David Sinclair and Michael Levin appear to study different biological problems. Sinclair investigates aging, epigenetic information and cellular rejuvenation. Levin investigates bioelectricity, morphogenesis, regeneration and collective cellular intelligence.
But viewed through the ceLLM framework, their work converges on the same underlying principle:
Life is governed by a nested intelligence network whose information is physically embodied in electrical state, timing, topology and molecular geometry.
Sinclair primarily examines the persistent information state inside the cell: how the genome remains largely intact while its regulatory interpretation becomes progressively disorganized. Levin primarily examines the collective runtime state between cells: how tissues use voltage, ion flow and electrical connectivity to remember anatomical goals and coordinate the construction and repair of whole bodies.
Sinclair is studying the fidelity of the stored model. Levin is studying how that model is executed collectively.
The Genome Is More Than a Parts List
DNA is often described as a blueprint, but that metaphor is incomplete. The protein-coding sequence specifies components, yet it does not independently determine when, where, how strongly or under what conditions those components will be used.
Much of that control resides in the non-coding regulatory genome and its three-dimensional organization:
- enhancers and silencers;
- promoters and insulators;
- splice-regulatory regions;
- transcription-factor binding sites;
- CTCF and cohesin boundaries;
- methylation patterns;
- histone states;
- chromatin loops;
- nuclear-lamina contacts.
In ceLLM language, these structures form an evolution-trained hardware matrix. Coding sequences define many of the components the cell can construct, while non-coding regulatory architecture supplies much of the routing logic, thresholds, weights and biases through which environmental information is interpreted.
The genome therefore contains both relatively stable evolutionary memory and a writable regulatory layer. Its sequence establishes the long-term prior, while methylation, chromatin folding and nuclear organization continually adjust how that prior is accessed.
DNA is the instrument. Regulatory architecture is the fingering system. Bioelectricity supplies the breath. The living phenotype is the tune.
The CACNA1C Study Exposes the Interface
The 2025 CACNA1C sleep study provides a particularly revealing example.
Researchers exposed 34 genotyped volunteers to standardized 700 MHz, 3.6 GHz or sham conditions before sleep. Only the 3.6 GHz exposure produced a significant change in spindle center frequency, and that response appeared in the T/C carriers of the intronic CACNA1C variant rs7304986—not in the matched T/T group.
The variant is located in a non-coding region. It does not itself substitute an amino acid in the CaV1.2 calcium-channel protein. Yet the two inherited regulatory contexts produced different bioelectrical responses to the same external stimulus.
That does not prove that rs7304986 directly altered channel expression, nor does it establish precisely how the field coupled to the system. The variant could be functioning as a regulatory element, altering chromatin context or tagging another causal variant through linkage. Nevertheless, the result demonstrates an important principle:
An identical environmental input can be processed differently when the inherited regulatory geometry is different—even when the relevant protein-coding sequence has not been changed by that variant.
That is exactly what we would expect if the non-coding genome participates in configuring the cell’s biological transfer function.
The external field was the input. The CACNA1C-associated regulatory context helped determine the system’s susceptibility. Calcium-dependent neural circuitry processed the perturbation. Sleep-spindle frequency was the measurable bioelectrical output.
This is where Sinclair’s and Levin’s levels meet: persistent genomic and epigenomic state determines how the live electrical network interprets an environmental prompt.
Sinclair and the Loss of Regulatory Fidelity
Sinclair’s information theory of aging begins from an important observation: aged cells generally retain the genetic information needed to produce more youthful function. The underlying sequence has not simply disappeared. What deteriorates is the cell’s ability to access and interpret that information with its earlier precision.
In Sinclair’s induced-changes-to-the-epigenome, or ICE, experiments, double-strand DNA breaks were introduced in a manner designed to disturb epigenetic organization without producing widespread sequence mutations. The resulting animals developed molecular and physiological features associated with aging. Expression of the partial-reprogramming factors OCT4, SOX2 and KLF4 subsequently restored portions of the youthful epigenetic state and improved several measured functions.
Earlier work in retinal ganglion cells showed that OSK expression could promote axonal regeneration and restore vision in mouse models, with the beneficial effects requiring the DNA demethylation machinery TET1 and TET2.
The central implication is profound:
At least some youthful biological information remains recoverable because it was not destroyed at the sequence level. The system lost high-fidelity access to it.
Under ceLLM, this resembles defragmentation—but not indiscriminate erasure.
The goal is not to remove every methyl group or epigenetic adaptation. Many marks are essential for cellular identity, development and survival. Instead, partial reprogramming appears capable of reorganizing portions of an increasingly disordered regulatory state while avoiding complete dedifferentiation.
The same genome can therefore produce an aged output or a more youthful output depending on the geometry and state through which it is interpreted.
Adaptation Can Accumulate as Fidelity Debt
During life, cells continuously modify themselves in response to inflammation, nutrient availability, injury, toxins, oxidative conditions, hormonal signals, mechanical forces and electrical activity.
Many of those changes may be locally rational. A cell responds to the conditions confronting it now—not to the organism’s desired lifespan decades in the future.
A regulatory adjustment that improves survival during the next five minutes may introduce a small long-term cost. Repeated adaptations can accumulate like patches applied to a running system. Each patch may solve an immediate problem, yet the growing collection of patches can distort the original routing architecture.
ceLLM describes this as fidelity debt or somatic overfitting:
- A perturbation demands compensation.
- The cell alters its regulatory state.
- That alteration changes how the next signal is interpreted.
- Additional compensation becomes necessary.
- The system gradually drifts away from its earlier high-fidelity attractor.
Aging, in this view, is not simply the passage of time or the accumulation of molecular debris. It is the progressive loss of precision with which cells interpret context and coordinate appropriate action.
Sinclair’s rejuvenation results suggest that portions of this drift are reversible because the evolutionary information needed for youthful operation remains latent within the system.
Levin and the Collective Runtime
Michael Levin’s work begins one scale higher.
Cells do not make decisions in isolation. Ion channels, pumps and gap junctions connect cells into bioelectric networks capable of storing state, processing information and coordinating activity over large anatomical distances.
These networks help regulate:
- polarity;
- organ identity;
- growth boundaries;
- differentiation;
- regeneration;
- cancer suppression;
- target morphology.
Levin’s planarian experiments demonstrate that transiently modifying bioelectric communication can change the anatomical form toward which the animal regenerates. A two-headed configuration can persist through later rounds of cutting without alteration of the DNA sequence. The tissue network has entered a different morphogenetic attractor.
This is not simply gene expression unfolding from a fixed blueprint. The cell collective is comparing present anatomy with a stored setpoint and coordinating actions intended to reduce the difference.
Levin’s bioelectric network is therefore the live, distributed execution environment. It integrates the decisions of many cellular inference engines into a larger-scale intelligence with goals that no individual cell could represent alone.
The Memory Stack of Life
The apparent disagreement between genetic and bioelectric explanations disappears when biology is viewed as a nested memory stack.
The DNA sequence and non-coding regulatory grammar contain deep evolutionary priors.
Methylation, histone state and chromatin topology preserve writable cellular history.
Ion channels, mitochondria and nuclear electrical state translate the immediate microenvironment into cellular decisions.
Gap-junction-connected bioelectric networks integrate those decisions across tissues.
Morphology, repair, metabolism and behavior are the resulting outputs.
Information moves in both directions. Genomic architecture influences ion-channel expression and electrical susceptibility. Bioelectric activity controls calcium signaling, transcription and epigenetic marking. Runtime modifies the persistent state, and the persistent state configures the next runtime.
Sinclair shows how the cellular interpreter can become detuned and partially retuned.
Levin shows how populations of those interpreters cooperate to remember and pursue large-scale anatomical goals.
The ceLLM Unification
The ceLLM framework adds the missing connection between these scales.
Each cell inherits an evolution-trained physical prior embodied in genomic sequence, regulatory architecture and chromatin organization. It receives a continuous prompt consisting of ionic, electrical, metabolic, redox and mechanical information. Its present geometry determines how that prompt is interpreted and which biological action becomes most probable.
The tissue then integrates millions of these local inferences through bioelectric and biochemical communication.
In high-fidelity biology, inherited priors, cellular interpretation and tissue-level goals remain coherently coupled. The correct signals arrive at the right thresholds, in the right sequence and at the right time.
In low-fidelity biology, that coupling deteriorates. The genome may remain intact, but access becomes noisy. Calcium signals become mistimed. Mitochondrial responses become less appropriately scaled. Epigenetic patches alter later interpretation. Cells remain active, yet their local decisions become progressively less aligned with the organism’s larger needs.
This creates the proposed meta-disease state: not one predetermined illness, but declining control fidelity that makes many pathological attractors more accessible.
The endpoint could be cancer, neurodegeneration, endocrine dysregulation, infertility, impaired repair or metabolic disease. Which endpoint appears depends on genotype, developmental history, tissue state, co-exposures and the location of the weakest regulatory boundary.
One Intelligence, Seen From Two Directions
Sinclair approaches biological intelligence from inside the nucleus outward. Levin approaches it from tissue geometry inward.
Sinclair asks:
How does a cell lose and recover accurate access to its youthful identity?
Levin asks:
How do cells combine their individual competencies into a collective mind capable of remembering and rebuilding a body?
ceLLM joins those questions:
What physical architecture allows each cell to interpret its local environment, retain experience and participate in a larger bioelectric intelligence?
The answer is not DNA alone and not bioelectricity alone. It is their continuous recursive coupling.
Evolution writes deep memory into genomic and regulatory geometry. Experience modifies that geometry. Bioelectricity queries it in real time. Cellular outputs reshape the tissue environment, and the tissue network feeds new information back to every cell.
Sinclair studies the fidelity of the instrument.
Levin studies the geometry of the orchestra.
ceLLM proposes that life emerges from the recursive conversation between them.
