills con-
tributed very little to my actions. Thus, the idea of a trained, adept body
isn™t quite to the point. Nevertheless, I would like you to entertain the
possibility, peculiar though it may seem, that my self-as-embodied bravely
and resourcefully orchestrated an autonomous descent for me.
It might be objected that the fact that someone does not de¬nitely want
to act otherwise, puts up no resistance to acting as the self-as-relational or
the self-as-embodied ordains, and does not come to regret going along
this way is necessary but not suf¬cient for autonomy. In addition, the
agent must really want to act as she does, and, as the following hypothetical
case shows, it cannot be a mere coincidence that her conduct meshes with
her authentic attributes.
Here is an example of such a coincidence. I often send donations to
an organization called City Harvest, which collects leftover food from
restaurants and grocery stores and uses it to feed destitute people. There
are quite a few charitable organizations in New York City that seek to
reduce hunger. But since I particularly approve of City Harvest™s methods
and its effectiveness, I see my donations as autonomous gestures, and I
intend to continue giving. Suppose, though, that late one night as I am
walking home, a City Harvest Robin Hood threatens me with a knife
Decentralizing Autonomy: Five Faces of Selfhood 35

and demands my wallet. Preferring safety over the meager contents of
my wallet, I relinquish my money. Although my action conforms with my
values and past practices, I don™t regard it as autonomous, for, however
worthy the cause, I was forced to hand over my cash. In the aftermath, I
might console myself by re¬‚ecting that I would have sent them at least
that much anyway. Still, since my control over this donation was severely
diminished, it was not autonomous.11 Rephrasing Isaiah Berlin™s famous
remark, people cannot be forced to be autonomous.
The autonomy of conduct elicited by interpersonal relations or issuing
from bodily processes may seem like “forced autonomy,” too. The same
objection, I would note, applies to conduct prompted by enculturation
or fueled by the workings of the unconscious. In each case, something
other than reason gives rise to action, and canonical mentalist, individu-
alist presuppositions about autonomous volition raise doubts about the
autonomy of such action. From that point of view, it seems that in these
cases, internal or internalized forces impel one to act. Yet there are some
striking ways in which the two personal experiences I have described
diverge from the Robin Hood mugging scenario.
Most striking is the fact that the nature of the compulsion, if compul-
sion is not a misnomer in the cases I set out, is entirely different. There is
no violence or threat of violence in my friends™ inadvertent aid control-
ling over my unruly appetite or my body™s carrying me to safety despite
the treacherous mountain terrain. Also signi¬cant is the fact that peo-
ple who are unwittingly keeping me within my dietary regimen are not
intentionally manipulating me, nor are they imposing a choice I would
eschew. Indeed, it might be a smart move to deliberately enlist some more
of these innocent accomplices. Nor is my clever, courageous body acting
against my interests or wishes “ there™s nothing about the post-accident
descent that I would have done differently if I had wasted time thinking
over the advantages and disadvantages of functional arms and hands or
plotting each step. In contrast, I would never choose to be terrorized into
ful¬lling my charitable obligations. Finally, the City Harvest Robin Hood
is a stranger, and this mugger could not have known whether I needed
that money for some other compelling purpose. But the self-as-relational
and the self-as-embodied are not strangers. Indeed, they are more than
acquaintances. They are aspects of me, aspects of my identity, aspects of
who I am.12 The same goes for the self-as-social and the self-as-divided.
All in all, then, the gulf between me and effective agentic power seems
far wider in the City Harvest Robin Hood scenario than it does in my
real-life experiences.
Diana Tietjens Meyers

For these reasons, I think it would be unwise to short-circuit inquiry
by declaring my experiences heteronomous. Still, it is not clear how
autonomy-defeating, alien motivations differ from autonomy-preserving,
authentic ones. The succeeding sections of this chapter propose a way to
distinguish the autonomous agency of the self-as-relational, as-embodied,
as-divided, and as-social from nonautonomous behavior arising from
these dimensions of the self.

III The Agentic Skills of the Five-dimensional Subject
Mere doings are not autonomous. Nor is aimless, self-defeating, or sub-
servient behaving.13 As a ¬rst step toward persuading you that conduct
stemming from the self-as-social, the self-as-relational, the self-as-divided,
or the self-as-embodied can be autonomous, I urge that attending to
these dimensions of selfhood brings to light some neglected agentic skills.
Moreover, I urge that these skills endow people with forms of practical
intelligence that can be seen to facilitate self-discovery, self-de¬nition,
and self-direction. If this is so, it seems to me that we cannot dismiss
the possibility that the self-as-unitary is not the preeminent arbiter of
Skills are forms of know-how. There are standards of performance “
in deep water there™s a big difference between a non-swimmer and a
swimmer, and in any pool there™s a big difference between an average
recreational swimmer and an Olympic contender. Skills can be taught,
and they can be practiced, cultivated, and improved. Even a rudimentary
skill like walking, which babies seem to pick up with minimal adult assis-
tance, is sometimes painstakingly taught “ for example, in the aftermath
of a severe spinal injury, the victim might need a physical therapist™s “tu-
toring” to regain the ability to walk. Skills can be exercised thoughtfully,
but they need not be. You can just swim without thinking about it, or you
can concentrate on perfecting your butter¬‚y stroke. Pro¬ciency enables
people who possess a skill to correct their own mistakes “ for example, a
pianist senses her lagging tempo and picks up the pace. Likewise, pro¬-
ciency enables people to adapt to varying circumstances “ for example, a
dietician customizes menus to suit particular nutritional needs. So, skills
are learnable, improvable, ¬‚exible, standard-governed abilities to engage
in different types of activity, and autonomy skills are skills that contribute
to self-discovery, self-de¬nition, and self-direction.
Turning to the social self and our experience of enculturation, it is nec-
essary to recall the role of cultures in people™s lives. A culture encodes a
Decentralizing Autonomy: Five Faces of Selfhood 37

collective intelligence “ the accumulated wisdom of a social group cou-
pled with its share of folly and falsity. Cultures prescribe ways to meet
ineluctable needs, they disseminate models of lives well lived, and they
furnish a worldview that enables people to experience life as meaningful.
The self-as-social is imbued with this collective intelligence. Of course, it is
often pointed out that from the standpoint of autonomy that is precisely
the trouble. The social self is too imbued with this collective intelligence
to be autonomous. In my view, however, this objection rests on an impov-
erished conception of culture, on a misunderstanding of enculturation,
and perhaps on a misunderstanding of autonomy as well.
To be autonomous, one needn™t be outlandish. For many people, liv-
ing autonomously means living a fairly conventional life.14 In these cases,
it is obvious how cultures help to secure conative resolve. Insofar as one™s
values and projects coincide with a culturally entrenched way of life, one™s
social context powerfully reinforces one™s resolve to live up to those values
and carry out those projects. Still, it is important to notice that lending
its imprimatur to certain ways of life is only a small part of a culture™s
involvement in volition. Cultures do not merely impart doctrines. They
also impart skills, including skills that enable people to seek and obtain
social approval and, if not approval, tolerance. Thus, a cultural envi-
ronment integrates the self-as-social in practices of self-revelation and
self-justi¬cation that afford opportunities to test one™s values and aspira-
tions and that solidify one™s resolve, whether or not social endorsement
is ultimately forthcoming.15
That cultures contribute to the capacity for social dissent may come as a
surprise if one pictures cultures as exquisitely coherent systems of beliefs
and practices embalmed in amber. But since a static culture is a dead
culture, thriving cultures have built-in mechanisms of change.16 To be a
cultural initiate, then, is to know how to use these mechanisms “ that is,
to know how to resist uncongenial cultural norms and defective cultural
values. Thus, cultures endow the self-as-social with resistance skills as well
as resolve skills, both of which are integral to autonomy.
Interpersonal relationships are also implicated in people™s capacity for
autonomy. Feminist consciousness-raising is a paradigm of self-discovery,
self-de¬nition, and self-direction. Through the synergy of pooled mem-
ories, dreams sparking off each other, and energizing solidarity, the rela-
tional selves participating in these groups became preternaturally smart,
visionary, and willful.17 In other words, the interpersonal skills of the self-
as-relational transformed women™s seemingly personal complaints into
political critique and oppositional activism. We ¬nd the same skills in
Diana Tietjens Meyers

play on a smaller scale. A friend may discern an intimate™s distress before
she consciously registers it herself, and sometimes the friend understands
the distress and grasps what needs to be done about it far better than the
sufferer does. Listening to a friend can jump-start autonomy or prevent
autonomy from ¬‚agging. Indeed, I doubt that autonomy can survive in an
interpersonal vacuum. If the tabloid press has any use at all, it is to record
an endless stream of evidence that opting for extremely attenuated social
relations is not especially conducive to autonomy “ people™s autonomy
skills get rusty and languish for want of interaction with others, and crazy
or vicious ideas take root more easily.
People frequently depend on friends to bolster their resolve to under-
take a daunting, but needed change of direction or to persevere despite
discouragement with a project. But receptivity to explicit suasion is by
no means the whole of the volitional structure of the self-as-relational.
As we have seen, constellations of companionship can in themselves con-
stitute individual resolve, and they need not be set up deliberately. As
with the collaborative insights of consciousness-raising groups, volitional
structures can arise through the dynamic of interaction.
Now, it might seem that I am describing a passive subject rather than
an autonomous subject “ someone who absorbs and yields to other peo-
ple™s ideas, not someone who is living by her own lights. But few, if any,
real people are relational sponges. Discriminating receptivity to others™
criticism, reassurance, and advice is a skill that can be de¬cient in either
of two respects “ one can have too little facility in distinguishing helpful
from unhelpful input or too little facility in assimilating the bene¬ts of
others™ perceptions. Here, it is worth remembering that relationships are
(or should be) cooperative endeavors. Individuals exercise interpersonal
skills that shape their relationships, and thus they have some control over
the trustworthiness of their associates as partners in autonomy. If people
express their values, needs, interests, and so forth in fashioning their
relationships, they not only act autonomously in maintaining these ties,
but they also minimize the risk of the relational self™s receptivity.18 Under
these circumstances, exercising discriminating receptivity to others is less
dangerous and more likely to augment autonomy.
Intelligence is not exclusively a property of the conscious mind. Un-
conscious thought processes work on abstract or personal problems while
we sleep; unconscious memory processes post reminders and revives
lost meanings; unconscious imaginative processes generate fantasies that
can open liberating possibilities; unconscious self-evaluative processes
issue warnings manifested in pangs of guilt and in outbreaks of anxiety,
Decentralizing Autonomy: Five Faces of Selfhood 39

frustration, boredom, agitation, confusion, and the like. Moreover, when
a person™s values, goals, and commitments become embedded in subcon-
scious desire and affect, that individual™s determination to stay the course
is less liable to falter.
The unconscious mind is plainly a resource for self-knowledge, self-
de¬nition, and self-direction. But it might be objected that none of the
phenomena I have singled out involve skills. They are merely psychocor-
poreal incidents that require interpretation and psychocorporeal struc-
tures that happen to back up authentic values, goals, or commitments.
Although the actual operation of these processes is mysterious and largely
beyond our control, some of them “ notably memory, imagination, and
self-evaluation “ are subject to deliberate cultivation. Also the operation
of these processes, though not always predictable, is patterned, and can be
judged more or less availing.19 These features of unconscious processes
bring them into the ambit of skills. But the case for the contribution
of the self-as-divided to autonomy need not depend on the admittedly
debatable claim that the unconscious mind is by itself a repository of
skills, for consciousness is as much a part of the self-as-divided as the
unconscious. Thus, it is this whole system “ the self-as-divided “ that is
skilled in selectively appropriating unconscious materials, and it seems
clear that the interpretive, re¬‚ective, and intention-forming skills of this
system are needed for autonomy. Now, it goes without saying that the
self-as-divided is not an unalloyed blessing from the standpoint of au-
tonomy. Unconscious processes can defeat autonomous plans, as well as
support them. But since there are numerous accounts in the psycholog-
ical literature of how autonomy is possible for a divided self, I shall not
pause to rehearse them here or to argue that a divided self can gain
Foucauldians will ¬nd nothing to quarrel with in the claim that the
body is a site of skills that perpetuate the social order but that also gen-
erate resistance to it. For Foucault, disciplinary regimes inscribe social
identities on the body by instilling styles of comportment and bodily rou-
tines that enforce these identities. Yet, deviations from these disciplines “
and deviations are inevitable, in Foucault™s view “ constitute opposition
to the status quo.
I mention Foucault because I agree that characteristic deportment and
demeanor express what one is like and because I agree that ingrained
bodily con¬gurations and habitual bodily practices help to preserve one™s
sense of self. It is well known how profoundly disorienting alienation from
the body brought on by physical pain, illness, or injury can be.21 It is also
Diana Tietjens Meyers

worth underscoring the fact that somatic discontent and damage can be
excellent barometers of injustice and potent catalysts for resistance.22
Feminists have called for an end to women™s sexual deprivations and
vulnerabilities, and advocates of workers™ rights have organized against
backbreaking and repetitive labor. In a related vein, I also agree with
Foucault that strategic refusals to replicate normalizing bodily conven-
tions can pose a sharp challenge to an oppressive social order. Never-
theless, I do not propose to rely on Foucault to make my case for the
embodied self™s contribution to autonomy, for the theory in which he
couches his insights about the self-as-embodied makes it hard to see how
individuals could assert control over their deviations from entrenched
disciplines and thus hard to see how they could autonomously rede¬ne
themselves or overcome oppression.
What is missing from Foucault™s account, as I understand it, is an appre-
ciation of the practical intelligence of the skilled embodied self. Think
of how subtle messages delivered through body language can be, and
remember that body language is a skill that people seldom exercise self-
consciously. Also, consider why self-defense training helps traumatized
sexual assault victims recover.23 It gives them reason to believe that they
are safer because they feel con¬dent that their bodies would assuredly
and forcefully react if they should ever be attacked again. Notice, how-
ever, that if these individuals had reason to fear that the self-as-embodied
would be prone to lose control and misuse its ¬ghting techniques, say,
by aggressing against loving partners or children in their care, this new
capability would not be much of a comfort. It is both because the self-
as-embodied has acquired a crucial form of practical intelligence “ not
just a batch of hand-to-hand combat moves “ and because self-defense
training increases the likelihood that the individual will act on her de-
sire to protect herself that this physical skill seems constitutive of victims™
autonomous agency.24 If the skills of body language and self-defense are
typical of the skills of the self-as-embodied, there is no reason to exclude
the self-as-embodied from autonomous volition.
In highlighting the agentic skills of the self-as-social, as-relational, as-
divided, and as-embodied, I am not denying that autonomous people
need the rational skills of the self-as-unitary. Autonomous people often
call upon instrumental reason to ¬gure out how to achieve their goals,
and they use abstract reasoning skills to notice, to assess, and sometimes
to resolve con¬‚icts within their value systems. What I am questioning is
that these skills suf¬ce to account for autonomy and that exercising these
skills is always necessary to achieve autonomy.
Decentralizing Autonomy: Five Faces of Selfhood 41

IV Reckoning with Anomalous Autonomy Phenomenology
At this point, it might be acknowledged that, like the self-as-unitary, the
self-as-social, as-relational, as-divided, and as-embodied are sites of agency-
enabling skills, yet it might be doubted that the skills I have inventoried
are autonomy skills. Since autonomy may seem to require deliberate self-
direction, and since so many of these skills operate with little or no con-
scious supervision, they may seem like poor candidates for inclusion.
In this section, I consider two ways to account for the autonomy of my
dietary control and my mountain descent without invoking these skills.
Speci¬cally, I ask whether the idea of retrospective autonomy or the idea
of personal style can circumnavigate the problem posed by the subcon-
scious, unmonitored functioning of these skills.
It seems undeniable that people sometimes spontaneously act in atyp-
ical ways, and that in retrospect they realize that this devil-may-care mo-
ment revealed a previously submerged, yet highly desirable, potential-
ity, one that the individual regrets not actualizing in the past and very
much wants to actualize more fully in the future. Limiting autonomy
to pre-authorized action would deny that such spontaneous departures
from critically examined and certi¬ed patterns of behavior could be au-
tonomous. Serendipity and surprise would be expelled from autonomous
life. Only after such anomalous behavior had been scrutinized and judged
to be expressive of authentic traits, affects, values, and desires could sim-
ilar future behavior count as autonomous. Since this exiguous view is
vulnerable to the familiar objections that autonomy valorizes bourgeois
planning and stability and masculinist rationalism,25 an account of ret-
rospective autonomy “ that is, critically re¬‚ecting on past conduct and
validating it after the fact “ is indispensable.
John Christman proposes a promising theory of retrospective valida-
tion. For Christman, an agent is autonomous with respect to a desire
“if the in¬‚uences and conditions that gave rise to the desire were factors
that the agent . . . would not have resisted had she attended to them,” and
the agent making the judgment about these in¬‚uences and conditions is
minimally rational and not self-deceived.26 In other words, authentic past
desires are desires that were formed by processes one now freely accepts.
I am not at all sure that Christman™s theory of retrospective autonomy
would certify my mountain descent and my gustatory inhibition as au-
tonomous. Of course, I™m glad I somehow got to be the sort of person
who could make her way down the snow¬eld, and I am not sorry that I
somehow became the sort of person who is sensitive to others™ opinions
Diana Tietjens Meyers

of me. I have to confess, though, that I don™t have a very clear idea of
what caused me to turn out this way. Moreover, when I speculate about
the in¬‚uences and conditions that may have given rise to my capacities
and character, I ¬nd much to criticize, for I grew up in a fairly typical
1950s Euro-American, middle class, patriarchal nuclear family. Within
the household precincts, my father™s authority was unquestioned and
unshared. He modeled decisiveness and competence, and he strictly sup-
pressed displays of weakness in his children. He set high standards, and he
bestowed approval sparingly. No doubt, these aspects of my upbringing
greatly in¬‚uenced the way I™ve turned out, and they seem especially rel-
evant to the examples of embodied autonomy and relational autonomy
I™ve proposed. Since I would not choose to raise children in a similarly
inegalitarian and censorious environment, however, it seems to follow
that neither of my cases quali¬es for autonomy.
Now, Christman might ask whether overall I disapprove of the way I was
raised, and I would readily acknowledge that I do not. In innumerable
respects, I was extremely fortunate to have had the parents I had and
the upbringing they gave me. But does that mean that everything I do
now is autonomous since my upbringing was not unequivocally bad? Or
does it mean that nothing I do now is autonomous since it is impossible
to distinguish those of my present actions that are caused primarily by
deplorable in¬‚uences or conditions that I now wish I had resisted from
those of my actions that are not so tainted. Evidently, our understanding
of desire and capability formation is far too crude to draw the distinctions
Christman™s account of retrospective autonomy requires.
But what if our social-psychological knowledge was less fragmentary
and conjectural? I would remain skeptical that a subtle and reliable the-
ory of desire and capability formation would settle whether the episodes
I have sketched were instances of autonomy. If this high-powered theory
revealed that the worst features of my childhood experience were the
principal factors responsible for my mountain descent and my avoidance
of unhealthy foods, I would wonder whether I should have resisted those
noxious in¬‚uences and opprobrious conditions. After all, had I resisted,
I might be maimed today, or my health might be rapidly deteriorating.
Perhaps I would be better advised to reject Christman™s account of retro-
spective autonomy, which could omit some actions that further my core
values and that further these values in ways that I have no reason to re-
What I would like to suggest here (and what I shall argue in Section V)
is that a model of autonomy that centralizes competency and authority
Decentralizing Autonomy: Five Faces of Selfhood 43

in the rational oversight functions of the self-as-unitary underestimates
the role of self-discovery and overestimates the role of self-de¬nition in
autonomy.27 Theories of this sort assign self-discovery an exclusively in-
strumental role. Self-examination is a necessary preliminary to critically
evaluating one™s past social background or one™s present attributes. Also,
one must recognize one™s temptations and weaknesses if one is to take
steps to counteract them and keep them from thwarting enactment of
authentic traits, affects, values, and desires. But why should self-discovery
be relegated to these ancillary roles in autonomy?
I suspect that the impulse to subsume self-discovery under self-
de¬nition is symptomatic of a misguided con¬‚ation of socialization with
indoctrination and the correlative con¬‚ation of self-determination with
self-creation. If I cannot actually cleanse myself of social input and create
myself from scratch, at least I can approximate this ideal by rationally
de¬ning myself “ that is, by exposing and evaluating my attributes and by
¬guring out how to accent my strengths and improve what I ¬nd lacking.
Now, I do not deny that there is a place for such “self-management” in the
autonomous life. That is why I have included the self-as-unitary among the
dimensions of the autonomous subject. In my view, however, it would be
a mistake to assume that self-de¬nition must always take precedence over
self-discovery in autonomous living and that self-de¬nition must always
take this cerebral form.
Perhaps it is not always necessary to authenticate traits, affects, values,
and desires in rational self-de¬nitional re¬‚ection. Perhaps people have
authentic traits, affects, values, and desires that they discover in acting.
Perhaps people autonomously de¬ne themselves in part by enacting these
discovered traits, affects, values, and desires.
It seems to me that a key virtue of “personal style” theories of autonomy
is that they do not over-emphasize self-de¬nition through critical re¬‚ec-
tion. For example, Richard Double™s “individual management style” the-
ory requires only that choices be in keeping with a person™s characteristic
decision-making method to count as autonomous. Different people have
different “individual management styles” “ some like to chart their course
in advance and work steadily toward their goals; others like to play the
odds and see where life takes them; some like to rely on personal precepts
and ideals to ¬gure out what to do; others like to turn to religion or some
other authority for guidance; and so forth.28 In Double™s view, to make
choices in one™s characteristic way, whatever that is, is to be autonomous.29
Double™s latitudinarian account neutralizes the charge that autonomy
is the province of dull plodders and hyper-rationalists by minimizing the
Diana Tietjens Meyers

role of re¬‚ective self-de¬nition in autonomy. Moreover, it honors indi-
vidual uniqueness. But can this go-with-the-¬‚ow theory pick out actions
stemming from authentic traits, affects, values, and desires?
Consider the experiences I described in Section II. In one sense, both
of them were, for me, altogether extraordinary. I had never before been
obliged to deprive myself of any culinary pleasures to speak of. I had never
been alone and seriously injured in a hazardous environment before.
Thus, it is dif¬cult to say what my “individual management style” is in such
predicaments. Since Double acknowledges that one might have different
characteristic ways of making different sorts of decisions “ one might
scrupulously calculate investment decisions, but see whatever movie hap-
pens to be playing at a convenient theater and time “ the indeterminacy
of one™s characteristic decision-making style in unprecedented situations
is not a trivial objection to his theory.30 Just when it seems autonomy mat-
ters most of all “ that is, in an emergency or other exceptional situation
in which what you do really counts “ Double™s theory falls silent.
But maybe I™m being unfair. Perhaps there was more continuity be-
tween my everyday individual management style and my responses to
these unprecedented situations than I have admitted. I do tend to trust
my body. I generally assume that I have enough strength, agility, and
coordination to carry me through, although you will be forgiven if you
are thinking that in light of my multiple hiking injuries this must be a
case of delusional overcon¬dence. So it seems that there was no man-
ifest con¬‚ict between my reliance on my physical competence on Mt.
Rainier and my usual decision-making practices. My dispersal of appetite
control, however, is rather aberrant for me. I™m generally pretty good at
self-discipline and pretty self-reliant about staying on course and living up
to commitments. Indeed, it bothers me a bit that I could not muster the
willpower to resist tempting foods entirely on my own. Thus, it seems that
if Double™s theory has anything to say about my experiences, it would pro-
nounce my mountain descent autonomous and my refusing forbidden
foods heteronomous.
I am not convinced that this conclusion would be right, however. I
suspect that insofar as it seems right, it is because the locus of control in
the Mt. Rainier case is within my individual unit “ my body is ontologically
part of me “ whereas in the dietary restriction case, the locus of control
extends beyond my individual unit “ other people are not ontologically
part of me. If it is possible, however, that, in the sense of identity that is
germane to the issue of autonomy, I am just as much a relational self as I
am an embodied self, this metaphysical truism is irrelevant.
Decentralizing Autonomy: Five Faces of Selfhood 45

On Double™s view, then, whether my relationally assisted compliance
with my diet is autonomous must turn on whether relying on others™
opinion of me to motivate myself is altogether alien to my quotidian in-
dividual management style, and of course it is not. Like most people,
I care what others think, and that in¬‚uences what I do. But notice that
Double™s theory now seems to be conferring autonomy a little too promis-
cuously. Since people™s individual management styles are typically quite
dense and ¬‚exible, it is hard to imagine a realistic case that Double™s
theory would decisively pronounce heteronomous. Only a person who is
living an out-and-out caricature of a particular individual management
style could act in ways that would be disquali¬ed as autonomous. For nor-
mal people, the distinction between the authentic and the inauthentic
The trope of the self-made man limns conceptions of self-de¬nition,
and its contrary, the trope of ¬nding yourself, limns conceptions of self-
discovery. What I have sought to show in my discussion of Christman™s and
Double™s views is that tipping the balance toward either of these images
yields an untenable account of autonomy. Of course, no serious student
of autonomy takes either of these tropes literally, and common sense tells
us that self-discovery and self-de¬nition are intertwined. Still, the prob-
lems I have pointed to in Christman™s and Double™s views suggest how very
dif¬cult it is to keep self-discovery and self-de¬nition in balance. Overem-
phasizing critical self-analysis and self-de¬nition disquali¬es conduct that
enacts traits, affects, values, and desires that could only be disavowed at
one™s peril. Overemphasizing self-discovery and uncritical self-acceptance
leaves us without resources to identify conditions of self-alienation and
acts of self-betrayal.

V Balancing Self-Discovery and Self-De¬nition
What makes us think that we ever enact authentic traits, affects, values,
and desires? How does one ever know what one really cares about? Which
commitments are one™s own? Whether one™s life accords with one™s true
People who are innocent of postmodernism, and many clinical psy-
chologists, associate autonomy with feelings of wholeness “ in colloquial
terms, feeling in touch with oneself, feeling at one with oneself, and feel-
ing right in one™s skin.31 From a phenomenological perspective, then,
what is distinctive about enacting authentic traits, affects, values, and de-
sires, is that doing so, whether in a particular situation or throughout
Diana Tietjens Meyers

one™s life, gives people the sense of wholeness that is characteristic of
autonomy. Of course, individuals are supplied with an almost constant
stream of visceral and affective feedback referencing their conduct. For
autonomous people, the predominant tenor of this feedback runs the
gamut from steady equanimity and low-key satisfaction to occasional in-
candescence and zingy exhilaration. In the aggregate, these positive feel-
ings anchor a con¬dent sense of who one is, of one™s worthiness, and
of one™s ability to translate one™s traits, affects, values, and desires into
acceptable conduct “ in short, a sense of wholeness.32
In contrast, for people who ¬nd some of their traits less than admirable,
who are not sure what matters to them, who are uneasy about their re-
lationships, or who feel overpowered by desires, much of this feedback
is far from reassuring. Emotional disquiet “ anxiety, confusion, anger,
humiliation, frustration, discouragement, exasperation, embarrassment,
guilt, shame, and so on “ impugns autonomy. And so does bodily distress “
restlessness, tensed muscles, headache, fatigue, tearfulness, palpitations,
and the like. Nor should we overlook the fact that complacency signals
inveterate inattentiveness, if not obtuseness that bespeaks questionable
autonomy, too. Dissonant cues such as these, together with af¬rming cues
on the satisfaction-exhilaration spectrum, comprise the expressive vocab-
ulary of the self-as-social, as-relational, as-divided, and as-embodied. This
visceral and affective vocabulary is a trenchant vehicle for communicating
avowal and disavowal and for advocating either persisting in or altering
one™s course. Autonomous people are attuned to and responsive to these
messages. Reports of self-alienation or poor ¬t between self and action
prompt self-monitoring, possibly leading to change.
Sometimes negative affective or visceral cues initiate an arduous pro-
cess of analysis and self-questioning that may ultimately persuade the
individual to craft a program of self-rede¬nition. Philosophers typically
focus on this enterprising sort of self-transformation. As often as not,
however, the import of these cues is assimilated and integrated into the
individual™s agentic infrastructure without conscious mediation, and the
individual™s subsequent conduct re¬‚ects this adjustment. In addition, it is
necessary to bear in mind that positive cues are as important as negative
ones. They let us know what we are doing right.
Now, someone might object that turning autonomy over to the social
self, the relational self, the divided self, and the embodied self blunts
autonomy™s critical edge. Many conformists feel ashamed when they ab-
rogate pointless customary practices. Aren™t their social selves defending
the status quo? Many U.S. mothers feel anxious and guilty when they work
Decentralizing Autonomy: Five Faces of Selfhood 47

outside the home. Aren™t their relational selves telling them to con¬ne
themselves to domesticity? Many sexists feel confused or angry when
women bring charges of sexual harassment. Aren™t their divided selves
arguing for women™s subordination? Many bigots™ bodies knot up when
they ¬nd themselves among African Americans. Aren™t their embodied
selves opposing racial integration? In short, doesn™t autonomy mandate
rationally probing these affective and visceral responses?
I hasten to point out that mobilizing critical reason by no means guar-
antees that these people will change their minds and adopt less reac-
tionary views. Critical reason does not confer sensitivity to affective and
visceral cues, nor does it ensure insight into their import. Still, it is unde-
niable that rationally examining these psychocorporeal responses would
at least provide an opportunity to grasp the harmfulness of the views
with which they are linked. So, let me reiterate that I have not excluded
the reasoning skills of the self-as-unitary from my conception of the au-
tonomous subject, and I have no doubt that rational re¬‚ection can be
salutary. Still, I would deny that rational re¬‚ection is always essential to
The objection under consideration and the suggestion that autonomy
cannot be salvaged without critical reason are premised on a mistaken
view of the relation between social doctrines and subjective responses
as well as an oversimpli¬ed view of subjectivity.33 There is no one-to-
one correspondence between social norms and practices, on the one
hand, and affective or visceral responses, on the other. An anxious, guilty
working mother need not read her feelings as an argument for 1950s
style homemaking. She could just as well read her feelings as an argument
for on-site daycare. Moreover, subjectivity is far from homogeneous. No
one™s affective and visceral responses are altogether harmonious. Even
the diehard bigot has probably had pleasant encounters with African
Americans now and again. Thus, autonomous individuals cannot escape
the need to negotiate the con¬‚icts among their affective and visceral
responses and to separate authentic traits, affects, values, and desires
from inauthentic ones.
Still, it must be admitted that many people have a prodigious capacity
to suppress disconcerting feelings, to rationalize misjudgments, and to
excuse blunders. That is why a sense of wholeness is not suf¬cient for
autonomy. That is why agentic skillfulness is necessary as well. There is
no reason, however, to limit our conception of this skillfulness to critical
reason, for to do so would be to ignore the complexity and scope of the
skills of the self-as-social, as-relational, as-divided, and as-embodied.
Diana Tietjens Meyers

Exercising the latter skills is an ongoing process of self-reading and
self-con¬guring. The skilled self-as-social registers convergences and
clashes with cultural norms, accounts for convictions and conduct when
appropriate, and revises these accounts as necessary. The skilled self-as-
relational elicits, internalizes, and deploys candid reactions and sympa-
thetic counsel from associates. The skilled self-as-divided retrieves, sym-
bolizes, and interprets subjective material. The skilled self-as-embodied
senses inclinations as well as needs, micro-manages itself to meet perfor-
mance standards, and maneuvers to achieve goals.
My suggestion is that autonomous people have a diverse, well-
developed, well-coordinated repertoire of agentic skills that they exercise
routinely and adeptly. Moreover, I am suggesting that in being repeatedly
enacted under the auspices of these agentic skills, a trait, affect, value, or
desire is reviewed and re-reviewed, and its authenticity is validated. If I am
right, it follows that the presumption that people™s cultural, relational,
intrapsychic, and bodily endowment is alien and must be overcome or
rationally mastered to attain autonomy is mistaken. Provided that peo-
ple have developed reasonable facility in exercising agentic skills, their
everyday choice-making and action authenticate or disown elements of
this endowment.
It is characteristic of skills that the greater one™s pro¬ciency, the more
rapidly and successfully one contends with variable conditions, recov-
ers from lapses, and corrects one™s mistakes. Like the agentic skills of
the self-as-unitary, the agentic skills of the self-as-social, as-relational, as-
divided, and as-embodied keep familiar traits, affects, values, and desires
in full view, disclose unrecognized attributes, notice problematic self-
enactments, and devise and carry out corrective measures. Self-discovery
is not exclusively an analytical introspective and interpretive project. We
discover much about who we are in doing what we do. Self-de¬nition is
not exclusively a project of critical re¬‚ection and recon¬guration. We
de¬ne ouselves as we act, and we cannot rede¬ne ourselves without al-
tering our patterns of action. Self-discovery and self-de¬nition can, but
need not be, intentional undertakings. Thanks to the agentic skills of the
self-as-social, as-relational, as-divided, and as-embodied, one may ¬nd out
who one is, and one may reaf¬rm, renew, revamp, recondition, or repair
oneself as one acts and interacts.
Neither living skillfully nor feeling whole suf¬ces for autonomy. The
requirement of agentic skillfulness counters the objection that one may
feel good about one™s life and yet remain utterly oblivious to the damage
one causes and deluded about the esteem one™s efforts and attainments
Decentralizing Autonomy: Five Faces of Selfhood 49

deserve. In virtue of agentic pro¬ciency, one™s self is being constituted
and reconstituted in an ongoing and intelligent way. Thus, there is no
reason to distrust positive affective and visceral feedback and no reason
to suspect that one™s sense of wholeness attests to rampant self-deception.
The requirement of feeling whole counters the objection that one can
act skillfully and live a lie. People who are estranged from themselves “
whether by choice, negligence, or ineptitude or because they are forced
to capitulate to an inimical social context “ lack this sense of wholeness.
Their lives fail to mesh with their authentic self, and they feel the loss.
Turning this point around, since exercising agentic skills well typically
confers a lively awareness of oneself and others together with a robust
sense of engagement with others while fully inhabiting oneself, it is no
wonder that someone who uses these skills adeptly would develop a sense
of wholeness.
Elsewhere, I have urged that autonomous people exercise a reper-
toire of skills to engage in self-discovery, self-de¬nition, and self-direction,
and that the authentic self is the evolving collocation of attributes that
emerges in this ongoing process of re¬‚ection, deliberation, and action.34
Here, I have argued that the agentic skills of the self-as-social, as-relational,
as-divided, and as-embodied, along with those of the self-as-unitary, be-
long among the re¬‚ective, deliberative, and volitional skills that com-
prise autonomy competency, for these agentic skills give rise to choices
and actions that tap authentic attributes. In exercising these skills, one
constitutes and enacts one™s authentic self.35
I readily concede, though, that none of the skills I identi¬ed in Sec-
tion III infallibly taps into authentic traits, affects, values, and desires,
and that no one can completely avoid waywardness and self-betrayal. But
privileging the reasoning skills of the self-as-unitary would not solve this
problem and would leave us with an account of autonomy that is inappli-
cable to a vast array of circumstances in which autonomy is badly needed.
Pro¬ciency with respect to agentic skills is a matter of degree. Most peo-
ple have a pretty good idea of how pro¬cient they are in various respects,
and they can work on improving weak skills if they want to. Thus, one™s
autonomy is a function both of one™s overall level of facility with respect
to autonomy competency and of how successfully one uses these skills
on any given occasion “ success being gauged by affective and visceral
All sorts of things can interfere with exercising autonomy competency,
or block it altogether. Some of these obstacles are peculiar to a particular
occasion or a particular person, but some of them are embedded in
Diana Tietjens Meyers

cultures and social structures. In the latter case, the value of autonomy,
together with the widespread desire to lead an autonomous life, provides
a prima facie reason to change those norms, practices, and institutions
that impede individual autonomy. It is at this juncture that autonomy
theory provides a platform for social critique. Although it is indisputable
that we must be satis¬ed with partially autonomous lives, we should not
reconcile ourselves to pervasively, intractably nonautonomous lives.

VI Decentralizing Autonomous Subjectivity and Agency
To avoid misunderstanding, it is necessary to call attention to a certain
arti¬ciality in my discussion. Because I have sought to link autonomy
skills to the self-as-social, as-relational, as-divided, and as-embodied, my
exposition might leave the false impression that these “selves” are com-
partmentalized agents of autonomy. But, on the contrary, an autonomous
person has a smoothly functioning repertoire of complementary auton-
omy skills, and adroitly calls on one or more of these skills as needed.
The unitary self, the social self, the relational self, the divided self, and
the embodied self are not ontologically distinct selves with no direct ac-
cess to each other.36 They merely “ and, I fear, cumbersomely “ signal
different sources of identity, different threats to autonomy, and different
autonomy-skill specialties.
Still, we should not be tempted to seek a reductionist account of the
autonomous subject or to pare down our view of autonomous agency just
because we have no theory that synthesizes the ¬ve conceptions of the self
that I have invoked. In fact, individuals have at their disposal a means to
accommodate the self-as-unitary, as-social, as-relational, as-divided, and
as-embodied in a single self-conception “ namely, the autobiographical
narrative. Unfolding life-stories weave together all of these disparate mo-
tifs with amazing ease.37 In fact, I am inclined to think that one rea-
son narrative accounts of selfhood have attracted so many exponents is
that they ¬nesse the incongruity of positing a ¬ve-dimensional self-as-
unitary, as-social, as-relational, as-divided, and as-embodied.38 However,
I also believe that to ¬nd such narrative accounts attractive is implicitly
to acknowledge the urgency of retaining all of these conceptions of the
self in some form.39 Thus, I see theories of the narrative self both as fur-
nishing a convenient way for ¬ve-dimensional subjects to articulate their
autonomy and as con¬rmation that autonomy theory must reckon with
Decentralizing Autonomy: Five Faces of Selfhood 51

1. Autonomy theorists do not agree, however, about what a theory of autonomy
should accomplish. David Velleman™s account, for example, seeks to distin-
guish “action from mere behavior and . . . from mere activity” (The Possibility
of Practical Reason. New York: Oxford University Press, 2000), p. 6.
2. This distinction is indispensable to feminist theory as well as to theories con-
cerned with other types of systematic social domination and subordination.
In this work, it is not enough for an account of autonomy to analyze the bases
for ascribing actions to individuals or for holding individuals responsible for
their actions. To account for both subordination and resistance to it, these
theories also need to be able to distinguish colonized consciousness and col-
laboration in one™s own subordination from emancipated consciousness and
autonomous choice and action.
3. Unity enters into accounts of autonomy in two different ways. In the Kantian
approach that John Rawls endorses, autonomy is traced to reasoning, the
hallmark of which is consistency “ that is, unity. In the Humean approach
that Harry Frankfurt endorses, autonomy depends on an integrated “ that
is, uni¬ed “ personality that need not be achieved through reason. In this
chapter, I use the term self-as-unitary to refer to the Kantian conception.
4. For discussion of this august legal tradition and the disturbing exception
to it that the courts have recently carved out for pregnant women and the
protection of the fetuses they are carrying, see Susan Bordo, “Are Mothers
Persons? Reproductive Rights and the Politics of Subjectivity.” In Unbearable
Weight (Berkeley, CA: University of California Press, 1993).
5. I want to emphasize, however, that illness, frailty, and disability by no means
preclude autonomy.
6. I defend this claim in “Narrative and Moral Life,” in Cheshire Calhoun, ed.,
Setting the Moral Compass (New York: Oxford University Press, 2003).
7. Here are a couple hints for thinking about how the self-as-social and the self-
as-divided might contribute to autonomy. As many social psychologists have
pointed out, enculturation commonly combines with social situations to as-
sume control over people™s conduct. Ingrained conventions of politeness, for
example, keep an assortment of sensitive topics out of many people™s conver-
sation at dinner parties. Often enough, the agent does not really want to do
anything different from what is customary. Also, many of you have probably
noticed, too, that the self-as-divided sometimes exhibits wonderful powers of
divination. I have often awakened from a deep sleep knowing exactly how
to deal with a vexing interpersonal or intellectual problem. Thankfully, my
unconscious mind has solved the problem for me, and I go ahead and do its
8. For a related discussion of interpersonal support in relation to agency, see
Susan J. Brison, “Outliving Oneself: Trauma, Memory, and Personal Auton-
omy.” In Diana Tietjens Meyers, ed. Feminists Rethink the Self (Boulder CO:
Westview Press, 1996).
9. As autobiographical narratives are wont to do, this story has spawned
subplots since I originally drafted this chapter. However, I shall not go into
Diana Tietjens Meyers

these complications, for they do not bear on the philosophical point implicit
in my earlier narrative.
For these observations about agency and the body, I am indebted to a con-
versation with Elise Springer in which she remarked that she thinks of some
practices of evaluation as being “in the body.” Her comment made a strong
impression on me and prompted me to think about the body as a locus of
George Sher argues that such actions are autonomous provided that the
agent is responsive to reasons, such as safety is more important than property
(“Liberal Neutrality and the Value of Autonomy,” Social Philosophy and Policy
12 (1995): 136“159). Whatever the merits of his arguments, however, it
would be question-begging in the context of the issues I am raising to agree
that coercion and autonomy are compatible.
This point also distinguishes my mountain descent as it actually happened
from the following scenario. Suppose that my accidents cause me to become
paralyzed with fear. Luckily, a Saint Bernard (recruited and trained to pa-
trol Mt. Rainier National Park in order to “downsize” the ranger force and
save money) ¬nds me. Toting a brandy cask and pulling a sled, the Saint
Bernard anesthetizes me with drink, nudges me onto the sled, and takes me
to the ranger station. Awakening later, I would undoubtedly thank the Saint
Bernard for hauling me to safety and medical treatment. But for the same
reasons that my Robin Hood-induced donations are not autonomous, and
for additional reasons that I develop later in this chapter, my descent would
not be autonomous.
But for a contrary view, see Paul Benson™s chapter (5), in which he argues
that “trivial” behavior can be autonomous. For Benson, one is autonomous
when one “takes ownership” of one™s actions. In contrast, my account accents
For a critique of the idea that autonomy requires eccentricity or rebellion
and defense of autonomous conventionality, see Diana Tietjens Meyers, Self,
Society, and Personal Choice (New York: Columbia University Press, 1989),
p. 75.
I think this line of thought adds weight to Paul Benson™s suggestion that
“normative competence” is necessary for autonomy (“Free Agency and Self-
worth,” Journal of Philosophy (1994): 650“668 at 660“663).
It goes without saying that cultures also have built-in mechanisms of self-
perpetuation. I discuss the tension between cultural stability and cultural
transformation as well as the tension between cultural stability and individual
autonomy in “Feminism and Women™s Autonomy: The Challenge of Female
Genital Cutting,” Metaphilosophy 31 (2000):469“491.
See Naomi Scheman, “Anger and the Politics of Naming,” in Engenderings
(New York: Routledge, 1993).
Notice that whereas the City Harvest Robin Hood neither knows nor cares
about his victims™ needs and values, philanthropically gung-ho individuals
will/should temper their expressions of enthusiasm for charitable giving
when interacting with friends who need their money for other purposes. If
they persist in extolling the virtues of giving, and if their thoughtless zeal
Decentralizing Autonomy: Five Faces of Selfhood 53

is making their less af¬‚uent friends feel like guilty outsiders, the latter are
justi¬ed in asking them to turn down the volume. Nevertheless, they may turn
their friends™ seeming insensitivity into an opportunity to explore whether
they have struck the right balance between personal need and helping needy
others. Unlike the City Harvest Robin Hood, whose solicitation methods
thwart interpersonal skills and preclude interpersonal exchange, friendship,
even when it goes awry, can foster the autonomy of the self-as-relational.
There is more similarity between unconscious thought processes and reason-
ing than philosophers usually acknowledge. If rationality, much less practical
rationality, is not reducible to formal logical deduction, the outcome of rea-
soning processes is not predictable either. Likewise, reasoning may or may
not turn out to have been availing.
Nancy Chodorow provides a helpful discussion of some of these theories in
“Toward a Relational Individualism: The Mediation of Self through Psycho-
analysis,” in Feminism and Psychoanalytic Theory (New Haven: Yale University
Press, 1989).
This phenomenon is familiar in medical settings. But it is worth noting that
Jon Krakauer™s account of summiting Mt. Everest features the prolonged
and relentless physical disruption and discomfort the climbers endured and
links it to the moral dislocation and curtailment of personal agency that
contributed to the fatalities on the mountain during his expedition (Into
Thin Air (New York: Villard, 1997). It would be interesting to learn whether
studies of famines and similar calamities bear out this line of thought.
Susan Babbitt discusses how non-propositional knowledge of oppression can
be lodged in and expressed by the body (“Feminism and Objective Interests:
The Role of Transformation Experiences in Rational Deliberation,” in Linda
Alcoff and Elizabeth Potter, eds., Feminist Epistemologies (New York: Routledge,
1993), pp. 257“259.
Susan Brison discusses the role of self-defense skills in restoring the autonomy
of sexual assault victims in “Outliving Oneself,” p. 31.
Another case I ¬nd illuminating is one that medical ethicists and physicians
address. Hospitals strongly encourage patients to make out living wills, and
many patients declare that they do not want extraordinary measures taken
to prolong their life if there is no realistic hope of recovery. However, when
the need for some extraordinary measure arises and the patient is able to
consent to it or not, it is standard practice to ask again whether the patient
wants to be treated. Many patients reverse themselves at this moment of crisis,
and medical practitioners defer to their decisions in order to respect their
autonomy. It seems to me highly dubious that these patients have rationally
reviewed their values and priorities and ¬gured out what was wrong with
their earlier decisions. On the contrary, it seems to me that for many people,
the authentic supremacy of the value of continued life is embedded in their
bodies, their self-as-embodied revolts against the dry rationalism of their
earlier judgment, and their request for treatment is the autonomous self-as-
embodied speaking.
See Kathryn Addelson. Moral Passages: Toward a Collectivist Moral Theory
(New York: Routledge, 1994), Chapter 5; Margaret Walker, “Getting Out of
Diana Tietjens Meyers

Line: Alternatives to Life as a Career,” in Mother Time (New York: Rowman
and Little¬eld), 1999, pp. 97“106.
John Christman, “Autonomy and Personal History,” Canadian Journal of Phi-
losophy 21 (1991): 1“24 at 22. I am pleased to see that in his chapter for this
volume and elsewhere, Christman revises his position. He now holds that
“what matters is the person™s relation to the attitude or characteristic given
its etiology rather than her attitude toward that etiology (simpliciter).” Thus,
a person might not feel alienated from a character trait despite feeling alien-
ated from the process through which the trait was formed, and enactments
of the trait would be autonomous. I do not have space to discuss Christman™s
current view in the detail it deserves. But I would like to point out that the
concerns I set forth about our inability to isolate the etiologies of our traits
seem to apply both to Christman™s earlier view and to the view he develops
Christman now stresses (1) that alienation and nonalienation are affective
states, and therefore that his account of autonomy does not rely solely on the
rationality of the self-as-unitary, and (2) that the rational re¬‚ection required
for autonomy is undertaken on a need-to-know basis, and therefore that his
account of autonomy does not stipulate that the autonomous self must be
uni¬ed (see his chapter in this volume).
Richard Double, “Two Types of Autonomy Accounts,” Canadian Journal of
Philosophy 22 (1992): 65“80 at 68“69.
Ibid., p. 69.
Ibid., p. 69.
For example, see Chodorow, op. cit., p. 159.
Some accounts of autonomy call attention to affective states, such as feeling
powerful (Jennifer Nedelsky, “Reconceiving Autonomy: Sources, Thoughts,
and Possibilities,” Yale Journal of Law and Feminism 1 (1989): 7“36 at 23“26);
Self-Worth (Paul Benson, “Feeling Crazy: Self-Worth and the Social Charac-
ter of Responsibility,” in Catriona Mackenzie and Natalie Stoljar, eds., Rela-
tional Autonomy (New York: Oxford University Press, 2000), pp. 72“80); and
self-trust (Trudy Govier, “Self-Trust, Autonomy, and Self-Esteem,” Hypatia 8
(1993): 99“120 at 104“109). They claim that these states of mind empower
people to choose and act autonomously. But since feeling powerful can be
overblown, and since self-worth and self-trust can be unwarranted, I am con-
vinced that feeling this way is not a good index of autonomy unless these
re¬‚exive attitudes stem from exercising autonomy skills well. Without this
backing, such attitudes may be a better index of social advantage or of effec-
tive defenses against severe disadvantage. Thus, I would examine the sources
of these feelings before I attributed autonomy to an individual.
I suspect that this objection is also fueled by a failure to honor individuality. If
autonomous individuals enact unique authentic selves, we should not expect
uniformity in autonomous lives. To be sure, some autonomous individuals
join with like-minded associates and rebel against social ills. Many others
subvert the system and enact dissident values in less public and dramatic
ways. But some autonomous individuals ¬nd ways to express their sense of
self within existing social constraints. A theory of autonomy cannot dictate
Decentralizing Autonomy: Five Faces of Selfhood 55

the traits, affects, values, and desires of the authentic self, nor can it anticipate
the trajectory of individual autonomous lives.
Meyers, “Narrative and Moral Life,” pp. 53 and 76; also see my “Intersectional
Identity and the Authentic Self? Opposites Attract!” in Relational Autonomy,
172“173 and my Gender in the Mirror: Cultural Imagery and Women™s Agency
(New York: Oxford University Press, 2002), Chapter 1.
It may seem that this view locks us into a vicious circle. If the authentic self
has no existence apart from a person™s exercising autonomy skills, how can
we tell which skills are autonomy skills? How can we tell which skills enable
one to discover and shape one™s authentic self and to enact authentic values
and desires? It seems to me that the requirement of feeling whole provides
the leverage we need to resist this objection. Agentic skills that promote this
positive sense of self count as autonomy skills.
For a complementary treatment of identity through time, see Susan James,
“Feminism in Philosophy of Mind: The Question of Personal Identity,” in
Miranda Fricker and Jennifer Hornsby, eds., The Cambridge Companion to Fem-
inism in Philosophy (Cambridge: Cambridge University Press, 2000).
For a discussion of autobiographical narrative and autonomous subject, see
J. David Velleman™s Chapter 3 in this volume.
For a example, Richard Rorty, “Freud and Moral Re¬‚ection,” in Joseph Smith
and William Kerrigan, eds., Pragmatism™s Freud (Baltimore: Johns Hopkins
University Press, 1986), p. 18; Marya Schechtman, The Constitution of Selves
(Ithaca NY: Cornell University Press, 1996), pp. 93“135; Margaret Walker,
Moral Understandings (New York: Routledge, 1998), pp. 106“129; Seyla Ben-
habib, “Sexual Difference and Collective Identities: The New Global Con-
stellation,” Signs 24 (1999): 335“361 at 341“350; Hilde Nelson, Damaged
Identities, Narrative Repair (Ithaca, NY: Cornell University Press, 2001) p. 15.
My concern about narrative accounts of the self is that they tend to obscure
autonomy competency “ that is, the extensive repertory skills needed to
achieve and renew autonomy “ or, in other words, the repertory of skills one
must exercise in order to be in a position to tell the story of an autonomous
protagonist. Autonomy competency is not reducible to story-telling facility.
One can be a beguiling raconteur without being autonomous. I develop this
line of thought in “Narrative and the Moral Life,” op. cit.
I am grateful to Susan Brison, John Christman, Hilde Lindemann Nelson,
Margaret Urban Walker, and an anonymous reviewer for Cambridge Uni-
versity Press for their helpful suggestions about earlier drafts of this chap-
ter. I presented it at the conference on Reasonably Autonomous Persons:
Rationality, Neutrality, and the Self, which was sponsored by Washington
University and the University of Missouri, St. Louis, as the Irving Thalberg
Memorial Lecture at University of Illinois, Chicago, and at a colloquium of
the Dalhousie University Philosophy Department, and I am indebted to these
audiences for their comments.

The Self as Narrator

J. David Velleman

Many philosophers have thought that human autonomy includes, or per-
haps even consists in, a capacity for self-constitution “ a capacity, that
is, to de¬ne or invent or create oneself.1 Unfortunately, self-constitution
sounds not just magical but paradoxical, as if the rabbit could go solo and
pull himself out of the hat. Suspicions about the very idea of this trick
have sometimes been allayed by appeal to the political analogy implicit in
the term “self-constitution”: a person is claimed to constitute himself in
the same way as a polity does, by writing, ratifying, and revising articles of
constitution.2 But a polity is constituted, in the ¬rst instance, by its con-
stituent persons, who are constituted antecedently to it; and suspicions
therefore remain about the idea of self-constitution at the level of the
individual person.
One philosopher has tried to save personal self-constitution from sus-
picions of paradox by freely admitting that it is a trick. A real rabbit can™t
pull himself out of a hat, according to this philosopher, but an illusory
rabbit can appear to do so: the secret of the trick is that the rabbit isn™t
real. We ask, “But if the rabbit isn™t real “ and there™s no magician, either “
then who is performing the trick?” He replies, “Why, of course: the hat.”
A rabbit can™t pull himself out of a hat, but a hat can make it appear that
a rabbit is pulling himself out of it.
Notwithstanding my frivolous analogy, I think that there is much to be
learned from this view of self-constitution, and so I propose to examine
it in detail and to offer my own variation on it. The author in question is
Daniel Dennett, and his view is that the autonomous person (the rabbit)
is an illusion conjured up by the human organism (the hat).3 In the end,
I will adopt most of Dennett™s view, except for the part about the rabbit™s
The Self as Narrator 57

being unreal. In my view, the rabbit really does pull himself out of the
hat, after all.

Dennett™s metaphor for this process is not sleight-of-hand but ¬ction.
In Dennett™s metaphor, the self is the non-existent author of a merely
¬ctional autobiography composed by the human organism, which neither
is nor embodies a real self.4 So understood, the self has the status of an
abstractum, a ¬ctional object that we “use as part of a theoretical apparatus
to understand, and predict, and make sense of, the behavior of some very
complicated things”5 “ namely, human beings, including ourselves.
Dennett compares the human™s autobiography to the spider™s web or
the beaver™s dam:

Our fundamental tactic of self-protection, self-control, and self-de¬nition is not
spinning webs or building dams, but telling stories, and more particularly con-
cocting and controlling the story we tell others “ and ourselves “ about who we
are. [. . .] These strings or streams of narrative issue forth as if from a single
source “ not just in the obvious physical sense of ¬‚owing from just one mouth,
or one pencil or pen, but in a more subtle sense: their effect on any audience is
to encourage them to (try to) posit a uni¬ed agent whose words they are, about
whom they are: in short, to posit a center of narrative gravity. [RS, 418]

The point of this last phrase is that an object™s physical center of gravity
can ¬gure in legitimate scienti¬c explanations but mustn™t be identi¬ed
with any physical part of the object:

That would be a category mistake. A center of gravity is just an abstractum. It is
just a ¬ctional object. But when I say it is a ¬ctional object, I do not mean to
disparage it; it is a wonderful ¬ctional object, and it has a perfectly legitimate
place within serious, sober, echt physical science. [CNG, 104]

Similarly, the “uni¬ed agent” conjured up by our narrative is a theoretical
abstraction, but it too has a legitimate place in a serious theory. Dennett
concludes the analogy as follows:

[W]e are virtuoso novelists, who ¬nd ourselves engaged in all sorts of behavior,
more or less uni¬ed, but sometimes disuni¬ed, and we always put the best “faces”
on it we can. We try to make all of our material cohere into a single good story.
And that story is our autobiography. The chief ¬ctional character at the center of
that autobiography is one™s self. And if you still want to know what the self really
is, you are making a category mistake. [CNG, 114]

What exactly is the category mistake that we make about the self, ac-
cording to Dennett? I shall ¬rst attempt to identify the mistake, and then
I™ll consider whether it really is a mistake. Speci¬cally, I™ll ask whether
J. David Velleman

Dennett himself can afford to call it a mistake, given the philosophical
commitments he undertakes in the course of diagnosing it. I shall argue
that in at least some respects, the conception of the self that Dennett calls
mistaken is in fact likely to be correct.
In arguing against Dennett™s diagnosis of this mistake, I shall not be
arguing against his positive conception of the self as the ¬ctive protagonist
of a person™s autobiography.6 On the contrary, I™ll argue that Dennett™s
positive conception of the self is largely right. My only disagreement with
Dennett will be that, whereas he regards an autobiography as ¬ctive and
consequently false in characterizing its protagonist, I regard it as both
¬ctive and true. We invent ourselves, I shall argue, but we really are the
characters whom we invent.

Dennett describes our mistaken conception as “the myth of selves as brain-
pearls, particular concrete, countable things rather than abstractions.”7
Sometimes he suggests that this myth mistakenly credits the self with
physical existence, as “a proper physical part of an organism or a brain.”8
But he also considers a version of the myth in which the self resides in
software rather than hardware, as “a supervisory brain program, a central
controller, or whatever.”9 Mostly, Dennett relies on metaphors that can
be read as alluding either to hardware or software: the “Oval Of¬ce in the
brain, housing a Highest Authority”10 or “the Cartesian Theater with its
Witness or Central Meaner”11 or “the central headquarters responsible
for organizing and directing all the subsidiary bureaucracies that keep
life and limb together.”12
Dennett cannot be faulted for describing the self in metaphorical
terms. His thesis, after all, is that the self is like one of those mythical beasts
that incorporate parts from different creatures and straddle boundaries
between different realms, in a way that de¬es literal description. Yet un-
less we understand what Dennett thinks is wrong with our conception
of the self, we cannot understand what he thinks is right about his own,
alternative conception. So we must look behind Dennett™s metaphors for
the error that they purport to reveal.
In Dennett™s view, our error about the self is to assume that the pro-
tagonist of a human being™s autobiography is identical with the author.
Dennett imagines that his own autobiography opens in the manner of
Moby Dick “ “Call me Dan” “ and he claims that this opening sentence
would prompt us to apply that name to “the theorists™ ¬ction created
by . . . well, not by me but by my brain [ . . . ].”13 In Dennett™s view, then,
the author of his autobiography is his brain, whereas the “me” whom we
The Self as Narrator 59

call Dan is a purely ¬ctional narrator, who is no more the real author
of the story than Ishmael is the author of the story that begins “Call me
Ishmael.” Dennett concludes:

Our tales are spun, but for the most part we don™t spin them; they spin us. Our
human consciousness, and our narrative selfhood, is their product, not their
source. [RS, 418]

But in what respect does the real source of Dennett™s autobiography
differ from the ¬ctional source that it conjures up for itself? Why should
Dan be compared to Ishmael rather than the author of a veridical auto-
biography, who really is identical with the protagonist of his story?

This question is especially pressing in light of the sophistication with
which Dennett is obliged to credit his real autobiographer. The brain that
composes Dennett™s autobiography has to be so clever as to approximate
the powers of its supposedly ¬ctional protagonist. We may therefore sus-
pect that Dennett, now in his capacity as philosopher, has tacitly posited
the existence of a real self to serve as the inventor of the supposedly
¬ctional one. Dennett anticipates and counters this suspicion:

Now, how can I make the claim that a self “ your own real self, for instance “ is
rather like a ¬ctional character? Aren™t all ¬ctional selves dependent for their very
creation on the existence of real selves? It may seem so, but I will argue that this is
an illusion. Let us go back to Ishmael. Ishmael is a ¬ctional character [. . .]. But,
one thinks, Ishmael was created by Melville, and Melville is a real character “ was
a real character “ a real self. Doesn™t this show that it takes a real self to create a
¬ctional self? I think not, but if I am to convince you, I must push you through
an exercise of the imagination. [CNG, 107]

The exercise mentioned here is to imagine a robot that emits a running
narration of its life, as the story of a character named Gilbert:

“Call me Gilbert,” it says. What follows is the apparent autobiography of this
¬ctional Gilbert. Now Gilbert is a ¬ctional, created self but its creator is no self.
Of course there were human designers who designed the machine, but they did
not design Gilbert. Gilbert is the product of a process in which there are no selves
at all. [Ibid.]

Dennett insists that he is not committed to crediting the robot with

That is, I am stipulating that this is not a conscious machine, not a “thinker.” It is
a dumb machine, but it does have the power to write a passable novel. [Ibid.]
J. David Velleman

[T]he robot™s brain, the robot™s computer, really knows nothing about the world;
it is not a self. It™s just a clanky computer. It doesn™t know what it™s doing. It doesn™t
even know that it™s creating this ¬ctional character. (The same is just as true of
your brain: it doesn™t know what it™s doing either.) [CNG, 108]

One might challenge this stipulation as self-contradictory. Stipulating a
“dumb machine” that writes a “passable novel,” one might think, is like
stipulating a blind man who sees. If someone sees, then he isn™t really
blind; and if something writes a passable novel, then it can™t be all that
dumb, no matter how loudly it may clank.14 How, then, can Dennett
claim that the computer generating Gilbert™s story doesn™t know what it™s
Part of the answer is that, according to Dennett, the computer isn™t
conscious; but I want to set aside the concept of consciousness, which is
only one aspect of selfhood. To be sure, Gilbert™s autobiographer por-
trays him as conscious, while Dennett denies that he really is. But the
robot™s claim to be conscious is not quite the same as his claim to be a
self. For as we have seen, claiming to be a self entails claiming not only
the status of “Witness,” who is the subject of experience, but also that
of “Central Meaner,” “central controller,” or “Highest Authority.”15 In-
deed, Dennett de¬nes a center of narrative gravity as a ¬ctional “uni¬ed
agent.”16 Leaving aside the question whether Gilbert™s autobiographer is
conscious, then, we can ask whether he really is a uni¬ed agent in the
sense that would satisfy the terms of this ¬ction.
Here again, one might think that Dennett™s stipulation is incoherent,
on the grounds that describing something as the author of a novel already
entails describing it as a uni¬ed agent. Yet I am willing to grant, for the
sake of argument, that a passable novel could be authored by a machine
endowed with no “Highest Authority,” “Central Meaner,” or other ironi-
cally capitalized locus of agency. What I suggest, however, is that Dennett
has equipped Gilbert™s and Dan™s autobiographers with more than the
mere capacity to produce passable novels, and that in doing so, he has
implicitly equipped them with enough of a self to be agents.
Dennett denies agency to the inventors of Gilbert and Dan primarily by
denying them agential unity. He defends this denial by citing the example
of a termite colony:

The revisionist case is that there really is no proper-self: none of the ¬ctive-selves “


. 2
( 13)