NOTE-265: What Happens When Someone Acts?

Contribution

The paper names the standard story of human action: desire and belief jointly cause an intention, the intention causes bodily movement, and with normal causal processes the result is an action with those states as its reasons. Velleman credits it loosely to Davidson and traces it back to Hobbes (p. 100 and n. 2). He argues that the story “fails to include an agent — or, more precisely, fails to cast the agent in his proper role” (p. 100). Reasons cause an intention and the intention causes movements, “but nobody — that is, no person — does anything” (p. 101). The paper then does three things:

1. It shows that appeals to causal normality (the deviant-chains literature) cannot rescue the story.
2. It argues that Frankfurt’s hierarchical account, “the best sustained attempt” at a reductive account of agent-causation (p. 110), fails for a structural reason.
3. It proposes a functionalist reduction in which the agent’s role is played by a single motive: the desire to act in accordance with reasons.

Key insight

Frankfurt looked for the mental state or event that constitutes the agent’s identifying with a motive. Velleman says the thing to reduce is instead the agent himself, the principal component of agent-causation (pp. 116–117). “Identifying with” is a relation one bears to something functionally distinct from oneself. So whatever plays the agent’s role cannot be anything the agent identifies with. It has to be what does the identifying, the party “always behind, and never solely in front of, the lens of critical reflection” (p. 119). In short, the agent’s contribution is the contribution of something functionally identical to him, not something he identifies with (p. 122).

Assumptions

• Naturalism about explanation. Occurrences, events and states, are “the basic elements of explanation in general” (p. 108). Chisholm-style primitive agent-causation is set aside (p. 110). “Agent-causation” is taken in the relaxed sense: agent-causation that is reducible to, or supervenes on, event-causation (pp. 108–109, n. 17).
• Conceptual target. The aim is to explicate “our concept of action” and not the reality, assuming “action” has a de dicto meaning and is not a natural-kind term (p. 106 and n. 16). Our practices of deliberation, rationalizing explanation and moral assessment are “designed for action as we conceive it to be” (p. 107).
• Not all actions are full-blooded. Behavior without the agent’s participation can still be an action, “albeit defective or borderline” (n. 5, p. 101). The explanandum is “human action par excellence”.
• The agent intermediates. He forms an intention under the influence of reasons, and he produces behavior pursuant to it. These are two roles that “the items themselves” (reasons, intentions) cannot play (p. 102).
• Frankfurt’s framing is adopted provisionally. The agent’s role is to reflect on motives competing for control of behavior and to settle the contest by taking sides, though Velleman says adjudication is “best understood as taking place among reasons” (p. 118).
• The problem of agency is independent of the mind–body problem (“finding an agent at work amid the workings of the mind”, p. 109) and of the problem of free will (n. 14, p. 105).

Key results

What the chapter argues, in order (pages are the 2015 reprint).

• The agent is missing (pp. 100–103). Two defences of the standard story are rejected.
  • “The agent is the sum of the components; a cake is not listed in its own recipe.” Velleman answers that the roles the agent plays are not played by anything in the story (p. 101).
  • “The agent is implicitly present as the subject of the states.” He answers that being the subject of causally related attitudes “does not amount to participation” (p. 103; Ginet quoted in n. 8).
• Frankfurt’s addict, as Velleman reports it (p. 103). Citing The Importance of What We Care About (CUP 1988), Velleman reports Frankfurt’s point that desires and beliefs can cause an intention “despite him”. The addict “may meaningfully make the analytically puzzling [statement] that the force moving him to take the drug is a force other than his own” (IWWCA p. 18), and may be “a helpless bystander to the forces that move him” (p. 21).
• Causal normality is not enough (pp. 103–105). The “deviant causal chains” literature (Davidson, Bishop, Peacocke and others, n. 10) diverted attention from simpler cases. The friendship example: latent grievances crystallize into a resolution to end a friendship, which gives “the hurtful edge” to my remarks. The desires exercised “the same causal powers that they exercise in ordinary cases” without any contribution from me, “it was my resentment speaking, not I” (p. 104). Therefore “no definition of causal normality will fix what ails the standard story” (p. 104). Note 12 claims the case survives Bishop’s “sensitivity” condition. Note 13 says responsibility remains, but as a failure to prevent or control the action.
• Chisholm’s challenge accepted (pp. 108–110). The goal is “to earn the right to make jokes about primitive agent-causation” by explaining how the agent’s causal role supervenes on events and states (p. 110). The problem of agency is “that of finding an agent at work amid the workings of the mind” (p. 109).
• Frankfurt’s account, as Velleman reconstructs it (pp. 110–112). Velleman cites these texts: “Freedom of the Will and the Concept of a Person”, “Three Concepts of Free Action”, “Identification and Externality”, “The Problem of Action” and “Identification and Wholeheartedness”, all in IWWCA (n. 25).
  • Frankfurt’s question narrows from what constitutes a person to the person “as an element in the causal order”. The essential difference “is to be found in the structure of a person’s will” (IWWCA 12), and inability to exercise the will “destroys him as a person” (p. 21) (n. 26).
  • The target is the difference between a person who “participates” in the operation of his will and one who is “a helpless bystander” (IWWCA 21–2). Elsewhere Frankfurt frames it as “activity or passivity with respect to … states of affairs” (p. 54) (n. 27).
  • The strategy is to find what is missing in defective action. What is missing when the agent is alienated from his “volition” is his “identifying” with it. So full-blooded action is action in which the agent identifies with the motives that actuate him (IWWCA 18 f., 54; n. 28).
  • The agent adds to his motives’ force “by intermediating among them … throwing his weight behind some of them rather than others” (p. 112).
  • The first candidate is a second-order desire to be actuated by the motive, with alienation being a desire not to be. Such higher-order desires “constitute his activity” (IWWCA 54) (p. 112).
• The regress, attributed to Watson and conceded by Frankfurt (pp. 112–113). “As Gary Watson has pointed out and Frankfurt has conceded” (Watson, “Free Agency”, in Free Will, OUP 1982; Frankfurt’s concession at IWWCA 65–6, which falls within “Identification and Externality”, pp. 58–68, per the book’s ch. 2 n. 3). An agent can be alienated from his second-order desires as from his first-order ones, so “the occurrence that supposedly constitutes the agent’s contributing to his behavior seems itself to stand in need of some further contribution from him” (p. 113).
  • Velleman presents this as Watson’s objection. In current literature it is usually called the regress or authority objection.
  • Note 29 adds that Velleman’s later interpretation of Frankfurt (“The Way of the Wanton”) is one “in which the problem pointed out by Watson doesn’t even arise”. He also thinks Frankfurt’s own attempted solution is “unsuccessful”.
• Watson’s values fare no better (p. 113). Watson’s replacement, action flowing from the agent’s “evaluational system”, on which alienated desires are “more or less radically independent of [his] evaluational systems” (Watson, “Free Agency”, 110 as printed), fails. One can be alienated from one’s values while they still grip one: “recoils from his own materialism or his own sense of sin” (example credited to Elizabeth Anderson, n. 32). If Watson restricts himself to the system, he “would simply be smuggling the concept of identification” into it (n. 33).
• Frankfurt’s decisive commitments fare no better (pp. 113–114). Velleman reports that Frankfurt “now suggests” that “decisions” or “decisive commitments” provide involvement, because “Decisions, unlike desires or attitudes, do not seem to be susceptible both to internality and to externality”, so invoking them “would appear to avoid … the difficulty” (IWWCA p. 68 n. 3). Velleman replies that the unwitting decision to end the friendship shows that decisions can be foreign too. A stipulated sense of “decisive commitment” (only commitments the agent actively makes) would make the solution “uninformative” (n. 35; a related criticism is credited to Christman 1991).
• Frankfurt’s refusal to iterate (pp. 114–115). Velleman quotes: “It makes no sense to ask whether someone identifies himself with his identification of himself, unless this is intended simply as asking whether his identification is wholehearted or complete” (IWWCA p. 54).
  • Velleman says this shows Frankfurt uses “identification” in a specialized sense. Ordinary identification, for instance with a character in a trashy novel, can happen to you without your participation.
  • He reads Frankfurt as assuming that identification with a motive “cannot happen at all without one’s participation”, though it may be halfhearted.
• The dilemma, which is the central negative result (pp. 115–116). If self-identification is a primitive that requires the agent’s participation, positing it “beg[s] the question of agent-causation”. So “self-identification” must be read as a redescription of the explanandum and reduced to desires, values or decisions. But any state specified without presupposing the agent’s participation is one “from which the agent can in principle dissociate himself”. Hence “a choice between begging the question and not answering it at all” (p. 116). Watson’s “system of values” and Frankfurt’s “decisive commitments” seem to succeed “only insofar as they smuggle such an assumption into the story” (p. 116).
• The diagnosis (p. 116). “Substituting one instance of agent-causation for another, as the target of reduction, does not advance the reductionist project.” The right move is to break agent-causation into components, and “the principal component of agent-causation is the agent himself” (pp. 116–117).
• Functional identity (p. 117). Look for mental events and states “functionally identical to the agent”, playing the causal role ordinary parlance attributes to him. The analogy is digestion: a person digests or fights infection in virtue of subsystems. So a person may be an agent “in the sense that there is an action-initiating system within him”.
• Why no identified-with item can be the agent (pp. 117–118). Frankfurt’s talk of “identifying oneself” with a motive and “making it one’s own” (IWWCA 18) tempts the thought that such motives become the agent. But identification requires one item to play the attitude identified with and another to play the agent identifying with it. The latter is the functional agent.
• The “identity apart” (p. 118). Frankfurt says adjudicating the contest entails occupying an “identity apart” from the motives (IWWCA 18). Velleman generalizes: the agent’s role is closed “to all potential objects” of critical reflection, at any level of the hierarchy, because it “must be played by whatever directs that process” (p. 119).
• Frankfurt’s omission (pp. 119–120). Frankfurt assumes the concerns animating reflection are embodied in particular second-order desires, such as the desire not to act out of anger. These figure only in a particular conflict and can themselves be reviewed one level up. What is needed is a motive “at work in all such instances of reflection”. The traditional picture of agency as a “neutral capacity” not animated by any motive must be given up (p. 120).
• The proposal (pp. 120–121). “What animates practical thought is a concern for acting in accordance with reasons”, embodied in a desire.
  • It may be a de dicto desire to act in accordance with reasons “so described”, or, as Velleman prefers, a desire to act on considerations of a kind that in fact constitute reasons, such as the desire “to do what makes sense, or what’s intelligible”, from Practical Reflection (1989) (n. 38–39).
  • Each agential function is performed by this desire: directing attention to motives, calculating their strengths, and “throwing weight”. When a desire appears to provide the strongest reason, the desire to act in accordance with reasons “becomes a motive to act on that desire”, reinforcing it.
  • “That part, I claim, is your desire to act in accordance with reasons, a desire that produces behavior in your name” (p. 121).
• Non-disownability (p. 122). The desire “cannot be disowned by an agent, although it can be disowned by the person in whom agency is embodied”. Rejecting it as a reason would manifest it. Truly suppressing it means suspending practical thought and hence agency. Conversely, its operation just is critical review, embrace or rejection, and reinforcement or suppression, so its contribution “can qualify as his — not because he identifies with it but rather because it is functionally identical to him” (p. 122).
• Weak motive prevailing (pp. 122–123). The account lets us say both that the agent makes the intrinsically weaker motive prevail and that “the contest always goes to the strongest combination of motives”. The agent’s weight is another motive, so the weaker motive plus it forms the strongest combination. The chapter closes: “Come to think of it, what else could an agent be?” (p. 123).

Claims

id	claim	strength	support
C1	The standard story omits the agent’s participation, not because it fails to name him but because nothing in it plays his intermediating role	informal argument	pp. 100–103; conceptual analysis of the roles of forming and executing intentions
C2	An agent can fail to participate even when desires, intention and behavior are linked by entirely normal causal processes, so no definition of causal normality rescues the standard story	informal argument from one case	Friendship example, p. 104; n. 12 claims it satisfies Bishop’s sensitivity condition, but this is asserted, not shown
C3	Frankfurt’s second-order-desire account of identification is open to the same objection as the standard story (the Watson regress), and Frankfurt conceded this	report + informal argument	Attribution to Watson 1982 and to Frankfurt IWWCA 65–6, p. 112; not checkable here against the originals
C4	Watson’s appeal to values and Frankfurt’s appeal to decisive commitments fail for the same reason: one can be alienated from values and from decisions	informal argument by counterexample	Materialism/sin example (p. 113, credited to Anderson); unwitting decision (p. 114)
C5	Any reduction of self-identification faces a dilemma: items specified independently of agent participation can be disowned; items specified to guarantee it beg the question	informal argument (the paper’s central negative claim)	pp. 115–116; it is general in form, but “can in principle dissociate himself” is argued only by the examples
C6	Whatever is functionally identical to the agent cannot be anything he identifies with, since identification relates two functionally distinct items	conceptual argument	pp. 117–118
C7	The agent’s role can be played only by a motive at work in critical reflection at every level: the desire to act in accordance with reasons	informal argument + proposal	pp. 119–121; existence of the motive is affirmed (“I believe so”, p. 120), not demonstrated; its content is left open (n. 38)
C8	That desire cannot be disowned while one remains an agent	conceptual argument	p. 122
C9	The account reconciles “the agent makes the weaker motive prevail” with “the strongest combination always wins”	conceptual argument	pp. 122–123
C10	The account “may well be realized in the world”	assertion (a hope)	p. 107; no empirical claim is made

Method

Conceptual analysis in the philosophy of action. There are case-based counterexamples (the addict; the assassin’s twitching trigger finger, p. 103; the friendship case), a question-begging dilemma posed against reductive rivals, and a functionalist reduction modelled on subsystem attribution (digestion, infection-fighting, p. 117). The explanandum is fixed as our concept of full-blooded action, not its reality (pp. 106–107).

Concepts

• Standard story of human action: desire + belief → intention → bodily movement, with normal causation (p. 100).
• Full-blooded action / human action par excellence, as against halfhearted, unwitting or defective action (pp. 101, 105). Velleman later replaces this with the action vs. activity distinction (PPR 2nd ed. ch. 2 n. 4).
• Agential participation: the agent’s intermediating between reasons and intention, and between intention and movement (p. 102).
• Agent-causation: Chisholm’s primitive notion vs. the relaxed, supervenient notion that Velleman adopts (pp. 108–109).
• Identification (Frankfurt, as Velleman reads him): the agent’s throwing his weight behind a motive. Velleman says Frankfurt uses the word in a specialized sense (p. 114).
• “Identity apart” (Frankfurt, IWWCA 18): the adjudicator’s distinctness from the contesting motives (p. 118).
• Decisive commitment (Frankfurt, IWWCA 68 n. 3): decisions as items not susceptible to both internality and externality (pp. 113–114).
• Functional identity with the agent: playing the causal role ordinarily attributed to the agent (p. 117).
• Desire to act in accordance with reasons: the motive that constitutes the agent functionally. In its favoured form it is a desire to do what makes sense (pp. 120–121).

Connections

• Within this batch.
  • Velleman’s “Identification and Identity” (2002) is the sequel. Its note 2 sends readers here for “the relevant literature” on objections to Frankfurt. Its note 62 says its alternative conception of the self is “developed more fully” here. The later essay drops agent-causation and turns to Frankfurt’s post-1990 claim that volitional necessities constitute a personal essence.
  • The SEP “Personal Autonomy” (Buss & Westlund) and IEP “Autonomy” (Dryden) entries should present the same regress/authority objection. Compare their attributions with this chapter’s, which credits the objection to Watson 1982 and has Frankfurt conceding it at IWWCA 65–6.
  • Frankfurt’s Tanner Lectures (2004/2005) postdate this chapter. The chapter explicitly excludes “The Faintest Passion” (n. 25, n. 34), so it says nothing about satisfaction as the late answer to the regress.
• LIT-117 (Enoch, “Agency, Shmagency”, Deferred). Enoch’s target is constitutivism, which he associates with Korsgaard and Velleman. This chapter is an early step on Velleman’s side: the book’s introduction says the chapter “argues that action must have a constitutive motive”, and “The Possibility of Practical Reason” (ch. 7) turns that into a constitutive aim (ch. 2 n. 27). The shmagency objection targets the later normative use of that aim, not this chapter’s reduction of agent-causation, which makes no normative claim.
• LIT-202 (Pettigrew, “Consent and the Formation of Preferences”). Its “Frankfurt-style case” is the 1969 alternate-possibilities device, not the hierarchical account discussed here. It is not a connection.
• Anthology of the SOTA. No ANTH document mentions Frankfurt, Velleman or higher-order desires.

Bearing on the record

• No THEORY in the record bears on this. If the record acquires a THEORY on hierarchical agency (“identification with higher-order desires constitutes the agent’s participation”), this chapter is a primary source for the against side: the Watson regress, as Velleman reports it, and the begging-or-not-answering dilemma. It is not a source for Frankfurt’s own claims, which it reports at second hand. Cite those as “per Velleman”.
• No ML-practice content. The one distant bearing is on talk of “agents” in AI. The chapter’s functionalist move, that the agent is whatever subsystem plays the reason-weighing role, is a clean philosophical model of an “agent” as a control subsystem. But the chapter makes no claim about artificial systems, and nothing here should be filed as practice.
• For filing. agency first, since this is the paper’s subject: action, autonomy, control. metaphysics second, since it is about agent-causation, reduction, supervenience and event-causation. Two tags are left out. identity is not justified, because personal identity is not discussed; that belongs to the 2002 essay. ethics is not justified either, since responsibility appears only in n. 13.

Note of 2026-10-16: THEORY-029 now cites LIT-295 as a source.

Limitations

• The positive proposal is a sketch, and the author revises it in this very book.
  • Ch. 2 n. 21 says the chapter’s “desire to act in accordance with reasons” is not meant de dicto. It adds that a de dicto reading would rule out weakness of will, an argument “I do not give”.
  • Ch. 2 n. 35 then identifies the relevant motive as “the agent’s desire to know what he is doing”.
  • Ch. 2 n. 4 says: “I now regard the terms ‘action’ and ‘activity’ as preferable” to full-blooded vs. defective action.
  • A reader citing the 1992 proposal should cite it with these later qualifications.
• The central counterexample carries a great deal of weight. The friendship case (C2) is the only case against causal normality. The claim that it survives sensitivity conditions (n. 12) is asserted, not shown.
• The dilemma (C5) is argued generally but supported only by examples. No principle is given for why every state specifiable independently of agent-causation must be disownable. The paper’s own positive proposal must escape the dilemma by being non-disownable “while remaining an agent” (p. 122). Whether that escape is fair, or whether it too presupposes what it explains, is not examined.
• Frankfurt is read only through The Importance of What We Care About (1988). “The Faintest Passion” (1991/92) is expressly excluded. So is everything later: wholeheartedness as satisfaction, volitional necessity, love. Nothing here assesses Frankfurt’s mature reply to the regress.
• “Identification and Wholeheartedness” (1987) is listed but not engaged. The “decisive commitment” view is quoted from the 1976 “Identification and Externality” footnote (IWWCA 68 n. 3).
• The originals were not read. Watson’s “Free Agency” is cited at “110” in the reprint’s n. 31, although n. 29 gives the essay’s span in Free Will as 205–20. One of the two is probably a typo or a different pagination (unverified). Frankfurt’s page references are to IWWCA and could not be checked here.
• The Mind pagination is not given. A reader needing Mind page numbers must consult the journal.

Open questions

• Does Frankfurt’s later appeal to satisfaction in “The Faintest Passion” escape Velleman’s dilemma, or does it fall on the “not answering” horn? Velleman’s later reading (“The Way of the Wanton”, n. 29), in which Watson’s problem “doesn’t even arise”, is the place to look, and it is not in this batch.
• Is the desire to act in accordance with reasons (or to make sense, or to know what one is doing) itself disownable by a person in a way that would restart the regress one level over? Velleman concedes “it can be disowned by the person in whom agency is embodied” (p. 122).
• Does the functional-identity strategy yield autonomy or only agency? The 2002 essay suggests the former, via the “aspectual” self, but this chapter does not argue it.
• Frankfurt’s reply to this line of criticism is known here only as the “Reply to J. David Velleman” in Contours of Agency (2002), pp. 124–128, per Crossref; it answers the 2002 essay, not this one. It is unread.

Corrections to the seeded skim

• none (there was no seed or dossier)
• The scoping report says the paper “replaces identification with a functional motive”. That is right, but it omits how Velleman frames the move. He calls his approach “simply a modification of” Frankfurt’s (p. 118). He also keeps Frankfurt’s assumption that the agent’s role is to adjudicate conflicts of motives, “for the moment” (p. 118). His target is Frankfurt’s reductive strategy: substituting one agent-causal phenomenon, self-identification, for another as the target of reduction (p. 116). He does not reject hierarchical structure as such, and he calls Frankfurt’s articles “the best sustained attempt” at a reductive explanation (p. 110).
• The chapter deliberately excludes Frankfurt’s later work. Footnote 25 says Velleman “shall not be discussing” “The Faintest Passion” (the 1991 APA Presidential Address), and note 34 repeats the exclusion. The Frankfurt examined here is the Frankfurt of The Importance of What We Care About (1988) only. The chapter cannot stand in for a critique of satisfaction, wholeheartedness-as-satisfaction or love. “Identification and Wholeheartedness” is listed in n. 25, but the text never quotes it. The “decisive commitment” passage is quoted from “Identification and Externality” (IWWCA p. 68 n. 3), not from the 1987 paper.
• “Frankfurt” appears 59 times and “identif” 57 times, per the scoping report. Those counts are for the whole book, not the chapter.* The chapter’s treatment of Frankfurt runs about eight pages (pp. 110–118), plus the closing proposal.
• The reprint does differ from the Mind text in at least its notes. The book’s note 1 says only “reprinted by permission of Oxford University Press” and records no revisions. But note 29 cross-refers to “The Way of the Wanton” (Chap. 11), a much later paper, so that note at least was rewritten for the second edition. Notes 3, 14 and 38 carry chapter cross-references to this book. Whether the body text was changed cannot be settled without the Mind printing, which I did not see. The book’s introductions add two retrospective qualifications, recorded under Limitations.