Source
- Mossio, Saborido & Moreno (2009), LIT-570, read in NOTE-444: §4.1 (closure, with Bénard cells and the candle flame as worked examples), §4.2 (organisational differentiation), endnote viii.
- Montévil & Mossio (2015), LIT-582, read in NOTE-468: Definitions 1–4, the remark that closure needs both slow and fast dependence, footnote 25 (closure versus a cycle of processes), and footnotes 4 and 27 (closure does not define the organism).
- Mossio & Bich (2014), LIT-1865, read in NOTE-1443: closure versus cycles (MS p. 15), the division of labour folded into closure (MS pp. 10–11), and dissipative structures (MS pp. 18–19).
The question
NOTE-094 reports the dispute. Closure accounts, autopoiesis and teleodynamics are each charged with failing to separate a living system from a self-organising dissipative structure such as a candle flame or a tornado. Until this batch the record held no answer. Prigogine (LIT-516, NOTE-424) gives the dissipative-structure criterion: an open system held far from equilibrium, maintained by flows, whose order is born from an instability. NOTE-424 records that a flame meets it on the same terms, and so, by the lecture’s own §1, do living systems. Friston (LIT-526) names the flame and excludes it in one sentence, by a criterion that THEORY-073 argues cannot do that work.
The claim, in three steps
1. Far-from-equilibrium self-maintenance is not the line. Organisms and flames both have it. That is Prigogine’s picture as NOTE-424 reads it, and Mossio et al. begin from it. A self-maintaining system is one in which a macroscopic pattern exerts “at least one ‘top-down’ constraint” that keeps up the flow enabling its microscopic dynamics (NOTE-444).
2. Closure in the broad sense is not the line either. Mossio et al. give the flame as a worked example. The flame keeps the temperature above the combustion threshold, vaporises wax and drives the convection that brings oxygen in, and the reactions in turn keep generating the flame (LIT-570, p. 9). That is “organizational closure, i.e. a circular causal relation between some macroscopic … pattern … and the microscopic … dynamics”. “Minimal self-maintaining systems, such as flames or hurricanes, do instantiate closure, but their components do not have functions.”
3. The line is differentiation. “Dissipative systems make a unique contribution to their own maintenance.” In a flame the many chemical components “all ‘converge’ to generate a single macroscopic pattern”, so no component makes a contribution of its own. Functions, and with them the organism’s kind of organisation, need “different and localizable patterns or structures, each making a specific contribution” (§4.2). Material complexity is not the same thing: a flame has many molecular species and is still undifferentiated. Endnote viii describes how a single-catalyst autocatalytic set becomes differentiated, and so functional, once it also makes a membrane.
The 2015 closure says the same thing in another vocabulary. This is the record’s reading, not either paper’s. Montévil and Mossio redefine closure. A constraint is something that acts on a process while its relevant aspects stay conserved at that process’s time scale (Definition 1). Closure is a set of constraints in which every member depends directly on another member and is generative for another (Definition 4), and such a set “must include” both a slow and a fast dependence. A single macroscopic pattern cannot satisfy that by itself: it would have to depend on and generate another constraint in the set, and there is no other. A cycle of transformations also fails, because its entities “do not act as constraints on each other” (footnote 25, the hydrologic cycle). So the 2015 closure already requires several distinct constraints, each with its own contribution and its own time scale. That is differentiation in the 2009 sense, built into closure. NOTE-468 and NOTE-444 both reach this reading and both mark it as an inference. The title states the line in the form common to the two papers.
2026-10-30. The folding is now the authors’ own, a year before the 2015 definitions. Mossio and Bich (LIT-1865, NOTE-1443) write that under closure “each component makes a specific and complementary contribution”, and that “by virtue of the interplay between division of labour and mutual dependence that it implies, closure captures a fundamental aspect of the idea of ‘organisation’” (MS pp. 10–11). They also withdraw step 2’s concession from 2009: dissipative structures “do not realise closure”, and would at most “generate a single macroscopic constraint (the structure itself)” (MS p. 19). That is the single-constraint half of the title in the authors’ words. They never mention distinct time scales among constraints, and they analyse no flame, so their verdict is asserted, which the promote_when says cannot settle the account. Moreno and Mossio’s Introduction (NOTE-1437, Skimmed) previews the same line, saying closure “fundamentally differs” from self-organisation (p. xxix), and leaves the argument to chapter 1, unread.
2026-10-20. The first half of this reading is now proved, not inferred: CLAIM-395 shows from Definition 4 that a closed set has at least two constraints, and, if each constraint is produced at the conserved scale of a generator in the set and not at its own, that it has both a slow and a fast dependence. The paper states the second property and does not prove it. The step from two constraints at distinct scales to the 2009 paper’s differentiation is still the record’s interpretation.
What this does not say
- It does not define life or the organism. “Although an organism necessarily realises closure, a system realising closure is not necessarily an organism” (LIT-582, footnote 27). The claim is that this condition, which organisms have, is what flame-like dissipative structures lack. It is a necessary condition that separates the two, not a sufficient condition for being alive. A protocell or an autocatalytic set with a membrane may meet it, and the account takes no view on whether such a system is alive.
- No paper shows that a flame fails the 2015 closure. Neither paper analyses one. Montévil and Mossio leave open whether “some specific classes of non-biological natural systems (such as, for instance, complex chemical systems) might be pertinently said to realise closure” (footnote 4). The obvious test is the candle itself. The wick is a constraint on the flow of fuel, but the flame does not make it. The wax pool is made by the flame’s heat, but it is fuel consumed by the process, not a constraint conserved through it. On that reading the flame has one constraint of its own and fails. The reading is the record’s, and promote_when asks for it to be done properly.
- “Differentiation” has no measure. The 2009 paper leaves “different and localizable” informal (NOTE-444, Limitations). The 2015 definitions make it checkable in principle, but they have been applied only to fragments and toy diagrams.
- It does not answer Baedke’s objection, which is a different question. Baedke (LIT-167, NOTE-148) argues that closure and persistence criteria are too permissive, because colonies, holobionts and ecosystems meet them. That objection is about which living individual counts as the organism, not about where life ends and flame begins. LIT-582’s tendency to closure and nested closures are its answer to that question, and they are labelled preliminary.
- It takes Nahas and Sachs’s first check. NOTE-094 asks any account of this kind to say which project it serves. This one is a claim about what organisms are, the philosophical project. It does not say that a model in biology must demarcate, which Nahas and Sachs argue it need not (C3 in NOTE-094).
- It is not about agency. Being an organism in this sense is not being an agent, and THEORY-071 adds two further conditions for that. Nor is it about any of the four unities of ADR-024.
Connections
- Friston’s criterion (LIT-526). It is an account of the same
demarcation, so it is this account’s natural rival. But the record
holds it as a LIT corrected by two critiques, with no THEORY stating
it, so no
rivalsis declared. The two share an ingredient. Friston says a blanket needs couplings that “change slowly with respect to the states”, and Montévil and Mossio say a constraint is conserved at the time scale of the process it acts on. Both put the line at a separation of time scales. They differ on production. For closure, the slow structures must be made by processes that the system’s other constraints shape. A blanket needs only to persist, and THEORY-073 argues it cannot represent a boundary the system makes. - Deacon (LIT-529, unread). Nahas and Sachs report his teleodynamics as the coupling of two mutually limiting dissipative processes (NOTE-094, §2.1). As reported, it also draws the line at more than one self-organising process constraining another, and it is the nearest neighbour of this account. The book is Deferred, so nothing here rests on it.
- THEORY-030 (Landauer). It notes that a structure held away from equilibrium “dissipates continually”. That standing cost belongs to flames and cells alike, which fits step 1. The thermodynamic theories in the record (THEORY-030, THEORY-026) price information processing. Neither draws or depends on this line, and no relation is declared.
- THEORY-071 (minimal agency) builds agency on an autonomous organisation of mutually enabling processes. It does not extend this account, and the reason is given there.
- “What Lives?” (LIT-211, NOTE-109). Its “Dissipative Self-Organizing Systems” and “Cognitive Autonomy” clusters map onto the two sides of this line, and this account says what separates them.
Note of 2026-10-30: what holds it back
The two analyses the promote_when asks for are undone: no canonical dissipative structure has been analysed under Definitions 1–4 of LIT-582, and no protocell model shown to realise closure under them. Mossio and Bich (NOTE-1443) answer the promote_when’s question about the authors’ view, folding the division of labour into closure, but deny dissipative structures closure without analysing one; Biological Autonomy’s chapters 1 and 3 (LIT-536) are unread beyond the Introduction (NOTE-1437).