NOTE-148: Baedke — The Organism (Cambridge Element)

Contribution

Baedke gives a history-and-philosophy account of the “return of the organism” in postgenomics, Evo-Devo, epigenetics, niche construction theory and microbiome research. He reduces the conceptual problem to two challenges: an Inward Challenge (what internal organization makes something an organism, as against other units) and an Outward Challenge (how to separate an organism from an environment it co-constructs) (p. 24). He argues that early-20th-century organicism and holism, and today’s organism-centred biology, both fail both challenges in the same way. They over-rely on self-maintenance (“Persistence Ain’t Enough”, pp. 28–34), and they turn reciprocal causation into organism–environment co-constitution (“Reciprocity Ain’t Co-constitution”, pp. 34–37). Either way the organism dissolves into a larger system. What he adds that is new is the “overcomer” characterization of organismal organization (pp. 38–45). He pairs it with the existing 2021 unrolled reciprocal-causation model as the relational half (pp. 46–50), and summarizes both in Box 3 (pp. 50–51).

Key insight

The persister criterion is too permissive to individuate organisms. Self-maintenance through closure of constraints also fits colonies, holobionts, symbioses and ecosystems, so applying it at whatever level meets it makes every contributing unit “a mere part” (p. 31). That leaves an unpalatable choice: humans are parts of holobionts (“we have never been individuals”), or “organisms can be made up of organisms” (fn. 23). Baedke’s fix is a stronger organizational criterion. Organisms can also give up a closure regime and reach a qualitatively new one, and they control that unstable phase with their own repertoires. He adds an epistemic, causal-tracing criterion for boundaries. The mereological consequence is that organismhood becomes a criterion that can rule a well-integrated whole (a colony, a holobiont) out as an organism while ruling its members in.

Assumptions

Premises the argument relies on, most of them explicit:

• An organism-centred biology needs a feature “only organisms have” (p. 32: “Are there any relevant features that only organisms have, and other units of life do not? An organism-centered biology should be able to answer this question affirmatively.”). This is the premise that makes the persister view’s generality a defect.
• Intuitive non-organisms are fixed points. The man o’ war (a siphonophore), the beehive, the holobiont and the ant colony “do not intuitively qualify as organisms” (p. 30). A criterion that admits them is counted as having failed. The intuition is used and not defended.
• Organisms cannot be composed of organisms, or at least that is a “problematic assumption” (fn. 23, p. 31). This is the anti-nesting premise behind the rejection of holobiont individuality.
• Individuation is epistemic and practice-driven. Boundaries are “meaningful epistemic boundaries” (pp. 34, 46; “epistemically meaningful boundaries”, p. 49), justified by tractability, intervention and modelling needs (Consequences, pp. 37–38; Needham’s argument, p. 37). He does not claim a mind-independent composition fact.
• Causal relations are interventionist. Arrows in the O/E model are invariant under “a range of counterfactual interventions on the organism state (see Woodward 2003)” (fn. 32, p. 47).
• Organism-level agency is real enough to impose structure. Organisms are “bounded loci of causation, as agents that set specific goals” and so make organism–environment relations asymmetrical (p. 50). He brackets organismal experience (fn. 28, p. 36; fn. 33, p. 49).
• Scope: the account is sketched, not developed: “a major endeavor, one that can hardly be developed in this Element alone” (p. 39).

Key results

Arguments and positions, with locations:

• Nexus concept (§2, Box 1, p. 5; §3.2, p. 20). “Organism” is a nexus linking an organizational cluster (life, (self-)organization, parthood, intrinsic purposiveness) and a relational cluster (symbiosis, ecological concepts, evolutionary concepts, reproduction) (Fig. 1). The kinds of biological individuality (physiological, metabolic, reproductive, ecological, evolutionary) are the “language” or “knots” that link each cluster to the organism. Integration is only partial: humans plus gut microbiota form “a metabolically and immunologically closed unit” but not “a unified reproductive or evolutionary unit” (p. 21), so “the organism is a nexus of conceptual integrations, but also of conceptual tensions”. Fn. 13 (p. 20): individuality can also be ascribed to non-organisms (genes, populations), and he denies that a single concept such as organizational closure (Moreno and Mossio 2015) can unify both dimensions.

• Against persisters (pp. 28–34). Godfrey-Smith 2013 (“Organisms are essentially persisters”) and S. E. Smith 2017 are the targets, along with the metabolic, thermodynamic and autopoiesis/closure views. Closure has been applied to symbioses (Bich 2019), ecosystems and eusocial colonies (Canciani et al. 2019). Smith applies persistence to siphonophores, beehives and holobionts. Mossio 2024 concedes that “organized systems are not necessarily organisms” (fn. 22). A second charge: persister views carry Kantian “preservation teleology” (Toepfer 2024), which makes all organismal agency serve self-maintenance. That is the view in Montévil et al. 2016, Virenque and Mossio 2024, Sultan et al. 2022 and Laland et al. 2019.

• Against co-constitution (pp. 34–37). Reciprocal causation is a feedback loop between separate relata, and co-constitution denies there are separate relata. Haldane, Uexküll, Waddington, Griffiths and Gray 2001, Sultan 2015, Walsh 2022 and Laland and Brown 2018 slide from the first to the second. Needham’s 1936 objection: an inseparable organism–environment system still has to be individuated from other systems (p. 37).

• Overcomers (pp. 38–45). The contrast is with Montévil et al. 2016, on which a system goes from closed regime to closed regime or dies. Baedke claims three things. (1) Destabilization need not lead toward death, because “deorganization and reorganization” are a non-pathological part of development and life cycles. (2) Organisms actively destabilize their own organization. (3) Change need not be “a continuous series of states of closure” (p. 39). Three cases illustrate this (pp. 40–43):

  • Anglerfish sexual parasitism: loss of adaptive-immunity genes, fusion into “a new metabolic and morphological whole”, and about 30% of females never find a mate (Pietsch 2009).
  • Elysia marginata autotomy: the head regenerates a whole body within about 20 days, possibly using kleptoplasty.
  • Harpegnathos saltator gamergates: brain volume −20%, ovaries 5×, lifespan from about six months to three years or more, and the change is reversible, including brain regrowth.

  Lessons drawn (p. 43): organisms tolerate “risky” variation, instability opens new phenotypic spaces, and “overcoming is not the same as persisting”. Further claimed instances: fused comb jellies, Oxalis pes-caprae autotomy, and plant developmental instability (p. 44).

• Relational individuation (pp. 46–50). This is the unrolled O/E model from LIT-110 (Figs. 7–8). There are ordered state series Oₙ and Eₙ, arrows for causal influence within and between the series, and highlighted “invariant paths”. Worked cases: coral reef niche construction (Fig. 8A), and ruminant herbivory with four series O (host), E (external environment), O′ (microbes) and E′ (rumen) (Fig. 8B). The verdict: treating the holobiont as one individual “would collapse the microbial role” to steps 3–5 of the five-step pathway (p. 49). Organism–environment relations are reciprocal but “usually … asymmetrical”, because environments are “causally dispersed and fragmented” (following Fábregas-Tejeda, forthcoming) (p. 50).

• Box 3 (pp. 50–51) is the framework in six bullets. Its final line: organisms are “a special kind of living beings that are able to actively and creatively tinker with themselves and with their environment”.

• §5 (pp. 52–55). Postgenomic organism-centred views undercut genetic determinism, but they generate “postgenomic determinism” (for example, blaming mothers) and “biosocial race” (racialized methylation and microbiome profiles). Historically, Ganzheitsbiologie had ties to Nazi ideology. Conclusion: organism-centred biology is “compatible with deterministic, exclusionary and racist views” (p. 3).

• §6 (pp. 55–58). Organism-centred biology should expand gene-centred biology, not replace it “in a quasi-Kuhnian manner” (p. 58). Open task: set explanatory standards for organismal explanations.

How it individuates organisms, and what that does to part–whole questions

• Criterion. Something is an organism if it is (i) a persister (necessary, shared with many living systems), and (ii) an overcomer, meaning it has its own strategies to induce, modulate and control destabilized organization. That feature is claimed to be organism-specific. It must also be (iii) identifiable as a bounded locus of causation in the diachronic O/E causal graph. No sufficient condition is stated formally, and the three conditions are never put together as a definition. Box 3 lists them as features.
• Holobionts. The persister view makes the holobiont the individual and the host “a mere part” (p. 31). Baedke rejects that as the default and treats host and microbes as distinct organisms, each with its own environment. The rumen, an organ of the host, is modelled as the microbes’ environment E′, a separate series from the host O (Fig. 8B, p. 48). So “environment of X” can pick out what is anatomically part of Y, and environmenthood is relative to an organism. He also concedes that host plus gut microbiota is a physiological and immunological individual without being an evolutionary one (p. 21). The holobiont is refused organism status on epistemic grounds (it obscures microbial niche construction) and on the anti-nesting premise. Its individuality of some other kind is not refused.
• Collectives. Colonies, hives, siphonophores and superorganisms are persisters, not organisms. The Harpegnathos colony is a “functionally immortal persister” that “however is not an organism”, while the individual ant is the overcomer (p. 43). Meyer-Abich’s graded superorganism holism is presented as the extreme failure mode (p. 31). Fn. 13 still allows collectives evolutionary individuality. That means the organism level and the individuality level can come apart, which the colony verdict relies on.
• Parts of organisms. Genes and cells are contextualized by the organismal whole (Cornerstone 1, p. 23; Russell: “The parts are the way in which the whole [organism] organizes itself”, p. 27). Evo-Devo must show that organisms bias variation “in ways … different from, say, how cells affect the production of variation in multicellular species” (p. 38). He does not say whether a part can become the organism. In the Elysia case the separated head is treated as the same organism surviving, and “the former body dies” (p. 42), without argument about which piece carries identity. The anglerfish (and comb-jelly) fusions produce “one unit” or “one individual” (pp. 40, 44), but he does not say whether that fused whole is one organism or two. Fn. 30 (p. 44) adds that overcoming can move an organism between individuality kinds: metabolic to ecological (sea slug with symbionts), or physiological to reproductive (ant).

Claims

id	claim	strength	support
C1	”Organism” is a nexus concept with organizational and relational dimensions, linked by kinds of biological individuality	informal argument	Historical survey §2; Box 1/Fig. 1 (p. 5); §3.2 (p. 20)
C2	Physiological and evolutionary individuality do not always align, so the nexus is only partly integrable	informal argument (from the literature)	Host–microbiota example (p. 21), citing Godfrey-Smith 2013 and Pradeu 2016
C3	The persister/self-maintenance criterion cannot single out organisms	informal argument	Closure has been applied to symbioses, ecosystems and colonies; Smith 2017’s own cases; Mossio 2024’s concession (pp. 30–31, fn. 22)
C4	Persister views impose “preservation teleology” and so impoverish the account of organismal agency	informal argument	Textual analysis of Montévil et al. 2016, Virenque and Mossio 2024, Sultan et al. 2022, Laland et al. 2019 (pp. 32–33)
C5	Early-20th-century and current organism-centred biology slide from reciprocal causation to co-constitution and lose the organism	informal argument (historical)	Quotations from Haldane, Uexküll, Waddington, Griffiths and Gray, Sultan, Walsh (pp. 35–36); fuller version in Baedke et al. 2021
C6	Organisms are “overcomers”: they can actively destabilize and reorganize their closure regime, and this is non-pathological	informal argument, illustrated	Three cases (pp. 40–43). Two of them, anglerfish immune-gene loss and colony-triggered gamergate transition, are lineage-level evolution or colony-triggered plasticity, not plainly the individual’s self-initiated tinkering
C7	Only organisms, and not all persisters, can tinker in this way	assertion	”I suggest … Only organisms can” (pp. 39–40); no survey of non-organismal persisters is offered
C8	An ant colony is a persister but not an organism; the individual ant is the overcomer	assertion	p. 43. The colony’s non-organism status is presupposed, not derived
C9	Organisms can be unambiguously individuated from their environment despite reciprocal causation, by tracing diachronic causal paths	informal argument, with two worked diagrams	Unrolled O/E model (Figs. 7–8, pp. 46–49), taken from Baedke et al. 2021 (LIT-110). “Unambiguously” is stronger than the epistemic, explanation-relative highlighting of paths the model actually supplies
C10	Treating a ruminant holobiont as one individual hides microbial niche construction (keeps only steps 3–5)	informal argument	Fig. 8B (p. 48), reading of Chiu and Gilbert 2020 (p. 49)
C11	Organism–environment causation is usually asymmetric because environments are causally dispersed	assertion (credited to Fábregas-Tejeda, forthcoming)	p. 50
C12	Overcoming may introduce much evolutionary novelty and is “far more common across the tree of life than one might assume”	assertion (speculative)	pp. 44–45; extra examples cited, not analysed
C13	Organism-centred biology is compatible with determinist and racist programmes, as it has been historically	informal argument (historical and sociological)	§5 (pp. 52–55), citing the epigenetics and microbiome race literature and Ganzheitsbiologie
C14	Organism-centred biology should expand, not replace, gene-centred biology	informal argument	§6 (pp. 57–58)

Concepts

• Nexus concept — “a concept that forms a core or interface in which epistemic and ontological facets of other concepts can be interlinked and connected in different ways” (p. 4).
• Organizational dimension / relational dimension — the internal organization of the organism, and its interactions with the abiotic environment and other organisms (Box 1). The Inward and Outward Challenges correspond to them.
• Biological individuality (in this account) — the family of individuality kinds (physiological, metabolic, immunological, reproductive, ecological, evolutionary) that serve as “knots” tying the organism to one or the other dimension (p. 20). It is not coextensive with organismality (fn. 13).
• Persister — following Godfrey-Smith 2013 and Smith 2017, a system that “uses energy to resist the forces of decay” by maintaining a functionally integrated whole. Necessary for organismality, not sufficient (Box 3).
• Overcomer — an organism considered as able to “deconstruct their wholeness, reject closure, and reorganize themselves”, with “special agential strategies to induce, modulate, and control phases of destabilized organization” (p. 40).
• Deorganization / reorganization — the non-pathological destabilization–restabilization cycle that marks overcomers (p. 39).
• Preservation teleology — Toepfer’s term for Kantian natural teleology “aimed at the conservation of an existing dynamic structure not at its transcending” (p. 32).
• Reciprocal causation vs ontological co-constitution — a feedback loop between separate relata, versus a single system with no meaningful organism–environment separation (p. 35).
• Unrolled / open-loop O/E model — two ordered series of organism and environment states, with causal arrows within and between them, and invariant paths highlighted for a given explanation (pp. 46–47, Fig. 7D).
• Relational niche construction — a change in organism–environment relations without changing the constitution of either (mice huddling) (p. 46, after Aaby and Ramsey 2022).

Connections

• LIT-110 (Baedke, Fábregas-Tejeda and Prieto 2021, “Unknotting reciprocal causation”): the relational half of this Element is that paper’s model, with its argument abridged (fn. 26, fn. 29, p. 46). This reading confirms LIT-110’s summary on the boundary-keeping point. The experiential niche construction half is explicitly bracketed here (fn. 28, fn. 33).
• LIT-156 (Hazelwood, “An Emerging Dilemma for Reciprocal Causation”, Phil. Sci.): not cited by the Element. The Hazelwood 2023 it does cite (p. 35, among others who discuss reciprocal causation) is a different paper: “Reciprocal Causation and Biological Practice”, Biology & Philosophy 38(1):5, doi 10.1007/s10539-023-09895-0. The record does not hold that paper (grep of literature.d found nothing). The Element therefore gives no answer to LIT-156’s dilemma about natural selection as an emergent cause. Its O/E model has only organism and environment states as relata, and selection never appears as a node. Whether that dodges the dilemma or just fails to reach it is unverified.
• LIT-168 (Wilson and Barker, SEP “Biological Individuals”): this is the same map, evolutionary versus physiological individuality, and the same catalogue of hard cases (holobionts, colonies, siphonophores). Baedke adds the claim that organismality is a separate question from individuality (fn. 13, following Prieto 2023, 2024). That is what lets him deny colonies organism status while allowing them individuality.
• LIT-157 (Bourrat, reproducees and reproducers) and LIT-188 (Griesemer and Shavit, scaffolding individuality): both give reproduction-centred or collectivity-centred accounts in which collectives become individuals. Griesemer’s “individuality is a special case of collectivity” runs opposite to Baedke’s colony verdict. The gamergate case (a worker switching from physiological to reproductive individual, fn. 30) is exactly the kind of case Bourrat’s reproducer and reproducee vocabulary is built to sort. Baedke does not engage either author (neither Bourrat nor Griesemer is in the reference list).
• LIT-049 and LIT-005 (Bourrat’s coarse-graining and interventional-information accounts of individuality): these are formal criteria for which level of a partition is an individual. Baedke offers nothing formal. His relational criterion is causal, though, with Woodward-style invariance under interventions (fn. 32), and he points to causal-graph formalization (Otsuka 2015, p. 50). So LIT-005’s move from observational to interventional individuality measures is the natural formal partner of his relational half, and nobody has done that work yet.
• LIT-124 (Varzi, SEP “Mereology”): Baedke’s “parts” are functional and organizational parts, contributors to a maintained whole. He states no parthood principles. His anti-nesting premise (fn. 23) and his organism-relative “environment” (the rumen as E′) are the only places where formal mereology would bite. Varzi’s entry excludes the systems-theory sense of “mereology” (its note 1), which is the sense Baedke works in.
• LIT-142 (Meincke, process view of pregnancy): Baedke cites Kingma 2020 and flags Gilbert 2014’s merging of mother, microbes and fetus “in one inextricable unit” as an intervention problem (pp. 37–38). He does not argue the one-or-two question itself.

Bearing on the record

• This is a nucleation (non-ML) work. It carries nothing for machine-learning practice and should not produce anything in the Anthology of the SOTA. The skim’s suggestion (NOTE-148) that the inward/outward split could frame “whether artificial systems are organism-like” is not in the text. Baedke says nothing about artificial systems, and the overcomer criterion is defined with biological closure (Montévil et al.) as its baseline.
• On the record’s mereology tag: the tag is justified. Parthood is one of the nexus concepts (Fig. 1). The Element’s substantive verdicts are part–whole verdicts: colonies and holobionts are not organisms, their members are, and anti-nesting (fn. 23). But the work belongs to the systems and biological sense of part–whole, not to formal mereology (LIT-124).
• In the theory.d scheme, the persister-versus-overcomer distinction is the reusable item. What it can support is a claim of the form “self-maintenance under-determines the level of organismality”. It cannot support “overcoming is unique to organisms”, which is asserted (C7).
• LIT-156’s LIT note says it “directly engages” LIT-110. This reading neither confirms nor refutes that, but it does establish that Baedke’s own 2025 Element cites a different Hazelwood paper and does not reply to the selection dilemma.

Limitations

• The positive framework is explicitly a sketch (pp. 39, 51). No necessary-and-sufficient conditions are stated, and Box 3 is a list of features.
• Uniqueness of overcoming is asserted, not shown. No non-organismal persister is examined and found unable to destabilize and reorganize itself. The one contrast (colony vs ant, p. 43) presupposes the verdict. A colony that loses its queen and reorganizes through a month-long tournament is plausibly itself undergoing deorganization and reorganization, and Baedke describes it only as persisting.
• Agency is attributed across timescales without comment. The anglerfish case is the loss of immune genes over evolutionary time in lineages (Swann et al. 2020). It is not something an individual organism does, yet it is presented as organisms “actively” exploring “a ‘risky’ space” (p. 41). The gamergate transition is triggered by the colony context (the queen’s absence). The sea-slug case’s controlling mechanisms “need to be studied in detail” (p. 42).
• Part–whole puzzles raised by the cases are left open: which piece of an autotomized slug is the organism, whether a fused anglerfish pair or merged comb jellies are one organism, and how “organisms can’t be made of organisms” squares with a fused pair being “one unit”.
• “Unambiguously individuate” (pp. 3, 46) claims more than the model gives. The model individuates relative to a chosen explanation, since only “the most relevant causal arrows” are highlighted (p. 47), and it presupposes the O and E series as inputs instead of deriving them.
• The historical survey leans heavily on the author’s own prior work (Baedke 2019a; Baedke and Fábregas-Tejeda 2023; Baedke et al. 2021, 2024a/b), which is fine for an Element but means the survey is not independent.
• A minor factual point outside the argument: “the human genome has only about 35,000 genes” (p. 13) reports the 2001 draft estimate, which was later revised down to about 20,000. It does not affect the argument.

Open questions

• Is there a non-organismal persister (a colony, a symbiosis, an ecosystem) that meets the overcomer description? A single well-documented case would refute C7. Colony-level reorganization after queen loss is the obvious candidate.
• Can the overcomer criterion be stated formally, for example over a closure-of-constraints model in the style of Montévil et al., so that it separates organism from colony on a worked example?
• Does the unrolled O/E model individuate organisms or presuppose them? That needs a case where the O series is contested, such as a holobiont or a fused anglerfish pair, worked out in the causal-graph formalization Baedke points to (Otsuka 2015). A formal individuality criterion in the style of LIT-005 or LIT-049 could be the test.
• Does the model survive Hazelwood’s (LIT-156) dilemma once selection is added as a relatum? It currently has none.
• If environmenthood is organism-relative (the rumen as E′), what is the mereological status of an organ that is part of one organism and environment of another? Is parthood itself relativized, or only the “environment” label?

Corrections to the seeded skim

• The skim (NOTE-148’s summary) says organisms “deliberately” destabilize their organization. Baedke never says “deliberately” and makes no claim about intention. His words are “actively and creatively tinker” and “agential strategies to induce, modulate, and control phases of destabilized organization” (pp. 39–40). “Deliberately” overstates him and should go.
• The skim’s summary credits the overcomer view with letting biology “individuate organisms as distinct causal agents”. In the text the two dimensions do different jobs. Overcoming answers the inward challenge: it is what singles organisms out from other self-maintaining systems. Individuation against the environment is done by the relational model, which traces diachronic causal influences and identifies organisms as “bounded loci of causation” (Box 3, p. 51). Also, organisms are persisters and overcomers, not overcomers instead of persisters (Box 3, p. 50).
• The skim says “a model is offered” on the relational side, as if it were new. It is not new. It is the model of Baedke, Fábregas-Tejeda and Prieto 2021, which is LIT-110 (p. 46: “let me draw on a conceptual and visual model … developed with Alejandro Fábregas-Tejeda and Guido Prieto”), re-presented with two worked cases (Fig. 8).
• The skim leaves out that “Parthood” is one of the concepts Figure 1 places in the organizational dimension of the nexus (p. 5). It also leaves out that the only new ontological verdicts in §4.3 are about part–whole cases: an ant colony “is not an organism” (p. 43), and ruminant microbes are distinct organisms, not parts of a holobiont individual (p. 49).
• The skim’s open question asked whether the overcomer criterion “really picks out organisms uniquely”. The answer is that it is not shown. Uniqueness is asserted (“I suggest that not all living systems … are able to actively tinker in this way … Only organisms can”, pp. 39–40), and the one contrast case (colony vs ant, p. 43) assumes that the colony is not an organism instead of deriving it.
• The dossier gives §4 as pp. 24–52 and §4.3 as “pp. 38ff”. The exact span of §4.3 is pp. 38–52, and §4.2 is pp. 28–38. This is a clarification; nothing contradicts the dossier.