Canonical · Relation to Adjacent Work

Major Transitions in Individuality

The empirical test of whether a stack can rewrite its own geometry, and where biology is caught mid-crossing

The companion page How Stacked Attractors Couple resolves a sharp question, can evolution rewrite a stack’s own geometry?, as mostly no. A stack’s depth geometry is fixed for its lifetime and cannot be edited from within: internal evolution is a formation-phase operation and cannot perform the compositional move that builds stack structure (§6.4); a stack is additive and non-substitutable (§5.2); and an attractor evolving to depend on the layer above is a cycle the stack forbids (§5.2/§5.3). What looks like a rewrite is the stack being lost (β-loss, §4.2) and a different one formed fresh, with identity not inherited (NFR, §4.4). One narrow edge stays open: a substructure caught mid-crossing the coordination–sovereign boundary while embedded (§7.2). This page tests that resolution against the biological literature built for exactly this question, the major evolutionary transitions in individuality. The verdict, stated precisely: major-transition biology is consistent with fixed token-stack geometry, provided the framework distinguishes token identity from evolutionary-lineage identity, which it does (§4.4 NFR, §4.5 RSR₄/RSR₅). Read with that distinction named, biology supports the resolution, and stresses it precisely where canon leaves it open, and where biology is itself genuinely unsettled.

The neighbor: transitions in individuality

Maynard Smith, J., & Szathmáry, E. (1995). The Major Transitions in Evolution. Oxford University Press. and Szathmáry, E., & Maynard Smith, J. (1995). The major evolutionary transitions. Nature, 374, 227–232. Authors, titles and 1995 dating stated with confidence; exact pagination and edition not independently re-verified here.

The framework’s stack-geometry question, can a lower attractor evolve so far that it changes what recursion depth it operates at?, is not a question biology has ignored. It is the central object of a mature field. A major transition in individuality is an evolutionary event in which formerly independent units become parts of a new, higher-level individual: independent replicators into chromosomes, prokaryotes into the eukaryotic cell, single cells into multicellular organisms, organisms into eusocial societies. The defining move each time is a change in the level of selection, the thing that used to be an individual becomes a component, and a new individual appears above it.

The phenomena line up exactly. “An entity changes what level it is an individual at” is, word for word, the framework’s “an attractor changes what recursion depth it operates at.” The transitions literature and the stack-geometry question are closely corresponding structural questions in two vocabularies, a strong correspondence, not an identity of theories. (The neighbor is concerned with evolutionary individuality, reproduction, fitness, heritability, and levels of selection; stack depth is a more general structural concept. The two rhyme; they are not definitionally the same.) And the literature’s own organizing principle rhymes with the framework’s: a transition is achieved only when “the disruptive effects of selection at the lower level” are suppressed, the lower unit gives up the autonomy that would let it defect. In the framework’s terms, giving up that autonomy does not by itself end sovereignty, because reproduction is not one of the four conditions (§7.2). What decides is Condition 3: whether the lower unit can still rebuild its own boundary from raw supply. Where it loses that ability and must take its boundary parts ready-made from the host, as an endosymbiont does, absorption is β-loss of its sovereignty. Where it keeps that ability, it remains a sovereign candidate inside the larger individual. This is the right neighbor against which to check the claim that geometry is not rewritten in place.

Demonstrations, a change of level is loss + reformation

The resolution’s core claim is that what looks like “a member rewrote the stack from within” is really two existing processes in sequence: the old arrangement is lost, and a different one forms fresh, its identity not inherited. If that is right, the most dramatic real cases of “something changed its level” should read as loss + reformation, not as in-place edits. They do.

Sagan [Margulis], L. (1967). On the origin of mitosing cells. Journal of Theoretical Biology. [verify volume/issue/pages and DOI against the primary source]

Endosymbiosis (the mitochondrion, the plastid). A once free-living bacterium, a sovereign in its own right, became a self-maintaining but non-sovereign component of the eukaryotic cell. This is not an in-place rewrite of a two-layer stack: the endosymbiont lost its sovereignty (it can no longer reconstitute its own boundary constituents, Condition 3, the reconstitution criterion, §7.2), and a new individual, the eukaryote, is what persists. One caveat keeps this honest: what was lost is not a single sovereign token sitting still for 1.5 billion years, but the independent-sovereignty status of a lineage across reproducing generations. The framework need not track one persistent token here; it tracks the lineage’s sovereignty status changing across serial re-ignitions (each generation a fresh IST ignition, related to its parent by RSR₄ template continuity, §4.5), and that status went from sovereign to non-sovereign component. Read at the carrier the framework actually names, lineage sovereignty, not a lone attractor-token, this is β-loss of independent standing plus reformation of a new higher individual, as the resolution says. This is treated in full on the mitochondrion page.

Murchison, E. P., et al. (2014). Transmissible dog cancer genome reveals the origin and history of an ancient cell lineage. Science, 343, 437–440. and Baez-Ortega, A., et al. (2019). Somatic evolution and global expansion of an ancient transmissible cancer lineage. Science, 365, eaau9923. Findings below are drawn from these primary reports; exact figures should be confirmed against them before any downstream use.

The sharpest apparent counter-example is a case where a subordinate part seems to become independent, and it is worth walking through, because it is the case that most looks like an in-place rewrite and turns out not to be. Canine transmissible venereal tumour (CTVT) is the oldest known somatic-cell lineage: it arose from a single dog roughly 11,000 years ago, went global about 500 years ago, and is now endemic in some 90 countries, transmitted as living tumour cells passed between dogs. It has accumulated on the order of 1.9 million somatic mutations while keeping a broadly stable genome. Tasmanian devil facial tumours, transmissible leukaemias in clams, and the immortalised HeLa line in the laboratory are the same shape.

A part broke out, and the case turns on which identity is being tracked. This is the sharpest example precisely because it forces the framework to say what carries identity, and the kernel already types this, so it is worth doing carefully rather than declaring victory. The famous fact about CTVT is that the living tumour lineage is physically continuous across hosts: dogs are not independently developing equivalent cancers; living cells are passed dog to dog. So there are several distinct carriers, and the framework reads each differently. (1) The founder dog’s stack: lost, the dog is long dead, its organism-sovereignty gone. (2) The tumour in any one host: a transient non-sovereign proliferation, formed and lost with that host. (3) The clonal cell lineage across 11,000 years: genuinely continuous, but continuous in the way the kernel calls trace/geometry propagation, not sovereign-token continuity. Under RSR₄ (§4.5) a pattern carried forward by a physical template (here, transmitted cells and their broadly stable genome) transmits geometry, not recursion: each proliferation in a new host is a fresh ignition biased by that inherited template, not a transfer of the founder cell’s original sovereignty. So the correct reading is not “identity was not inherited, full stop” it is that lineage continuity is real and token-sovereignty continuity is not, and RSR₅ (§4.5) decides which by trace continuity, not configurational similarity. The dog’s stack was not rewritten in place; a lower-layer cell lineage broke out and now propagates as its own thing. The point the example actually makes is that the framework must, and does, name the carrier before it renders a verdict.

Kirk, D. L. (2005). A twelve-step program for evolving multicellularity and a division of labor. BioEssays, 27, 299–310. and Hanschen, E. R., et al. (2016), and Matt, G., & Umen, J. (2016), on the volvocine algae. Volvocine claims below are the field consensus (Chlamydomonas → Gonium → … → Volvox; germ–soma via co-option of existing pathways); exact authorship/dates not independently re-verified here.

The germ–soma transition (volvocine green algae). The volvocine algae run from the single-celled Chlamydomonas through colonial forms (Gonium, Pandorina, Eudorina) to Volvox, which separates a germ line from sterile somatic cells. Comparative genomics finds the unicellular→multicellular transition needed “a surprisingly low amount of genomic innovation,” achieved largely by co-opting existing genes. In the framework’s terms: formerly independent cells gave up independent reproduction (the soma), and a new individual formed above them. Giving up reproduction does not by itself end a cell’s sovereignty, because reproduction is not one of the four conditions (§7.2). The test is Condition 3, whether the cell can still rebuild its own boundary from raw supply, and it is decided cell type by cell type. A sterile somatic cell that still makes its own membrane proteins remains a sovereign candidate beneath the organism. A mammalian red blood cell loses its “ribosomes, organelles, and mRNA” as it matures (Warren, A. J. 2017, Blood 129(5), 544–545), can no longer rebuild its membrane proteins, and fails. So this transition has a different shape from endosymbiosis. There the absorbed partner lost the ability to rebuild its own boundary; here a cell can keep it.

Stresses, the cases caught mid-crossing

The resolution does not claim to close everything. Its honest residual is a substructure caught mid-crossing the coordination–sovereign boundary while embedded, whose momentary relation to the layer above, degenerate stack, nascent colony, or neither, canon declines to force (§7.2, the boundary-crossing regime). The striking result of checking against biology is that this is not a contrived edge: nature is caught in the act, and biologists are unsettled in exactly the same place.

Coale, T. H., et al. (2024). Nitrogen-fixing organelle in a marine alga. Science, 384, 217–222 (science.adk1075). and Marchant, H. K., commentary, and the reporting of >350 host-encoded proteins imported by UCYN-A. Coale et al. 2024 stated with confidence; the exact protein count and pagination should be confirmed against the paper.

Two organelles are, right now, in the middle of the crossing, and they arrived there independently, which turns the residual from a hypothetical into a repeatable natural experiment.

The nitroplast (UCYN-A in Braarudosphaera bigelowii). In 2024 a cyanobacterial symbiont was argued to have “evolved from a symbiont to a eukaryotic organelle” a nitrogen-fixing organelle, the nitroplast, on the strength of the host’s cell importing hundreds of proteins into it and the symbiont having lost biosynthetic pathways it once ran. This is a substructure embedded in a sovereign and mid-crossing the sovereign→non-sovereign-component boundary. Keep the three levels apart. Observation: UCYN-A shows extensive host integration and dependence, with hundreds of host proteins imported. Biological interpretation: the investigators argue it qualifies as an organelle rather than merely a symbiont. Framework mapping (ours): the configuration lies near the sovereignty/component boundary of §7.2. That third statement is a mapping we draw, not a discovery biology made of §7.2, but the field’s own “symbiont or organelle yet?” difficulty and the boundary-crossing regime are difficult in the same place, which is the correspondence worth reporting.

The Paulinella chromatophore. A second, independent primary endosymbiosis, only 90–140 million years old, against the ~1.5-billion-year-old primary plastid, caught early enough that it is studied precisely as “the early stages” of an organelle forming. The existence of two independently evolved intermediate cases supports the decision not to force a binary classification during an organizational transition: these systems occupy intermediate organizational states that preserve features of both endosymbiont and integrated organelle, twice over and by separate routes. [verify the 90–140 Myr dating and the independent-origin claim against the primary literature]

Strassmann, J. E., & Queller, D. C., and collaborators, on Dictyostelium discoideum as a model for social evolution. The ~80% spore / ~20% stalk figure and the facultative-cheater result are the field consensus; exact figures should be confirmed against the primary reports.

Facultative multicellularity (Dictyostelium). The social amoeba lives as independent single cells when fed, and on starvation aggregates by cAMP signalling into a multicellular fruiting body in which roughly 80% of cells become reproductive spores and about 20% form a sterile, dying stalk. The same cells cross the individual–part boundary reversibly, every cycle. This is the best adversarial test on the page, and it too is a question of which carrier, there are at least three. (1) The individual amoebae: persist across cycles, each its own lower-layer entity. (2) The fruiting-body collective: a token that is assembled fresh each cycle and lost each cycle, C₁ ≠ C₂, loss + reformation at the level that matters. (3) The developmental program that can generate a collective at all: persists across cycles, but persists as RSR₄ (§4.5) artifact-borne geometry carried in the amoebae, not as a persisting collective-token. The framework does not decide the collective is a new identity because NFR predicts it; it decides by trace continuity under RSR₅ (§4.5): each fruiting body is built fresh with no trace-continuity to the prior one, so each is a new token, while the amoebae, whose own recursion is unbroken across the cycle, generate their own continuous history (THT, §3.12). Name the carrier and the case is clean at every level; leave it unnamed and it looks like a contradiction. The altruistic stalk is also a live picture of the transitions literature’s central tension, and “cheater” genotypes that overproduce spores are exactly the lower-level selection a completed transition must suppress.

The living transect, and the “is the colony the individual?” debate. The volvocine series (single cell → undifferentiated colony → germ–soma organism) is a standing row of intermediates, some genuinely ambiguous about whether the individual is the cell or the colony. Siphonophores (whose zooids sit between “organs” and “individuals”) and the eusocial-superorganism debate are the same ambiguity at other levels. In the framework these are the transitional STACK/COLONY indeterminacy, and the fact that biology has not settled them either is the point, not a failure of either account. [verify siphonophore and superorganism specifics before any strong claim]

Which identity is carried, and how the framework decides

Every case above turns on one question: what object is carrying identity? A change of level always leaves something continuous and something discontinuous, and “identity lost” is meaningless until the carrier is named. The framework names four, and, the load-bearing point, it decides continuity for each by a single criterion fixed in advance, not by reading back whatever NFR would predict.

The four carriers. A token is one sovereign-attractor instance (a founder cell, a particular fruiting body, one endosymbiont). A lineage is a chain of serial re-ignitions related by template continuity, each generation a fresh IST ignition biased by the parent’s inherited geometry (RSR₄, §4.5). A collective is a STACK/COLONY token (§5.2, §7.2). A stack geometry is the depth structure itself. These are not interchangeable, and the framework already keeps them apart.

One criterion, applied before the answer is known. Whether two configurations are the same identity is decided by RSR₅ (§4.5): restoration (identity preserved) requires trace continuity via ALPHA (§8.3); replacement (new identity) is anything without it, and “a replacement may be configurationally identical to the lost original and still not be it.” Continuity is thus decided by trace, not by resemblance and not by convenience. That forecloses the failure the neighbor’s reviewers rightly fear, that “whenever something reforms, call it a new identity” becomes unfalsifiable. It is not: the trace either continues or it does not, independently of the verdict one wants.

The four cases, one ruler. Applied uniformly, without changing the rule after seeing each answer:

Endosymbiosis: lineage-sovereignty status changes (sovereign → component); no token persists 1.5 Gyr; new higher token forms. Replacement, not restoration.
CTVT: founder dog-stack lost; per-host tumour transient; clonal cell lineage continuous as RSR₄ template propagation, not sovereign-token continuity.
Dictyostelium: amoebae persist (unbroken recursion, own history under THT); each fruiting body a new token by RSR₅; developmental program persists as RSR₄ geometry.
Nitroplast / Paulinella: token caught mid-crossing the §7.2 boundary; verdict reserved by design until the crossing completes.

The same rules classify all four without the criterion being adjusted after the fact. That is the result worth advertising, stronger than the analogy to evolutionary biology it grew from: these four are an adversarial test suite for the framework’s identity machinery, and the machinery passes on a criterion set before the cases were read.

The residual, made concrete

The resolution admits one expository gap: canon names no single account of a dissolved stack’s members recomposing laterally into a colony. Lichens supply a candidate instance, a fungal and an algal (or cyanobacterial) partner that can, under the right conditions, be separated and re-synthesise the association. That is members dissociating and re-associating side by side rather than top-to-bottom, a candidate picture of lateral recomposition. Held at arm’s length, though: reassembling A + B after separating them demonstrates compositional assembly, which existing composition may already describe; it does not by itself establish a distinct lateral-recomposition operator. So this is an observation to sit with, not yet pressure for a named construct, the prior question is whether anything already in canon covers it. [verify lichen resynthesis claims (Ahmadjian and successors) against the primary literature]

In every major biological case examined here, transmissible cancers most vividly, an apparent change of level resolves more naturally as break-out plus a fresh individual than as an identity-preserving rewrite of an existing hierarchy. No clean in-place rewrite turned up; a single clean instance would reopen the question. And the one edge canon leaves open, a substructure caught mid-crossing the coordination–sovereign boundary, is where biology occupies genuinely intermediate organizational states right now, in the nitroplast and in Paulinella. The framework is open precisely where the organization is intermediate, not underspecified where it is clear.

What this page claims, and what it leaves open. The claim is narrow: read against a mature empirical field, the stack-geometry resolution holds, a change of level is loss + reformation, and the sole residual (mid-crossing indeterminacy) corresponds to cases biology itself has not resolved. Threads deliberately left open:

The absence is evidence. No clean case turned up of a subordinate part smoothly becoming a co-equal peer while the whole persisted with its identity intact, an actual in-place geometry rewrite. That absence is itself support for fixed geometry; a single clean instance would be the thing that reopens it, and is the case a reader should hunt for.
Lateral recomposition wants a name. The lichen-style “dissolve, then recompose side by side” path is real biology with no dedicated construct; whether it deserves one is a live question for the kernel’s own amendment process, not a claim made here.
The mid-crossing verdict stays reserved. Nothing on this page forces a status on the nitroplast or Paulinella; §7.2 declines to, and so does this page. When such a configuration finishes crossing, it will read cleanly, sovereign, or non-sovereign component, and not before.

This page introduces no construct and modifies no canon; it locates the framework against the transitions-in-individuality literature and tests one downstream resolution against it.

For the resolution this page tests, why a stack’s depth geometry is fixed for its lifetime, and the exact shape of the narrow residual, see How Stacked Attractors Couple →

Adjacent-work assessments state where Principia Attractum agrees with and departs from neighboring phenomena and frameworks. They introduce no constructs and modify no canon; they locate the framework relative to its field. The reading of the transitions above as β-loss plus reformation, rather than in-place geometry rewriting, is a downstream application of the framework’s STACK (§5.2), POD (§6.4), sovereignty (§7.2), NFR (§4.4), the RSR pack’s artifact-propagation and restoration-vs-replacement rules (RSR₄/RSR₅, §4.5), ALPHA trace continuity (§8.3) and THT (§3.12) canon, not an addition to it. The token/lineage/collective/stack identity-carrier reading is a labeling of distinctions already fixed by those constructs, not a new type system. Biological and historical claims carry [verify] marks or confidence hedges and should be confirmed against the cited primary sources before publication.

← Back to Relation to Adjacent Work