THEORY-220: A far-from-equilibrium system that contributes to its own maintenance has functions relative to itself, with no history or observer needed, and one that switches among self-maintaining processes has indications of success that can be false by its own lights

Source

Bickhard (2000), LIT-1029, read in NOTE-729: the sections Autonomy (the ladder from minimal to recursive self-maintenance), Self-Maintenance and Function (serving and having a function, dynamic presupposition, against etiology) and Recursive Self-Maintenance and Representation (interactive content, system-detectable error).

What was actually shown

Nothing is measured. The paper argues from cases and from two premises: emergence must be causally efficacious, and only a system’s current state is efficacious.

Function. A far-from-equilibrium process ceases if it reaches equilibrium, so it needs exchanges with its environment. Some such processes contribute to their own continuation. A candle flame keeps itself above combustion temperature, vaporises wax and drives the convection that brings oxygen. Those contributions are functional “for that system”. Function is relative to the system served, which is why a parasite’s gut is functional for the parasite and dysfunctional for the host. Having a function, as distinct from serving one, is being of a type that the whole’s processes presuppose to meet some condition; the paper sketches this and defers it to Christensen & Bickhard.

Against etiology. On a history-based account a molecule-for-molecule duplicate of a lion has no functions, so function is not a property of current state and is causally epiphenomenal. History explains why a kidney exists, not what its having a function is.

Representation. A system that maintains its self-maintenance across changing conditions must switch among processes, and so must differentiate environments. Differentiations alone have no content. Content belongs to the indication that a further interaction will succeed after a given differentiation, and is the implicit set of conditions under which it would. The indication can be false, and the system can register that from an internal outcome. Information semantics, Fodor’s asymmetric dependence and Dretske’s account are argued to make error, or error the system can detect, impossible.

What this does not say

• It does not say a flame is alive, or an agent. It gives the flame functions and nothing more. The organism line is not drawn.
• It does not say what ought to persist. The norm is persisting versus ceasing for a given system. “Normative” here is not value, which agrees with CLAIM-075.
• It does not define having a function or malfunction rigorously. The typing that carries malfunction is a sketch, deferred to another paper.
• It does not say anything about designed or artificial systems beyond one closing sentence that their design will have to meet these conditions. The application to simulations below is the record’s.
• It does not give an individuality condition. The system whose persistence fixes the functions is taken as given.

Connections

• CLAIM-288 (modeller-fixed objectives cannot show self-constitution). Corroborated, not grounded. On this account a function is relative to the system whose far-from-equilibrium persistence it serves, and a function that exists only in an ascription or explanation is not a natural function at all. A reward the modeller writes is not any operator’s persistence condition, so it confers no function on anything in the simulation. This is a second route to the claim’s premise, beside THEORY-071 and LIT-1017. If this account were rejected the premise would still stand on those.
• QUESTION-087 (an intrinsic standard for malfunction or capture). A partial answer, yes. A self-reproducing authority structure serves functions for itself, and whether it is dysfunctional for the organization is fixed by the organization’s own persistence conditions, as the parasite’s gut is fixed by the host’s. A part the whole’s processes presuppose to meet a condition, and that fails to, malfunctions. The limit: persistence is the only standard. Captured authority that keeps the organization alive is not dysfunctional here, so the account cannot tell entrenched authority from merited trust when both keep the organization going. That is the gap CLAIM-075 marks.
• CLAIM-284 (self as endogenous governance). Not supported. Recursive self-maintenance is governing among regimes, the second of CLAIM-025’s layers. Transforming the organization that governs does not appear, and selves and authority are not discussed. No relation is declared.
• THEORY-072 (organisms versus flames). A disagreement about functions, not about the demarcation. One of THEORY-072’s sources holds that a flame’s components have no functions; this account gives the flame functions it serves, though no typed parts that have them. Its infrastructure tiers, structure stable on a slower time scale that directs energy and switches processes, sit near THEORY-072’s differentiation, but the paper gives no closure condition. THEORY-072’s title is not contradicted, since this account draws no organism line. No rivals is declared, because the two accounts answer different questions and could both be right about their own.
• THEORY-071 (minimal agency) and LIT-1017 (Di Paolo). The same bacterium, and a near match between self-maintenance and recursive self-maintenance here and robustness and adaptivity in Di Paolo. The difference is where norms begin. Di Paolo denies a merely conserving system any norm it can use; this account gives it functions but puts error the system can detect only at the recursive level, which is about where Di Paolo puts norms for the system. THEORY-071’s individuality and asymmetry conditions have no counterpart here.

Note of 2026-10-18: a contribution measure of the kind the promote_when asks for, not yet passing its test

A16 of the exchange of 2026-10-17, under the owner’s question about normativity from first principles (QUESTION-128), proposed the model this entry’s promote_when asks for, contribution to self-maintenance defined from the dynamics alone:

Δ_p(T) = Pr(τ_O > T | do(p intact)) − Pr(τ_O > T | do(p disabled)),

with τ_O the time at which the specified organization stops being maintained. A16 called it “a functional contribution measure, not yet a moral value or a proof of normativity”, dependent on “how (\mathcal O), the intervention, and organizational failure are specified”.

It does not yet meet the promote_when’s own test. Any component whose removal shortens persistence scores positive, so Δ_p as stated cannot tell functions “from mere causal contributions to persistence that no one counts as functional”. It was not applied to the flame, the protocell or the designed controller. THEORY-628 (Di Paolo) is a further objection to the step A16 takes from self-maintenance to “endogenous conditions of success and failure”: persistence against cessation is an all-or-nothing norm “available to an observer, not to the system”, and a graded norm needs adaptivity. Δ_p measures the all-or-nothing norm. The status is unchanged.

Note of 2026-10-29: what holds it back

No model defines contribution to self-maintenance from the dynamics in a way that tells functions from mere causal contributions to persistence. The one the record holds, A16’s Δ_p (the note of 2026-10-18 above), scores every component whose removal shortens persistence as positive, and it was applied to none of the flame, the protocell or the controller. Having a function, which malfunction needs, is deferred by the source to Christensen & Bickhard (2002), which the record does not hold.