Canonical · Sovereignty

What Is an Attractor

The thing that holds itself together

In this framework an attractor is a recursive structure that produces the conditions for its own persistence (reappearance of identity). The definition may feel severe or unlikely, but it is observable and quite common as a natural phenomenon. To carry its own regime forward — to persist as long as it can, and when it cannot, to leave a record from which a new one ignites in its place — is the canonical pursuit of this framework.

“Attractor” is already a working term across biology, neuroscience, and ecology, not a mathematical monopoly. Principia Attractum borrows the word but reserves it for a different object — it does not sharpen the dynamical-systems notion so much as repurpose the term. In ordinary nonlinear dynamics an attractor is only a set that trajectories converge to: no boundary, no self-production, no maintenance cost, no history. The object defined here requires all four. So this is a substitution, not a refinement of the same concept, and it is honest to say so plainly — the four-condition test below is what makes the attractor/attractlet distinction possible.

The plain idea

Start with the word. An attractor recursively returns to its own regime — which is exactly why it looks “attracted” to itself, and why the name fits. That appearance is the easy half. What makes something an attractor here, rather than a look-alike, is where the return comes from — produced from within and paid for, not imposed from outside.

Everything that lasts is held together somehow. The question is by what. There are two answers, and the whole idea turns on the difference between them.

A structure is exogenous when what holds it together comes from outside it. Something external supplies the order: builds the boundary, feeds the supply, corrects the drift. The structure persists as long as that external hand keeps working, and stops when it withdraws. The stability is real, but it is borrowed.

A structure is endogenous when the thing holding it together is the structure itself. It produces its own order: it closes its own loop, makes and defends its own boundary, carries its own history, and pays the cost of continuing out of its own activity. Nothing outside is tending it. The stability is its own.

In this framework, an attractor is an endogenous structure. That is the whole of the idea, before any of the machinery. It is not merely stable and not merely persistent; it is self-sustaining, because the source of its persistence is inside the boundary rather than outside it. Be careful with the word ‘self-sustaining,’ though: it does not sustain its existence out of nothing. It sustains its identity, by re-establishing its own boundary and paying its own maintenance, out of the throughput passing through it. Cut off the throughput and it stops like anything else; what is its own is not the supply but the ordering the supply is put to. The four conditions below are just what "endogenous" turns out to require when you state it precisely.

The distinction is not about needing nothing. Every structure draws on the world for raw energy and material; that never stops. Endogenous does not mean self-sufficient. It means the structuring is done from within. A living cell eats and breathes, and still it is the cell that builds the cell. That is the honest picture of an attractor.

This is a stricter word than in ordinary nonlinear dynamics, where any state a system settles into can be called an attractor. Here the term carries a sovereignty requirement: to earn the name, a structure must pass a four-part test, and passing it is all-or-nothing.

A note on the word “attractor.” We use this word in two ways, wide and narrow, and this page means the narrow one. Here is the difference, and why it isn’t a contradiction.

The wide sense. Any structure that holds itself together through its own recursion counts as an attractor. The taxonomy sorts these into three classes, ranked by how much a structure’s own activity has managed to build: base (it holds a simple loop), coordination (it organizes several parts together), and sovereign (it makes and defends its own boundary and pays its own way). All three are real attractors. Base and coordination are just the earlier stages — recursive, but not yet all the way to sovereign.

The narrow sense. On this page, “attractor” means only the top class: the sovereign one — the structure that passes the four-condition test below. So when you read “an attractor” here, read “a sovereign attractor.”

That is why phrases like “a base attractor” or “a coordination attractor” are not slip-ups. They are the wide sense being used on purpose, naming the earlier classes. Whenever it actually matters which sense we mean, we write sovereign attractor out in full. It is one word used at two zoom levels, not two definitions fighting each other.

One word never stretches either way: an attractlet. That is the opposite category — a look-alike held together from outside, with no recursion of its own — and it is never an attractor in either sense.

A worked example

Kauffman's autocatalytic set

Stuart Kauffman asks what the least thing is that could count as alive, and answers with a collectively autocatalytic set: a web of molecules in which every member is produced by reactions the set itself catalyzes. No single molecule makes itself. The set, taken as a whole, makes the set.

Read against the four conditions, it lands on three and requires an argument on the second. The mutual catalysis is the closed loop (Condition 1). Its boundary is the reaction-closure itself: the inside is the set of species the closure keeps regenerating, the outside is everything it does not, and the set makes and holds that distinction with no membrane required (Condition 3). And it runs only while fed a throughput of raw precursors, paying its way the whole time (Condition 4).

Condition 2 is the one that does not come for free, and we should not wave it through. A bare autocatalytic set’s composition is its state, and calling that state “a compressed record of its own recursion” risks relabelling rather than establishing anything — the same redescription-without-content move we warn against elsewhere. Whether such a set recovers from a shock is set-specific: many autocatalytic sets are fragile and collapse rather than re-ignite. So the honest reading is that Kauffman’s object is a clean reference case for Conditions 1, 3, and 4, and a candidate for Condition 2 only for those sets whose composition demonstrably re-establishes itself after bounded perturbation. Condition 2 is exactly the added requirement this framework carries beyond the autocatalytic minimum (see the note below); it is not something Kauffman’s bare set is guaranteed to satisfy.

Note that Condition 3 is Boundary Retention, not membrane retention. A bare set in a well-stirred tank has no wall, yet it still holds an identity-bearing boundary, because the boundary is which reactions close on themselves, not a physical container. That is what lets Kauffman's minimal object stand as the reference case without upgrading it.

The molecules arrive from outside; the ordering of them does not. That is the endogenous line exactly. The attractor is the autocatalytic set itself, not any particular organism: a living cell is what such a set looks like scaled up, later wrapping its self-made closure in a self-made membrane, and every cell runs on one. We adopt the bare set as the reference case.

After S. A. Kauffman, The Origins of Order: Self-Organization and Selection in Evolution (Oxford University Press, 1993), on collectively autocatalytic sets and the origin of self-maintaining life.

The four conditions of sovereignty

A configuration is a sovereign attractor if, and only if, all four of the following hold at once. They are not weighted and they do not trade off; a structure that passes three and fails one is not "mostly" an attractor. It is not one.

1Recursion Lock

The structure sustains itself through closed-loop feedback, with no external driving required to keep the loop turning. Its present state feeds its next state, and that loop has closed. This is the condition that separates a genuine attractor from anything driven from outside.

2Internal Recurcline Persistence

The structure maintains its own identity through its own dynamics, and can recover from bounded perturbation on its own. It carries recurcline, the pressure of return read from the compressed record of its own recursion, and that pull toward re-closure is what lets it absorb a shock and re-establish itself rather than dissolve.

3Boundary Retention

The structure produces and maintains its own identity-bearing boundary. This is the decisive condition, and the one most mimics fail. Having a boundary is not enough; the boundary must be self-made. A structure whose edge is supplied from elsewhere can be elaborate and long-lived and is still not sovereign.

4Maintenance-Bearing Continuation

The structure bears the ongoing cost of its own continuation. Persistence is never free here; a sovereign attractor runs at cost, always, and it pays that cost itself. There is no admissible regime in which it persists for nothing and no shortcut around the toll. The cost may be paid in any currency the substrate uses, energy, attention, computation, effort, but it is paid, and paid from within.

The nearest prior theory: autopoiesis. These four conditions track Maturana and Varela’s autopoiesis closely, and it would be evasive not to say so. A closed self-producing loop (Condition 1), a self-made identity-bearing boundary (Condition 3), and maintenance paid from within (Condition 4) are substantially the autopoietic conditions. The debt is real and we state it plainly. The delta is Condition 2: autopoiesis does not require a stored, compressed record of the system’s own recursion. That is what this framework adds — the recurcline — and it is not decoration. It is precisely what makes the attractor/attractlet basin-indistinguishability problem tractable: two structures can share an identical return-shape after a nudge, and it is the presence or absence of an internally stored recursion-record, not the basin, that sorts the self-maintaining case from the externally scaffolded one. Prigogine’s dissipative structures are the other neighbour in this lineage (order sustained by throughflow); autopoiesis is the closer one, and Condition 2 is the honest statement of what is new here. This is a claim about lineage, not a stipulation — so it is fair to hold us to it.

Why binary. Sovereignty is all-or-nothing at the level of kind. Partial satisfaction does not produce partial sovereignty. A configuration is a sovereign attractor or it is not, and the sub-classifications below refine what kind of attractor it is, never how much of one. "Weak" describes robustness, not existence.

Kinds of sovereign attractor

All four kinds below are sovereign — each satisfies all four conditions. They differ in how the sovereignty is realized or how much margin it carries. (These are sub-classifications within the sovereign case, not to be confused with the three taxonomic classes — base, coordination, sovereign — which sort attractors by whether they reach sovereignty at all.)

Attractor (canonical)

A single sovereign attractor satisfying all four conditions in their standard form. The default when no further qualification is needed.

Weak Attractor

Fully sovereign, but with small margins: its perturbation tolerance, boundary integrity, or maintenance reserve is thin enough that a minor shock could tip it into irreversible loss. Still an attractor; "weak" qualifies the robustness, not the fact.

Diminished Weak Attractor

Still sovereign, but on a trajectory toward losing one of the conditions. What marks it is the direction of travel: it is heading either back toward health or toward loss. The trajectory is the diagnostic.

Colony Attractor (collective sovereignty)

A sovereign attractor whose sovereignty is realized collectively, across a lateral group of members that are each themselves sovereign. The colony holds its boundary, pays its maintenance, and keeps its identity as a group, surviving the turnover of individual members. A group of non-sovereign parts is not a colony; the members must each be sovereign in their own right.

Where the hard cases fall. The binary is clean as stated, but Condition 3 (a self-made boundary) has genuine intermediates in the world — symbiont-scaffolded boundaries, templated replication, obligate mutualism — where “self-made” is a matter of degree rather than a clean yes/no. A sharp binary must adjudicate these somewhere, and it is honest to admit the page does not yet state the cut. The governing rule is that the boundary-producing loop must close within the candidate: a structure whose boundary is produced only by an external partner is scaffolded, not sovereign; a mutualism in which each partner’s closure genuinely re-produces the other is a candidate colony case, judged partner by partner. Until a hard case is adjudicated against that rule explicitly, the binary should be read as a definition, not as a verdict already delivered on the borderline instances.

How sovereignty is lost

An established attractor can fall. Irreversible loss occurs when the identity-bearing boundary is violated past recovery; the attractor is gone, and anything that returns must start over from scratch rather than resume, because identity is not inherited across that break. Short of that, an attractor can slide: its self-boundary can falter, its maintenance supply can fall short of the ongoing cost, or it can regress to a mere coordination structure that looks organized but no longer holds itself together.

The contrast: the thing that looks like an attractor but is held together from outside is an attractlet, the companion page to this one. The four-part test that separates the two is set out in full on what makes an attractor sovereign. Both an attractor and an attractlet have a basin, the shape of the return after a nudge, which is exactly why a basin cannot tell them apart. See also the bootstrapping interval metrics (including recurcline, which only sovereign attractors exhibit) and the six ontic primitives that sovereignty presupposes.

← Back to Principia Attractum, HTML Edition