NOTE-1497: Metzinger — M-Autonomy

Contribution

A shorter and sharper version of the 2013 paper (LIT-1942), which its note 1 names as the “more comprehensive publication”. Three things are new. It decomposes M-autonomy into four capacities of second-order mental action and three degrees. It takes a “moderate, nuanced position”: mind-wandering loses the conscious knowledge of the ability, not the ability (p. 282). And it argues that the onset of every period of control “may therefore, necessarily, involve an element of misrepresentation” (p. 291). It ties the EAM to the first-person perspective and to personhood, which it calls “a concept of social ontology” (p. 284).

Key insight

A capacity can be kept and yet not be usable, because the system does not represent that it has it. “One can only deliberately and autonomously exert an ability if that ability is explicitly represented in one’s phenomenal self-model” (p. 280). The representation of absence is not the absence of representation (p. 283, after Brentano and Dennett): during mind-wandering the self-model does not represent that control is missing; it simply fails to represent control. Hence the feeling that one was in charge all along, and hence, on return, the misattribution of the return to oneself.

Assumptions

• M-autonomy is functional and multiply realizable (p. 271, note 2).
• A subdoxastic account. Attentional agency is more basic than cognitive agency and needs no rationality or propositional attitudes (p. 277).
• Veto is central “simply because if you cannot terminate your very own activity, then you cannot be said to be autonomous in any interesting sense” (p. 278).
• Domain-specific identity of mind-wandering with its neural correlate (pp. 286–287).
• The 2013 sampling figures, reused (pp. 292–294).

Key results

• Four capacities of second-order mental action (p. 276): “The imposing of rules on one’s own mental behaviour; explicit goal-selection, goal-commitment, goal-permanence; satisfaction of rationality constraints or rational guidance; intentional inhibition, suspension, or termination of an ongoing process.”
• Three degrees (p. 280): M-autonomy “comes in three major degrees: the phenomenally represented knowledge that oneself currently possesses this specific ability, executed attentional self-control, and cognitive self-control”.
• Tuning out keeps M-autonomy (p. 280): if we “voluntarily indulge in a fantasy while all the time knowing that we could terminate this inner activity at any instant — then we possess M-autonomy”.
• The moderate position (p. 282): “what we have lost is a specific form of knowledge, and not the ability itself … We are still persons, because we have the relevant potential.” Restated in the conclusion as “an epistemic deficit, a lack of conscious self-knowledge: an absence of representation that is not represented as an absence” (p. 296).
• The return is caused unconsciously (p. 283): “in most cases the re-emergence of an EAM will have to be caused by an unconscious event, perhaps by chance, perhaps based on an implicit knowledge about the relevant potential”.
• Confabulated self-causation (pp. 290–291): “the phenomenology of M-autonomy does not justify the claim that the functional property of M-autonomy was actually present”; “I simply must have been a mental agent in the preceding conscious moment too, because I had the ability, the potential, all the time. The transition must have been self-caused, because a ‘representational bridge’ has to be built to earlier instances of M-autonomy.”
• Meditation (pp. 288–289): focused practice has “cyclically recurring losses of M-autonomy, plus an equally recurring second-order mental action”; advanced open monitoring is “meta-awareness without an EAM”, “effortless and non-agentive”.
• The estimate (pp. 292–295), as in 2013: “not autonomous mental subjects for roughly two thirds of their conscious lifetime”. Its minutes give 52% (NOTE-1495).

Claims

id	claim	strength	support
C1	M-autonomy is the capacity for second-order mental action, centred on veto	moderate as a definition	conceptual (pp. 276–280)
C2	Mind-wandering loses the represented knowledge of the ability, not the ability	weak; asserted	argued from the sense that one could have stopped, against the absence of an EAM (pp. 282–283)
C3	An ability can be deliberately exercised only if represented in the self-model	weak; stated as a principle	no argument beyond its statement and the tuning-out case (p. 280)
C4	The return of control is usually caused by an unconscious event	weak	inference from C2 (p. 283)
C5	The return is represented as self-caused even when it was not, to keep the autobiographical self-model coherent	weak; hypothesis	Schooler et al. 2011 and Wegner 2002 cited (pp. 290–292)
C6	A process becomes personal-level by integration into an EAM and a sociocultural context of mutual ascription	moderate as a proposal	conceptual (pp. 284, 296)
C7	Humans lack M-autonomy for about two thirds of conscious life	weak; misstated	own minutes give about half (pp. 292–295)

Concepts

• Second-order mental action — mental action whose satisfaction conditions are “successfully influencing other mental actions or mental behaviours” (p. 277); “vertical” control.
• Actions and behaviours (p. 272) — both have satisfaction conditions; only actions have conscious goal-representation playing “a central causal role” and can be “terminated, suspended, intentionally inhibited”; behaviours involve “ownership without agency”.
• Principle of Autonomy by Phenomenal Self-Representation (p. 280) — quoted above.
• Epistemic autoregulation (pp. 275, 282) — “the active control of one’s own epistemic states”; the EAM “helps a self-conscious system in selecting and determining what it will know”.
• Epistemic self-causation (p. 290) — the phenomenal property of actively constituting oneself as a thinker, at the moment of “coming to”.
• Personhood (p. 284) — “a concept of social ontology”, “constituted not in brains, but in societies”; M-autonomy and the EAM are candidate necessary conditions.

Connections

It extends LIT-1942 and leaves out the SRB and the UI (note 1). It adds predictive processing (Hohwy 2013; Clark 2015) for the difference between attending to breath and attending to thought: the cognitive self-model predicts “much more abstract aspects of reality, in a wider temporal frame of reference, and not ongoing events on the sensory sheet” (p. 287), the theme LIT-1945 develops. Its EAM is the self-model theory’s (LIT-568, NOTE-460; THEORY-064) specialised to the knowing self, as in LIT-1871.

Limitations

• The central principle is stated, not argued. “Autonomy by Phenomenal Self-Representation” is introduced in one sentence (p. 280). The paper also says inhibition “can be an entirely subpersonal process” (LIT-1942, p. 13), which is control exercised without the representation; the principle is saved only by “deliberately and autonomously”. So the principle may be true by definition of “deliberately”.
• The moderate position has no test. Nothing in the paper separates “the ability is kept and unrepresented” from “the ability is lost”. LIT-1945’s H1 and H3 would.
• The confabulation hypothesis is unfalsified as stated. The return is “really triggered by unconscious events” in “many such cases” (p. 291); which cases, and how one would tell, is not said.
• The estimate repeats 2013’s arithmetic and its gap.

Open questions

• Does a capacity for mental control that is not represented in the self-model stay available, in the sense that a cue to the representation alone restores it? If so, the representation is a condition of efficacy; if not, the ability itself lapsed (QUESTION-367).
• In which returns of control did the person’s control system cause the return, and in which was it caused otherwise and represented as self-caused?

Bearing on the record’s own argument

Same structure, not shown same mechanism, unless said otherwise.

• CLAIM-047. The moderate position is this claim at the scale of mental action: during mind-wandering the agent possesses the regime of second-order mental control, “We are still persons, because we have the relevant potential” (p. 282), and the regime is not effective. CLAIM-047 says possession and efficacy differ and leaves what makes the difference to CLAIM-027. Metzinger names a candidate: representation of the capacity in the self-model (p. 280).
• CLAIM-027 and TERM-023. The candidate is a relation, not a content. Nothing in the competing mental processes changes when the EAM collapses or returns; what changes is whether the veto can be exercised. Of TERM-023’s dimensions it is accessibility, and the veto itself is TERM-017’s suspension. The four capacities (p. 276) also include precedence-like (“rules on one’s own mental behaviour”) and persistence-like (“goal-permanence”) relations. The support verdict is ARG-051, Proposed: alike in structure, the mechanism not shown, and the intervention that holds the processes fixed not run.
• CLAIM-122. The return of control is the cleaner half: “caused by an unconscious event, perhaps by chance” (p. 283), with no change in what the person pursues proposed. But no intervention is reported, so it is not a ground.
• CLAIM-156. Meditation as “cultivating M-autonomy” (p. 288) is a change in which mental processes may be stopped, not in what is valued, which is the claim’s route at the scale of thought. Two cautions. The paper’s advanced stage is “meta-awareness without an EAM” (p. 289), so the mature form exercises no veto at all, and if mindfulness integrates by suspension, in that stage it is suspension with no suspender represented. And a report of noticing and redirecting “may be false” (p. 289), so self-reported decentering cannot by itself show the mechanism the claim names. Nothing here touches discounting, so the claim’s defeater is not reached.
• CLAIM-022. Ownership and efficacy come apart both ways here. Mind-wandering has “ownership without agency” (p. 272); the regained control is attributed to oneself whether or not one caused it (p. 291); and whether a thought is the person’s is settled by integration into the EAM and by social ascription (p. 296), not by what was effective. The claim’s disclaimer holds in this literature too.
• CLAIM-523. Meta-awareness is “a necessary precondition for second-order mental action” (p. 277): at the personal level, legibility before volitional efficacy. The claim says the three achievements have no fixed order across organizations; for deliberate mental control in one mind Metzinger asserts one, and allows subpersonal correction without it.
• QUESTION-364. Personhood is “constituted not in brains, but in societies” (p. 284), and the representation that makes a process the person’s is scaffolded by “normative practices of mutually ascribing personhood” (p. 296). The standard by which a process is read as one’s own is supplied from outside the agent. LIT-1945 adds that it is installed before autobiographical memory.
• CLAIM-355 (not this reader’s line). Pp. 290–292 propose a specific self-model misdescription of who controls what: an unconsciously triggered return represented as “I generated this insight myself!”. It is offered as a hypothesis with no case, so it does not meet what the claim’s note of 2026-10-29 asks for.
• QUESTION-367 is filed from this reading: whether representing a capacity is a condition of its efficacy, so a relation the will-organization maintains, or only its accompaniment.
• No instruction for machine-learning practice.