NOTE-1532: Juarrero — Context Changes Everything

Contribution

A book-length synthesis, not a result. It restates the constraint vocabulary of Dynamics in Action (1999, LIT-1990, which it cites for enabling constraints, second-order context-dependent constraints and hermeneutics) and extends it from intentional action to “complex systems, living and nonliving” (p. 23): convection, lasers, homeostasis, ecosystems, economies, cultures. Its claim is ontological: coherent dynamics generated by constraints are “real and novel, interactional types of entities” (p. 72) with top-down powers. The author calls it “a work of speculative metaphysics” (p. 32). There are no models, theorems or data of the author’s own; the empirical chapter (15) interprets two published neuroscience studies.

Key insight

Causation by constraint. A constraint changes what can happen and how likely it is without transferring energy: “Constraints are entities, processes, events, relations, or conditions that raise or lower barriers to energy flow without directly transferring kinetic energy” (p. 40). Timing a swing’s kick “does not impart additional energy to the swing over and beyond the kick itself” (p. 46). On that basis the book claims that a whole can act on its parts without overdetermination or a breach of causal closure (pp. 25, 46, 186): bottom-up enabling constraints make a coherent dynamic, and its constitutive constraints then “do double duty as governing constraints” (p. 79), confining components “to a subregion of the system’s overall possibility space” (p. 80).

Assumptions

• Open systems far from equilibrium throughout: “The one prerequisite: conditions must be open and away from thermal equilibrium” (p. 233).
• Probability as the measure of constraint. Constraints are “encoded as probabilities” (p. 46); context-dependent constraints are “measured with conditional probability: X is more or less likely to happen, given the presence of constraint Y, in context Z” (p. 67). “Selection” and “filtering” are to be “translated into the language of conditional probability” (p. 82).
• Realism about possibility spaces and order parameters. Possibility spaces “are not solely epistemic; they are real” (p. 47); Pattee’s “statistical” loss of detail tracks real coherence, not only epistemic simplification (pp. 96–97).
• Wholes are not other than parts. Governing constraints “are not spatiotemporally other than the components and constraints from which they were formed and which they control” (p. 81; again pp. 84, 95, 233).
• Constraints transfer no energy, so their effects escape conservation and closure arguments (p. 46). The book does not argue for this beyond the examples.

Key results

The kinds of constraint (chs. 3, 4, 6)

• Context-independent constraints (gradients, fields, buffers, isolation) “take or keep conditions away from equiprobability” (p. 67).
• Context-dependent constraints “take conditions away from independence (Gatlin 1972)” by making processes “mutually conditional on one another and on the history and context in which they formed” (p. 67). Kauffman’s buttons, Huygens’s clocks and Bénard cells are the examples (pp. 67–69).
• Enabling constraints (Pattee 1973; Salthe 1985; Juarrero 1999) “irreversibly link and couple previously separate … entities at the same scale as the constraints” and “drive parts-to-whole phase transitions” (p. 71).
• Constitutive/governing constraints (Mossio 2013; Moreno and Mossio 2016) hold the result together, and act top down as “second-order context-dependent constraints (Juarrero 1999)” (p. 81), “often as negative feedback loops” (p. 79), as “distributed mechanisms of control” embodied in order parameters (p. 80).
• Temporal constraints make one event conditional on an earlier one; sequences become units (“chunked into a context-independent constraint, a ritual or protocol”, p. 74), and complex systems become historical: “They carry their history on their backs” (p. 75).

Time scales and levels (chs. 6, 12)

• Coherence makes slower patterns. Entrained pendulums’ “overarching wave patterns … are longer and slower than the swings of the individual pendulums”, and “Different cycling rates at different scales of organization will play a key role in refashioning a new understanding of the boundaries of identity” (p. 68).
• Inside a coherent dynamic, coupling is faster. Its interactions become “more tightly coupled, their feedback loops and relaxation times shorter, faster, and more regular” than the environment’s or the elements’ (p. 72).
• Boundaries are rate differences, not material. “Rate differences between full predator–prey cycles and the life span of the individual animals entrained in those cycles constitute the boundaries between parts and wholes” (p. 77).
• Levels are rate-demarcated (Salthe 1985, 72): “Boundaries between levels of organization and constraint are therefore typically marked by sharp discontinuities in order parameters and average process and relaxation rates”, and “overarching hierarchical levels of organization are usually slower than their components” (p. 183).
• Why levels interact by constraint, quoting Salthe: processes at different levels cannot interact energetically “because [faster] lower-level dynamics would have been completed before feedback from the [slower] upper level arrived” (p. 187). Whole-to-part relations are “rate independent” (Salthe 1985, 72): governing constraints “inform and influence” components “without participating in them dynamically” (p. 188).
• Control is lost when the rates converge. The asymmetry “can be destroyed by either excessively accelerating the dynamics of an embedding environment or slowing down the rates of the focal-level system beyond its tolerance range” (p. 189); “Lack of asymmetry in cycling rates … reflects the absence of control constraint by one on the other, regardless of signal amplitude” (p. 190). A component sped up becomes “invisible” to its context; the tumour is the example (p. 190).
• Modularity and near-decomposability come from Simon, but the book holds that “this matryoshka doll image … is misleading”, since hypercycles and closure “confound the clean material separability of layers”, so decomposability is analysed “dynamically, as differences … in phases, rates, scales” (p. 183).

Closure (ch. 7)

• Montévil and Mossio’s closure of constraints is quoted from Moreno and Mossio 2015 (pp. 20–21) and their figure reproduced (pp. 93–95, fig. 7.1). The book adds that the closed loop C₂–C₄ “spans (and is conserved throughout) a longer interval, τ2 through τ4” and “persists despite turnover in its components” (p. 94), and glosses it as “a virtual governor … a hypercatalyst” (p. 93).
• It reports their view that physical and chemical dissipative structures “still depend on the externally set … boundary conditions” and are “not yet self-constraining” (p. 92), and accepts it: closure of constraints is what makes a system “self-determining” (p. 94). Yet it also says “Dissipative processes like the Rayleigh–Bénard cells and BZ reactions described earlier are autopoietic” (p. 91), which neither Maturana and Varela nor Montévil and Mossio say.

Many-to-one transitions and analog control (ch. 11)

• Coherence-making is “the generation of such Many-to-One functions” (p. 158): different physical inputs recoded as the same “effective input” relative to order parameters (Patten and Auble 1980; pp. 156–157).
• Top-down control is analog: order parameters take continuous values, so control “in reference to continuous values supports a wider range of behavior” (p. 168) and is “more tolerant of error” (p. 169). Ashby’s law of requisite variety is invoked (p. 168). The examples are fuzzy-logic thermostats and anti-lock brakes (pp. 168–169).
• The analog layer “supervenes on precise and accurate digital input–output relations” (p. 175).
• Qualia are hypothesized as tags on recoded integration that speed action (pp. 163–166). Stated as a hypothesis.

The neuroscience (ch. 15)

• Churchland et al. 2012 and Shenoy et al. 2011: preparatory activity occupies a low-dimensional subregion whose end state predicts reaction time and movement; population activity rotates (pp. 224–227).
• Mante, Sussillo et al. 2013: in PFC a line attractor (“axis of choice”) and a context-dependent “selection vector” integrate only the cued evidence (pp. 227–232).
• The book reads the subspace as “top-down governing constraints” (p. 230), the axis of choice as “the neural correlate of a prior intention” (pp. 229–230) and its persistence as “the standing cause of the response” (p. 230). The studies’ own evidence, as the chapter reports it, is dimensionality reduction, prediction, and a trained-network simulation; no intervention on the population pattern is described.

Claims

id	claim	strength	support
C1	Constraints bring about effects without transferring energy, so top-down influence by constraint does not overdetermine or break closure	weak	asserted with examples (swing, seesaw, vasculature; pp. 40, 46, 82); no argument that the constraint’s own maintenance transfers no energy
C2	Context-dependent constraints make processes conditionally dependent and drive phase transitions to coherent dynamics	moderate	textbook cases (Bénard, Huygens, Kauffman’s buttons, BZ), described, not modelled (pp. 67–69, 90–91)
C3	The constitutive constraints of a coherent dynamic govern its components top down by confining them to a subregion of state space	moderate as a description; weak as a causal claim	order-parameter and attractor cases (pp. 79–84); ch. 15’s correlational and model evidence
C4	Levels are demarcated by rate discontinuities, higher levels are slower, and control between levels is lost when the rates converge	moderate	hierarchy theory (Salthe 1985, Allen and Starr 1982, Allen, O’Neill and Hoekstra 1984, Simon), with institutional and oncological examples (pp. 183–190)
C5	Control by continuous order parameters (analog) is more robust and energy-efficient than digital control	weak	analogy (thermostats, ABS, Dyson) and speculation about brains (pp. 168–175)
C6	Coherent dynamics carry the marks of the conditions of their formation in their present structure	weak	examples (snowflakes, graphite and diamond; pp. 26–27, 76); no condition on when this holds
C7	Closure of constraints, not of processes, is what makes a system self-determining	moderate	adopted from Montévil and Mossio (pp. 92–95)

Concepts

• Constraint — anything that raises or lowers barriers to energy flow without transferring kinetic energy (p. 40). No time-scale condition.
• Context-independent / context-dependent constraint — away from equiprobability / away from independence (Gatlin; p. 67).
• Enabling constraint — couples separate entities at the same scale, bottom up (p. 71).
• Constitutive / governing constraint — the constraint regime of the coherent whole, acting top down as a “second-order context-dependent constraint” (p. 81).
• Interactional type — a coherent, multiply realizable dynamic generated by constraints; “Types are not things” (p. 27).
• Effective input / output — physical signals as recoded relative to a system’s order parameters (Patten and Auble; pp. 156–158).
• Analog control — control by continuous order parameters (pp. 167–175).
• Profile — a possibility space’s “probability contour” or “dynamic signature” (p. 46).

Connections

• Juarrero 1999 (LIT-1990), cited for enabling constraints (p. 71), second-order context-dependent constraints (p. 81), the content of intentions (p. 179) and hermeneutics (p. 163).
• Montévil and Mossio 2015; Moreno and Mossio 2015, 2016 (LIT-582, NOTE-468): closure of constraints, adopted (pp. 92–95).
• Gatlin 1972: the two kinds of constraint as departures from equiprobability and from independence.
• Salthe 1985, Allen and Starr 1982, Pattee 1973, Simon 1969: the hierarchy theory behind chapter 12.
• Haken: “slaving”, called “unfortunate terminology” (p. 81); used for lasers and cities, with no time-scale argument.
• Mante, Sussillo et al. 2013; Churchland et al. 2012 (ch. 15).
• Not cited in the chapters read or found by search: Kane, Bishop, Atmanspacher, contextual emergence, Fenichel or Tikhonov, any work on control of chaos. The book cited by Froese is the 1999 one.

Bearing on the record

• CLAIM-478: philosophical prior art for both directions, without the mechanism. Organization makes separation: entrained pendulums produce a pattern “longer and slower” than their swings (p. 68), and higher levels are “usually slower than their components” (p. 183). Separation makes control: inter-level influence is by constraint because the faster level completes before the slower one’s feedback arrives (p. 187), and “Lack of asymmetry in cycling rates … reflects the absence of control constraint” (p. 190). That is close to the claim’s thesis and to its defeater’s negation. But the book has no slow manifold, no singular perturbation and no critical slowing; its source is hierarchy theory (Salthe), and its constraint (p. 40) is not relative to a time scale, unlike LIT-582’s. It also says that within a coherent dynamic the internal loops get faster (p. 72), which fits the claim (fast internal relaxation is what settles a relation) but which the book does not connect to the slow whole. Bousquet et al.’s critical separation (LIT-1292) is the formal counterpart of its p. 189: an asymmetry that fails when the rates come together. THEORY-1306 states what the book holds here.
• CLAIM-533: the 2023 book has nothing on sensitivity or chaos, and credits control to confinement. Its control is governing constraints holding components to a subregion “through cascades of negative feedback” (p. 101; pp. 79–80, 167), measured against Ashby’s requisite variety (p. 168). Amplification appears only as the bottom-up route to a phase transition (p. 25). That fits the claim’s “sensitivity is not control”, and Jennings’s emphasis on constraint over amplification, but it is not an argument for either.
• Froese’s objection (NOTE-1521), read against the 2023 book. Froese says it makes no difference to the subject’s efficacy whether one credits an order parameter or the neural substrate. The 2023 book grants his premise: governing constraints “are not spatiotemporally other than the components” (p. 81). Its answer to overdetermination is that constraints transfer no energy (pp. 25, 46). That answers the closure worry, not Froese’s question about whose efficacy it is. And the book is not libertarian: “The price of such creativity and uniqueness is indetermination and uncertainty” (p. 235), with no account of an agent’s control over that indetermination. So Froese’s diagnosis fits the 2023 book; whether it fits the 1999 book is unread.
• TERM-070: Juarrero’s context-dependent constraint is not the record’s constraint. Hers is Gatlin’s: a departure from independence, a conditional dependence among processes (p. 67), with no condition that it be conserved at the time scale of the process it acts on. TERM-070’s constraint is Montévil and Mossio’s, which has that condition. The two meet only in chapter 7, where the book adopts TERM-070’s closure word for word (pp. 93–94) and adds the time-scale reading of the loop (p. 94). Her “governing constraint” is the nearer match: it is a whole’s regime acting on parts, and her chapter 12 makes it slower. A reader citing Juarrero for “constraint” in TERM-070’s sense should cite chapter 7 only.
• TERM-100 and CLAIM-532: self-determination here is closure, so maintenance, and the book’s claim about history is the opposite of CASE-191’s result. Self-determination is “self-constraint” by closure (pp. 93–94), a present organization; the book offers no formation condition. But it holds that coherent dynamics “carry their history on their backs” (p. 75) and that constrained trajectories “make manifest the context and the moment in which they were created” (p. 76), with graphite and diamond as the case. CASE-191 builds two agents with different formation histories that reach the same organization, which no knockout separates afterwards. The book’s own multiple realizability (one type, many routes) cuts the same way as the case. Its examples differ in kind (different attractors); the case is one attractor reached two ways.
• THEORY-1314 (Jennings, contextual emergence): a related view with a different boundary. The book does not use “contextual emergence”. It shares the Bénard case and says convection cells “depend on externally set boundary conditions” (p. 91). It even has the skull: “a central nervous system with a brain protected by a skull and capable of supporting context-dependent coordination might have evolved” (p. 168). But its boundaries are rate differences, not material ones (“boundaries need not be tangible, material structures”, p. 77), and it holds that the governing regime is not other than the parts (p. 81). So it does not supply the irreducible boundary that THEORY-1314’s promote_when asks for, and its neural evidence (ch. 15) is of the same correlational and model kind.
• As a stand-in for LIT-1990. It restates the 1999 constraint vocabulary and cites the 1999 book for its key terms, so it is a fair source for what Juarrero means by enabling, governing and second-order constraints. It is not a source for the 1999 book’s action theory (intentions as constraints on behavior), which chapters 13 and 15 touch and this reading did not cover in full.

Limitations

• Speculative and unformalized by its author’s account. Claims C1, C5 and C6 are argued by example.
• The energy-free claim (C1) is not examined: a governing constraint is maintained by dissipating processes (p. 82: vasculature), so whether its influence involves no energy transfer depends on how the system is cut.
• Inconsistency on autopoiesis (p. 91 against p. 92).
• The neuroscience is read as top-down causation where the studies as reported give prediction and model fits.
• Half the chapters were not read (see history), including the philosophical case study (ch. 14, Shapiro, Gillett and Lange), the persistence and entrenchment chapters (9, 10) and the update of the backstory (13).

Open questions

• What, in the book’s terms, separates a constraint that is slower than the process it constrains from one that is not? Its swing-timing example (p. 46) is a phase relation, not a slow variable; does it fall under CLAIM-478 or against it?
• Does the 1999 book’s account of intentions as context-sensitive constraints say more about control against noise than this one does?
• Chapters 9 and 10 (persistence, entrenchment) may bear on CLAIM-331 and CLAIM-477; unread here.