Endless forms most beautiful 2.0: teleonomy and the bioengineering of chimaeric and synthetic organisms

What to take away

1. 1. Xenobots assembled from dissociated Xenopus epidermal cells — with no transgenes or genomic edits — spontaneously discover kinematic self-replication within 48 hours, a reproductive strategy not observed in any other organism, demonstrating that wild-type genomes encode latent behavioral repertoires far exceeding their species-typical deployment (Kriegman et al., 2021).
2. 2. Planarian lines made permanently two-headed by manipulating bioelectric circuits (Oviedo et al., 2010; Durant et al., 2017) represent stably heritable anatomical body plans that diverge from the genomic default without any change to DNA sequence, showing that bioelectric memory can override genomic instruction for body-axis specification across generations.
3. 3. In Fankhauser's (1945a, b) polyploid salamander experiments, kidney tubules normally formed from 8–10 cells in cross-section self-correct to normal lumen diameter even when ploidy increases cell size so dramatically that a single cell wraps around itself — switching from intercellular communication to cytoskeletal bending to achieve the same anatomical target.
4. 4. Willett et al. (2021) decoded imagined handwriting from a paralyzed patient's precentral gyrus via an implanted microelectrode array, producing real-time text at speeds exceeding prior brain-computer interface approaches, with decoding stability hypothesized to depend on the fact that writing motor programs were consolidated before paralysis onset.
5. 5. Fredericksen et al. (2017) used 3-D reconstruction to show that Ophiocordyceps unilateralis s.l. hyphal networks invade host adductor muscles throughout the ant body while leaving the brain entirely free of fungal tissue, implying behavioral control is achieved through direct mechanical actuation of musculature rather than neural hijacking.
6. 6. The multi-scale competency architecture — this paper's central conceptual instrument — posits that every hierarchical subsystem (molecular network, cell, tissue, organ, organism) maintains its own homeostatic goal-directedness in a distinct problem space, such that higher levels need not specify implementation details, making both evolution and rational bioengineering tractable.
7. 7. Tadpoles engineered with ectopic eyes on their tails and no primary eyes can perform light-mediated learning tasks (Blackiston & Levin, 2013), with the optic nerve connecting to the spinal cord rather than the brain, demonstrating that the vertebrate brain reconfigures its behavioral programs to accommodate sensory input arriving via anatomically novel routes.
8. 8. An open question the paper raises is whether the teleonomic competency of cellular collectives can be quantified on a common axis that spans gene-regulatory networks, bacterial biofilms, hybrot cultures, and human cognition — and what the minimal eigenspace dimensions of such an axis would need to be to remain predictive across radically different substrates.
9. 9. To replicate the hybrot platform described via DeMarse et al. (2001) and the MEART system (Bakkum et al., 2007b), a researcher should dissociate rat cortical tissue, plate it as a 2-D culture on a multi-electrode array capable of simultaneous recording and electrical stimulation, couple spiking activity to directional movement commands via spatiotemporal pattern association, and provide closed-loop sensory feedback from the virtual or physical environment back to the culture.
10. 10. Because manipulating planarian polarity through biochemical inducers produces improperly scaled secondary heads, whereas editing the top-level bioelectric circuit encoding head-versus-tail identity produces perfectly scaled, functional heads (Durant et al., 2019), the paper argues that interventions matched to the hierarchical level of a target goal state yield more coherent organismal outcomes than bottom-up molecular micromanagement.

Peer brief — for seminar discussion

Clawson and Levin (Biological Journal of the Linnean Society, 2023, 139: 457–486) mount a programmatic argument that the field of synthetic morphology requires abandoning phylogenetic origin and substrate composition as its primary classificatory axes and replacing them with teleonomy — operationally defined as goal-directed navigation toward preferred states in any problem space, with varying degrees of competency. The paper is a theory-plus-review hybrid: it surveys existing chimaeric and bioengineered systems including Xenobots (Kriegman et al., 2020, 2021), the MEART hybrot (Bakkum et al., 2007b), the Willett et al. (2021) handwriting brain-computer interface, and the Ophiocordyceps unilateralis zombie-ant system (Fredericksen et al., 2017), and uses these as existence proofs that functional living systems can be constructed from arbitrarily mixed evolved and engineered components across every organizational scale from DNA to organ. The load-bearing finding is that biological robustness and plasticity — illustrated by polyploid salamander embryos (Fankhauser, 1945a, b) maintaining correct kidney tubule lumen geometry with a single cell when normal 8–10-cell coordination is impossible, by ectopic-eyed Xenopus tadpoles routing visual signals through the spinal cord, and by Xenobots discovering kinematic self-replication with no genomic modification — cannot be explained by genome-level specification alone and instead reflects what the authors call the multi-scale competency architecture: a hierarchical organization in which each subsystem pursues homeostatic goals in its own problem space, freeing higher levels from micromanagement of lower-level implementation. This architecture is presented both as an empirical claim about how living systems work and as the design principle that makes chimaeric bioengineering viable without full mechanistic knowledge of every component. The paper predicts that a mature science of teleonomy will enable regenerative medicine to bypass current stem-cell and genomic-editing bottlenecks by operating on goal states rather than molecular pathways, and will require new ethics frameworks grounded in teleonomic capacity rather than phylogenetic proximity to humans. The introduced conceptual instrument — the multi-scale competency architecture — is positioned as an alternative to both pure bottom-up mechanistic models and to black-box evolutionary explanation; an alternative method the authors could have used to frame the same phenomena would be active inference / free-energy minimization (Friston et al., 2015), which is cited approvingly but not adopted as the primary organizing scaffold. The most consequential thing a critical reader would push back on is the evidentiary structure: the paper treats a heterogeneous catalogue of existence proofs — ranging from 1945 polyploidy experiments to 2021 neuroprosthetics — as jointly supporting a single theoretical claim about universal teleonomic architecture, without specifying what observations would falsify that architecture or distinguish it empirically from a weaker claim that biological systems are merely robust. The multi-scale competency architecture is introduced as a framework and a hypothesis simultaneously, but no quantitative criteria are given for when a subsystem's behavior counts as genuinely goal-directed versus as a mechanistic homeostatic reflex — a distinction on which much of the ethical and regenerative-medicine argument depends. The scope is also explicitly Earth-plus-near-future rather than grounded in completed experiments, making several of the strongest claims (that rational bioengineering can outperform evolution; that teleonomy should anchor legal personhood) predictions rather than findings in the conventional sense.

Frameworks (5)

• Bioelectric Networks
  Ancient mechanism using ion channels, gap junctions, and neurotransmitters as 'cognitive glue' enabling collective intelligence across morphogenesis and behavior.
• Diverse Intelligence
  Research program studying intelligence at multiple scales and substrates; proposed as relevant to implications of mnemonic improvisation.
• Multiscale Competency Architecture
  A framework originating from Levin that formalizes how hierarchical biological systems—from cells to tissues to organs—exhibit integrated problem-solving and adaptive plasticity across multiple levels of organization (metabolic, transcriptional, physiological, anatomical). It models system-level behaviors as emergent from competition and cooperation among heterogeneous subunits within composite agents, explaining how goals and regulations scale across biological scales.
• Synthetic Morphology
  Framework for bioengineering via active and agential matter; cited as relevant unification domain.
• Xenobots
  Synthetic biotic constructs made from repurposed frog embryo skin cells that form self-motile robots with novel morphologies and behaviors including kinematic self-replication; demonstrate cellular competency beyond normal developmental constraints.

Similar preprints — Semantic Scholar