Basal Cognition, Platonic Ingression, and the Limits of Biological Intelligence: A Critical Assessment of Michael Levin’s Theoretical Framework

Framing the Stakes

Biologist Michael Levin’s framework for understanding evolution and cognition (Source: Levin’s website, thoughtforms.life) represents one of biology’s most ambitious attempts to reframe life itself as fundamentally intelligent. That ambition carries philosophical burdens that survive scrutiny unevenly, and, more troublingly, it carries structural liabilities that make it directly exploitable by advocates of intelligent design. His core claims about genotype to phenotype conversion, Platonic pattern ingression, and causal emergence through agent fusion offer genuinely novel experimental approaches while simultaneously importing metaphysical commitments that exceed what his empirical work can support and that echo, with uncomfortable precision, the teleological arguments the Discovery Institute has deployed against evolutionary biology for decades.

This is not a minor risk. When a prominent biologist argues that mechanistic explanation will always be insufficient to account for goal directedness, that physical systems are “pointers” to patterns in a nonphysical realm, and that “we are not fundamentally physical beings,” he is advancing claims that the Discovery Institute can and will (and already has! 12+ articles citing Michael Levin @ Discovery Institute’s Website) harvest selectively, stripped of their experimental context, as scientific validation of supernatural teleology. The response to Levin’s framework cannot be dismissal of his experimental achievements, which are genuine. It must be precise philosophical surgery: separating what the data actually support from what the metaphysical superstructure demands, and naming the exact points where the framework becomes a liability for scientific naturalism.

Morphogenesis as Problem Solving: Reframing That Does Not Escape Mechanism

Levin’s first major claim is that genotype to phenotype conversion is “intelligent” and that DNA functions as “prompts” rather than blueprints, a reframing grounded in decades of empirical work in developmental bioelectricity. His 2023 Springer contribution argues that the cells with which evolution works are not passive components: they derive from ancestral unicellular organisms with rich behavioral repertoires, and they retain those repertoires throughout multicellular development. The multiscale competency architecture he proposes holds that biological organization exhibits regulative plasticity at every level, from individual cells to tissues to whole organisms, with each scale solving problems appropriate to its context.​

The empirical foundation is genuinely substantial and should be acknowledged as such before any critique proceeds. Levin’s laboratory has demonstrated that manipulating bioelectric gradients can induce planarian flatworms to regenerate with two heads, and, crucially, this morphological state persists without continued intervention. The finding implies that bioelectric patterns encode anatomical target states, what Levin calls “set points,” that the organism’s developmental machinery works to achieve and maintain. His 2025 paper with Hartl in Trends in Genetics extends this by framing the genome as a generative model rather than a hardwired algorithm: morphogenetic machinery actively decodes genetic information in contextually appropriate ways, interpreting rather than merely executing instructions.​

The critical question, and the one where the framework first becomes vulnerable, is whether this reframing requires cognitive vocabulary, or whether it represents sophisticated mechanism described in a different register. A mainstream developmental biologist examining the same two headed planarian result would say: “This demonstrates unforeseen plasticity in gene regulatory networks and bioelectric circuits. We need better mechanistic models of this complex dynamical system.” Levin’s move is stronger and more philosophically committed: he claims that mechanistic explanation is, and will always be, insufficient to explain goal directedness, and that teleological explanation is therefore not merely useful but necessary. This is not a direct inference from the data. It is a philosophical claim about the limits of a class of explanations, and it requires independent philosophical argument to establish, argument that Levin does not provide and that the data cannot provide.​

The Stanford Encyclopedia of Philosophy draws a sharp distinction between teleology, genuine goal directedness involving intentions and representations, and teleonomy, goal directed patterns fully describable without attributing mental states. Colin Pittendrigh coined “teleonomy” precisely to allow biologists to speak of biological purpose without metaphysical commitment. Levin’s framework systematically blurs this distinction. Cellular “competencies” could represent genuine cognition or sophisticated teleonomy, and no experiment Levin has conducted can adjudicate between these interpretations. The two headed planarian is equally consistent with both. This is not a gap to be filled by future experiments; it is a conceptual underdetermination that more data cannot resolve.​

The Discovery Institute attack surface here is direct. When Levin asserts that mechanism will always fail to explain goal directedness, he is advancing a functional equivalent of Michael Behe’s irreducible complexity argument: a claim that certain biological phenomena lie beyond the reach of reductive explanation, necessitating appeal to a higher organizing principle. Behe located that principle in an intelligent designer; Levin locates it in “Platonic space.” The logical structure is identical. Skeptics are right to find this troubling not because Levin’s experiments are flawed, they are not, but because his philosophical interpretation of those experiments creates an unnecessary opening for exactly the arguments evolutionary biology has fought to close.

Falsifiability assessment: The claim that morphogenesis exhibits problem solving is testable in the weak sense. Specific predictions about bioelectric manipulation outcomes can be verified or falsified, and Levin’s record here is impressive. But the stronger philosophical claim that this constitutes cognition rather than mechanism is unfalsifiable in principle, because any successful prediction can be claimed as evidence of “intelligence,” while any failure can be attributed to incomplete understanding of the “cognitive” system. This is the same immunization strategy Karl Popper identified as the hallmark of pseudoscience: the theory reshapes itself to accommodate any outcome, yielding no genuine empirical risk.

Natural Induction: Genuine Theoretical Innovation with Hidden Premises

The “evolution over agential material” thesis receives formal treatment in Levin’s collaboration with Richard Watson, whose 2025 Royal Society paper argues that natural induction constitutes a distinct mechanism of adaptation from natural selection. The core insight is elegant: dynamical systems with adjustable connections that yield slightly under stress spontaneously reorganize in ways functionally equivalent to associative learning in neural networks. If evolution builds on organisms that already exhibit learning like dynamics, then natural selection may function not as the source of novel adaptations but as a memory mechanism that stabilizes innovations discovered through natural induction.​

This is a genuinely novel theoretical contribution that deserves serious engagement rather than dismissal. It inverts the standard neo Darwinian picture in a way that generates testable predictions about evolutionary rates and the kinds of adaptations that should emerge in systems with different network architectures, predictions that extend beyond the framework’s original explanatory domain. In Imre Lakatos’s terms, this is the hallmark of a progressive research program: it generates novel empirical content.​

However, critical examination reveals hidden premises that substantially weaken the argument. The framework assumes that biological gene regulatory networks, developmental systems, and ecosystems actually instantiate the relevant network dynamics with sufficient fidelity for associative learning to occur. This is an empirical claim about biological implementation, not a mathematical theorem, and it remains entirely open. The mathematical possibility of natural induction does not establish its biological actuality. The formal demonstration that systems with Hebbian like dynamics can exhibit learning precedes selection behavior says nothing about whether evolved gene regulatory networks are the kinds of systems in which this actually occurs. The gap between mathematical existence proofs and biological mechanism is precisely where the argument’s weight rests, and it has not been bridged.

A thermodynamic monism alternative deserves prominent placement here as a deflationary competitor. Ilya Prigogine’s framework of dissipative structures demonstrates that highly complex, organism like behavior can spontaneously emerge from thermodynamic self organization in far from equilibrium systems, without requiring cognitive vocabulary at any level of description. Living systems navigate toward states of minimum entropy production through purely physical constraint propagation. The phenomena Levin attributes to “intelligence,” regulative plasticity, error correcting development, contextual response to perturbation, are precisely what Prigogine’s framework predicts for open thermodynamic systems operating far from equilibrium. No Platonic space is required; no cognitive vocabulary is needed; no interaction problem arises. This alternative is not merely available in principle. It is the default scientific explanation, and Levin’s framework must earn its cognitive vocabulary against this baseline, not simply assert its necessity.​

The Platonic Ingression Hypothesis: The Framework’s Most Vulnerable and Most Dangerous Claim

Levin’s third cluster of claims constitutes the philosophical core of the framework and its most consequential vulnerability. The proposal, articulated in his blog and forthcoming “Ingressing Minds” paper, holds that physical systems like machines, computers, embryos, biobots are pointers to patterns in a Platonic space. They are interfaces through which these patterns ingress into the physical world. Mathematics provides the precedent: mathematical truths are discovered, not invented, suggesting a structured, ordered, non physical space that also contains higher agency patterns that we call kinds of minds.​

Levin makes the ontological commitment explicit: “The relationship between mind and brain is proposed to be the same as the relationship between Platonic patterns and the morphogenetic outcomes they guide.” This is not a metaphor. It is substance dualism of a Platonic variety, asserting the existence of a non physical realm of forms that causally influences physical reality by guiding biological development and cognitive organization. The philosophical objections to this view are severe, well established, and not adequately addressed anywhere in Levin’s published or public work.​

The Interaction Problem Has Not Been Dissolved

The most immediate and fundamental objection is the interaction problem that has vexed dualist theories since Descartes: how does a non physical space causally interact with physical systems? Princess Elisabeth of Bohemia posed this question to Descartes in 1643, and it has never received a satisfactory dualist answer in the three and a half centuries since.

Levin acknowledges the challenge in public question and answer sessions: “All interactions are ‘unsolved’, including that of matter affecting matter. Causation and interaction in physics has a lot of open questions still.” This response is philosophically evasive and should be recognized as such. It conflates two entirely different problems. The epistemological difficulty of providing a fully satisfying philosophical account of physical causation, a genuine and interesting problem in the philosophy of physics, is not the same as the ontological problem of specifying how an entity that, by definition, lacks spatiotemporal location and physical properties could produce spatiotemporally located physical effects. We can acknowledge uncertainty about the deep metaphysics of physical causation while still insisting that physical causes and effects share a substrate in a way that makes their interaction at least intelligible. Platonic patterns and physical biological systems do not share a substrate on Levin’s own account, and his deflection to the general difficulty of causation papers over this crucial distinction.​

Mathematical Platonists face a version of this problem in the form of Benacerraf’s epistemological challenge: how can we have knowledge of causally inert abstract objects if knowledge requires causal contact with its objects? But Levin’s extension of Platonism to agential minds dramatically intensifies the difficulty beyond what mathematical Platonism faces. Mathematical truths plausibly constrain physical systems through necessitation relationships. The value of Euler’s number limits what compounding growth functions are physically possible. But minds, as Levin conceives them, must actively guide behavior. They must exercise genuine causal influence, not mere formal constraint. This requires a mechanism of interaction that Levin does not provide and cannot provide, because providing it would require specifying physical properties of the Platonic realm that would undermine its non physical status.​

Mathematical Structuralism Offers a Deflationary Alternative That Levin Has Not Refuted

Michael Resnik and Stewart Shapiro’s mathematical structuralism provides an alternative framework that could accommodate all of Levin’s empirical observations without incurring any of the metaphysical costs. On structuralist accounts, mathematical objects are positions in structures rather than independently existing Platonic entities. Physical systems can instantiate mathematical structures without those structures existing independently in a transcendent realm.​

Shapiro’s “ante rem” structuralism grants structures a form of reality but treats them as abstractions from possible instantiations, not as prior, causally efficacious entities waiting to be accessed. If biological systems exhibit mathematical order, if morphogenetic outcomes follow patterns describable in mathematical terms, this could reflect their instantiation of abstract structures rather than their “pointing to” patterns in another realm. The same morphological outcomes would result from the same physical processes. The same experimental predictions would follow. The framework’s empirical content would be preserved entirely. The only thing lost is the transcendent Platonic realm itself, which Occam’s razor identifies as the correct response when an entity’s elimination leaves explanatory work undisturbed.​

Levin has not refuted this alternative. He has not identified a single empirical observation or experimental prediction that his Platonic ingression hypothesis makes but that mathematical structuralism cannot accommodate. Until he does, the parsimony principle demands the structuralist account.

The Chris Fields Contradiction Exposes Levin’s Position as Metaphysical Preference

A particularly revealing tension emerged in public discussions of Levin’s framework: his collaborator Chris Fields, whose work Levin cites in support, explicitly defines the “Platonic realm” as the physical world itself, identifying abstract patterns with physical constraint structures rather than transcendent entities. This directly contradicts Levin’s dualist interpretation. When confronted with this, Levin responded: “I am allowed to contradict even Chris.”​

The significance of this exchange is not merely interpersonal. If Fields can explain the same experimental phenomena by identifying “Platonic space” with physical reality, then Levin’s transcendent interpretation is not empirically forced. It is metaphysical preference. The phenomena do not require the dualist interpretation. They are equally well described under a thoroughly physicalist account. When a theory’s most distinctive and controversial commitments turn out to be optional add ons that leave the empirical content unchanged, those commitments are candidates for elimination, not celebration.

Process philosophers in the tradition of Whitehead’s naturalistic interpreters offer a further alternative. Donald Sherburne’s “Whitehead without God” and related Aristotelian positions locate what Whitehead called “eternal objects,” the forms that structures instantiate, not in a transcendent realm but in the flux of worldly actualities themselves, as immanent patterns of the physical world rather than externally imposed templates. This coupling versus ingression distinction is philosophically decisive: physical systems may not “point to” external patterns but simply instantiate structural possibilities through their internal dynamics. The patterns are real, as mathematical structures are real, but they do not exist in a separate realm awaiting ingression into physics.​

The Discovery Institute Attack Surface: Making the Parallel Explicit

The structural parallels between Levin’s philosophical claims and intelligent design arguments deserve explicit treatment, because Levin’s empirical credibility makes them more dangerous to scientific naturalism than the Discovery Institute’s own arguments, which lack that credibility.

Levin claims that mechanism will always be insufficient to explain goal directedness, a functional equivalent of irreducible complexity. He claims that physical systems are “pointers” to organizing patterns in a nonphysical realm, a functional equivalent of Stephen Meyer’s “specified complexity” requiring an external information source. He explicitly denies that “we are fundamentally physical beings,” a direct assertion of substance dualism. He proposes a “Functional Agency Ratchet” that links learning to emergence through a mechanism he cannot fully specify, a functional equivalent of the “front loading” arguments some intelligent design theorists use to attribute evolutionary outcomes to prior design.

None of this means Levin is a creationist or that his experimental work is invalid. The point is structural and rhetorical. Discovery Institute fellows do not need Levin to agree with them. They need only quote him selectively: a prominent Harvard affiliated biologist asserting that mechanism cannot explain biological goal directedness, that physical systems access nonphysical organizing patterns, and that organisms are not fundamentally physical. The experimental context that qualifies these claims evaporates in citation; the metaphysical commitments remain. Levin’s framework, treated as literal ontology rather than heuristic scaffolding, hands the Discovery Institute exactly the scientific credibility it has spent decades trying to manufacture.

The responsible response is not to suppress Levin’s experimental findings, which deserve wide dissemination. It is to insist, loudly and precisely, that his metaphysical interpretations of those findings are neither required by the data nor philosophically defensible, and to name the fallacies involved with enough specificity that selective quotation becomes identifiable as distortion.

The Prisoner’s Dilemma Simulation and the Limits of Causal Emergence

The IEEE paper on Prisoner’s Dilemma with merge and split dynamics is a more empirically tractable component of the framework. By breaking the standard assumption of fixed players and allowing agents to merge or split based on fitness consequences, Levin’s team demonstrated that multicellular tissues outperform unicellular tissues in the game theoretic sense, providing a formal model for one plausible evolutionary pathway to multicellularity.​

The theoretical question is whether this demonstrates causal emergence in Erik Hoel’s sense: that macroscales can exercise “stronger causal relationships” than their underlying microscales. Hoel’s formalism uses effective information as a metric to show that macro level descriptions can reduce causal noise. The demonstration is formally interesting. But Joe Dewhurst’s critique in Thought identifies the decisive limitation: the simulation demonstrates only weak or epistemic emergence, meaning better descriptions that reduce noise from the modeler’s perspective, not ontologically novel causal powers that exist independently of description. Dewhurst is direct: “Anyone hoping for a demonstration of the strong emergence of ontologically novel causal powers should not be too excited.”

Levin needs strong causal emergence, not the epistemic variety. His framework requires genuine downward causation: collective intelligence directing cellular behavior in ways not fully explicable by cellular level physics. But Jaegwon Kim’s exclusion argument presents the fundamental challenge: if physical causal closure holds, meaning every physical effect has a sufficient physical cause, then higher level causes are either redundant or they violate closure. Either macro level cognition adds something that physical causes do not provide, which violates causal closure, or it does not add anything genuinely distinct, in which case it is epiphenomenal and the “intelligence” language is decorative. Levin’s framework requires the first horn of this dilemma, which puts it in direct conflict with the most well supported principle in contemporary philosophy of mind and physics.​

The motte and bailey fallacy operates throughout this section of Levin’s argument. The motte, the easily defended position, is that macro level descriptions are useful for prediction and control, which is true and uncontroversial. The bailey, the much stronger position that Levin needs but cannot defend, is that this usefulness reflects genuine ontological causal powers at the macro level that are not reducible to or fully explained by micro level physics. The argument migrates between these positions depending on the objection, occupying the motte when the bailey is attacked and advancing the bailey when the motte succeeds.

The Three Paradigm Challenges: Strength Varies by Several Orders of Magnitude

Levin articulates three paradigm challenges whose epistemic statuses vary enormously and should be distinguished clearly.

Challenge A, genome determinism is wrong, succeeds as an argument against simplistic blueprint models of development. The bowtie architecture Levin describes, where evolutionary experience is encoded in engrams that are actively and contextually interpreted rather than mechanically executed, is a genuine conceptual advance over the most naïve versions of genetic determinism. The evo devo research program has independently established substantial genotype phenotype indirectness using entirely mechanistic vocabulary. Levin’s contribution is to emphasize the regulative and contextual dimensions of this process. The cognitive reinterpretation of this indirectness remains underdetermined by the data, but the challenge to blueprint determinism succeeds independently of that reinterpretation. Verdict: Empirically productive reframing; cognitive interpretation remains underdetermined.

Challenge B, minds are found only in brains, and this is wrong, correctly identifies neural chauvinism in mainstream cognitive science. The historical conflation of cognition with neural implementation has produced artificial constraints on the study of intelligence in biological systems. Levin’s work with Xenobots, with planaria, and with Barresi and others on bioelectric computation provides genuine experimental evidence that information processing relevant to behavioral outcomes occurs in non neural biological substrates. However, the challenge faces the Mark of the Cognitive problem identified by Fred Adams and Ken Aizawa: what distinguishes genuine cognition from complex but non cognitive information processing? Adams and Aizawa argue that genuine cognition involves intrinsic, non derived content, representations with meaning independent of the interpreter’s purposes, while many processes we loosely call “cognitive” involve only extrinsic, observer derived content. Levin’s pragmatic response, that cognitive claims are just engineering protocol claims, assertions about what models are useful for prediction and control, sidesteps this problem by definitional fiat, but at the cost of emptying “cognition” of the meaning he needs it to carry. If cognition is just whatever makes prediction and control more effective, the term loses the normative and metaphysical weight that Levin’s stronger claims require. Verdict: Valid critique of neural exclusivity; positive account remains underspecified and vulnerable to the coupling constitution fallacy.

Challenge C, we are not fundamentally physical beings, and this is wrong, is the framework’s most radical claim and its least defensible. The proposal that minds are the patterns, with their ingression into the physical world enabled by the pointers of natural or synthetic bodies, constitutes substance dualism of a Platonic variety. It asserts the causal efficacy of nonphysical entities, contradicts the methodological naturalism that grounds the experimental work on which the framework depends, faces the interaction problem described above, and is empirically indistinguishable from thoroughgoing physicalism for any observation we can currently make or anticipate making. There is no experiment that could distinguish between “cells accessing Platonic patterns” and “cells instantiating mathematical structures through their physical dynamics.” A theory with no distinguishing empirical consequences is, by Karl Popper’s criterion, not science. By Lakatos’s criterion, it is a degenerating protective belt that absorbs anomalies without generating novel predictions. Verdict: Philosophical speculation substantially exceeding empirical warrant; substance dualism with heightened interaction problem. Direct Discovery Institute liability.

Socratic Questions That Expose the Framework’s Structural Vulnerabilities

Five questions expose structural vulnerabilities that no amount of additional data can resolve, because they concern the logical architecture of the framework rather than its empirical content.

The individuation problem: If morphological patterns exist in Platonic space prior to physical instantiation, what individuates them? What makes the “two headed planarian pattern” a distinct form rather than a region of a continuous possibility space? Without a principled individuation criterion, the Platonic realm becomes a formless repository from which any outcome can be post hoc claimed as an instantiated pattern, a framework immune to disconfirmation by design.

The selection problem: How does a physical system “select” which Platonic pattern to instantiate? If the selection is determined by physical facts, the Platonic realm is causally inert and epiphenomenal, doing no real explanatory work. If the selection is not determined by physical facts, we have causally uncaused selection by abstract entities, which is incoherent. Neither horn is acceptable.

The parsimony problem: What explanatory work does “Platonic ingression” do that mathematical structuralism cannot accomplish? If the answer is “none,” Occam’s razor requires the ontologically leaner framework. If the answer is “it explains the causal efficacy of forms,” the interaction problem returns with full force.

The recursion problem: If minds are Platonic patterns, what is the ontological status of the “pointing” relationship between physical systems and those patterns? Is it itself a Platonic pattern? If yes, we need a third level pointer to explain the second level pointing, which requires a fourth level, and so on without termination. If no, the “pointing” relationship must be physical, which relocates all the explanatory work back into physics and eliminates the need for the Platonic realm.

The experimental indistinguishability problem: What observation would distinguish “cells solving problems through collective cognition by accessing Platonic patterns” from “cells exhibiting complex mechanistic behavior that we find it useful to model in cognitive terms”? If no such observation is possible even in principle, the cognitive interpretation is not a scientific hypothesis. It is a metaphysical gloss applied to data that are equally consistent with its negation.

Policy Implications: Precaution Without Premature Attribution

The framework’s implications for bioethics, artificial intelligence ethics, and medical practice are real and significant, but adopting them without stronger philosophical grounding risks serious harms.

If biological tissues have goals in a morally relevant sense, surgical intervention might require forms of consent currently reserved for conscious patients. The Nuffield Council on Bioethics has already flagged neural organoid research as requiring special ethical consideration as organoid complexity increases, a reasonable precaution that does not depend on accepting Levin’s specific metaphysical claims. But Levin’s framework, taken at face value, would extend comparable moral consideration to wound healing, tumor regression, and routine cellular repair, generating obligations with no coherent limit.​

For artificial intelligence ethics, Levin’s substrate independent cognition thesis could support premature attribution of moral status to sufficiently complex artificial systems. If cognition exists on a continuum and is substrate independent, then any sufficiently integrated information processing system may warrant moral consideration. The risk is not that this claim is obviously false, it may prove correct in some form, but that premature attribution, driven by unfalsifiable theoretical commitments rather than empirically grounded criteria, could, in the words of one AI safety researcher, “hamper our ability to control systems or shut them down if we need to.” The appropriate principle is to avoid creating systems whose moral status is genuinely unclear, and to insist on empirically grounded criteria for moral status attribution rather than criteria derived from unfalsifiable metaphysics.​

The epistemic risk may be the most significant of all. If Levin’s framework achieves mainstream acceptance without adequate critical examination, it could redirect research resources toward unfalsifiable questions about Platonic space and cognitive ingression and away from tractable mechanistic investigations of bioelectric signaling, gene regulatory dynamics, and developmental constraint. The framework’s explanatory flexibility, any morphological outcome can be interpreted as “accessing forms” at some depth, any failure to predict as “incomplete understanding of the cognitive system”, makes it resistant to falsification in exactly the way Lakatos identified as characteristic of degenerating research programs: the protective belt absorbs anomalies while the hard core remains empirically insulated.​

What Survives Scrutiny, and What Does Not

Michael Levin’s experimental program has produced replicable, surprising, genuinely important results. Two headed planaria. Xenobots exhibiting collective behavior not encoded in any individual component. Bioelectric pattern manipulation inducing stable morphological changes. These findings challenge simplistic genetic determinism and demand more sophisticated models of developmental plasticity and biological information processing. They represent real scientific achievements.

The theoretical framework built atop these achievements is stratified, with different layers surviving scrutiny to different degrees. The claim that biological organization exhibits regulative plasticity at multiple scales is well supported by Levin’s own data and by independent evo devo research. The claim that cognitive modeling frameworks are useful for generating predictions about this plasticity is pragmatically defensible and heuristically productive; the Teknisk University of Denmark xenobot replication studies, for instance, benefited from Levin style predictive framing. The claim that this usefulness reflects genuine cognition rather than sophisticated mechanism is underdetermined by the data and requires independent philosophical argument. The claim that minds exist as patterns in a nonphysical Platonic realm that physically guides biological organization is substance dualism, faces the interaction problem, is empirically indistinguishable from physicalism, and constitutes a direct Discovery Institute liability.

What survives all scrutiny is this: biological systems exhibit remarkable functional competency, the ability to achieve outcomes through flexible, error correcting, contextually sensitive processes. Whether this constitutes cognition in a metaphysically robust sense depends entirely on how the term is defined. Levin’s pragmatic definition, cognition as whatever makes prediction and control more effective, is genuinely useful for engineering and experimental design while being too thin to support the strong ontological claims his framework advances at its most ambitious.

The framework is best understood, and most productively deployed, as a heuristic: a way of generating novel experiments and cross domain insights, not a literal description of biological ontology. Treated as heuristic, it has proven enormously productive. Treated as metaphysics, it overreaches its evidence and opens scientific flanks that methodological naturalism requires keeping closed. The appropriate intellectual response is to extract the heuristic value while maintaining precise and explicit skepticism about the Platonic superstructure, following Kant’s carefully drawn distinction between the regulative and constitutive uses of teleological concepts: teleology as “a regulative, or heuristic, principle and not a constitutive one; a guide to the conduct of inquiry rather than to the nature of reality.”​

Levin’s data deserve celebration. His ontology deserves a much harder conversation than it has so far received.

References

[1] Levin, M. (2023). Darwin’s Agential Materials. Cellular and Molecular Life Sciences, 80(6), 142. DOI: 10.1007/s00018-023-04790-z. https://link.springer.com/article/10.1007/s00018-023-04790-z

[2] Hartl, B., Levin, M. (2025). What does evolution make? Trends in Genetics, 41(6), 480-496. DOI: 10.1016/j.tig.2025.04.002. https://www.cell.com/trends/genetics/fulltext/S0168-9525(25)00077-0

[3] Levin, M. Darwin’s Agential Materials, PMC: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10167196/

[4] Hartl & Levin (2025), Trends in Genetics, as cited above.

[5] “Michael Levin’s Platonic Biology.” https://medium.com/@AIchats/michael-levins-platonic-biology-fcadcb67c3bf

[6] “Teleological Notions in Biology.” https://plato.stanford.edu/entries/teleology-biology/

[7] Watson, R.A., Levin, M., Lewens, T. (2025). Evolution by Natural Induction. Interface Focus, 15(6), 20250025. DOI: 10.1098/rsfs.2025.0025. https://royalsocietypublishing.org/rsfs/article/15/6/20250025

[8] “Imre Lakatos.” https://plato.stanford.edu/entries/lakatos/

[9] “Toward the Relational Formulation of Biological Thermodynamics.” https://pmc.ncbi.nlm.nih.gov/articles/PMC10814957/

[10] Levin, M. “Platonic Space: where cognitive and morphological patterns come from.” https://thoughtforms.life/platonic-space-where-cognitive-and-morphological-patterns-come-from-besides-genetics-and-environment/

[11] Levin, M. Q&A responses. https://thoughtforms.life/qa-from-the-internet-and-recent-presentations-4/

[12] “Platonism in Metaphysics.” https://plato.stanford.edu/entries/platonism/

[13] “Structuralism in the Philosophy of Mathematics.” https://plato.stanford.edu/entries/structuralism-mathematics/

[14] “Process Philosophy.” https://plato.stanford.edu/entries/process-philosophy/

[15] Shreesha, L., Pigozzi, F., Goldstein, A., Levin, M. (2025). Extending Iterated, Spatialized Prisoner’s Dilemma. IEEE Transactions on Molecular, Biological, and Multi-Scale Communications, 11(2). DOI: 10.1109/TMBMC.2025.3562358. https://ieeexplore.ieee.org/document/10970107/

[16] Dewhurst, J. (2021). Causal emergence from effective information: Neither causal nor emergent? Thought: A Journal of Philosophy. DOI: 10.1002/tht3.489

[17] Hoel, E. “A Primer on Causal Emergence.” https://www.theintrinsicperspective.com/p/a-primer-on-causal-emergence

[18] “Emergent Properties.” https://plato.stanford.edu/entries/properties-emergent/

[19] “Research Using Neural Organoids.” https://www.nuffieldbioethics.org/publication/research-using-neural-organoids-ethical-considerations/

[20] “Moral Dimensions of Synthetic Biological Intelligence.” https://pmc.ncbi.nlm.nih.gov/articles/PMC10796793/

[21] Buckley, C.L., et al. (2024). Natural Induction: Spontaneous Adaptive Organisation. Entropy, 26, 765. DOI: 10.3390/e26090765. https://www.mdpi.com/1099-4300/26/9/765

[22] “Ingressing Minds: Reflections on Levin’s Platonic Research Program.” https://footnotes2plato.com/2025/02/06/ingressing-minds-causal-patterns-beyond-genetics-and-environment-reflections-on-michael-levins-platonic-research-program-for-biology/

[23] “Emergence.” https://iep.utm.edu/emergence/

[24] “AI and Agency: Karl Friston” (critique of FEP). https://philosophyofbrains.com/2025/11/17/ai-and-agency-karl-friston.aspx

[25] Levin, M. “Ingressing Minds: Causal Patterns Beyond Genetics and Environment.” https://www.researchgate.net/publication/388757135

[26] Search Listings Results of 12+ Articles Citing Michael Levin in support of Intelligent Design on Discovery Institute’s Website https://scienceandculture.com/?s=Michael+Levin