NOTE-266: Autonomy

Contribution

A textbook overview of autonomy in four registers: history (§1), personal autonomy (§2), feminist philosophy (§3), and political theory and bioethics (§4). For this batch its value is §2.a, which presents the hierarchical model as “the basis upon which the mainstream discussion builds and reacts against”. It sets out its standard objections under memorable names (Manipulation, Regress or Incompleteness, Ab Initio) and follows the line of critique into relational autonomy.

Key insight

The hierarchical model is procedural and content-neutral. It asks only whether a first-order desire is endorsed by a higher-order volition, not where that volition came from or what it says. Every standard objection targets one of those two neutralities:

• History is ignored, so it cannot tell a hypnotised agent from an autonomous one (Manipulation).
• The source of the endorser’s authority is left open (Regress, Ab Initio).
• A unified, wholehearted self is presupposed, which feminist and mestiza critics deny is the only form of autonomous agency.

Assumptions

The premises and authorities the entry works from:

• Personal autonomy is distinct from moral autonomy (Kant) and from political autonomy. Contemporary accounts are “metaphysically neutral” and “procedurally individualist” (§2).
• Content-neutral vs. substantive is the main division (§2). Hierarchical, historical and coherentist accounts are content-neutral. Wolf, Benson, Korsgaard and Hill are substantive.
• Autonomy can be a bare capacity or an ideal (preamble), and it can be assessed locally (for actions) or globally (for agents).
• Authorities: Frankfurt 1988 (collection) and Dworkin 1988/1989 for the hierarchy; Watson 1975, Mele 2001, Christman 1991, Thalberg 1978 and Friedman 1986 for the objections; Meyers 2005, Barvosa-Carter 2007 and Oshana 2005 against wholeheartedness; Mackenzie & Stoljar 2000 and Nedelsky 1989 for relational autonomy.

Key results

What the article argues or reports.

• §1 History.
  • Self-mastery in Plato and Aristotle.
  • Kant’s moral autonomy versus heteronomy.
  • Romantic, Millian and existentialist individualisation.
  • Lévinas’s preference for heteronomy.
  • Maslow, Rogers, Kohlberg, and Gilligan’s critique.
  • Contemporary work is content-neutral and avoids developmental ideals.
• §2.a.i The hierarchical model.
  • It begins with Frankfurt and G. Dworkin in the 1970s. Their “insight” is that decisions are worth protecting if rooted in the agent’s values and commitments, which contrasts the weak-willed person with the self-consciously valuing person.
  • Statement: “an agent is autonomous with respect to an action on the condition that his or her first-order desire to commit the act is sanctioned by a second-order volition endorsing the first-order desire (see Frankfurt 1988, 12-25)”.
  • The model is neutral about the origins of higher-order desires, and so admits socially and relationally constituted ones. “The cause of such desires does not matter, solely the agent’s identification with them (Frankfurt 1988, 53-54).”
  • Dworkin calls the higher-order preferences the “true self” (Dworkin 1989, 59).
• §2.a.ii Criticisms.
  1. Manipulation. Frankfurt’s account is “ahistorical”, so a hypnotist who installs second-order desires is not excluded (Mele 2001, 144–173). Christman 1991 makes autonomy depend on “the means and historical process by which an agent reaches certain decisions”, so that a brainwashed desire is nonautonomous. Its difficulties are an agent who freely gives up her autonomy, and an agent who endorses a desire but not the means by which it was forced on her (Taylor 2005, 10–12).
  2. Regress or Incompleteness. If second-order volitions need third-order ones, and so on, there is an infinite regress. If they are autonomous for some other reason, the model is incomplete (Watson 1975). Frankfurt acknowledges “no theoretical limit” to the series but ends it with a “decisive commitment” (Frankfurt 1988, 21). The entry objects that terminating the series is “arbitrary if there no reason behind it (Watson 1975)”.
     • Frankfurt’s reply, in “Identification and Wholeheartedness”: a decisive commitment is one made “without reservation”, where the agent feels no reason to continue deliberating (1988, 168–9). Stopping there is “hardly arbitrary”. Possible error is a feature of all deliberation. Deciding seeks “to overcome or to supersede a condition of inner division and to make himself into an integrated whole” (174). It establishes “a constraint by which other preferences and decisions are to be guided” (175).
     • Against wholeheartedness: Meyers 2005 argues for a decentred, fivefold self. Barvosa-Carter 2007 argues that ambivalence is inescapable for mixed-race agents with conflicting inheritances. Oshana 2005 makes the point for life in a racist society.
  3. Ab Initio. It is “a puzzle how decisive commitments or higher-order desires acquire their authority without themselves being endorsed”. If the source of autonomy cannot itself be reflectively endorsed, autonomy originates in something non-autonomous, outside the model.
  4. Priority of higher orders. The model gives “unjustified ontological priority” to higher levels (Thalberg 1978). It begs the question by positing an uncaused “true self” at the top (Friedman). Friedman 1986 proposes integrating desires of all orders instead.
• §2.a.iii Coherentist accounts. “For Frankfurt, we identify with a lower level desire if we have a second order volition endorsing it. However, our second order volitions don’t necessarily represent us — we may have no reason for them, which Frankfurt acknowledges.”
  • Ekstrom 1993: a self is a character of beliefs and preferences endorsed through reflection, where preferences are first-level desires “to be good”, “similar to a second order volition for Frankfurt”. A preference is authorised by coherence with the rest. This is meant to avoid the Regress and Ab Initio problems.
  • Bratman 2007, 5: plans and intentions partly constitute personal identity, and decisions are self-governing relative to them.
• §2.b Substantive accounts. Wolf’s normative competence (strong). Benson’s normative competence (strong, 1991), and his weak substantive condition of self-worth or self-trust. Korsgaard’s practical identities, and Hill’s version.
• §3 Feminist.
  • Critics: Gilligan, Benjamin, and Hoagland (“a thoroughly noxious concept”). Friedman and Christman reply that procedural autonomy is not substantive independence.
  • Relational autonomy distinguishes two levels of relationality: social sources of values, and social commitments themselves. On Friedman’s view one can autonomously be a hermit. Donchin denies that one can opt out.
• §4 Political and bioethics.
  • Kant’s principle of right. Mill’s anti-paternalism. Gaus’s liberal principle. Christman & Anderson on liberal neutrality. Sandel’s communitarian critique. Okin and Weiss. Fineman on dependency. Forst’s five conceptions of autonomy. Sen and Nussbaum’s capabilities.
  • Belmont Report. Beauchamp & Childress’s respect for autonomy. Disability rights (“Nothing about us without us”). Donchin’s weak vs. strong relational autonomy in health care.

What the entry attributes to the paywalled originals

Section numbers are the entry’s; page numbers are the entry’s citations to The Importance of What We Care About (1988). Which essay each page belongs to is inferred as described in the corrections.

• Frankfurt 1971 (“Freedom of the Will…”, reprint pp. 11–25):

  • autonomy with respect to an action = a first-order desire “sanctioned by a second-order volition endorsing” it (§2.a.i, pp. 12–25);
  • “no theoretical limit” to the series of higher-order desires, which ends in a “decisive commitment” to a first-order desire (§2.a.ii, p. 21);
  • we identify with a lower-level desire if a second-order volition endorses it, but we “may have no reason” for our second-order volitions, “which Frankfurt acknowledges” (§2.a.iii, no page).

  Not reported: wantons, persons, freedom of will vs. freedom of action, or the definition of a second-order volition as wanting a desire to be one’s will.

• Frankfurt, pp. 53–54 (probably “Three Concepts of Free Action: II”; original date not given in either bibliography): the cause of higher-order desires does not matter, only identification (§2.a.i).

• Frankfurt 1976 (“Identification and Externality”): nothing.

• Frankfurt 1987 (“Identification and Wholeheartedness”, named in §2.a.ii):

  • decisive commitment = made without reservation, with no felt reason to go on deliberating (pp. 168–9);
  • stopping there is not arbitrary, and fallibility is common to all deliberation;
  • decision aims to overcome inner division and make oneself “an integrated whole” (p. 174);
  • it sets “a constraint by which other preferences and decisions are to be guided” (p. 175).

• Frankfurt 1992 and 1994 (“The Faintest Passion”; “Autonomy, Necessity, and Love”): nothing. No satisfaction, volitional necessity, caring or love.

• Watson 1975 (“Free Agency”):

  • the regress: second-order volitions would need third-order ones, or else the model is incomplete (§2.a.ii);
  • ending the series by decisive commitment is arbitrary without a reason (§2.a.ii).

  Not reported: valuing vs. desiring, Watson’s evaluational system, and any positive view of Watson’s.

• Bratman 1996, 2004: nothing. Only Bratman 2007, 5 (plans partly constitute personal identity; §2.a.iii).

Claims

id	claim	strength	support
C1	Hierarchical model: autonomy with respect to an action = first-order desire sanctioned by an endorsing second-order volition; origin irrelevant	second-hand report	§2.a.i, citing Frankfurt 1988, 12–25 and 53–54 by page only
C2	Hierarchical accounts cannot distinguish a manipulated agent from an autonomous one	informal argument	§2.a.ii; hypnotist case; Mele 2001 cited
C3	Christman’s historical model fixes manipulation but mishandles freely chosen loss of autonomy and endorsed-but-forced desires	second-hand report	§2.a.ii; Christman 1991; Taylor 2005
C4	The regress/incompleteness objection is Watson’s (1975), and so is the charge that stopping by decisive commitment is arbitrary	second-hand report	§2.a.ii; no page in Watson; the SEP entry does not credit Watson with the regress
C5	Frankfurt’s 1987 reply: a decisive commitment, made without reservation, non-arbitrarily ends the series and integrates the self	second-hand report with page citations	§2.a.ii; pp. 168–9, 174, 175 with short quotations
C6	Ab initio: if autonomy’s source cannot be endorsed, autonomy originates in the non-autonomous	informal argument	§2.a.ii; stated as a puzzle, no citation
C7	Wholeheartedness misdescribes ambivalent agents (oppressed, mixed traditions)	second-hand report	§2.a.ii; Meyers 2005, Barvosa-Carter 2007, Oshana 2005
C8	Frankfurt’s and Dworkin’s original concern was which individual freedom should be protected from paternalism	assertion	§2.a.i; no citation; doubtful for Frankfurt (see corrections)
C9	Bratman: long-range plans partly constitute personal identity	second-hand report	§2.a.iii; Bratman 2007, 5
C10	Procedural autonomy does not imply substantive independence, so feminist critiques of “autonomy” miss it	second-hand report	§3.a; Friedman 2000 (not in bibliography), Christman 1995

Concepts

• Content-neutral (procedural) vs. substantive autonomy. Whether the account fixes any value the autonomous agent must hold (§2).
• Hierarchical procedural account. Autonomy is endorsement of a first-order desire by a second-order volition (§2.a.i).
• Problem of Manipulation. The ahistorical endorsement condition cannot exclude installed endorsements (§2.a.ii).
• Regress or Incompleteness Problem. Either an infinite regress of higher-order volitions, or authority from some non-hierarchical source (§2.a.ii).
• Ab Initio Problem. The un-endorsed source of autonomy is itself non-autonomous (§2.a.ii).
• Decisive commitment / wholeheartedness. A commitment made without reservation that ends deliberation and integrates the self (per “Identification and Wholeheartedness”).
• Coherentist account. Ekstrom’s and Bratman’s accounts, as opposed to the hierarchical model (§2.a.iii). This is narrower than the SEP “Personal Autonomy” entry’s “coherentist”, which includes the hierarchical model.
• Relational autonomy. Autonomy compatible with, or partly constituted by, social relations. It works at two levels: social sources of values, and social commitments themselves (§3.b).

Connections

• Within this batch.

  • The SEP “Personal Autonomy” entry (Buss & Westlund). It is the argumentative counterpart and disagrees on several points:

    • (a) It reads Watson 1975 as a rival evaluative-judgment account and never as the source of the regress. This article reads Watson only as the regress critic.
    • (b) It carries Frankfurt’s reply past decisive commitment to “satisfaction” (1992). This article stops at 1987.
    • (c) “Coherentist” means different things in the two entries.
    • (d) The SEP entry states the hierarchy with “highest-order desires”, this one with “second-order volition”.

    They agree that the hierarchical condition is ahistorical and content-neutral, that manipulation defeats it, and that Christman’s history-sensitive condition is the main repair.

  • The SEP PAP entry (Robb). It covers the other face of Frankfurt 1969–1971: responsibility without alternatives, and identification as a compatibilist account.

  • Velleman’s “Identification and Identity”. It treats the 1976–1992 sequence that this article leaves out.

  • Frankfurt’s Tanner Lectures. His own late statement of wholeheartedness and satisfaction.

• Adaptive preferences (LIT-202). Pettigrew’s mechanism-based criterion for when shaped preferences ground consent is a descendant of the historical (Christman) line against manipulation reported in §2.a.ii. The article’s §4.b on informed consent in bioethics is the applied setting LIT-202 addresses.
• Capability approach (LIT-289, LIT-283). §4.a links political autonomy to Sen’s and Nussbaum’s capabilities in one sentence (Sen 1999, Nussbaum 2006). The link is asserted, not developed.
• Self-determination theory (LIT-560, LIT-558). In this entry’s classes, SDT’s autonomy is coherentist in its 2006 criterion (reflective endorsement “by the whole self”, LIT-560, p. 1560), historical through internalization (a value taken in under control is introjected, under autonomy support integrated; THEORY-047), and substantive, since some values are said to be “not integrateable” (LIT-558, p. 246). It takes the side of the fourth objection reported here, against the priority of higher orders (Friedman’s integration model): SDT places autonomy in integration, graded, not in the order of an endorsing attitude (THEORY-054).
• Anthology of the SOTA. No ANTH document mentions autonomy, Frankfurt or free will.

Bearing on the record

• No THEORY in the record bears on this. If one is written on hierarchical agency, this article is the most convenient second-hand source for the names of the objections (Manipulation, Regress/Incompleteness, Ab Initio) and for the 1987 reply with page references. Frankfurt’s later view, Watson’s positive view and Bratman’s views need other sources.
• No ML-practice content. §4.b’s informed-consent and respect-for-autonomy material is bioethics, not ML practice. Nothing belongs in the Anthology.
• For filing. agency first (personal autonomy as self-governance); ethics (moral autonomy, bioethics); society-and-governance (§4: liberalism, paternalism, political autonomy, Belmont Report). identity is not justified: “true self” and Bratman’s plan-constituted identity get a sentence each.

Note of 2026-10-16: THEORY-029, THEORY-040 and THEORY-120 now cite LIT-297 as a source.

Limitations

• It is short and introductory. §2.a, the part that matters here, is about 1.4k words.
• Frankfurt is cited by pages of a collection, with no essay titles except the 1987 one, so claims cannot be traced to specific papers without another bibliography.
• It stops at 1987 on Frankfurt and at 2007 on the literature overall. Frankfurt’s satisfaction (1992), volitional necessity and love (1994, 1999) are absent, and so is Watson’s valuing/desiring distinction.
• The desire/volition slippage in the regress paragraph obscures what, exactly, regresses.
• The attribution of an anti-paternalist motive to Frankfurt (C8) is unsupported.
• The page is undated; the dates here come from the site’s CMS and the Wayback Machine.

Open questions

• Does the regress objection belong to Watson 1975, as this article says? The SEP entry presents the regress anonymously and reads Watson as offering an alternative locus of authority. Only the original, or a reliable reconstruction such as Velleman 2002, can settle what Watson argued.
• Does decisive commitment stop the regress non-arbitrarily, or does it relocate the ab initio problem to the commitment? The article states both the reply and the ab initio puzzle but does not adjudicate.
• Can wholeheartedness be a condition of autonomy without excluding ambivalent agents (Barvosa-Carter, Oshana)?

Corrections to the seeded skim

• none (there was no seed or dossier)
• The scoping report’s “n.d.” can be narrowed. The IEP’s WordPress record gives post date 2010-11-21 and last modification 2013-12-25, and the Wayback Machine has the page by 2010-11-25. The body cites nothing later than 2007.
• The scoping report says the article covers “wholeheartedness and Bratman”, and that “the planning-agency view is reachable… through the IEP entry”. Bratman is one sentence in §2.a.iii: “our personal identity is partly constituted by the organizing and coordinating function of our long-range plans and intentions (Bratman 2007, 5)”. The view is filed as a coherentist account “similar” to Ekstrom’s. Nothing from Bratman 1996 (identification as a decision to treat a desire as reason-giving) or Bratman 2004 (self-governing policies; the Frankfurt–Watson exchange) is reported. Neither paper is in the bibliography. Wholeheartedness is covered, but only as defined in the 1987 paper. “Satisfaction”, caring and love do not appear.
• Citation looseness. Every Frankfurt citation is “Frankfurt 1988, pp.”, to The Importance of What We Care About. The bibliography lists only the book, so a reader cannot tell which essay is meant. Using the reprint pagination given in the SEP “Personal Autonomy” bibliography:
  • pp. 12–25 and p. 21 fall in “Freedom of the Will and the Concept of a Person” (1971; reprint pp. 11–25).
  • pp. 53–54, cited for “the cause of such desires does not matter, solely the agent’s identification with them”, fall in “Three Concepts of Free Action: II” (reprint pp. 47–57), not in the 1971 paper. That is unverified: it is inferred from page ranges.
  • pp. 168–9, 174 and 175 fall in “Identification and Wholeheartedness” (1987; reprint pp. 159–76), which the text names.
• Slippage between “second-order desire” and “second-order volition”. The hierarchy is first stated with a second-order volition sanctioning the first-order desire. The regress paragraph then says first-order desires must be “endorsed by second-order desires”. The SEP entry says “highest-order desires”. The distinction matters in Frankfurt’s own scheme (per the scoping report’s summary of 1971), and the article does not keep it.
• Minor bibliography defects.
  • “Benjamin, Benjamin” should be Benjamin, Jessica.
  • “Friedman 2000, 37–46” is cited in §3.a but absent from the bibliography.
  • The 1971 paper, Frankfurt’s 1987 paper and Watson 1975 appear only as “Frankfurt 1988” (the book) and as Watson, J. Phil. 72.
• Framing of Frankfurt’s motive. §2.a.i says Frankfurt’s and Dworkin’s 1970s “concern was to give an account of what kind of individual freedom ought to be protected”, which is an anti-paternalism framing. No citation supports this for Frankfurt. The SEP entries in this batch place Frankfurt 1971 in the free-will and moral-responsibility literature (Robb §5.1) and in the metaphysics of agency (Buss & Westlund §2). The framing fits Dworkin (1972 “Paternalism” is cited) better than Frankfurt. This is not verifiable from the original here.