Programming Life

Five threads, one claim: voltage is a control variable for anatomy.

The next 32 slides weave through five themes. Each one earns its place by either grounding the claim in formal language, putting it on trial in a wet lab, or showing what falls out the other side as a tool, a company or a clinic.

The Bioelectric Code of Life — eight speakers, one day.

Before we get into the foundations: if any of what follows lands, there is a live SEMF workshop that goes much deeper — bridging fundamental theory with real applications of bioelectric control and morphogenesis. bioelectricitycourse.com

Joel Dietz

MIT · CIMC

Intellectual history · mathematical biology · morphospace theory

Michael Levin

Tufts · Harvard Wyss

Developmental biology · bioelectricity · regenerative medicine

Emmett Shear

Softmax

AI alignment · complex systems · technology entrepreneurship

Derek Lomas

TU Delft · Playpower

Positive AI · education · cognitive science

Adam Goldstein

Hum Labs

Multi-agent RL · AI alignment · collective intelligence

Adam Safron

Tufts Allen Center

Resonance · consciousness · world modeling

Patrick McMillen

Tufts

Developmental biology · collective intelligence · bioelectricity

Wesley Clawson

Tufts

Embodied cognition · neuroscience of learning · collective intelligence

The field has grown faster than any one researcher can track.

ProbKnow is a probabilistic knowledge graph for bioelectricity research — papers, researchers, organizations and concepts, all linked by inferred relationships.

• Surfaces hidden connections between labs, papers and ideas the citation graph misses.
• Lets a non-expert ask "who works on planarian regeneration AND uses GEVIs?" and get a navigable answer.
• Updates as new papers land, so the map stays current as the field accelerates.

Try it live → probknow.com

ProbKnow knowledge graph · live exploration of the bioelectricity literature.

Ten words that do most of the heavy lifting.

It took ~170 years to be sure cells generate their own voltage.

Five experiments settled the basic question — do cells produce electricity at all? Hover any card for the detail.

Plate from Galvani's De Viribus Electricitatis (1791) · animated reconstruction of the bimetal-arc circuit closing.

D'Arcy Thompson: form is geometry, not just inheritance.

In On Growth and Form, Thompson showed you could turn one species' body into another's by warping a coordinate grid — not by changing the parts list.

That image is a conceptual precursor to bioelectricity research a century later: there is some field, some tissue-scale instruction, that helps decide shape on top of the cellular hardware.

The genome lists ingredients. Something else does the warping.

Original · On Growth and Form, 1917 · Figs. 150–153

Same idea, animated · Wolfram Demonstrations

The math arrived 25 years before the biology.

In 1952, less than two years before his death, Turing showed that two diffusing chemicals with the right nonlinear interaction can spontaneously break a uniform tissue into stable spatial patterns — spots, stripes, waves. From nothing but the math.

• Activator + inhibitor diffusing at different rates — the activator self-amplifies locally, the inhibitor diffuses faster and suppresses neighbours. A uniform field becomes a regular pattern.
• Predicted patterns ~25–40 years before any biological mechanism was confirmed.
• Vindicated in marine angelfish stripes (Kondo & Asai, Nature 1995), mouse digit spacing (Sheth et al., Science 2012), hair-follicle spacing (Sick et al., Science 2006), and fish skin pigmentation.

Gray-Scott reaction-diffusion simulation (activator V × inhibitor U, du=0.2 / dv=0.1, f=0.055 / k=0.062) — a noisy central seed self-organizes into Turing spots. Rendered from fractastical/morpholib · 1952-turing-morpho.py.

If form is geometry, then there is a space of all forms — and most of it is empty.

Raup parameterised gastropod shells with just three numbers — W (whorl expansion rate), D (distance from the coiling axis), and T (translation rate along the axis). Sweep the parameters and you generate the full theoretical morphospace of coiled shells.

The striking finding: sampled gastropod shells occupy only a small region of that space. Most of the cube is geometrically possible but biologically unused.

Form lives in a space. Evolution and development pick a trajectory through it.

Raup 1966 · the original W × D × T cube

Raup 1967 · n=405 shell occupancy, W×D plane

Same parameters, animated · Wolfram

Cells build bodies with a small set of repeatable tricks.

Before bioelectricity enters the story, here are the four classical mechanisms developmental biology uses to explain how a sheet of cells becomes an organ. Bioelectricity sits on top of all of them.

Cells sort themselves by adhesion (Steinberg, 1963 / Glazier-Graner CPM, 1992); voltage prepatterns sit on top of the result.

Cells run on voltage — and they share it across tissue.

Three ingredients turn single-cell voltages into a tissue-scale instruction. Every result on the next five slides is a read or a write on this layer.

The same answer, in silicon: local rules → global form. i what's an NCA? If Turing showed that local diffusion rules can spontaneously produce stable patterns, a recent line of work shows the same thing with learned local rules. Neural Cellular Automata are tiny neural networks where every "cell" runs the same update rule on its own state plus its neighbours. Train them, and they grow lizards from a single seed pixel, regenerate after damage, classify MNIST digits collectively, or self-organise complex textures — all from purely local interactions.

LIVE · distill.pub/2020/growing-ca open in new tab ↗

Growing NCA · Mordvintsev, Randazzo, Niklasson & Levin, Distill 2020 — live embed. Scroll inside the frame to see lizards grown from a single seed pixel and regenerated after damage.

Self-Organising Textures · Niklasson, Mordvintsev, Randazzo & Levin, Distill 2021 — same architecture, trained to produce texture statistics. Reaction-diffusion's modern descendant.

Why this slide is here

Bioelectric V_mem patterns are doing the analog version of what NCAs do in silico — every cell runs the same local rule on its own membrane voltage + its neighbours' (gap-junction-coupled). Substrate differs (ion channels vs learned weights); the math is the same. Levin co-authors the Distill series.

Four generations of probes for reading V_mem — and one calibration trick.

Every result on the next slides depends on a measurement technique. The field grew when the tools moved from one cell at a time to a glowing map of a whole embryo.

Each new instrument opened a new biology.

Bioelectricity has always been a measurement-limited field — from the first detection of an animal current to whole-embryo voltage movies.

From a relative thermometer to an absolute V_mem ruler.

Brightness drifts with loading, expression, bleaching and laser power. Lifetime — the nanoseconds the dye spends in the excited state — doesn't. Read the lifetime of the same dye/GEVI and you get voltage in actual millivolts: "this cell is at −42 mV."

Are dissociated neural cultures trainable at the network level?

Wesley Clawson brings closed-loop electrophysiology into dissociated cultures to ask whether learning rules — hebbian, homeostatic, burst-dependent STDP, network-level attractor reshaping, credit assignment — emerge in tissue stripped of body and brain. Two decades of work say cultures are highly adaptable, possibly controllable collectives. The open question is which rule dominates.

MaxLab Live Scope · CMOS-MEA grid (left) shows spatial spike events; right panel scrolls per-burst frame counts as the network fires.

The setup — dissociated cultures grown on multi-electrode arrays, driven through closed-loop hardware (cf. DishBrain · Cortical Labs · FinalSpark on slide 23). Wesley's twist: ask whether the bioelectric repertoire itself changes under stimulation — not just spike patterns, but the rules.

The face appears as a voltage map before any features form.

Using voltage-sensitive dye, the Levin lab imaged the surface of Xenopus embryos as the head developed. Hyper- and depolarized regions pre-figured the eyes, mouth, and nostrils — sometimes hours before the cells got there.

Disrupt the voltage pre-pattern and you get craniofacial mispatterning (colloquially the "Picasso tadpole" phenotype) — features in the wrong places.

Bioelectric prepatterns predict and influence the craniofacial pattern that follows.

Voltage map · CC2-DMPE + DiBAC₄(3)

Anatomy follows · brightfield time-lapse

Modern follow-up · 2026 · post-wound wave-vector flow

Two heads, same DNA — and the change persists.

Cut a planarian flatworm in half and it normally regenerates one head and one tail. Briefly perturb its bioelectric circuit before cutting and it can grow two heads instead — and in Durant et al. (2017) that the altered head/tail target morphology is maintained across subsequent re-amputations via stable gap-junctional state changes, even though the genome was never touched.

• Body shape is partly stored in tissue-level voltage gradients (V_mem)
• That store is rewritable and the rewrite can persist across regeneration
• It behaves like pattern memory: long-term, addressable, content-specific

Live regeneration footage

The bioelectric edit · two-headed phenotype

If a body plan is a graph, regeneration is a graph-rewriting rule.

4 · The corpus the grammar was trained on PlanformDB · Lobo / Malone / Levin · ~16 decades of cut-and-graft

Real output of 1900-planformDB_parser.py from fractastical/morpholib. The 2015 evolutionary search ran against this corpus — every documented experiment, every observed morphology, machine-readable.

If form is partly software, you can recompile it — the xenobot.

Take embryonic skin cells from Xenopus laevis, remove the rest of the embryo, let the cells reaggregate. They do not become frog skin.

They self-organise into a novel anatomy: 2020 — first xenobots, designed by an evolutionary algorithm and assembled by hand (Kriegman et al., PNAS). 2021 — ciliated, self-propelling forms (Blackiston et al., Sci. Robotics) and kinematic self-replication, where xenobots gather loose cells into new xenobots (Kriegman et al., PNAS).

The genome doesn't encode a frog. It encodes capacity.

Xenobot navigating a field of loose cells. Source: M. Levin / Allen Discovery Center, Tufts.

Anthrobots interact with human tissue in culture.

Made from adult human tracheal cells, anthrobots aggregate into "super-bot" clusters. In in vitro assays they move across cultured human neurons, congregate near injuries, and are associated with improved closure in neuronal-wound assays.

Because they're built from the same cells as the tissue around them, they natively share its bioelectric and biochemical signalling — making them candidate research tools for studying repair and tissue dynamics.

Wound-closure assays Regenerative-medicine probe Tissue dynamics

Super-bot cluster (green) on cultured human neurons (red), next to a neural wound. Source: Gumuskaya et al. (2024), Levin lab.

Specify the anatomy. Let the cells compile it.

Stop micromanaging tissues; hand cells a target morphology and let them execute. Lobo & Levin built a literal compiler — a graphical model where you draw the body plan you want and the planarian regenerates it. A 24-hour drug cocktail in a wearable silicone BioDome then carries the same idea into adult Xenopus: 18 months of patterned limb outgrowth from one intervention.

● Compiler · Lobo & Levin · planarian regeneration model

Specify the target body plan in the editor (here: triple-head). The planarian regenerates that shape. The compiler turned "anatomical compiler" from metaphor into a literal piece of software.

● Compiler in vivo · Murugan et al. 2022 · BioDome on adult Xenopus

Adult X. laevis hindlimb amputation → silicone BioDome cap loaded with a 5-drug cocktail (BDNF · 1,4-DPCA · RD5 · GH · RA) → 24 h drug delivery → cap off. Bone, vasculature, innervation and motor function all show up over the next 18 months. Murugan et al., Sci. Adv. 2022.

Same substrate. Different goal.

The same excitable-cell layer that builds bodies can also be wired up to compute. Bioelectricity becomes the I/O channel — voltage in, voltage out — and the cells do the work in between.

• DishBrain (Kagan et al., Neuron 2022) — ~800 K mouse + human-iPSC (induced pluripotent stem cells: adult cells reprogrammed back to a stem-cell state) cortical neurons on a high-density MEA (multi-electrode array — a chip with hundreds of micro-electrodes the cells grow on, used for both reading and stimulating activity), taught to play Pong via closed-loop electrical feedback — the authors interpret the training scheme through the Free-Energy Principle (Friston's framework: systems learn by minimizing prediction error / "surprise"), though the implementation itself is heuristic closed-loop feedback.
• Cortical Labs · CL1 (Melbourne, launched 2025) — billed as the first commercially available biological computer; cloud-accessible "Synthetic Biological Intelligence" platform building on the lab's earlier DishBrain hardware (partner access from 2023–24).
• FinalSpark Neuroplatform (Vevey, Switzerland; Neuroplatform publicly described 2023) — remote API access to ~16 living brain organoids (mm-scale 3D self-organized neural tissue grown from stem cells — "mini-brains"); Wetware-as-a-Service.
• Brainoware (Cai et al., Nature Electronics 2023) — brain organoid used as the reservoir in reservoir computing (a network whose fixed, rich dynamics do most of the work; only a small readout layer is trained); speech-recognition demo.

Why this slide is here: every prior slide treated bioelectricity as a morphogenetic signal. These groups treat the same circuitry as a computational one — a parallel frontier with shared physics, shared tools (MEAs, GEVIs, optogenetic ion channels), and a very different end-game.

Closed-loop MEA + cultured cortical neurons · the substrate underneath every group on this slide.

Cortical Labs CL1 (2025) running DOOM · 200 K living human neurons driving the gameplay. Credit: Cortical Labs.

Different EEG bands = different scales of prediction.

If cognition is hierarchical inference, each oscillatory band carries a different layer of the predictive stack: gamma is the ascending error signal, beta the descending prediction, alpha integrates them into a frame, theta brings agency in, delta sets the slow context. Bioelectricity isn't just resting potentials — it's computational potentials at multiple harmonics simultaneously.

The same logic, across very different cells.

A small zoo of model systems — each stress-tests the same claim (tissue-level voltage instructs anatomy) in a different cell type and body plan.

Xenopus laevis

Frog embryo skin & cardiac cells

Face-of-a-frog craniofacial mapping (2011); xenobots assembled from skin (2020); ciliated swimmers (2021); kinematic self-replication (2021).

Planaria

Neoblast stem-cell flatworms

Two-headed regeneration via gap-junction perturbation (2010–13); pattern memory persists through fission (Durant 2017).

Human airway

Tracheal epithelial cells

Anthrobots — adult human cells self-organize into motile multicellular bots and heal neuronal wounds in vitro (Gumuskaya 2024).

Human cortex

iPSC-derived neurons

The substrate anthrobots interact with — provides the readout for behavioural / wound-healing assays in the 2024 work.

Axolotl & tadpole

Regeneration models

Limb-regrowth with bioelectric drug cocktails on Xenopus tadpoles (Murugan / Levin 2022); axolotl as a comparative reference.

Cell lines & cancer

Melanoma, glia, MDCK

Bioelectric perturbation triggers a neoplastic-like conversion of melanocytes in Xenopus embryos (Blackiston et al. 2011); cancer as a tissue-voltage failure mode.

An AI brain + a robot body, running cell experiments around the clock.

Closed-loop autonomous labs have already proven themselves in chemistry (LBNL's A-Lab) and pharma (Strateos, Emerald Cloud). The Levin lab brought the same idea to living cells.

• MomBot — the Tufts prototype. First proof a bioelectric experiment could be designed, run and scored on a closed loop with no human in the seat.
• One of the few AI + wet-lab systems explicitly aimed at living-cell collective behaviour, not just chemistry or pharma screens.
• Pairs an AI experiment-design "brain" with robotic plate-handling, imaging and stimulation — every result feeds the next hypothesis.

Closed-loop wet-lab automation · the substrate MomBot pioneered for living-cell experiments.

The field is distributed.

Raup's morphospace, generalized.

If voltage maps can navigate cells through the space of possible bodies, where does that space live? Levin's working answer: a structured, non-physical space of forms, patterns and higher agencies that biological systems (and minds, and AI) appear to discover, not invent.

Attractor landscape

Anatomies are stable basins the system gravitates toward. Bioelectric circuits encode which.

Active inference

Cells minimize surprise against a target morphology — Friston's framework, applied to shape.

Cognitive platonism

Forms are real as operational constraints, not as transcendent objects (Dodig-Crnkovic).

Convergence

Biology, cognition and large-scale ML all seem to land on overlapping patches of this space.

Full glossary & symposium slides → /platonic-space

Generalized · Ballell et al. 2022 dinosaur teeth + Marshall et al. 2019 caecilian skulls

Conceptual lineage · Goethe → D'Arcy Thompson → Raup → Wright → Turing

V_mem wave propagating across a tissue grid · inspired by BETSE bioelectric simulations (Pietak & Levin 2016).

Open a K⁺ channel. Stop a tumour.

The bioelectricity literature, graphed in real time.

ProbKnow ingests new papers, links them to ~2,700 testable predictions, and lights up the hypothesis lanes as evidence arrives. Live tool → probknow.com.

One idea, eight steps, a hundred years.

Each step makes the same claim sharper: genes are the parts list, but something at the tissue scale helps decide what gets built — and bioelectricity is a major part of that something.

Where each step comes from

• 1780–1952 — Discovery of bioelectricity
  Galvani, De Viribus Electricitatis in Motu Musculari Commentarius (1791). Volta, Phil. Trans. R. Soc. (1800) on the voltaic pile. Du Bois-Reymond, Untersuchungen über thierische Elektricität (1848–84). Bernstein, "Untersuchungen zur Thermodynamik der bioelektrischen Ströme," Pflügers Archiv (1902). Hodgkin & Huxley, "A quantitative description of membrane current and its application to conduction and excitation in nerve," J. Physiol. 117 (1952). Galvani plate: public domain via De Viribus Electricitatis Tab. I.
• 1917 — D'Arcy Thompson
  On Growth and Form, Cambridge University Press. Animation by S. M. Blinder, Wolfram Demonstrations Project.
• 1952 — Alan Turing
  Turing AM, "The Chemical Basis of Morphogenesis," Philosophical Transactions of the Royal Society B 237 (1952). Confirmations: Kondo S & Asai R, "A reaction-diffusion wave on the skin of the marine angelfish Pomacanthus," Nature 376 (1995); Sick S et al., "WNT and DKK determine hair follicle spacing through a reaction-diffusion mechanism," Science 314 (2006); Sheth R et al., "Hox genes regulate digit patterning by controlling the wavelength of a Turing-type mechanism," Science 338 (2012). Bioelectric extension: Pietak A & Levin M, "Exploring instructive physiological signaling with the Bioelectric Tissue Simulation Engine (BETSE)," Frontiers in Bioengineering & Biotechnology 4: 55 (2016).
• 1966 — David Raup
  "Geometric Analysis of Shell Coiling," J. Paleontology. Animation by Phil Ramsden, Wolfram Demonstrations Project.
• Mechanism primer
  Levin, "Bioelectric signaling: Reprogrammable circuits underlying embryogenesis, regeneration, and cancer," Cell (2021). Levin & Martyniuk, "The bioelectric code," BioSystems (2018).
• 2011 — Face of a Frog
  Vandenberg LN, Morrie RD, Adams DS, "V-ATPase-dependent ectodermal voltage and pH regionalization are required for craniofacial morphogenesis," Developmental Dynamics 240 (2011). Earlier mechanistic work: Adams DS et al., Development (2006).
• 2010–17 — Two-headed planaria
  Oviedo NJ et al., Dev. Biol. (2010); Beane WS et al., Chem. & Biol. (2011); Durant F et al., Biophys. J. (2017) — long-term maintenance of the altered head/tail target morphology across re-amputation, via stable gap-junctional state changes.
• 2013–present — Use case: bioelectric oncology (Gentile lab)
  Lansu K & Gentile S, "Potassium channel activation inhibits proliferation of breast cancer cells by activating a senescence program," Cell Death & Disease 4:e652 (2013) — NS1643 / hERG activator induces p21-mediated senescence in MDA-MB-231 triple-negative breast cancer; subsequent Gentile-lab work extends the program to HCT-116 colorectal cancer and to mouse xenograft tumour-mass reduction. Upstream basis: Blackiston DJ, Adams DS, Lemire JM, Lobikin M, Levin M, "Transmembrane potential of GlyCl-expressing instructor cells induces a neoplastic-like conversion of melanocytes via a serotonergic pathway," Disease Models & Mechanisms 4 (2011); Blackiston DJ & Levin M, "Bioelectric controls of cell proliferation: ion channels, membrane voltage and the cell cycle," Cell Cycle 10 (2011); Chernet BT & Levin M, "Endogenous voltage potentials and the microenvironment," J. Clinical & Experimental Oncology S1 (2013); Chernet BT, Adams DS, Pai VP, Levin M, "Use of genetically encoded, light-gated ion translocators to control tumorigenesis," Oncotarget 7 (2016) — KRAS-driven tumour suppression via forced hyperpolarization. Framing: Tuszynski J, Tilli TM & Levin M, "Ion channel and neurotransmitter modulators as electroceutical approaches to the control of cancer," Current Pharmaceutical Design 23 (2017). Adjacent drug-repurposing line: Brackenbury WJ & Djamgoz MBA on Nav-channel-driven cancer invasion (J. Physiol. 573, 2006); Djamgoz MBA & Onkal R on Nav blockers as anti-metastatic agents (Recent Patents on Anti-Cancer Drug Discovery 8, 2013).
• 2020–21 — Xenobots
  Kriegman S et al., "A scalable pipeline for designing reconfigurable organisms," PNAS 117 (2020); Blackiston D et al., "A cellular platform for the development of synthetic living machines," Sci. Robotics 6 (2021); Kriegman S et al., "Kinematic self-replication in reconfigurable organisms," PNAS 118 (2021).
• 2024 — Anthrobots
  Gumuskaya G et al., "Motile living biobots self-construct from adult human somatic progenitor seed cells," Advanced Science 11 (2024).
• Cell palette — additional sources
  Blackiston DJ, Adams DS, Lemire JM, Lobikin M, Levin M, "Transmembrane potential of GlyCl-expressing instructor cells induces a neoplastic-like conversion of melanocytes via a serotonergic pathway," Disease Models & Mechanisms 4 (2011); Blackiston DJ & Levin M, "Bioelectric controls of cell proliferation: ion channels, membrane voltage and the cell cycle," Cell Cycle 10 (2011). Murugan NJ et al. (Levin lab), "Acute multidrug delivery via a wearable bioreactor facilitates long-term limb regeneration and functional recovery in adult Xenopus laevis," Science Advances 8 (2022) — bioelectric / drug-cocktail tadpole limb regrowth.
• Tools — MomBot · ProbKnow
  MomBot: Levin lab autonomous wet-lab prototype, Tufts (Allen Discovery Center). ProbKnow probabilistic knowledge graph: probknow.com.
• Latent platonic space
  Levin M & Hazan H, symposium & working notes (2025) on the latent platonic space of forms and agencies; see Levin, "Self-improvising memory: a perspective on memories as agential, dynamically reinterpreting cognitive glue," Entropy 26 (2024) for the surrounding cognitive-platonism framing. Glossary: /platonic-space.
• Biocomputers — DishBrain · CL1 · FinalSpark · Brainoware
  Kagan BJ, Kitchen AC, Tran NT, et al., "In vitro neurons learn and exhibit sentience when embodied in a simulated game-world," Neuron 110, 3952–3969 (2022) — the original DishBrain Pong result. Cortical Labs CL1 (Melbourne, launched 2025) — billed as the first commercially available biological computer; builds on the lab's earlier DishBrain hardware (partner access from 2023-24); see corticallabs.com. FinalSpark Neuroplatform — remote-access bio-computing on living brain organoids; Vevey, Switzerland; Neuroplatform publicly described 2023; see finalspark.com. Cai H, Ao Z, Tian C, et al., "Brain organoid reservoir computing for artificial intelligence," Nature Electronics 6, 1032–1039 (2023) — Brainoware speech-recognition demo.
• Bioelectric Code of Life — live workshop
  SEMF Workshop Series, 7 March 2026 (online). Frontier seminar bridging bioelectric theory with applications. Registration & programme: bioelectricitycourse.com.
• Original animations & reconstructions
  Slide 06 (frog-leg twitch): original inline SVG depicting Galvani's bimetal-arc circuit closure (1791). Slide 10 (cell sorting): inline SVG inspired by Steinberg's differential adhesion hypothesis, Science (1963), and the Glazier-Graner cellular Potts model, Phys. Rev. Lett. (1992); Morpheus M0021 reference implementation. Slide 11 (mechanism): original inline SVG of ions / gap-junction equilibration / V_mem → anatomy. Slide 13 (imaging-toolkit voltage trace) and Slide 14 (timeline scrubber): original CSS/SVG. Slide 22 (anatomical compiler): original inline SVG of the BioDome regeneration timeline (amputation → BioDome → blastema → patterned limb). Slide 25 (cell-palette montage): stitched from CC-licensed lab clips referenced above. Slide 29 (V_mem wave on tissue grid): inline SVG inspired by BETSE — Pietak A & Levin M, "Exploring instructive physiological signaling with the Bioelectric Tissue Simulation Engine," Frontiers in Bioengineering and Biotechnology 4: 55 (2016), CC BY; source code BSD-2 at gitlab.com/betse/betse.
• Levin Lab
  drmichaellevin.org · thoughtforms.life
• Bioelectricity Nexus
  bioelectricitynexus.com — curated hub for papers, researchers, and tools.

Over to Mike Levin.

A century-old answer to "why isn't morphology already mathematical?"

Thompson saw, before any of the modern tools existed, exactly the obstacle this deck has been pushing against — and pointed at the way out. Three sentences from page 642:

"The details in which the figure differs from its mathematical prototype are more important and more interesting than the features in which it agrees."

"No chain hangs in a perfect catenary and no raindrop is a perfect sphere; and this for the simple reason that forces and resistances other than the main one are inevitably at work."

"The same is true of organic form, but it is for the mathematician to unravel the conflicting forces which are at work together."

Nine core hypotheses · prior, evidence weight, posterior

The whole vocabulary, in one cell-shaped sticker.

Click any term · opens the full /glossary at that entry

• Agency →
• Anatomical Homeostasis →
• Basal Cognition →
• Bioelectric Code →
• Bioelectricity →
• Collective Intelligence →
• Connexins / Innexins →
• Depolarization / Hyperpol. →
• Diverse Intelligence →
• Electroceuticals →
• Gap Junctions →
• Ion Channels →
• Morphogenetic Code →
• Morphospace →
• Pattern Memory →
• Stigmergy →
• Synthetic Morphology →
• V_mem (Membrane Voltage) →
• Xenobots →