Memory of Form: Bioelectric Computation and Agential Matter

For the last seventy years, the biological sciences and the technology sector have operated under a shared, unquestioned metaphor: that biology is like classical computing. In this framework, we assume a strict separation between hardware and software. DNA is the 'source code', or a read-only memory drive, and the cells are the passive 'bricks' that simply execute that code.

This mechanistic view has driven major advancements in genetics, but it is currently losing its theoretical merit. When we try to build complex synthetic biology, or when we attempt to reverse-engineer diseases like cancer or severe birth defects, the "DNA-as-software" model breaks down. If every cell in your body contains the exact same genetic code, how does a stem cell know whether to build the rigid structure of a femur or the delicate, excitable tissue of a retina? More importantly, when an organ is damaged, how does the tissue know what shape to rebuild?

In Emergent Self-Organization, through Nikta Fakhri’s non-equilibrium physics, we established living tissue is a thermodynamic active material, like an engine fueled by energy to stave off entropy. In Bioelectic Failsafe, relying on Adam Cohen’s topological biophysics, we saw how that active material uses spontaneous symmetry-breaking to build robust bioelectric walls, creating protection against biological noise.

Now we must look to developmental biology, specifically the pioneering research of Dr. Michael Levin at Tufts University, to answer a final question: What are those bioelectric boundaries actually protecting?

They are protecting memory.  Not just the high-speed neural memory we associate with a human brain (which is likely an evolutionary offshoot of this exact process) but the ancient, basal foundation of cognition itself: morphological memory, the bioelectric computation of shape.

Bioelectric Maps: The Memory of the Planarian

To understand morphological memory, we must look at the Planaria, a highly regenerative flatworm. If you bisect a planarian worm horizontally the tail half will grow a new head, and the head half will grow a new tail, resulting in two perfectly formed worms.

Under the classical genetic model, we assume the DNA simply triggers a "rebuild" sequence. But Michael Levin’s lab asked a deeper question: How does the tissue know where the missing head is supposed to go? And more crucially, how does it know when to stop growing once the head is finished?

Levin’s team discovered that the worm’s shape is stored in the bioelectric network, not its genome. Every cell in the worm's body has ion channels that pump charged particles (potassium, sodium, calcium) across the cellular membrane, creating specific voltage gradients. Levin's lab found that before the worm even begins to physically rebuild the missing head, a bioelectric "pre-pattern" flashes across the tissue. The electrical network literally draws the shape of the required anatomy before a single stem cell moves to build it.

The proof of this mechanism came when Levin’s team altered the bioelectric gradient of the tissue without touching the worm's DNA at all. By temporarily opening and closing specific ion channels using localized drugs, they altered the bioelectric memory of the tissue. When they cut the worm, it grew a second head on its tail end.

Even further, when they took that two-headed worm and cut it again in plain water (with no further bioelectric manipulation), it continued to grow two heads. The bioelectric network had permanently rewritten its target morphology. The DNA remained unchanged, but the bioelectric software had been reprogrammed.

This can inform a new understanding of topological protection. Within a unified framework, we can hypothesize that the domain walls Cohen observed in excitable media are also physical registers of this morphological memory and store the "target state" of the organism.

Agential Problem Solving: The Computation of Shape

If the tissue stores its own target morphology, it changes how we view the cells themselves. They can be seen as agential material that navigates a problem space to achieve a specific geometric goal.

Levin demonstrated this by grafting the primordium of an eye onto the tail of a blind tadpole. Under the classical model of top-down control, this should be a useless appendage. The brain has no genetic programming that dictates, "If an eye appears on the tail, route a new optic nerve down the spine."

Yet, the eye actively sought out the spinal cord, established a synaptic connection, and the tadpole could successfully see and navigate its environment.

In Levin’s framework, this is evidence of morphological plasticity. The cells negotiated this solution dynamically without a centralized master plan to achieve anatomical homeostasis.

But how does a physical material negotiate a structural change of this magnitude?

We can answer this by bridging Levin’s biology with the thermodynamics of active matter. In the language of physics, for a material to escape a rigid historical state and reorganize, it must undergo functional decoherence, experiencing a phase transition that breaks its original symmetry to establish new topological boundaries.

In Levin’s framework, the tail cells are not permanently locked into their identity as "tail tissue." They possess the inherent bioelectric plasticity to navigate toward a new functional goal.

Anthrobots and the Computation of Shape

The biggest validation of the agential material model comes from Levin’s recent work with synthetic biology, specifically the creation of Anthrobots.

Led by researcher Gizem Gumuskaya, the lab extracted adult human tracheal cells. In a human airway, these cells are heavily constrained by their mechanical environment, using their hair-like cilia to perform a single rigid job: sweeping away mucus. Gumuskaya stripped these cells of their native environment and placed them in a novel context.

The lab did not genetically edit the cells or insert new DNA to turn them into robots.

Stripped of their top-down boundaries, the cells spontaneously aggregated into motile, multicellular spheres. Their cilia, no longer pressed against a windpipe, pointed outward and became oars, allowing the structures to swim and navigate mazes. Astoundingly, when placed on a scratched layer of human neurons, they spontaneously induced the neurons to heal the gap.

How do unedited windpipe cells know how to build a swimming, wound-healing robot?

We can answer this by bridging Levin’s biology with the thermodynamics of active matter. In classical physics, a system will naturally relax into its most stable, lowest-energy configuration based on its physical constraints like water flowing to pool at the bottom of a newly formed valley.

The human genome does not hardcode a rigid "windpipe." Instead, it provides the hardware for a bioelectric network that constantly seeks stability across a dynamic landscape. When pressed against a human airway, the most stable configuration for these cells is a flat sheet. But when those mechanical boundaries were removed, the landscape changed.

Freed from their historical constraints, the cells spontaneously relaxed into a new stable configuration: a sphere. In the language of physics, the system found a new thermodynamic minimum. In Levin’s framework, the cells navigated a dynamic latent morphospace to find a new morphological attractor. By reconciling the thermodynamics of the tissue with the bioelectric memory of the cells, the Anthrobots self-organized into a novel, functional geometry that evolution never explicitly planned for.

Redefining Intelligence

The convergence of Fakhri’s active matter, Cohen’s topological boundaries, and Levin’s bioelectric memory creates a unified, predictive framework for emergent complexity.

We must expand our definition of intelligence. Intelligence is not solely the domain of neurons firing in a primate neocortex. Intelligence is the ability of a system to navigate a complex space toward a target state, overcoming unpredictable obstacles along the way. In the case of living tissue, that space is anatomical.

Biology does not separate the computer from the machine. Its media are thermodynamic substrates that compute their own optimal geometry through spontaneous symmetry-breaking, storing the memory of their target shape in the topological structure of its voltage.

Understanding this is not merely an academic exercise in biophysics. This is likely the roadmap for the next century of deep technology. By translating the underlying physics of how complex systems self-organize, we move away from the brute-force, destructive interventions of classical engineering. The future of medicine, computation, and infrastructure belongs to systems that do not fight the thermodynamic noise of reality, but actively compute with it.