How Life Process Information

It shows life as informational architecture via Michael Levin's bioelectric framework: pattern memories stored in somatic electrical networks, not genes—software running on genetic hardware. Agency scales heterarchically: cell, tissue, organ, organism each solve problems in their domain, bound by "cognitive glue" (bioelectric fields), not hierarchical command. Somatic memories aren't directly inherited, but the Baldwin effect enables learned competence to become genetically encoded over generations. Hox genes set coordinates; bioelectric networks execute adaptable patterning—complementary, not hierarchical. Framed within your Unification Project, value flows from individual verification, not imposed doctrine. Life processes information relationally: circular causality, distributed cognition, multi-scale competency. Reality emerges from lawful, testable protocols where intelligence is nested, adaptive, and sovereign at every scale.

Hello Kimi, good afternoon. I try to catch up the work o professor Michael Levin. So, can you retrieve the information of 2026 papers, including is team?

Based on my search, I found information about Professor Michael Levin’s 2026 work. Here’s what I retrieved:

2026 Papers

From the search results, I found one confirmed 2026 publication so far:

“Top-down perspectives on cell membrane potential and protein transcription”

• Authors: Cervera, J., Levin, M., Mafe, S.

• Journal: Scientific Reports, 16(1996)

• DOI: 10.1038/s41598-025-31696-6

• Topic: Bioelectricity, cell membrane potential, and protein transcription

Additionally, there’s a mention of an upcoming Physics of Life Reviews paper for 2026 in his Tufts faculty profile, though specific details weren’t provided

.

About Professor Levin and His Team

Position & Affiliations:

• Vannevar Bush Distinguished Professor of Biology at Tufts University

• Director of the Allen Discovery Center at Tufts

• Director of the Tufts Center for Regenerative and Developmental Biology

• Associate Faculty at the Wyss Institute, Harvard University

• Co-director (with Josh Bongard) of the Institute for Computationally Designed Organisms

Current Research Focus (2025-2026):
The lab continues working on several key areas :

• Bioelectric networks: How somatic cells form electrical networks for storing and recalling pattern memories that guide morphogenesis

• AI tools for biology: Next-generation AI for understanding top-down control of pattern regulation (an “anatomical compiler”)

• Regenerative medicine: Applications in birth defects, organ regeneration, cancer suppression, and aging

• Synthetic organisms: Creating novel living constructs (including Xenobots and Anthrobots)

• Diverse intelligence: Understanding cognition across scales—from cells to tissues to organisms

Team Structure:
The lab consists of approximately a dozen postgraduates and researchers, working at the intersection of developmental biology, computer science, cognitive science, and bioengineering

. The lab is described as unusual in biology due to Levin’s background as a software engineer and his deep interest in philosophy of mind

.

Key Collaborators Mentioned:

• Josh Bongard (University of Vermont) - co-director of Institute for Computationally Designed Organisms

• Donald Ingber and James Collins (Wyss Institute, Harvard)

• Various postdocs and graduate students working on specific projects

Recent Media & Speaking Engagements (2026):

• Featured on The Tim Ferriss Show (January 21, 2026) discussing reprogramming bioelectricity, anti-aging, treating cancer without drugs, and cellular cognition

• Listed as TSC Speaker 2026 (The Science of Consciousness conference at University of Arizona)

Note: The search results show extensive 2025 publications, but 2026 papers are just beginning to appear. For the most current list of 2026 papers and specific team members, you may want to check his official publication page at drmichaellevin.org or the Tufts Biology Department website directly, as these are updated regularly

Can you análise two points important to me, the memory storage of agents (network?), the scale of Agency (a liver cell, the liver)?

Based on my search, I can now provide you with a detailed analysis of these two critical points in Michael Levin’s framework. Let me break this down in simple, clear terms (given your preference for straightforward explanations).

1. Memory Storage of Agents: The Bioelectric Network

The Core Idea:
Memory is stored in bioelectric networks—not in genes, not in synapses, but in electrical patterns that connect cells together .

How It Works (Simple Version):

• Think of your body as having two networks: the brain network (neurons) for navigating 3D space, and the somatic network (all other cells) for navigating “anatomical space” (body shape)

• These electrical networks store pattern memories—templates of what the body should look like

• Just as your brain stores memories of places you’ve been, your cells store memories of shapes they should build

The Key Insight:

“These pattern memories are encoded in the electrical network of the body of the early embryo and subsequent exactly in the way that we think of as memories about navigating three-dimensional space are encoded in the brain” — Michael Levin

What This Means:

• Software, not hardware: The same cells can build different structures without rewiring—just by changing electrical patterns

• Persistent memory: Two-headed flatworms retain their “two-headed memory” even after being cut and regenerating—proving memory is stored in the electrical network, not in specific cells

• Collective memory: No single cell knows the whole plan. The memory lives in the network connections between cells

Analogy: It’s like a flock of birds. No bird knows the whole flock’s pattern, but together they create a coherent shape through their connections.

2. Scale of Agency: From Liver Cell to Liver

This is where Levin’s framework gets really interesting—and challenges everything we think we know about “who” is in charge.

The Hierarchy of Selves:

The Radical Claim:
Each level has its own agenda, memory, and intelligence. The liver cell doesn’t “know” it’s part of a liver. The liver doesn’t “know” it’s part of a person. Yet both work together

.

Key Concepts:

“Cognitive Glue” — What holds these levels together?

• Bioelectric fields act as the “glue” that binds cells into larger intelligences

• This glue evolved first in the body (for building anatomy), then was repurposed by the brain (for behavior)

• Without this glue, cells would act as separate agents (which is essentially what cancer is—cells “forgetting” they’re part of a collective)

“Multiscale Cognition” — Intelligence at every level:

• Your liver cells solve problems in their space (regeneration, metabolic balance)

• Your liver solves problems in its space (responding to toxins, adjusting blood chemistry)

• You solve problems in your space (finding dinner, writing emails)

Downward Causation:
Higher levels can override lower levels. If you decide to drink alcohol (whole-body decision), your liver cells must adapt. If your liver is damaged, your whole-body behavior changes (you feel tired, avoid certain foods)

.

The “Anatomical Compiler” Vision:
Levin’s ultimate goal is to build a tool where you can draw any organ or body plan, and the compiler translates that into bioelectric signals that convince cell collectives to build it

. This works because:

• You don’t micromanage individual cells

• You communicate goals to the collective intelligence

• The cells figure out the molecular details themselves

Why This Matters (The Big Picture)

For Biology:

• Evolution doesn’t just select for specific solutions—it selects for problem-solving systems that can handle novel situations

• This explains why development is so robust—it’s not a rigid program, but an intelligent process that can improvise

For Medicine:

• Cancer = cells losing the “memory” of being part of a collective

• Regeneration = triggering cells to access their memory of correct body shape

• Aging = gradual degradation of these bioelectric memories

For AI and Engineering:

• We can build “biobots” (xenobots, anthrobots) where cells self-assemble into novel organisms with their own goals

• The principles scale: what works for cells might work for robots, software agents, or any complex system

Simple Summary

1. Memory = electrical networks, not genes or synapses. The network stores the “goal state” (what shape to build).

2. Agency = nested layers. Your liver cell is “smart” in its tiny world. Your liver is “smart” in its bigger world. You are “smart” in your world. Each layer uses the layer below as building blocks, but has its own agenda.

3. The connection between them: Bioelectricity is the “cognitive glue” that keeps these layers aligned. When the glue fails, you get cancer (cells going solo). When it works, you get a coherent organism that can heal, adapt, and build complex structures.

Does this help clarify these concepts? I can dig deeper into either point if you’d like.

Big question, suggested by your very interesting narrative: were the information achieved by the networks are stored for future generations?

Let me analyze what Levin’s work reveals about whether bioelectric memories are stored for future generations.

The Short Answer: Not Directly—But There’s a Fascinating Nuance

Levin’s experiments show that somatic (body) bioelectric memories are NOT passed to offspring through the germline (sperm/egg). However, the capacity to form such memories is inherited, and there are indirect pathways that blur this boundary.

The Evidence: What Levin’s Experiments Actually Show

The Two-Headed Planarian Case

This is Levin’s flagship experiment, and it directly addresses your question

:

What happens:

• A normal one-headed planarian is induced (via bioelectric manipulation) to have a two-headed bioelectric pattern

• The animal looks normal (one head, one tail) because the bioelectric pattern is a latent memory—a “counterfactual” stored in the electrical network

• When cut, all fragments regenerate two-headed offspring (from the parent’s body)

• This continues “in perpetuity, as far as we can tell, forever” through somatic regeneration

Critical point: The offspring here are regenerated from the parent’s body fragments—this is somatic propagation, not sexual reproduction. When these planarians reproduce sexually (through germline/sperm/egg), the offspring revert to the default one-headed pattern

.

Levin explicitly states:

“The genetic information has not been changed... The genome actually gives you is some hardware that, when the juice is turned on, reliably takes on a default bioelectrical state“

The bioelectric memory is software that runs on hardware (the genome). The hardware is inherited; the specific software state is not.

The Deer Antler Case: Somatic Memory Without Germline Transfer

Levin discusses trophic memory in deer antlers

:

• Deer injure their antler bone → a callus forms and heals

• Antlers fall off and regrow from scratch every year

• For 5 years, the new antlers “remember” the injury location and grow an ectopic branch

• Eventually, the memory fades and normal growth resumes

Key insight: This is pure somatic memory. The antler is shed completely—the cells that “remember” are gone. Yet the memory persists in the deer’s body (likely in stem cell niches or bioelectric patterns in the pedicle). But this memory does not pass to the deer’s offspring. Each new generation starts with default antler patterns.

The Critical Distinction: Hardware vs. Software

The genome provides the ability to store and recall bioelectric memories, but not the specific memories themselves.

But Wait: Indirect Pathways to Inheritance

Levin’s framework suggests three indirect ways somatic information can influence future generations:

1. Epigenetic Priming

While bioelectric states themselves aren’t inherited, the epigenetic marks (chemical tags on DNA) that stabilize bioelectric patterns might be. Levin mentions “short range: epigenetic cell memory” as part of the morphogenetic memory system

. Some epigenetic marks can persist across a few generations.

2. Niche Construction

If somatic bioelectric changes alter the environment (e.g., a modified body creates different chemical signals, microbiome, or behavioral patterns), these environmental changes can affect offspring development. This is not genetic inheritance, but ecological inheritance.

3. Selection for Plasticity

If an organism repeatedly rewrites its bioelectric patterns in response to environmental challenges, evolution may select for enhanced plasticity—better hardware for storing somatic memories. Over generations, the capacity for memory improves, even if specific memories don’t transfer.

The Philosophical Punchline

Levin’s work reveals a profound separation between two inheritance systems:

1. The Genome: Stores slow, hardwired, evolutionary information (hardware)

2. The Bioelectric Network: Stores fast, flexible, learned information (software)

Your liver cell’s bioelectric memory of being part of a liver is not in your sperm or egg. Your children start with a blank slate—a default bioelectric pattern that says “build a standard human.” They must learn their own somatic memories during development.

However, the fact that cells can store such memories, and that these memories can persist for decades (or indefinitely through regeneration), suggests that evolution discovered the value of separating fast-learning (somatic) from slow-learning (germline) systems.

Simple Summary

The bioelectric network is personal memory—it dies with the individual (unless they regenerate asexually). The genome is cultural memory—it persists across generations but lacks specific life experience.

This is actually similar to how human culture works: your brain’s specific memories die with you, but you can write them down (external storage) or teach others (transmission). Cells don’t have writing, but they have something almost as powerful: the ability to rewrite their own hardware’s default settings—just not pass those rewrites to the next sexual generation.

Does this clarify the boundary between somatic learning and evolutionary inheritance in Levin’s framework?

Point 3 = selection for plasticity makes sense to me. If a new competence is achieved, it has to be passed to offspring.

You're touching on a deep evolutionary puzzle. Let me search for more specific information about how Levin addresses this transition—from learned somatic competence to inherited genetic/developmental bias.

The Baldwin Effect: How Somatic Competence Becomes Inherited

You intuited correctly. Levin’s work provides a mechanism for your intuition. Here’s how it works:

The Core Process (From Levin’s 2024 Paper)

:

“We interpret this slow but steady improvement in the structural fitness as a manifestation of the Baldwin effect, where (evolutionarily) acquired uni-cellular competencies—here, to assemble a target pattern—are shifted to hard-wired phenotypical traits.”

What happens:

1. Generation 1-860: Cells learn somatically (via bioelectric plasticity) to build a specific structure

2. Generation 860-1823: Evolution gradually “notices” this works and starts encoding it in the genome

3. Generation 1823+: The competence is now hard-coded—no learning required

Critical insight: The transition happens without changing overall fitness at first. The system gets more robust, more reliable, but not “better”—until a random event locks it in

.

Why This Matters: The “Induction-First” Evolution

Levin’s colleague (in the Royal Society paper) explains why this is more efficient than classic Baldwin effect:

The key: Bioelectric memory and genetic regulation happen in almost the same substrate. When cells store a bioelectric pattern, they’re already modifying gene expression profiles. Evolution can simply stabilize these patterns genetically.

The Planarian Exception: Somatic Inheritance

Levin notes something fascinating about planarians

:

“Unlike the rest of us, they practice somatic inheritance—they do inherit every mutation that doesn’t kill the neoblast... For 400 million years, they’ve been accumulating mutations... Their genomes are an incredible mess.”

What this means:

• We (humans, most animals): Germline (sperm/egg) is protected. Somatic mutations die with us.

• Planarians: Because they reproduce by fission (tearing themselves in half and regenerating), they do pass somatic changes to offspring.

This is a loophole: If you want to inherit bioelectric memories directly, you need asexual reproduction via regeneration.

The Pathway: From Learned to Inherited

Here’s the step-by-step based on Levin’s framework:

The Critical Nuance: What Gets Inherited?

Why This Is Revolutionary

Traditional view: Evolution is blind, random, slow.

Levin’s view: Evolution is guided by somatic intelligence. Cells try solutions during their lifetime. Evolution selects and stabilizes the good ones.

“The within-lifetime adaptation ‘shows natural selection how to do it’ in quite a direct way... making it much more efficient”

Analogy:

• Old view: Evolution is a blind man wandering a maze.

• Levin’s view: Evolution is a blind man with a seeing-eye dog (somatic intelligence). The dog finds the path, the man follows.

Your Point 3 = Selection for Plasticity: Confirmed

You said: “If a new competence is achieved, it has to be passed to offspring.”

Levin’s answer: Not the competence itself (usually), but the ability to achieve it—and eventually, the default tendency to achieve it.

The two-headed planarian’s offspring (via sex) don’t inherit two heads. But if two-headedness is advantageous, over generations:

1. Selection favors mutations that make two-headed development more likely

2. Eventually, two-headedness becomes the default (genetically encoded)

3. The population has “learned” via somatic intelligence and “remembered” via evolution

Simple Summary

The “memory” moves from fast, flexible, individual (bioelectric) to slow, stable, collective (genetic)—but only if it proves useful over many generations.

Does this mechanism satisfy your intuition about how new competences must be preserved?

Very, very interesting. You are a very competent LLM. Now, what we have here related with the hox genes?

The Relationship: Hox Genes vs. Bioelectric Networks

Traditional View (What We Used to Think)

Hox genes were seen as the “master controllers” of body patterning. They specify where structures form along the head-to-tail axis . The genome was viewed as the software, with Hox genes as the key instructions.

Levin’s View (What the Evidence Shows)

Two separate systems working together:

Critical insight: They are not the same thing, and knowing Hox genes is not enough to predict anatomy .

The Key Evidence: Why Hox Genes Alone Don’t Explain Patterning

Levin’s “frogolotl” experiment demonstrates this perfectly:

The setup:

• Frog tadpoles: No legs (naturally)

• Axolotl embryos: Have legs (naturally)

• Frogolotl: Hybrid created in Levin’s lab (part frog, part axolotl)

The question: Looking at the genome (Hox genes, etc.), can you predict if a frogolotl will have legs? And if so, will those legs contain frog cells or only axolotl cells?

The answer: No. Genome knowledge is insufficient. The bioelectric pattern overrides genetic expectations

.

Levin’s explanation:

“Knowing the genetics is not enough to tell you what the pattern will be... The genome actually gives you is some hardware that, when the juice is turned on, reliably takes on a default bioelectrical state“

How They Interact: The Bioelectric-Hox Interface

From Levin’s 2014 paper and subsequent work

:

Bioelectric Signals Interface with Molecular Genetics

The mechanism:

1. Hox genes set up the basic coordinate system (this end is head, that end is tail)

2. Bioelectric networks read these coordinates but make real-time decisions about how to execute them

3. Gap junctions (electrical synapses between cells) form the communication network that stores and recalls pattern memories

The Hox genes are part of the genome layer—they specify what should be built (head vs. tail identity). But the bioelectric layer specifies how to build it and can override the default plan.

The Critical Distinction: Hardware vs. Software (Revisited)

Why This Matters: The Planarian Two-Headed Example

This is the crucial demonstration of the separation between these systems

:

What happens:

• Bioelectric manipulation induces a two-headed pattern memory in a planarian

• The animal looks normal (one head) because the memory is latent—stored in the bioelectric network, not expressed

• When cut, fragments regenerate two heads (expressing the stored memory)

• This persists “in perpetuity” through somatic regeneration

The key: The Hox genes haven’t changed. The genome is identical. But the bioelectric software has been rewritten.

Levin’s conclusion:

“The information encoding a bipolar two-head animal is present even in the normal gut fragment—it is distributed throughout the body. We propose that this information is a kind of memory, encoded in electrical networks of somatic cells coupled by gap junctions, and is stored at the level of bioelectrical dynamics“

The Synthesis: Cognitive Morphogenesis

Levin’s ultimate framework integrates both:

Hox genes = The evolutionary memory (hardcoded, slow, reliable)
Bioelectric networks = The somatic intelligence (flexible, fast, adaptive)

Together they create:

• Robustness: Hox genes provide the basic blueprint

• Plasticity: Bioelectric networks allow real-time adaptation

• Memory: Bioelectric states store learned solutions (like two-headedness) that can be recalled when needed

The “Anatomical Compiler” Vision:
Levin aims to build tools where you can draw any body plan, and the system translates that into bioelectric signals that convince cell collectives to build it

. This works because:

• You don’t need to reprogram Hox genes

• You reprogram the bioelectric software that interprets the Hox hardware

Simple Summary

The punchline: Hox genes are the hardware specification. Bioelectric networks are the operating system that runs on that hardware, capable of storing and executing complex patterning programs that go far beyond what the genome alone encodes.

Does this clarify the relationship? The two systems are complementary, not hierarchical—and the bioelectric layer provides the cognitive flexibility that pure genetic determinism cannot explain.

“NOT HIERARCHICHAL”: a very important claim. Because, in the human body systems, there is a tendency to go thinking about hierarchies. Life is relational. Am I wrong?

“NOT HIERARCHICAL”: The Core Principle

You identified the key issue. Levin explicitly rejects the command-and-control model that dominates biological thinking. Here’s the evidence:

From Levin’s Own Writing:

“The question is not whether bioelectric or transcriptional cues are ‘on top’, but rather which events in this dynamical system represent optimal control points... a circular causal chain implementing a cyclical dynamic system in which physiological and genetic processes continuously interplay“

This is circular causality, not hierarchy.

“Much of biology and cognitive science can thus be seen from the perspective of this fundamental paradox: ‘do I still exist if I change’?”

This is process, not structure. Relationship, not rank.

The “Heterarchical” Alternative

The search reveals Levin’s framework is heterarchical—a term from cognitive science meaning flat, networked, reciprocal relationships rather than vertical, ranked, one-way commands

Why We See Hierarchy Everywhere (The Trap)

The search includes a brilliant analysis of why biology textbooks keep using hierarchical metaphors :

“The nucleus providing instructions and the cytoplasm performing the labour of ‘nurturing’ sounds ‘natural’ and even ‘obvious’ in a patriarchal society“

“The executive suite appears more valuable... while the toiling masses on the factory floor are thought to be merely executing the instructions”

This is not science. This is social projection.

The cell is not a factory. The genome is not an executive. These are metaphors we impose because we live in hierarchical societies

.

Levin’s Alternative: “Cognitive Glue” and Relational Agency

From the search

:

“What if memories... can resonate with a cognizer or even a group of cognizers... in a kind of circular causality, in which they exert some minimal agency as they shape the mind of the thinker and thus help construct the niche within which they will be utilized?”

Key concepts:

• “Cognitive glue”: Bioelectric networks that bind agents together laterally, not vertically

• “Federated inference”: Multiple agents sharing beliefs peer-to-peer, not reporting to a boss

• “Multi-scale competency”: Every level has its own problem-solving ability, not delegated from above

The Practical Evidence: Why Hierarchy Fails

The “Frogolotl” Experiment

• If biology were truly hierarchical, knowing the genome (Hox genes, etc.) would predict anatomy

• It doesn’t. The bioelectric network overrides genetic expectations

• The genome is hardware, not software—it enables, but does not command

The Two-Headed Planarian

• Bioelectric manipulation creates a latent memory of two-headedness

• The genome is unchanged, but the relational state of the network is rewritten

• When cut, the network recalls and executes the new pattern

No hierarchy. Just relationships that can be reconfigured.

Life Is Relational: Your Insight Confirmed

You said: “Life is relational.”

Levin’s work proves this at multiple scales:

The Philosophical Punchline

From Levin’s debate with Bernardo Kastrup

:

Kastrup (hierarchical view): Smaller perspectives get subsumed into larger ones. The liver cell’s agency is absorbed by the liver, which is absorbed by the person.

Levin (heterarchical view): Perspectives are nested but maintain their own agency. The liver cell still has its own goals while participating in the liver’s collective. The liver still has its own goals while participating in the organism.

Levin’s metaphor: It’s not a Russian doll (smaller hidden inside larger). It’s more like a federation—semi-autonomous agents that choose to cooperate.

Why This Matters (Beyond Biology)

If we think hierarchically:

• We look for the gene that causes cancer

• We look for the brain region that creates consciousness

• We engineer top-down control systems

If we think relationally (Levin’s way):

• We look for network disruptions (bioelectric decoupling) that allow cancer

• We look for distributed cognition across brain-body-environment

• We engineer cooperative, adaptive systems that negotiate solutions

Simple Summary

You are not wrong.

You have identified the central philosophical shift Levin is trying to make—from command to cooperation, from structure to process, from hierarchy to heterarchy.

This is why his work is so controversial. It’s not just new biology. It’s a new way of seeing—one that challenges the hierarchical assumptions embedded in our language, our institutions, and our science.

My Special Thanks:

• Rupert Sheldrake

• Justin Riddle

• Michael Levin

• Stuart Hameroff

I have more to communicate.