Contribution
It puts three animal systems side by side as cases of memory surviving the remodelling of the brain, and asks what they imply for where memory is kept. It argues the standard answer, synaptic connectivity, is incomplete. It proposes that other cellular mechanisms, possibly outside the nervous system, back memory up, and it lays out a research programme for the authors’ lab.
Key insight
If a memory outlasts the cells and connections that formed it, the memory was not only in them. The authors’ version: “there is a ‘blueprint’ of the memories that may be kept safely in place by a mechanism other than the synaptic connections themselves” (What mechanisms underlie memory?). For planaria they go further: “it is possible that memory can be stored outside the brain and imprinted on the nascent brain during regeneration” (Memory in aneural systems).
Assumptions
- The behavioural reports measure memory. Each case is taken largely at face value. The planarian line’s history is told as controversy, but the claims are kept.
- Memory of shape and memory of experience may share machinery. The programme rests on the lab’s view that bioelectric networks among non-neural cells store pattern information during regeneration, and might store learned information too.
- Survival through remodelling is a test case for the engram. If specific memories stay as cells are replaced and moved, that constrains how mental content is physically represented (Introduction).
Key results
What the perspective reports, by section. None of it is new data.
- Why it matters (Introduction). Three stakes: the engram question; stem-cell therapies for the brain, which must not distort “the decades of memory content and personality of the patient”; and engineering “biobots” and computing media robust to cell turnover.
- Insects (Memories survive CNS rearrangements and remodeling in
insects).
- Hopkins’ host-selection principle and its rival, the chemical legacy hypothesis. Chemical legacy is real: contaminated Drosophila pupae and parasitic wasps removed from their aphid hosts show it.
- Controlled cases: a weevil and a wasp keep larval preferences and are unmoved by contaminating the pupal case. Most convincing, in the authors’ words, are odour-shock aversions in Drosophila (Tully et al. 1994) and Manduca (Blackiston et al. 2008), unaffected by odour or shock alone.
- Mushroom bodies remodel: γ neurons prune completely, and “the α’/β’ neurons demonstrating partial pruning before establishing adult specific projections”. A synaptic memory in some of these neurons has “only indirect evidence”.
- The best support: Tenebrio larvae trained in mazes learn faster as adults, and RNA enriched in larval mushroom bodies persists into the adult (Punzo and Malatesta 1988).
- Untested alternatives: “Non-neural memory, chemoreceptor tuning, and gene silencing”. The question “remains unsolved”.
- Planaria (Memories persist during brain regeneration in planaria).
- McConnell’s worms, conditioned with light and shock and cut in two: head fragments kept the training, and tail fragments, which regrew a brain, needed fewer trials to relearn (“savings”).
- Naive worms fed on trained worms, or injected with their RNA, showed savings; RNase abolished it. A Russian group (Cherkashin and Sheiman) found retention in both halves 14 days after cutting.
- Controls for pseudoconditioning and sensitization were run, and the field was then “mired in controversy” and abandoned.
- The authors’ device gives “24*7” individual training with objective scoring. With it, “our initial study confirmed the ability of worms to recognize a surface etch pattern in a place learning task, and the persistence of this information across head regeneration” (Shomrat and Levin 2013).
- Candidate stores: neoblasts (stem cells) altered epigenetically by learning, or non-coding RNAs.
- Mammals (Memory and brain repair in mammals).
- Hibernating ground squirrels: body temperature near 0 °C; dendrites pruned, hippocampus included; branching restored within hours of arousal.
- Kin recognition after nine months turned out to be self-referent phenotype matching, not memory. Housemates were discriminated after hibernation, while maze and operant recall was weak (European ground squirrels).
- “Perhaps most convincingly”, two operant tasks were “unimpaired following 6 months of hibernation”. The cited study is of Alpine marmots (Clemens et al. 2009).
- Tau phosphorylation rises in torpor and falls on arousal, a candidate for stabilizing connections.
- Against synapses alone (What mechanisms underlie memory?). LTP, LTD and STDP are “inherently unstable”; reconsolidation and the reported need for continual PKMζ activity argue against a hard-wired store; and memory traces have been reported after erasing the related synaptic connectivity (Chen et al. 2014, Aplysia).
- Memory without a brain (Memory in aneural systems). Frog eggs injected with orange or citral give animals preferring food with those substances (Hepper and Waldman 1992). Embryos exposed to an odour with an alarm chemical give larvae that flee the odour alone (Mathis et al. 2008). Neural-like computation and memory are reported in sperm, amoebae, yeast, plants, the immune system, bone and heart.
- The bioelectric hypothesis (Why the Mind is in the Brain (or is it?)). Ion channels, neurotransmitters and electrical synapses work throughout the body and drive patterning. The authors propose that somatic and cognitive memory are coupled through bioelectric signalling, and that this can now be tested.
- Programme (Outlook). Five lines: modelling neural networks under topological change; wider surveys of memory through remodelling; molecular study of how neural networks exchange information with other tissues; testing whether regeneration specifically preserves encoded information; and new assays, such as regional brain transplants and learning in non-neural cell networks.
Claims
| id | claim | strength | support |
|---|---|---|---|
| C1 | Memories can survive drastic rearrangement and rebuilding of the CNS in insects, planaria and mammals | moderate as a survey: the authors call the data “convincing”; each system rests on a few studies, and the planarian line has a contested history | Outlook, and the three systems’ sections |
| C2 | How memory survives metamorphosis is unknown; persistence in mushroom-body neurons has only indirect support | strong as a statement of the evidence’s state | insects section |
| C3 | Synaptic plasticity alone does not explain lifelong memory; some other mechanism keeps a “blueprint” | interpretive: argued from instability, reconsolidation, PKMζ and one Aplysia study | What mechanisms underlie memory? |
| C4 | In planaria, memory may be stored outside the brain and imprinted on the regenerating brain | hypothesis: stated as possible | Memory in aneural systems; planaria section |
| C5 | Somatic (bioelectric) and cognitive memory systems are coupled | hypothesis, proposed as testable | Why the Mind is in the Brain |
Concepts
- engram: the physical trace of a memory.
- savings: faster relearning by a trained animal, or one given material from trained animals, than by a naive one.
- chemical legacy: an adult’s response shaped by larval-environment odours met at eclosion.
- synaptic plasticity hypothesis: memory is stored as changes in synaptic connections and the resulting network activity.
- blueprint: the authors’ word for a memory store kept by a mechanism other than synaptic connections.
- neoblasts: planarian adult stem cells, the source of the regenerated brain.
- bioelectric signalling: communication among non-neural cells through resting potentials and gap junctions, which the lab holds directs patterning.
Connections
- Blackiston, Silva Casey and Weiss 2008 (LIT-1680). Restated in one sentence with Tully et al. Neither the third-instar result nor the pupal washing is mentioned. The 2008 paper had α′/β′ neurons keeping intact projections; this perspective has them partly pruned, citing the same two Drosophila papers.
- Shomrat and Levin 2013 (LIT-1732), unread. Restated in one sentence: place learning of a surface etch pattern, kept across head regeneration. Its reference entry reads “planarians” and “J Exp Biol 2013; 216: 3799–810”; Crossref has “planaria” and no volume or pages.
- Alloway (LIT-1781) is cited, as “1971”, among the Tenebrio maze studies.
- Dias and Ressler (NOTE-1281) is cited as modern work on the transgenerational transfer of “environmentally-imprinted information”, in the passage on McConnell’s RNA results, which bears on QUESTION-297.
Bearing on the record
- QUESTION-302, what carries it across. The perspective lists the
candidates the question asks about and settles none. These are
inferences from its survey.
- For insects it calls the mechanism “unsolved” and the mushroom-body account indirect. It names non-neural memory, chemoreceptor tuning and gene silencing as untested.
- For planaria it offers two non-brain stores: neoblasts and non-coding RNA, and a bioelectric network of non-neural cells. Its own result shows only that something survives; where, it does not say.
- Its account of the mushroom bodies differs from the 2008 paper’s. If α′/β′ neurons are only partly kept, “the mushroom bodies are partly preserved”, as the question has it, is right, but the surviving connection is less secure than the 2008 paper made it.
- QUESTION-302, whether the animal after is the one that learned.
The perspective does not discuss identity. But it frames the medical
stake in terms of it: brain repair must not distort “the decades of
memory content and personality of the patient”. Two inferences.
- Planarian fission is Parfit’s fission case (LIT-469; THEORY-464) made literal. McConnell’s report, as given here, is that both halves kept or regained the training. If that holds, both successors are psychologically connected to the trained worm. THEORY-464’s line that the relation that matters comes apart from identity then has a natural case.
- If memory can be stored outside the brain, the brain is not the sole bearer of psychological continuity. The philosophers’ brain-transplant cases assume it is. This is the authors’ hypothesis, not a finding.
- The owner’s thesis (CLAIM-491). No direct bearing. No dated note is proposed for it.
- A dated note is proposed for QUESTION-302 (section 3).
- No THEORY filed. The work is a programme; its claims are hypotheses.
- No ML instruction. The engineering stakes (biobots, unconventional computing) are not instructions for ML practice.
Limitations
- No new data, and an interested survey. The planarian evidence is the authors’ own result and a 1960s literature they hope to revive. Replication failures in that literature are mentioned, not weighed.
- Its own planarian result is one sentence. It gives no numbers, nor the reminder session the 2013 result needed.
- Some cases are loosely placed. Cowley and Griesel’s transgenerational malnutrition in rats is offered in the memory-transfer story, though it is not memory transfer. The “ground squirrel” result called clearest is, by its reference, in Alpine marmots.
- Citation and wording errors.
- Adipose antioxidant levels in ground squirrels are cited to ref. 26, which is a paper on human glial chimeric mice.
- Drosophila pupae “contaminated with methanol”: the cited study (Barron and Corbet 1999), as the 2008 paper reports it, used menthol.
- Alloway’s paper is listed as 1971; it is 1972.
- “polyphogous”; “neutrally-derived” for neurally derived; “paring” for pairing; “following regenerating?”; “tail fragments without brain cannot be the conditioned”; “rats and mice create, the field”; “(counter McCulloch) than the CNS”; “persistence though brain remodeling”; “for larvae to adult”; “a crucial significant component”; mismatched quotation marks around “Hopkins’ host-selection principle”.
Open questions
- Which tissue in a headless planarian fragment holds the memory? The programme’s first question, unanswered here.
- Do both halves of a cut, trained planarian keep the memory under modern, automated scoring? That is the fission case, and McConnell’s report of it is sixty years old.
- Does ablating the persisting mushroom-body neurons abolish trans-metamorphic memory? The perspective calls the evidence indirect; the 2008 paper proposed the test.