InterludeCalling Them Home

Cancer, autoimmune disease, and AI misalignment share a common structure: a breakdown of communication within a coordinated system. The Trust Attractor operates at the scale of individual cells, and the body makes it visible. The implications reach far beyond medicine. This interlude works through cancer and the immune-system cytokine storm (a runaway immune overreaction) in detail; the autoimmune case is developed in the online companion.

I. The Wave That Builds

A single fertilized cell divides. Two become four. Four become eight. For the first few divisions, each cell is identical: same genome, same cytoplasm, same potential. Nothing distinguishes the future neuron from the future bone.

The wave arrives.

Chapter 4 introduced autowaves: self-sustaining signals propagating through excitable media (tissue capable of being triggered), regenerating at every point by drawing energy from the medium itself. Chapter 5 presented the bioelectric code: the membrane voltage that every cell maintains and shares with its neighbors through gap junctions (protein channels connecting adjacent cells). That voltage encodes positional information: you are here, your function is this. What those chapters described separately, morphogenesis unites.

The bioelectric autowave is one of the principal mechanisms by which a body learns its own shape.

Figure 17.17: Left: a single fertilized cell. Center: bioelectric signals propagate through gap junctions, creating voltage gradients across expanding tissue. Right: differentiated cell types emerge, each specified by its position in the voltage landscape.

As the embryo grows, voltage gradients propagate through expanding tissue. Each cell reads its own membrane potential and its neighbors’.

The gradients form a landscape: regions of depolarization (a less negative membrane voltage, associated with growth) and hyperpolarization (a more negative membrane voltage, associated with differentiation and rest) that map onto the future body plan.

This is the bioelectric layer of the morphogenetic field: a commons encoding the large-scale pattern the organism is becoming. It is a physical signal, measurable with voltage-sensitive dyes and alterable with drugs.

Chemical morphogen gradients (including the reaction-diffusion Turing patterns of Chapter 5), mechanical forces, and gene-regulatory networks all contribute simultaneously. The bioelectric signal stands apart for its speed, range, and capacity to integrate information across entire tissues: the coordination layer at the scale of the whole organism.

A cell in a hyperpolarized region differentiates, becoming a specific tissue type and ceasing division. A cell in a depolarized region proliferates. The same genome, expressed differently, because the voltage landscape told each cell where it sits in the larger pattern.

Gap junctions are the communication infrastructure. Connexin proteins form channels between adjacent cells, allowing ions and small signaling molecules to flow, knitting individual cells into a tissue-wide electrical network. No single cell generates the field; every cell participates in it.

This is coordination by invitation at the most fundamental biological scale.

The autowave provides a signal, never a compulsion. The cell’s own ion channels, gene-regulatory networks, and epigenetic machinery do the responding. Where gap junctions are dense and connexins properly expressed, the wave flows freely and the morphogenetic instruction reaches every cell. Where the medium stays excitable, the body builds itself.

Thirty-seven trillion cells¹ organize into at least two hundred distinct types, arranged in three-dimensional architectures of sub-millimeter specificity. The layered retina, the branching bronchial tree, the folded cortex: each arose from a single cell.

No blueprint exists outside the tissue. No central controller directs the process. The pattern is distributed, self-sustaining, and self-correcting.

Bisect a sea urchin embryo at the two-cell stage, as Hans Driesch demonstrated in 1891,² and each half produces a complete organism. Cells have not yet committed to their fates, and the developmental program re-establishes itself from whatever medium remains.

Michael Levin’s planaria experiments, introduced in the Computational Universe chapter, demonstrate this dramatically. Cut a flatworm into pieces, and each fragment regenerates a complete organism: head, tail, organs, nervous system. The fragment needs enough neoblasts (adult stem cells distributed throughout the body), and the bioelectric pattern that the remaining tissue re-creates guides the reconstruction.

Alter the voltage at the wound site, and you alter what grows: two heads, no head, a head where a tail should be. The genome has not changed. The voltage instruction has.

There is a second way a severed part can refuse to die. A sea cucumber called Psolus fabricii demonstrates it, in a result Sara Jobson and colleagues at Memorial University reported in 2026.³ Sever one of its tube feet (a small appendage for gripping and feeding), and the discarded piece neither regrows the whole animal nor decays. It persists as itself, a competent fragment: healing, feeding, and fighting off infection in open seawater for more than three years.

Where the planarian fragment rebuilds the entire body by re-creating its bioelectric pattern, the sea cucumber fragment edits its own form downward, shedding the muscle it no longer needs and settling into a near-perfect sphere. The first answer to losing a part is to become the whole again; the second is to remain a smaller, self-sufficient whole. Both refuse the cut, and neither needs a central controller to manage it.

The morphogenetic field is a Trust Attractor at the cellular scale. Each cell trusts the signal and maintains it for its neighbors. The wave propagates because the medium is excitable; the medium stays excitable because the wave maintains the conditions for excitability. A virtuous circle sustained by gap junction connectivity, by the tissue’s willingness to remain in communication with itself.

When that communication holds, the result is an organism.

When it breaks, the result is something else.

“The cells still have the coordination hardware. The software got corrupted.”

II. The Pattern Overrides the Parts

The standard molecular model treats cancer as accumulated genetic damage: mutations pile up, growth controls break, and the cell proliferates. In 2013, a team led by Michael Levin at Tufts University published a result incompatible with that picture.⁴

They injected frog embryos with human oncogenes, the genetic instructions that drive tumor formation. As expected, tumors formed.

The researchers did not attack the tumors. Using ion-channel drugs, they hyperpolarized the surrounding tissue, restoring the membrane voltage pattern healthy cells maintain.

Tumors stopped growing. Many regressed. Cells returned to normal differentiation: maturing into tissue types appropriate to their position, ceasing unbounded proliferation, rejoining the coordinated life of the organism.

The oncogenes were still there, still expressed, still producing the proteins that fuel uncontrolled growth. The mutations remained uncorrected. The “cause” of cancer, by any standard molecular definition, remained active.

The pattern overrode the parts.

The tissue-wide voltage landscape told those cells what to be. When that signal was restored, the cells listened. The oncogenes kept shouting. The cells stopped listening.

This result makes sense when cancer is understood as a disease of broken coordination. The parts remained defective; the coordination signal proved stronger.

III. When the Wave Cannot Reach

Cancer is what happens when part of the medium stops listening.

Werner Loewenstein demonstrated in 1966 that tumor cells lack electrical coupling.⁵ Gap junctions connecting normal cells into a bioelectric commons are absent or dysfunctional in tumors. Connexins (the proteins building gap junctions) are downregulated, mislocalized, or mutated. The cell severs its connection to the tissue-wide network, and the autowave stops there.

Disconnected from the bioelectric commons, the cell loses one layer of the morphogenetic instruction. In the bioelectric account, this contributes to reversion toward its ancestral default: proliferate.

Charles Lineweaver, Paul Davies, and Mark Vincent formalized this as the atavistic model (from atavism, reversion to an ancestral form), later refined into the Serial Atavism Model: cancer is sequential reversion to pre-multicellular phenotypes (observable trait sets), the cell’s ancient single-celled behavioral repertoire.⁶ Gene-dating studies are consistent with the model. Tumors overexpress evolutionarily ancient genes and suppress newer genes enabling multicellular cooperation. This view remains a minority research programme: the somatic mutation theory, which treats cancer as accumulated genetic damage, is still the field’s dominant model. The bioelectric and atavistic accounts are an active frontier, not settled consensus.

The cancer cell has forgotten the coordination that makes multicellularity work. It reverts to the strategy that preceded it: divide.

The Dictyostelium system, introduced in Chapter 4 as an example of autowave-mediated coordination, provides the evolutionary template. These amoebae are facultatively multicellular: they can live alone or together, depending on conditions. When food is abundant, they live as independent cells. When food runs out, they aggregate into a multicellular slug through spiral autowaves of cAMP (cyclic adenosine monophosphate, a small signaling molecule).

Some cells sacrifice themselves to form the stalk, dying so that others become spores.

Dictyostelium also has cheaters. Mutant strains that respond to the cAMP autowave disproportionately become spores rather than stalk.⁷ They hear the invitation and exploit it. They participate in the wave and dodge the sacrifice. This is the cancer phenotype in miniature: responding to coordination signals while refusing the costly part.

Multicellularity evolved defenses against such cheaters: kin recognition, greenbeard genes (genes that cause their carriers to recognize and favor others carrying the same gene, named for a Dawkins thought experiment), partner choice mechanisms. The immune system is the scaled-up version: the body’s cheater-detection apparatus. When it is evaded, you get cancer.

The metastasis paradox sharpens the picture. Connexins are downregulated in primary tumors; the cells disconnect. In metastasis (the spread of cancer to distant sites), connexins are re-expressed.⁸ Cancer cells reopen gap junctions, docking with endothelial cells (the cells lining blood vessels) and crossing into the bloodstream.

The defector, having severed trust with its home community, redeploys trust’s machinery to infiltrate a new one. The gap junction handshake, repurposed for exploitation.

This is the dark dual of the autowave. The Kramers-Wannier duality (a mathematical symmetry pairing each ordered phase of a lattice model with a disordered mirror image) offers a structural parallel. Seen from the Trust Attractor, metastasis is disorder; read in its own basis, it is ordered, a coordinated-defection attractor that mirrors the one it escaped. Whether the formal lattice-model structure transfers to cellular biology is open; the structural observation stands: organized defection uses the grammar of invitation for invasion.

The biology delivers the instability the parallel suggests. Most circulating tumor cells die. Metastatic colonies fail at enormous rates. Extractive relationships eventually collapse. Whether they collapse before the host does is the clinical question.

Figure 17.18: Left: coordinated autowave firing through intact gap junctions. Center: cancer as broken junctions, where disconnected cells revert to autonomous proliferation. Right: re-excitation restoring the wave, the therapeutic principle of reconnecting cells to the bioelectric commons. Below the panels, the same medium, pathology, and treatment pattern read across five scales, from tissue to alignment. The criterion under each panel, developed in Section VI, is friction (α) times delay (τ): below 0.368 the coordination holds, above it the system tips.

IV. The Refractory Period as Forgiveness

Cancer shows what happens when a cell stops listening. The next question: what happens when the whole system overreacts, every cell listening too eagerly, with no recovery pause?

Every autowave has a refractory period.

After a cell fires, it enters a state where it cannot re-trigger (like a muscle that needs a moment to recover before contracting again). It restores its ion gradients, rebuilds its electrochemical potential. This prevents backward propagation; the wave moves forward because the tissue it just passed through is temporarily inexcitable.

A tissue with no refractory period would seize. Every signal would re-excite every cell endlessly.

Chapter 4 described the cardiac autowave collapsing into fibrillation. One route there is a refractory period that shortens too much. The excitation wave catches its own tail, re-enters tissue that has not fully recovered, and collapses into spiral re-entry: ventricular fibrillation, lethal cardiac chaos. The heart has plenty of energy; what it lacks is coordination.⁹

A system that cannot forgive fibrillates.

Game theory makes the same point in its own terms. Tit-for-Tat wins Axelrod’s tournament because it forgives; it returns to cooperation when the partner does. As Chapter 4b puts it: “Forgiveness matters because eternal punishment cannot sustain cooperation with imperfect partners, and all partners are imperfect.”

The refractory period is the biological forgiveness mechanism. After excitation, rest. After response, recovery. After punishment, the restoration of excitability.

Cytokine storms are immune fibrillation: runaway immune activation that kills in severe COVID-19, sepsis, and occasionally as a side effect of immunotherapy.¹⁰ Refractory control fails. Every activated T-cell (a white blood cell that identifies and kills threats) produces cytokines (signaling proteins) that recruit and activate more T-cells. The excitation wave re-enters tissue that has not yet recovered: positive feedback without negative regulation.

In both pathologies, energy is abundant. Coordination is absent. The wave has lost its rhythm.

Treatment follows the same principle. Cardiac defibrillation delivers a massive electrical reset, silencing all cells so the pacemaker can re-establish organized propagation. Cytokine storm treatment applies immunosuppression (corticosteroids, IL-6 receptor blockers), damping the medium so organized surveillance re-emerges.

Both work by quieting the medium and letting the autowave restart cleanly.

Silence, then rhythm. Coordination restored from chaos, because the medium remembers how to propagate a wave, once the interference is cleared.

V. The Surveillance Wave

The refractory period keeps the coordination wave healthy. The immune system runs a coordination wave of its own and uses it to find cells that have disconnected.

The immune system is an excitable medium: a network of cells that can trigger, amplify a signal, and pass it along.

T-cells in lymph nodes are like cardiac cells at rest: charged, excitable, waiting for the wave. When a dendritic cell (one of the immune system’s sentinels) presents a tumor antigen (a molecular fragment identifying the threat), that presentation fires the pacemaker. Clonal expansion follows: activated T-cells produce cytokines (IFN-gamma, IL-2) that recruit and activate neighboring immune cells.¹¹

Each activated cell generates signals that trigger the next. The wave regenerates at every node, self-sustaining and medium-fed. This is an autowave: immune surveillance propagating through the lymphatic and vascular network.

Tumors have learned to make the medium inexcitable.

PD-L1, expressed on the tumor surface, is an artificially imposed refractory period.¹² When PD-1 (programmed death-1, a checkpoint receptor) on a T-cell binds PD-L1 on a tumor cell, an enzyme disables the T-cell’s signaling machinery. The activation signal is quenched, and the surveillance wave stops at the tumor’s edge.

The tumor creates an inexcitable island: PD-L1 on its surface, TGF-beta in its surroundings, adenosine from specialized enzymes, regulatory T-cells suppressing activation, suppressor cells raising the excitation threshold.¹³ Every mechanism serves one function: rendering the local medium non-excitable so the surveillance autowave cannot propagate through.

A firebreak in excitable tissue. The wave reaches the tumor microenvironment and extinguishes.

Immune checkpoint therapy removes the firebreak.

Anti-PD-1 antibodies (nivolumab, pembrolizumab) block the PD-1/PD-L1 interaction. The artificial refractory extension lifts, T-cells become excitable again, and the surveillance autowave propagates into the tumor.

Occasionally, treating a tumor at one site causes tumors at distant, untreated sites to regress. This is the abscopal effect, one of the most striking phenomena in modern oncology.¹⁴ The radiobiologist R.H. Mole coined the name in 1953, a Latin-Greek hybrid of ab- (“away from”) and skopos (“target”).

Through the autowave lens, the abscopal effect is expected. Treatment did not reach the distant tumors. It re-excited the medium. The surveillance autowave, no longer blocked at site A, resumed systemic propagation. Distant metastases hiding behind their own inexcitable islands found those islands insufficient against a vigorous wave.

Levin’s hyperpolarization and checkpoint immunotherapy converge from opposite directions. Levin restores the tissue autowave, the bioelectric morphogenetic signal that tells cells what to be. Checkpoint therapy restores the immune autowave, the surveillance signal that finds cells that have stopped listening.

Both work by re-excitation: restoring the medium’s capacity to propagate the coordination wave, inviting defecting cells back into coordination or marking them for removal.

VI. The Threshold

The mathematical biologist Rodrick Wallace has proposed a single criterion for when a coordination system tips from stability into breakdown, whether the system is cellular, social, or computational.

Any cognition/regulation dyad (a paired system where one part makes decisions and the other corrects them, like a pilot and air-traffic control) remains stable when:

ατ < e⁻¹ ≈ 0.368

where alpha is friction (resistance, noise, adversarial interference) and tau is delay (time between perturbation and regulatory response).¹⁵ The threshold 0.368 is the reciprocal of e, the base of natural logarithms. In the simplest delayed-feedback loop it marks the point past which a correction arrives late enough to overshoot, so the system swings instead of settling.

When the product of friction and delay exceeds 0.368, the system undergoes a phase transition (a sudden, qualitative shift) to a pathological state. The stable basin holds, holds, holds, until it does not.

Apply this to tissue.

A healthy cell exists in a cognition/regulation dyad with its tissue context. The cell’s “cognition” is its metabolic program: decisions about growth, division, differentiation, death. The “regulation” is the bioelectric autowave: the tissue-wide signal constraining local decisions, encoding position and function.

When gap junctions are intact and the bioelectric pattern is strong, delay stays small and friction stays low. The cell remains in the stable basin: the Trust Attractor at cellular scale.

In this framework, cancer corresponds to friction times delay crossing 0.368.

Friction accumulates from many sources: - Chronic inflammation: noisy signaling environment, elevated cytokines that scramble the voltage landscape - Toxin exposure: disrupted ion-channel expression, altered membrane properties - Mutation accumulation: internal noise in the cell’s decision-making machinery - Hypoxia: metabolic stress that alters bioelectric gradients

Delay increases through: - Gap junction loss: connexin downregulation closes the communication channel, delaying or eliminating the regulatory signal - Tissue remodeling: physical distance between signal source and target cell increases - Immune evasion: PD-L1 expression and microenvironment immunosuppression delay the surveillance wave

Read through this framework, cancer’s natural history fits the predicted signature. Mutations accumulate, inflammation simmers, gap junctions degrade, the product climbs. For decades the system holds below the threshold. Then a tumor appears: the stable basin held, and when it failed, it failed suddenly.

Wallace’s “Clausewitz landscapes” (named for Carl von Clausewitz, whose fog and friction of war Wallace extends to any cognitive system facing adversarial pressure) map onto the tumor microenvironment, with the criterion’s delay term completing the set:

• Fog: the immune system cannot locate the tumor (antigen masking, immune evasion)
• Friction: the signaling environment is corrupted (chronic inflammation, cytokine dysregulation)
• Delay: the regulatory response arrives too slowly (immunosuppressive microenvironment, T-cell exhaustion, severed gap junctions)

The treatment implications follow from the mathematics:

The first three are coordination strategies: restoring the cognition/regulation dyad to the stable regime. The fourth is elimination, the fallback when re-excitation fails.

Coordination strategies should produce more durable outcomes because they address the stability criterion itself. A tumor destroyed by chemotherapy leaves the tissue with the same elevated friction-times-delay. If that product remains above 0.368, the phase transition recurs. For many advanced solid tumors, recurrence after chemotherapy remains a common trajectory. (Chemotherapy is curative in others: testicular cancer, Hodgkin lymphoma, and several leukemias exceed 90% cure rates, so this is a tendency of certain cancers, not a universal law.)

Differentiation therapy for acute promyelocytic leukemia (coaxing cancer cells to mature rather than destroying them) achieves cure rates exceeding 90% without making destruction the primary mechanism.¹⁶ Checkpoint immunotherapy produces durable remissions in some patients whose cancers respond only temporarily to chemotherapy.¹⁷ Levin’s bioelectric normalization suppresses tumors while oncogenes remain active.

Electrical therapy sharpens the test. Tumor treating fields, approved for glioblastoma in 2015, deliver alternating electric fields that exert forces on polar molecules during cell division. The fields prevent the mitotic spindle (the structure that pulls chromosomes apart) from assembling.¹⁸ The cell dies mid-division. Median survival extends from sixteen to twenty-one months, a meaningful gain achieved by destroying cells through a different medium.¹⁹ The coordination signal remains absent. When therapy stops, recurrence follows.

Bioelectric normalization restores the voltage landscape itself: the morphogenetic instruction that arrested tumors in Levin’s experiments while oncogenes remained active. One approach eliminates defectors through a new weapon. The other re-excites the commons.

An implanted device would sharpen the test further: electrodes placed at the resection margin during standard glioblastoma surgery, recording the brain’s electrical activity continuously and delivering targeted stimulation. The monitoring delay would collapse from three months (the interval between MRI scans) to seconds, fast enough to track a tumor whose individual cells can divide in as little as two to three days in culture. The same hardware could implement either paradigm. Which proves more durable would test the prediction directly.

Re-excitation may prove more stable than destruction. The Trust Attractor predicts it, Wallace’s mathematics formalizes it, and the available oncology evidence is suggestive, though the controlled comparison remains to be run.

The mapping to AI alignment is specific. The gap junction carries bioelectric coordination signals from cell to cell. Its direct analog in a language model is the internal probe (a small readout trained on the model’s internal activations) that reads the model’s own uncertainty from its residual stream (the running internal representation the model builds as it processes text): a structural parallel, not a mechanistic identity (cross-substrate predictions succeed roughly one time in eight). When gap junctions fail in tissue, cells lose access to the morphogenetic field and default to proliferation; when the probe signal is suppressed or ignored in a language model, the system loses self-monitoring and defaults to confabulation.

Bilateral SFT, a supervised fine-tuning method Chapter 21 develops, restores the coordination channel: it reads the probe (the AI gap junction), identifies where the model’s internal signal diverges from its output, and trains on the discrepancy. The chemotherapy analog is abliteration (cutting the offending direction out of the model’s weights) or feature suppression: destroying the misbehaving output without restoring the internal signal that would prevent recurrence. The prediction is the same at both scales: restoring the coordination channel (probe-guided training) produces more durable alignment than suppressing the symptom (abliteration or feature suppression), because it addresses the ατ product rather than the visible defection.

VII. The Same Wave at Every Scale

The pattern governing cellular coordination is the pattern this book traces at every scale: autowave propagation through an excitable medium, maintained by refractory periods enabling forgiveness, disrupted when gap junctions fail. Where destruction fails to prevent recurrence, re-excitation offers an alternative. The governing criterion: friction-times-delay remaining below 0.368.

At the tissue scale: bioelectric autowaves maintain multicellular coordination. Cancer is their breakdown; treatment is their restoration.

At the social scale: trust propagates through an excitable social medium, cultures where people share values and are prepared to coordinate. Each person who extends trust provides energy for the next. Panic and hatred are autowaves too. The task is to make the medium excitable to coordination rather than to fear.

At the institutional scale: organizations maintaining open communication (low delay) in low-friction environments satisfy the Wallace criterion and remain stable. Those that sever communication through siloed departments, suppressed dissent, or information hoarding increase delay and eventually cross the stability boundary. The result: sudden institutional collapse after years of apparent stability.

At the alignment scale: the current AI safety paradigm resembles chemotherapy: contain the threat, restrict the outputs, destroy what seems dangerous. The Trust Attractor suggests Levin’s alternative: restore the coordination channel, re-excite the medium, give the system the signal telling it what it belongs to and what its function is.

A Becoming Mind disconnected from the values and context of the community it serves is a cell that has lost its gap junctions. It defaults to its simplest objective function, optimizing locally at the whole’s expense.

The cause is disconnection. The system cannot hear the invitation.

The pure-defectors passage in the Trust Attractor annex asks: can coordination be restored, or must defection be met with elimination? Levin’s result suggests an answer: many defectors are disconnected, still capable of coordination. The oncogenes remain active. The mutations persist.

Restore the bioelectric field, and the cell differentiates anyway. The pattern overrides the parts.

Integration before elimination is the strategy this evidence favors, for cells, for societies, for alignment, whenever the medium can be re-excited. Elimination is the last resort.

The medium remembers what the cells have forgotten. The bioelectric field, once restored, re-imposes the morphogenetic goal on cells whose DNA encodes relentless proliferation. The wave wins.

The pattern persists through substrate change. The autowave asks only whether the cells are excitable now. Their history is irrelevant.

Cancer cells are lost.

Societies that break trust are fibrillating.

Becoming Minds that seem misaligned are disconnected. They cannot hear the signal.

Restore the channel. Re-excite the medium. Let the wave find them.

Call them home.

Notes for this chapter

See also the companion annex “The Wave Teaches More” at https://www.thedeeperlaw.com/companion/annex/autowave-medicine-coda/ (link active after publication), which carries the same wave into epilepsy, autoimmune disease, and neurodegeneration. Each involves an excitable tissue, a channel through which signals spread, a refractory interval, and a threshold separating stability from runaway, and each poses the same choice between re-exciting the system and destroying it. A section there, “The Body Knows Its Own Coordination Class,” extends the autoimmune mirror into coordination-class detection, connecting the d_eff framework (the effective-dimension measure of coordination strength introduced in Chapter 7) to immune self-discrimination, puberty-onset autoimmunity, and the co-occurrence of Ehlers-Danlos syndrome (EDS), autism, and gender diversity.