Michael Levin’s “Self-Improvising Memory”: Why the Agential Engram Hypothesis Is Not Established Neuroscience

**Editorial opinion** ## What is known, what is not, and the risks For generations, academic neuroscience has anchored its model of memory in the synaptic trace: when an organism acquires new information, specific ensembles of neurons in the hippocampus and cerebral cortex physically strengthen their synaptic connections, encoding engrams. An archived post from the bhzine channel (#382) reviewed a provocative paper by renowned developmental biologist Michael Levin of Tufts University titled “Self-Improvising Memory.” The post suggested a radical alternative: memory is not a passive warehouse of static traces, but an active, improvisational process driven by agential biological subsystems capable of storing information in subcellular structures across the entire body. The post also cited insect metamorphosis, planarian regeneration, and anecdotal claims of memory transfer via organ transplantation. What does empirical science actually say about these concepts, and where does rigorous biology diverge from speculative philosophy? - **What is known:** In 2024, Michael Levin published a conceptual perspective essay in the peer-reviewed journal *Entropy* entitled “Self-Improvising Memory: A Perspective on Memories as Agential, Dynamically Reinterpreting Cognitive Glue” [S-001]. Levin presents a theoretical framework proposing that biological memory preserves adaptive salience and meaning rather than low-level fidelity for an ever-transforming self. In 2008, Douglas Blackiston and colleagues demonstrated in tobacco hornworm moths (*Manduca sexta*) that an associative aversive odor-conditioning response acquired during the late larval stage can persist through metamorphosis into adulthood, mediated by the survival of specific mushroom-body neurons [S-002]. Furthermore, Shomrat and Levin (2013) demonstrated in planarians (*Dugesia japonica*) that decapitated flatworms, after 14 days of complete head regeneration, exhibit accelerated retraining to a familiar illuminated environment compared to naive controls [S-003]. - **What is not known:** Whether non-neural subcellular architectures (cytoskeleton, organelle networks, somatic bioelectric gradients) can autonomously store autobiographical or declarative memory in mammals; the exact biophysical mechanisms of non-synaptic engram readout; and whether invertebrate habituation models scale to human cognition. - **Risks and limitations:** Levin’s paper is explicitly framed as “A Perspective” (a theoretical essay), not clinical human neuroscience. Extrapolating morphogenetic paradigms into human medical protocols or claiming that “neural networks do not store memory” contradicts mainstream neurobiology. Claims regarding “donor personality and memory transfer via heart transplantation” are scientifically unfounded: extensive medical research explains these anecdotal reports through the neurotoxicity of immunosuppressive medications (glucocorticoids, calcineurin inhibitors) and acute psychological stress [S-004]. The concept of the “senome” remains a theoretical construct of biological pan-cognitivism [S-005]. ## The hook What if our brains do not actually store memories like hard drives, but constantly reconstruct them anew on the fly from somatic clues scattered across our tissues? While the concept sounds like pure science fiction, distinguishing frontier developmental biology from popular pseudoscience is crucial for any grounded biohacker. ## In brief - **Levin’s core premise:** Memory in living systems is not a high-fidelity filing cabinet, but an agential process prioritizing salience and meaning across somatic transitions [S-001]. - **Moths and caterpillars:** Adult moths can indeed retain simple associative odor aversions learned by late-stage caterpillars, but this occurs via surviving mushroom-body neurons, not because the brain completely dissolved [S-002]. - **Planarian regeneration:** Decapitated flatworms with regenerated heads relearn familiar environmental contexts faster, though the exact somatic substrate remains scientifically debated [S-003]. - **The organ transplant myth:** Clinical science has found zero empirical evidence for “cellular memory transfer” in heart transplant recipients; reported behavioral shifts stem from immunosuppressants and psychological confabulation [S-004]. - **The “senome” abstraction:** Viewing every cell as an autonomous cognitive agent is an intriguing philosophical paradigm, but not an established neuroscientific consensus [S-005]. ## Michael Levin’s radical thesis: memory as dynamic cognitive glue Dr. Michael Levin’s pioneering research on bioelectric signaling in morphogenesis and xenobots has fundamentally challenged how science views cellular cooperation. In his 2024 *Entropy* essay [S-001], Levin tackled the philosophical persistence paradox: how does an organism maintain a stable sense of identity across time when its cellular components, synapses, and molecules are continuously remodeled? Levin proposes several paradigm-shifting concepts: 1. **Agential multi-scale perspectives:** Every tier of biological organization—from molecular networks to somatic tissues and neural circuits—acts as an agential system navigating its own problem space. 2. **Salience over fidelity:** Evolution does not require biological memory to record an exact, pixel-perfect copy of the past. Instead, it optimizes for adaptive salience—extracting the actionable gist of past experience to guide future survival in radically altered embodiments. 3. **Neural networks as interpreters:** Levin speculates that central neural networks might function less as permanent storage drives and more as decoding engines that learn to interpret somatic engrams embodied across subcellular components. While this perspective offers a brilliant conceptual framework for synthetic biology and robotics, it is vital to recognize its epistemic status: it is a theoretical hypothesis, not clinical neuroscience. ## Can a butterfly remember being a caterpillar? Unpacking the hornworm experiments A popular scientific legend often claims: “A caterpillar dissolves into total cellular soup inside the chrysalis, yet the emergent butterfly remembers its entire life.” What did controlled empirical experiments actually reveal? In a benchmark study by Douglas Blackiston and colleagues (Blackiston et al., 2008, *PLoS ONE*), researchers tested associative learning in tobacco hornworm moths (*Manduca sexta*) [S-002]. Larvae were conditioned to avoid ethyl acetate odor paired with mild electric shocks. The experimental findings were precise: - Caterpillars trained during early instars (3rd instar) completely lost the conditioned aversion after metamorphosis. - Memory retention into adulthood was observed **exclusively in caterpillars trained during the late 5th instar**—just prior to pupation. - Neuroanatomical tracing confirmed that the larval nervous system does not fully liquefy: key neuroblasts and structures within the **mushroom bodies** (the primary associative olfactory centers in insects) persist through metamorphosis, migrating directly into the adult brain. Thus, memory persistence across metamorphosis relies on the physical survival of specialized neural circuits rather than mystical brainless memory. ## Planarian head regeneration: separating 2013 rigor from 1960s folklore The second prominent example involves freshwater planarians, flatworms capable of regenerating a complete organism from tiny amputated fragments. Here, science must strictly delineate modern reproducible data from historical artifacts: - **The 1960s McConnell controversy:** In the 1960s, James V. McConnell claimed that conditioned planarians could transfer memory to naive worms through cannibalism via “memory RNA.” Rigorous follow-up studies in independent laboratories failed to replicate these findings, revealing them as artifacts of light-induced sensitization and uncalibrated handling. - **The Shomrat & Levin automated paradigm (2013):** Shomrat and Levin revisited the question with computer-automated laser tracking to eliminate experimenter bias [S-003]. Planarians (*Dugesia japonica*) were habituated to explore an illuminated rough-textured petri dish. After decapitation and 14 days of complete head regeneration, worms displayed accelerated relearning compared to naive controls after a brief reminder session. Levin hypothesizes that bioelectric gradients and neoblasts preserve somatically encoded experience. However, until the precise biophysical substrate is characterized, extrapolating flatworm habituation to human declarative memory remains an unsubstantiated leap. ## The myth of “cellular memory” in organ transplant recipients The most problematic claim in the archived post was the suggestion that organ transplant recipients acquire donor memories, preferences, and emotions. Evidence-based medicine and transplantation science have repeatedly debunked this notion [S-004]: 1. **Absence of neurobiological circuitry:** Peripheral organs such as hearts and kidneys contain autonomic nerve plexuses that regulate hemodynamic tone and contractility. They entirely lack the pyramidal cell architecture, hippocampal indexing, and dendritic plasticity required to encode declarative memories or personality traits. 2. **Immunosuppressant neurotoxicity:** Post-transplant patients depend on lifelong regimens of immunosuppressants, including high-dose corticosteroids (prednisone) and calcineurin inhibitors (tacrolimus, cyclosporine). These agents cross the blood-brain barrier and frequently cause mood swings, vivid dreams, altered gustatory perception, and neurocognitive disturbances. 3. **Psychological confabulation:** Experiencing end-stage organ failure and surviving life-saving surgery induces profound existential stress. When recipients learn biographical details about deceased donors, the human mind instinctively weaves confabulatory narratives, attributing normal post-operative emotional changes to the donor’s “spirit.” Controlled double-blind studies confirm that when patients have no information about their donors, zero specific memory transfer occurs. ## The “senome” hypothesis: biological pan-cognitivism vs. academic consensus Archived post #382 also referenced the “senome” as an ultra-high-dimensional biological reservoir for distributed memory, referencing the work of Baluška, Miller, and Reber (2023) [S-005]. The “senome” (derived from sentience and sensation) framework posits that basic cognition is universal across all living matter (the Cellular Basis of Consciousness). In this view, every somatic cell possesses sensory agency, storing experiential history via cytoskeletal conformational changes and epigenetic modifications. However, mainstream neurobiology cautions against semantic dilution. In physics and biochemistry, a system’s tendency to retain structural changes from previous stimuli is called **hysteresis** or molecular plasticity. Labeling all cellular hysteresis as “memory” and all receptor transduction as “cognition” provides a poetic metaphor, but conflates molecular signaling with the complex network dynamics of hippocampal-cortical memory consolidation. ## Author’s personal experience | Practicing biohacker > **Personal experience with cognitive tracking and mental models of memory** > In my years of biohacking and cognitive performance optimization, evaluating theoretical paradigms against rigorous personal tracking has always served as an essential filter against wishful thinking [personal experience]. > - Relying on vague concepts of “somatic cellular memory” to learn complex intellectual material produces zero measurable gains: without systematic synaptic consolidation (spaced repetition, active recall, deep restorative slow-wave sleep), no bodily tuning compensates for deficiencies in cortical network encoding [personal experience]. > - Nonetheless, Michael Levin’s metaphor of adaptive memory reinterpretation offers genuine psychological value: understanding that the brain does not store an unchangeable video recording of trauma or past errors, but rather dynamically reframes memories to serve the present self, is a powerful tool for cognitive reframing and mental resilience [working hypothesis]. > - Objective biomarkers (heart rate variability, deep sleep percentages on EEG/wearables) show direct correlation with retention rates, confirming that while the brain is integrated within a holistic body, the hippocampus and neocortex remain the true physical organs of memory [from practice]. ## What requires clarification in the archived post Archived post #382 introduced a fascinating topic, but presented speculative theories in a manner that general readers could easily confuse with established clinical facts: 1. **The absolute claim that “neural networks do not store memory”:** Contemporary optogenetics (e.g., Susumu Tonegawa’s engram research) has conclusively demonstrated that direct optical stimulation of specific hippocampal and cortical engram neurons reactivates complex, specific episodic memories. Levin’s proposal is a theoretical alternative regarding functional allocation, not an empirical refutation of synaptic memory. 2. **Overgeneralization of insect metamorphosis:** Blackiston (2008) documented the persistence of one elemental olfactory aversion conditioned immediately prior to pupation, mediated by surviving mushroom-body cells. It does not prove that butterflies remember the broad life experience of caterpillars. 3. **Uncritical inclusion of organ transplant memory:** Citing organ transplant memory stories without clarifying their debunked, anecdotal status misleads readers regarding human biology. 4. **Presentation of the “senome”:** The senome was described as an established neuroscientific discovery, whereas it is actually a speculative theoretical construct within the niche field of biological pan-cognitivism. ## What to do next How to navigate Michael Levin’s ideas and cutting-edge memory science in daily practice: 1. **Distinguish frontier philosophy from applied biology:** Read Levin’s work as brilliant frontier essays that broaden our conceptual horizons, but do not mistake speculative developmental biology for verified medical protocols. 2. **Rely on evidence-based consolidation levers:** Optimize long-term memory using proven neurobiological mechanisms: prioritize 7–9 hours of sleep with adequate slow-wave architecture, engage in regular aerobic exercise to boost BDNF, and practice active recall and spaced retrieval. 3. **Maintain a healthy skeptical filter:** Critically dismiss esoteric claims regarding somatic cellular memory or organ personality transfer, regardless of how compelling the storytelling may appear. ## Question for readers Do you perceive your memory as an immutable recording of the past, or as a living sketchpad that your mind continuously edits and updates depending on your current life context? Have you noticed your childhood memories evolving over time? ## What was verified and what was meant | Claim | Status | Source | |---|---|---| | Michael Levin’s hypothesis frames memory as an agential tool preserving adaptive salience rather than exact data fidelity | theoretical hypothesis / conceptual perspective in a peer-reviewed journal (not established clinical neuroscience) | S-001 | | Tobacco hornworm moths retain associative olfactory aversions learned as caterpillars via surviving mushroom-body neurons | verified by laboratory experiment in Manduca sexta (simple conditioned reflex, not personality transfer) | S-002 | | Decapitated planarians with regenerated heads exhibit accelerated relearning to a familiar illuminated environment | verified by automated behavioral testing in Dugesia japonica (somatic storage mechanism remains debated) | S-003 | | Claims of memory and personality transfer to recipients following heart or organ transplantation (“cellular memory”) | debunked by evidence-based medicine (anecdotal case reports; driven by immunosuppressive neurotoxicity and psychology) | S-004 | | The “senome” concept as a universal subcellular reservoir for distributed biological memory | speculative theoretical framework of biological pan-cognitivism (not mainstream neuroscientific consensus) | S-005 | Author: Serge Luboff. Reviewer: unassigned. Medical review: not conducted. This material is purely educational, does not constitute medical advice, and is not a substitute for professional healthcare consultation.