paper
active
2023
paper:s00018-023-04790-z

Darwin's agential materials: evolutionary implications of multiscale competency in developmental biology

ByMichael LevinⓘAllen Discovery Center At Tufts University, Allen Discovery Center, Tufts University + 10 more
DOI 10.1007/s00018-023-04790-z
Bioengineering TeleonomyCollective MorphogenesisMultiscale agency & collective intelligenceCollective intelligence & bioengineering morphogenesiscanalizationBasal Cognitionion channel drugsCellular CompetencyCyberneticsexcitable mediumgenetic assimilationhomeorhesistop-down controlvirtual governor

TL;DR

Cellular collectives operating between the genotype and anatomical phenotype constitute an agential substrate that fundamentally reshapes the evolutionary search process—this is the central claim of Levin's review, which introduces the multiscale competency architecture (MCA) as the organizing framework. Cells, tissues, and organs exhibit regulative plasticity across metabolic, transcriptional, physiological, and anatomical problem spaces because metazoan cells descend from unicellular ancestors with rich behavioral repertoires; evolution therefore searches not the astronomically rugged space of genomic microstates but the smoother space of behavior-shaping signals that exploit these pre-existing competencies. Concrete examples anchor the argument: Xenopus laevis frog skin cells liberated from developmental context spontaneously form self-motile Xenobots capable of kinematic self-replication, a mode unknown elsewhere in the tree of life; polyploid newt kidney tubules achieve normal diameter through a single giant cell wrapping around itself rather than the usual eight-to-ten-cell arrangement, demonstrating real-time downward causation without genomic change; and a 2-day bioelectric intervention targeting planarian ion channels permanently resets head-number patterning, with gap junctional blockade shown to recapitulate 100–150 million years of morphospace divergence across flatworm species. A computational model by Shreesha and Levin (2023) in Entropy directly demonstrates that higher cellular competency levels accelerate evolutionary search and initiate a self-reinforcing ratchet in which improved competency makes structural genomes harder to read by selection, further driving investment in problem-solving capacity. The paper argues this implies that biological evolvability is not a property of genetic architecture alone but emerges from the computational intelligence of the morphogenetic layer, explaining the speed and robustness of evolutionary change and motivating top-down intervention strategies for regenerative medicine and synthetic bioengineering.

What to take away

  1. 1. The multiscale competency architecture (MCA) proposes that cells, tissues, and organs each function as cybernetic problem-solving agents navigating distinct problem spaces (metabolic, transcriptional, physiological, anatomical), meaning evolution searches behavior-shaping signals rather than raw genomic microstates.
  2. 2. Xenopus laevis frog epithelial skin cells, when freed from instructive signals of neighboring cells, self-organize into Xenobots—motile constructs capable of kinematic self-replication by corralling loose cells into clumps that form the next generation, a reproductive mode not observed in any known natural organism.
  3. 3. Human adult tracheal progenitor cells similarly form self-propelled biobots with the demonstrated ability to traverse and heal neural wounds in vitro, extending the Xenobot platform to a clinically relevant human cell type.
  4. 4. Polyploid newts with artificially doubled genome copies still produce kidney tubules of normal diameter: when cells are so large that the standard eight-to-ten-cell arrangement is impossible, a single cell wraps around itself using cytoskeletal bending instead of cell-cell communication, achieving the same macroscale target on developmental timescales without any evolutionary change.
  5. 5. A brief 2-day physiological intervention targeting planarian bioelectric circuits permanently resets the head-number pattern memory, and gap junctional blockade alone recapitulates approximately 100–150 million years of morphospace divergence across flatworm species in genetically wild-type animals.
  6. 6. A machine learning algorithm was required to discover an intervention that breaks the normally coin-toss concordance of melanocyte conversion in Xenopus embryos—without the algorithm, the bioelectric serotonergic control system coordinates all melanocytes in a body to make the same stochastic decision (either all convert or none do), illustrating the non-linear, system-level nature of bioelectric control.
  7. 7. The computational model by Shreesha and Levin (2023, Entropy 25:131) using an artificial embryogeny framework directly demonstrates that increasing cellular competency levels accelerates evolutionary search speed and initiates a ratchet: higher competency makes the structural genome increasingly opaque to selection, channeling further evolutionary pressure toward improving problem-solving capacity rather than hardwiring phenotypes.
  8. 8. Tadpoles in which native eyes are prevented from forming and an ectopic eye is placed on the tail can learn and perform visual behavioral tasks using the tail-eye, which connects to the spinal cord rather than the brain, demonstrating functional sensory-motor adaptation that does not require generations of evolutionary selection.
  9. 9. An open hypothesis the paper raises is whether cells resolving novel physiological stressors—as in planaria rapidly upregulating a small number of genes to achieve barium insensitivity after potassium channel blockade, a stressor with no evolutionary precedent—could sometimes identify which genomic loci to edit for adaptive outcomes, providing a potential mechanistic path for credit-assignment-based Lamarckian-like effects.
  10. 10. To test MCA predictions experimentally, a replicable methodological approach is to extract biological components from their normal context (e.g., dissociating frog embryo skin cells or fragmenting planaria after brief ion-channel-targeting treatments) and then assay their behavior in novel problem spaces, specifically probing for goal-directed outcomes not predictable from default developmental trajectories.

Peer brief — for seminar discussion

Levin's review, published in Cellular and Molecular Life Sciences (2023, doi:10.1007/s00018-023-04790-z), argues that the layer of developmental physiology sitting between genotype and anatomical phenotype is not a passive relay but an agential computational medium whose problem-solving competencies fundamentally restructure how evolutionary search operates. The paper introduces the multiscale competency architecture (MCA) as its organizing framework—the claim that cells, tissues, and organs each constitute cybernetic agents navigating distinct problem spaces (metabolic, transcriptional, physiological, anatomical) using homeostatic feedback loops as their atomic unit of intelligence. Rather than treating morphogenesis as open-loop emergent complexity driven by local rules, MCA casts it as goal-directed pattern completion in anatomical morphospace, directly analogous to pattern completion in neural connectionist systems. The load-bearing finding is that evolution consequently searches not the astronomically high-dimensional space of genomic microstates but the substantially smoother, lower-dimensional space of behavior-shaping signals that exploit pre-existing cellular competencies. Multiple empirical cases are marshaled: Xenopus laevis skin cells freed from developmental context form self-motile Xenobots capable of kinematic self-replication unknown elsewhere in the tree of life; polyploid newt kidney tubules achieve normal macroscale diameter in real time when a single giant cell substitutes cytoskeletal wrapping for the standard eight-to-ten-cell arrangement; a 2-day ion-channel-targeting bioelectric intervention permanently resets planarian head-number pattern memory and can recapitulate roughly 100–150 million years of flatworm morphospace divergence in genetically wild-type animals; and tadpoles with ectopic tail-eyes connecting to the spinal cord rather than the brain perform successful visual learning tasks without any evolutionary adaptation. A computational model (Shreesha and Levin, Entropy 2023) using an artificial embryogeny platform directly demonstrates that higher MCA-level competency accelerates evolutionary search and initiates a self-reinforcing ratchet: competency makes structural genomes increasingly invisible to selection, redirecting evolutionary pressure toward improving problem-solving capacity itself, which the paper predicts explains why planaria—whose genomes are described as mixoploid chimeras accumulated over 400-plus million years of somatic inheritance—have no known morphologically abnormal mutant strains despite decades of attempts. The implications are threefold: MCA smooths fitness landscapes (making many otherwise deleterious mutations neutral), facilitates credit assignment during evolutionary search, and explains why organisms function as general-purpose problem-solving machines rather than niche-specific phenotypic specialists. An alternative framing the paper could have employed is a dynamical systems or Evo-Devo attractor landscape approach, which Levin explicitly sets aside in favor of the cybernetic/connectionist framing precisely because attractor models remain open-loop and do not accommodate top-down goal rewriting. A critical reader would push back on the evidential structure: most of the empirical cases are cited as supporting illustrations rather than as prospective tests of MCA-specific quantitative predictions. The claim that evolution searches 'behavior-shaping signal space' rather than genomic microstate space is conceptually compelling but currently lacks a rigorous formal definition of what counts as that signal space, how its dimensionality is measured, or how one would falsify the assertion that it is systematically smoother than genotype space for any given lineage. The competency ratchet hypothesis in particular rests heavily on the single Shreesha-and-Levin computational model, which uses a stylized artificial embryogeny setup whose mapping onto real developmental systems remains to be established. Whether the MCA framework adds explanatory content beyond well-developed concepts like developmental systems theory, facilitated variation, or genetic assimilation—or whether it repackages them in connectionist language—is a question seminars should press directly.

Methods (1)

  • ion channel drugs
    Pharmacological modulation of ion channels (e.g., barium for K+ channels) used to perturb morphogenesis.

Frameworks (2)

  • Basal Cognition
    An interdisciplinary research framework that reconceptualizes intelligence as observer-relative problem-solving competencies existing on a continuum from simple to highly complex, extending cognition beyond neural systems to pre-neural and non-neural substrates including microbial control loops, plants, tissues, and cellular collectives. It grounds the study of evolutionary and developmental origins of cognitive and behavioral capacities by linking information processing at the chemical and cellular level to classical cognition, and provides philosophical foundations for understanding agency and goal-directedness in systems without nervous systems.
  • Cybernetics
    A mathematical and engineering framework for understanding goal-directed behavior in systems through feedback control mechanisms. Cybernetics formalizes how systems maintain purposive behavior and self-regulation, with applications spanning biology (morphogenetic control), behavioral science, and artificial systems; it provides rigorous language to analyze teleological processes without invoking teleology.

Findings (21)

Claims (21)

Hypotheses (2)

Questions (5)

Related work— refs + corpus + external arXiv

Cited / in-corpus / arXiv badges show which signals surfaced each row. Multi-source rows weighted higher.

+27 more

Similar preprints — Semantic Scholar

Cited by (5)

Cross-corpus bridges (12)

same_concept_as · Nomic cosine

External markdown files that talk about the same concept as this entity.

  • aboutblank_kb
    What are the features of morphogenesis that enable evolution to search difficult phenotype space efficiently despite pleiotropy, degeneracy, and redundancy?questions/what-are-the-features-of-morphogenesis-that-enable.md0.851
  • aboutblank_kb
    How do the computational properties of biological software enable evolution to search a rugged phenotypic space effectively?questions/how-do-the-computational-properties-of-biological-software.md0.843
  • aboutblank_kb
    What mechanisms allow coordination of cell activity toward one consistent morphology?questions/what-mechanisms-allow-coordination-of-cell-activity-toward.md0.842
  • aboutblank_kb
    How can evolution navigate the rugged search space of morphogenesis given its complexity, pleiotropy, degeneracy, and redundancy?questions/how-can-evolution-navigate-the-rugged-search-space.md0.837
  • aboutblank_kb
    What are the mechanisms by which single-cell competencies scale up into collective intelligence at tissue and organismal levels?questions/what-are-the-mechanisms-by-which-singlecell-competencies.md0.833
  • aboutblank_kb
    What is the nature of the target morphology representation that guides cellular collective behavior during development and regeneration?questions/what-is-the-nature-of-the-target-morphology.md0.833
  • aboutblank_kb
    How do bioelectric mechanisms enable evolution to efficiently explore the genotype space without exhaustive search?questions/how-do-bioelectric-mechanisms-enable-evolution-to-efficiently.md0.828
  • aboutblank_kb
    What mechanisms allow the coordination of cell activity toward one consistent morphology across multicellular collectives?questions/what-mechanisms-allow-the-coordination-of-cell-activity.md0.827
  • aboutblank_kb
    What mechanisms allow cells to scale up from individual competencies to collective problem-solving in larger and more complex problem spaces?questions/what-mechanisms-allow-cells-to-scale-up-from.md0.826
  • aboutblank_kb
    Cellular Collective Intelligence Research Programframeworks/cellular-collective-intelligence-research-program.md0.823
  • aboutblank_kb
    Vol.:(0123456789)papers/cleaned/s00018-023-04790-z.md0.821
  • aboutblank_kb
    What computations, algorithms or dynamics enable cellular collectives to respond adaptively, reaching the same form and function despite radical induced changes of circumstances?questions/what-computations-algorithms-or-dynamics-enable-cellular-collectives.md0.820