Contribution
It names a problem and a research programme rather than a result. The problem: enactive and “embodied dynamicist” accounts give motivation a role only by identifying its efficacy with that of a system-level property (a global value constraint, an order parameter), “while assuming operational equivalency”, which “push[es] the problem of reification to a higher level of description” (abstract; §4). The programme: treat behaviour’s underdetermination by its material basis as where motivation shows up, and measure it as unpredictability, by information-theoretic entropy of neural, physiological and behavioural signals (§5.2.1).
Key insight
The difference a motivation makes cannot be seen as a motivated difference. It can only show up indirectly, as “a special subset of material changes that are not attributable to the activity’s material basis” (§5.2). Hence the working hypothesis: “The more an agent’s embodied activity is motivated, the less that activity is determined by its material basis” (§5.2). The effect is undirected, so direction has to come from somewhere else. Froese places it in the body’s attunement, built by evolution, development and learning: “the unfolding of future behavior can then be largely offloaded into the affordances and constraints of the agent–environment system” (§5.2.3).
Assumptions
- Axiom 1, motivational efficacy: “An agent’s motivations, as such, make a difference to the material basis of the agent’s behavior” (§5.1). Its support is “our first-person perspective”.
- Axiom 2, incomplete materiality: “It is impossible to measure how motivations, as such, make a difference to the material basis of behavior” (§5.1).
- Axiom 3, underdetermined materiality: “An agent’s behavior is underdetermined by its material basis” (§5.1). Adopted “in line with libertarian positions”, with quantum physics cited against “a clockwork universe”. Its extent “remains to be established”.
- Motivations are taken as given. The paper is about efficacy, “while taking the presence of motivations, such as value, for granted” (§2).
- Irruptions are small. They “would have to be exceptionally small-scale; otherwise, their consequences at the material level would have presumably already been noticed” (§5.2.2).
Key results
There are no theorems, models or data of the author’s own. What the paper establishes is a set of definitions and an argument.
- The diagnosis (§§3–4). Di Paolo, Rohde and De Jaegher’s enactive value moves value from a local appraisal to “a global constraint on all processes”, but keeps “identifying value with an aspect of the material living system”. Thompson and Varela’s consciousness-as-order-parameter fares no better: “it ultimately does not seem to matter whether we assign the subject’s efficacy to a collective property (order parameter) or to the efficacy of a material property (neural substrate)”. “Embodied dynamicism”, which appeals to “emergent top-down constraints in nonlinear dynamical systems”, shares the problem; Juarrero 1999 (ref. 56) and Kelso 1995 (ref. 57) are among its examples (§4).
- The libertarian double challenge (§4). Reject “the complete material determination of agent behavior” while not “falling into the opposite extreme of completely random agent behavior”. Irruption theory is offered as “a novel systematic framework that solves this double challenge”.
- Three questions, three theses (§5.2). How is an irruption quantified (irruption thesis: as material underdetermination, by entropy)? How does it make a difference (scalability thesis: “a poised system such that it amplifies microscopic irruptions to macroscopic fluctuations”)? How does it lead to appropriate behaviour (attunement thesis: “an attuned system such that it responds to scaled up irruptions in a context-sensitive and adaptive manner”)?
- Scalability (§5.2.2). Jonas’s cone balanced on its tip; chaos, defined as sensitivity “to microscopic, nearly infinitely small, differences in initial conditions” (ref. 86, which is Langton’s “edge of chaos” paper); 1/f noise; self-organized criticality; global-workspace “ignition”. “Future work will need to work out a more detailed neurophysiological account of this scalability.”
- Attunement (§5.2.3), the load-bearing passage. “Irruptions cannot directly control a motivated activity’s material basis; they can only serve to counteract existing material constraints by increasing their underdetermination. Essentially, the changes induced by irruptions are arbitrary with respect to those material constraints … they can loosen the organization of the current behavior but cannot force the formation of the behavior that is to come.”
- The self-optimization model as the answer (§5.2.3). A Hopfield network is reset at intervals to “an arbitrary configuration” and allowed to relax, with Hebbian learning; it “will undergo a process of generalization over the set of visited equilibria” and comes to reach “otherwise unvisited, deeper equilibria” (Watson, Buckley and Mills 2011, ref. 102; Fig. 4). Froese proposes that motivations, expressed as irruptions, “could serve the role of the ‘resets’”, which “holds potential to provide a novel systems theoretic answer to the main concern raised against libertarian interpretations … how an agent’s behavior can be effective even when it is underdetermined”.
- Evidence read as support (§5.2.1). Levels of consciousness track neural entropy (the “entropic brain”; dreamless sleep low, psychedelics high). Sergent et al. 2021’s increased variability in near-threshold auditory classification. Greater neural entropy with more task degrees of freedom. Schurger et al.’s finding that movement onset follows increased stochastic fluctuation, read as answering their challenge of making such fluctuations attributable to the subject.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Accounts that identify motivation’s efficacy with an order parameter or global constraint only move the reification problem up a level | moderate | argument, §4; that an order parameter does no causal work of its own is asserted with ref. 52, not shown |
| C2 | If motivations make a difference not reducible to the material basis, that difference must appear as increased underdetermination of the material activity | moderate | follows from Axioms 1–3 by the paper’s definition (§5.2); the axioms are posited |
| C3 | Irruptions can be indirectly quantified as information-theoretic entropy of neurophysiological signals | weak | proposed (§5.2.1); no measure is specified that separates irruption from other unpredictability |
| C4 | Findings that consciousness, near-threshold perception and task load go with higher neural entropy can be read as more motivational involvement | weak | reinterpretation of cited work (§5.2.1), with the added assumption that consciousness bounds motivational involvement from below |
| C5 | Chaos, 1/f noise and criticality could scale microscopic irruptions to behaviour | weak | named as candidates (§5.2.2); “future work” |
| C6 | Irruptions are undirected: they loosen constraints and cannot fix the next behaviour | strong (as a commitment of the theory) | stated, §5.2.3; it follows from Axiom 2 and the definition |
| C7 | Irruption together with self-optimization answers how underdetermined behaviour can be effective | weak | proposed by analogy with Watson et al.’s model (§5.2.3), in which the resets are arbitrary |
Concepts
- Motivated activity — “any embodied activity that is directed at an end”, including regulation; all cognitive activity counts (§1).
- Motivation-involving account — one on which motivation “as such” participates in generating behaviour, not via a material surrogate (§4).
- Irruption — the increase in material underdetermination associated with motivational involvement; from the Latin for a sudden bursting in, recalling Ashby’s “breaks” (§5.2).
- Embodied dynamicism — Thompson’s label for accounts that explain motivated activity by emergent top-down constraints in nonlinear dynamics (§4).
- Relaxed naturalism — nature counted as including an agent’s motivations as well as its materiality (§6).
Connections
- Fuchs, In Defense of the Human Being (ref. 39) supplies the starting position: freedom needs only that “actions are physically underdetermined” (§5).
- Ashby’s homeostat (ref. 69): adaptation by instability-triggered random “breaks”. Irruptions are modelled on these.
- Watson, Buckley and Mills 2011 (ref. 102) and Froese’s own self-optimization papers (refs. 11, 103–106).
- Juarrero 1999 (ref. 56; LIT-1990) is cited among the embodied-dynamicist accounts that the diagnosis targets. Kane is not cited. Neither is any work on control of chaos.
- Later work: Froese 2024’s absorption (Entropy 26:288, not held), and Sánchez-Cañizares 2024 and 2026 (LIT-1132; LIT-1992, NOTE-1506). Hölken’s critique (LIT-1957) is unread here.
Bearing on the record
- CLAIM-533: a held source that says sensitivity is not control, in its own terms. The scalability thesis is the Kane-style move the claim addresses: microscopic indeterminacy amplified by chaos or criticality into behaviour. The attunement thesis then concedes the claim’s central point: amplified irruptions “cannot directly control”, are “arbitrary with respect to those material constraints”, and “cannot force the formation of the behavior that is to come” (§5.2.3). Froese’s answer to where direction comes from is not a channel. It is attunement: a body shaped so that perturbations relax into good states. That answer has a consequence the claim predicts. In the self-optimization model the resets are “an arbitrary configuration”, so on the model offered nothing distinguishes a motivated irruption from noise: replace the irruptions with random resets and the model does the same. The agent’s share against noise, which the claim says amplification cannot supply, is not supplied here either. This is a second independent statement of the claim’s gap, after Bishop’s SEP entry, and from a libertarian.
- CLAIM-478: a rival reading of constraints. Froese’s §4 targets the family of views CLAIM-478 belongs to: top-down constraints and order parameters as what organizes behaviour. His objection is not that they fail to constrain. It is that crediting motivation to them makes motivation’s efficacy a material surrogate’s. That bears on whether CLAIM-478’s constraints can carry agency, not on whether they make structure, which is what CLAIM-478 claims. Noted, not a defeater.
- CLAIM-354 and QUESTION-071. On this theory an agent’s motivations act by widening a system’s options, so a manipulator’s injected noise and the agent’s irruption are, by Axiom 2, indistinguishable in any measurement. That is QUESTION-071’s present-structure problem in an extreme form, at the level of a single act. The paper does not raise it.
- THEORY-328 and THEORY-040 (the luck objection). The paper claims to answer “the main concern raised against libertarian interpretations”, the randomness horn. What it offers locates the effectiveness in attunement, which is the agent’s history and not the indeterminacy. That is the compatibilist’s resource, not a reply to luck. THEORY-1321 files this as Proposed.
- No instruction for machine-learning practice.
Limitations
- No model or measurement of its own. Every empirical point is a reinterpretation of others’ findings.
- Entropy does not separate irruption from noise. The paper offers no way to tell motivated unpredictability from any other. By Axiom 2 it cannot, except through correlation with first- and second-person reports (§5.2.1).
- The chaos citation is to Langton’s “Computation at the edge of chaos”, not to a source on sensitive dependence.
- The attunement model’s resets are arbitrary, so the model shows that random perturbation plus learning can improve an attractor landscape. It does not show that motivation adds anything to that.
- Froese says himself that the mechanism “may ultimately turn out to be conceptually impossible even, in principle” (§6).
Open questions
- What measurement would distinguish a motivated irruption from noise entering the same variables? The paper’s own axioms seem to rule one out.
- In the self-optimization model, does it matter when the resets come or where they land? If it does, the timing of resets is a channel, and the question becomes whose it is.
Note of 2026-11-03 (later): Juarrero 2023
Froese’s diagnosis fits Juarrero’s 2023 book as well as the 1999 one he cites (LIT-1995, NOTE-1532): it grants that governing constraints are “not spatiotemporally other than the components” (p. 81), and its reply to overdetermination, that constraints transfer no energy (pp. 25, 46), answers the closure worry, not the question of whose efficacy it is. It is not libertarian; it accepts “indetermination and uncertainty” (p. 235). Whether this holds of the 1999 book stays unread.