NOTE-094: Nahas & Sachs — What's at stake in naturalizing teleology

Contribution

Before this paper, disputes among organism-centred accounts of teleology were argued as if there were one “problem of teleology” with one set of success criteria. After it, there is a stated distinction between two metatheoretical orientations: advancing biology (§3.1) and settling the naturalism debate (§3.3). Each imposes different constraints on an account. There is also a demonstration that two live critiques (Cusimano and Sterner on closure; García-Valdecasas on the OA versus Deacon) and one strong claim (McShea and Babcock’s field theory as the only legitimate explanation) each run the two orientations together.

Key insight

Whether an account of teleology must demarcate genuine goal-directedness from mere self-organisation depends on what the account is for. A scientific model can leave the boundary indefinite and be better for it. A metaphysical account of how mind and value fit into nature may have to draw it cleanly, and it needs no operationalisation. Accounts that look like rivals may therefore be compatible answers to different questions.

Assumptions

Premises and authorities:

• Scientific models routinely trade truth for empirical adequacy and usefulness (Potochnik 2017; Wimsatt 2007; Godfrey-Smith 2001), and false models can lead to truer theories (§3.1, §3.3).
• Philosophical disputes are “almost never resolved”, and their criteria are unclear (§3.2).
• Naturalism, in whatever version, imposes only “general epistemic norms”, such as rigorous argument and productive insight, on a philosophical account of teleology (§3.3).
• Being a natural kind and being explanatorily useful come apart in both directions (§3.3).
• A parallel from biological individuality: the theoretical dispute need not be resolved for empirical work to proceed (Kovaka 2015; Love 2018) (§3.3; fn. 13).

Key results

What it argues, by section:

• §2.1 Organizational approaches. Kant: organisms are “self-organized beings”, in which each part is cause and effect of another. Kauffman 2000 relates constraints to work (the cannon, the car engine). The OA’s “closure of constraints” (Montévil and Mossio 2015): a constraint channels energy to produce or maintain another, circularly, in a system thermodynamically open and organisationally closed. On this account function is any part that contributes to the maintenance of the whole. Deacon’s teleodynamics: two mutually limiting dissipative (morphodynamic) processes. His “autogen” model (Deacon, Srivastava and Bacigalupi 2014) also realises minimal interpretation, which is biosemiotic.
• §2.2 Behavioural approaches. Walsh: organisms are agents, defined through goal, repertoire and affordance. Agency “is the gross behavioral capacity of an organism to bias its repertoire in response to what its conditions afford for attaining its goals” (Fulda 2017: 84). Agency is an explanans for “why” questions, complementary to mechanistic “how” questions. It is compared to fluid viscosity (fn. 7). McShea’s field theory: goal-directedness is persistence plus plasticity, comes in degrees, and is fixed by an external field. The theory applies equally to bacteria and heat-seeking missiles (fn. 18).
• §3.1 The scientific orientation. The limits of the Modern Synthesis (developmental plasticity, inclusive inheritance, niche construction; the EES) motivate an “agency perspective” (Sultan, Moczek and Walsh 2021). Advancing biology requires operationalisations and models that may be idealised and ontologically agnostic. The failure of reductionism does not imply that teleological explanation will be needed.
• §3.2 The philosophical orientation. Liberal naturalism (McDowell; De Caro and Macarthur) defines itself against a restrictive scientific naturalism that omits biology. A scientific naturalism that accepted organisms as intrinsically teleological agents would have resources for intentionality, normativity and value (Okrent 2018; Rouse 2015, 2022; Thompson 2007; Deacon). That would “invite a revision” of liberal naturalism.
• §3.3. A philosophical account “incurs no obligation to operationalize its terms”. A scientific one “does incur the obligation to show how naturalized teleology is operationalizable”. The naturalism debate needs claims that are “broadly true”, not claims that are “concretely useful”.
• §4.1. Demarcation (candle flames and tornadoes versus organisms) is “unavoidable and not particularly problematic” for models. Cusimano and Sterner 2020 do not draw the conclusion their own argument supports. McShea and Babcock 2021 reify a model’s silence into a denial of the intrinsic/extrinsic distinction.
• §4.2. García-Valdecasas 2021 prefers teleodynamics to the OA because it yields a beneficiary individual and a hierarchical second-order constraint. That preference rests on an Aristotelian conception of teleology and on a critique of modernity (fn. 21, Jonas). It is not a scientific point. He accepts that the OA satisfies a Kantian conception, so the preference for Aristotle over Kant “would need to be argued, rather than assumed”. The authors think the OA does account for second-order constraints (fn. 22), and that Kantian realism about teleology is available (fn. 25; Gambarotto and Nahas 2022).
• §5. A plurality of accounts is “tempting to suggest”: Deacon for naturalism; Walsh’s ecological approach and the OA for biology, “even if they remain largely untested”.

Claims

id	claim	strength	support
C1	Naturalizing teleology serves at least two relatively autonomous projects (scientific; naturalism/metaphysics) with different success criteria	informal argument	§§3.1–3.3
C2	A philosophical account of teleology need not be operationalisable; a scientific one must be	informal argument	§3.3, resting on philosophy-of-modelling premises (Potochnik, Wimsatt)
C3	Models of teleology (closure of constraints, field theory) cannot and need not demarcate teleological from other complex systems	informal argument	§4.1, against Cusimano & Sterner 2020; the caveat in fn. 16
C4	McShea & Babcock commit “pernicious reification” and beg the question by defining goals externally	informal argument	§4.1
C5	García-Valdecasas’s preference for teleodynamics rests on an unargued Aristotelian conception and not on biology	informal argument (textual analysis)	§4.2; fns. 21, 22, 25
C6	Accounts that seem in conflict may be compatible answers to different questions	assertion, illustrated once	§1, §5; the only worked case is the demarcation dispute
C7	A plurality of accounts is likely needed (Deacon for naturalism; Walsh and the OA for biology)	suggestion	§5, “tempting to suggest”; skepticism about a single account (§3.3)
C8	An expanded scientific naturalism accepting organismal teleology would invite a revision of liberal naturalism	informal argument	§3.2

Method

Metatheoretical analysis: survey the accounts (§2), distinguish the orientations that motivate them (§3), then re-read one dispute under each orientation (§4) to show that the verdict depends on the orientation. It uses no formal or empirical method.

Concepts

• organism-centred teleology: the view that teleology is “the intrinsic goal-directedness of whole organisms insofar as they are agents that act on their own behalf”, not only the function of parts (§1).
• closure of constraints: the OA’s organisational criterion. Constraints mutually produce and maintain one another in a system open to energy and matter and closed in its constraints (§2.1).
• teleodynamics / morphodynamics: Deacon’s levels. Morphodynamics is self-organising dissipation; teleodynamics is the coupling of mutually limiting morphodynamic processes that preserves system integrity (§2.1, §4.2).
• agency (Walsh/Fulda): “the gross behavioral capacity of an organism to bias its repertoire in response to what its conditions afford for attaining its goals” (§2.2).
• field theory (McShea): goal-directedness as persistence and plasticity relative to an external field; graded, and neutral between organisms and artefacts (§2.2).
• metatheoretical orientation: the question an account is meant to answer, which fixes its success criteria (§1, §3).
• liberal / scientific naturalism: nature as more than what the sciences describe (values, norms, intentionality) versus nature as what the best sciences describe (§3.2).

Connections

• LIT-167 (Baedke, The Organism), read. Baedke argues that closure and persister criteria are too permissive: they take in colonies, symbioses and ecosystems. He adds that they impose a “preservation teleology” that impoverishes organismal agency (NOTE-148). That is a demarcation complaint of the kind Nahas and Sachs would ask to have indexed to a project, scientific or metaphysical. Baedke’s complaint is about individuating organisms, which is arguably a third project. Both cite Walsh, Mossio and Sultan.
• LIT-168 (SEP, Biological Individuals), read. Its §6.3 (autonomous agency: Moreno and Mossio 2015; Walsh; the cognitive-metaphor accounts) covers the same organisational-agency literature. Nahas and Sachs’s fn. 13 draws the parallel with biological individuality explicitly, via Kovaka and Love.
• LIT-157 (Bourrat, reproducees and reproducers), read. It is a biology-facing account of Darwinian individuality with operational “recipes” (NOTE-100). It is an example of the scientific orientation’s demand for operationalisation, which Nahas and Sachs say need not also settle metaphysics.
• LIT-120 (Bruckner), read. It uses a loose, behavioural notion of autopoiesis (after Thompson 2007) to ground welfare subjecthood. That is a third, normative use of organismal teleology. The paper does not engage the demarcation problem that Nahas and Sachs report for autopoiesis (Deacon and Cashman 2013) and for closure (candle flames).
• LIT-110 (Baedke et al., reciprocal causation) and LIT-156 (Hazelwood, a dilemma for reciprocal causation), both Deferred. Nahas and Sachs cite Baedke, Fábregas-Tejeda and Prieto 2021 (the LIT-110 paper) only for the history of organicism’s sidelining (§3.1), not for reciprocal causation. The skim’s link through Walsh is therefore thinner than it suggested.
• LIT-211 (Bender et al., definitions of life), read (NOTE-109). Its “Cognitive Autonomy” and “Dissipative Self-Organizing Systems” clusters map onto the OA/Deacon side and the dissipative-structure side of the demarcation dispute.
• The dissipative-structure literature, now filed. Prigogine’s Nobel lecture (LIT-516), Turing (LIT-535) and the BZ reaction (LIT-531) supply the canonical dissipative structures, and none of them draws a line between those and organisms; Prigogine counts “a living system” among them. Of the readings that try, only Friston (LIT-526) names the candle flame, asserting in one sentence that it “cannot possess a Markov blanket”. Deacon (LIT-529), whose chapter with Cashman is the source of the candle-flame case here, and Moreno & Mossio (LIT-536) are filed unread.

Agency. The paper holds no account of agency of its own. It presupposes that organism-centred teleology treats organisms as agents acting on their own behalf. It surveys three candidate accounts: organisational (closure, where agency is grounded in self-maintenance), semiotic (Deacon, where teleology is tied to interpretation and to continuity between life and mind), and behavioural/ecological (Walsh and Fulda’s repertoire-biasing, McShea’s graded field-relative goal-directedness). Its thesis is that which account is right depends on whether agency is wanted as a biological explanans or as a metaphysical ground for mind and normativity in nature. The tag is justified as primary.

Bearing on the record

• Two THEORY documents now bear on this paper’s questions. THEORY-072 is the record’s first answer to the demarcation question: organisms differ from dissipative structures such as a candle flame by closure among several differentiated constraints that produce one another at different time scales, not by far-from-equilibrium self-maintenance or by circular self-maintenance alone. THEORY-071 is an account of minimal agency: individuality, interactional asymmetry and self-generated normativity. This paper supplies the first check on both: state which project the account serves, and do not treat a model’s failure to demarcate as a metaphysical verdict (or the reverse).
• For ML practice: nothing direct, and no ANTH document is affected. There is a transferable caution for debates over whether ML systems are “agents” or “goal-directed”. A graded, field-relative measure like McShea’s applies to artefacts by design, and its success as a measure would not settle intrinsic agency. This is my extrapolation; the paper does not discuss AI.
• NOTE-094 should add McShea’s field theory and the §4.1 targets, soften “dissolves some disputes” to one dispute, and record that the read was of the accepted manuscript.

Limitations

• Only one dispute (demarcation) is worked through, so the general claim that metatheoretical clarity “resolves and dissolves” debates is illustrated, not established.
• The science/philosophy split is asserted as “relatively autonomous” but never bounded. The authors concede that the relation is “far more complex” than McShea and Babcock assume (fn. 19), and that closure could play either role (fn. 16), without saying how the two sets of criteria interact when they conflict.
• The naturalism orientation is characterised almost entirely through McDowell-style liberal naturalism. Other philosophical uses of teleology are not separated out, such as normative ones (welfare, moral status: see LIT-120) or individuation.
• Positive recommendations (Deacon for naturalism; Walsh and the OA for biology) are offered without argument beyond the section’s framing.
• Read from the accepted manuscript. The VoR’s final wording and bibliography were not seen, and the manuscript has citation slips (see corrections).

Open questions

• Can one account satisfy both sets of criteria? The authors are “admittedly skeptical” (§3.3) but leave it open. An account that is both operationalised in a successful research programme and defended as metaphysically correct would settle it.
• Is the OA’s closure a mere model or an ontological claim? The authors flag this as “in urgent need of elaboration” (fn. 16).
• Are there more than two orientations (normative, individuational), and do they impose further, conflicting criteria?

Corrections to the seeded skim

• Not open access; the read is of the accepted manuscript. The batch brief lists the paper as open access. Springer, Unpaywall and Semantic Scholar all show it closed (see read:), so the dossier’s account of access was right.
• The behavioural approaches include McShea’s field theory, which the skim omits, and it is a main target. §2.2 presents Walsh’s agency (goal, repertoire, affordance; Fulda 2017’s definition) and McShea’s field theory, on which goal-directedness is persistence plus plasticity, is graded and is fixed by an external field (Lee and McShea 2020). In §4.1 McShea and Babcock 2021 are charged with “pernicious reification”. Their failure to distinguish intrinsic from extrinsic purposiveness is treated as a fact about the world, and they claim to be “the only legitimate explanation”. The charge also includes begging the question by defining goals only in terms of external fields.
• §4.1’s targets and thesis are more specific than the skim says. The skim reports only the García-Valdecasas critique. §4.1 first argues against Cusimano and Sterner 2020 that closure of constraints, as a model, is unlikely ever to meet their objectivity criterion and need not. Demarcation problems in modelling are “local to the specific explanatory project”. Its broad applicability may even be an advantage. The authors add a caveat (manuscript fn. 16, VoR n. 14): closure could also be used for ontological claims, and there the worry “would take on a different hue”.
• Pluralism is suggested, not argued. The skim’s §5 (“suggests a plurality … Deacon for the naturalism debate; Walsh and the OA for advancing biology”) is accurate in content, but the authors frame it as “tempting to suggest”. They say Walsh and the OA “remain largely untested”. In §3.3 they “do not take a stand on whether or not a single account … could satisfy both scientific and philosophical desiderata (though we are admittedly skeptical)”. The NOTE summary’s “suggests pluralism” is right; “dissolves some disputes” should be “one dispute, in two contexts”.
• The philosophical project is specifically about naturalism and mind. The paper’s second question is “what role might a teleological conception of organisms have in the philosophical debate about naturalism, particularly the possibility for finding a place for mind in nature?” (§1). §3.2’s argument is that an expanded scientific naturalism that accepts organismal agency would undercut liberal naturalism’s complaint (McDowell 1996) and “invite a revision” of it. The skim’s “how teleology changes our picture of nature and ourselves” is vaguer than the paper.
• Manuscript citation slips (unverified in the VoR). In the manuscript, “Moreno and Bich (2017)” in §4.2 should be Mossio and Bich 2017. “McShea and Babcock (2021)” is cited but not in the bibliography, and neither are Kovaka 2015, Love 2018, Pezzulo and Levin 2016, Deacon 2012/2014, Bernstein 2002, Lovibond 2006 or Sachs 2012. The VoR notes correct some author-date forms, so these may be fixed in print.
• agency tag: justified, as the primary tag. The paper’s object is the organism-centred view that organisms are intrinsically goal-directed “agents that act on their own behalf” (§1). It sets out Walsh’s and Fulda’s definitions of agency and McShea’s graded goal-directedness (§2.2), and it asks what an account of agency is for.