Brain Preservation

Brain preservation asks a narrower question than cryonics, and the narrowness is the whole point. Cryonics claims that a preserved person can eventually be resuscitated. Brain preservation claims only that the structural information specifying a person can be kept from decaying, leaving open — deliberately, and for centuries if necessary — whether anyone will ever be able to do anything with it. The framing under test is: if the connectome survives, the person survives. That is a claim about what a person is made of, and it is stated here as a hypothesis because that is what it is. See Cryonics for the resuscitation claim, which this brief does not make and which uses different chemistry to avoid making it.

Established The subject is unusual because the engineering is far ahead of the theory. A demonstration exists — a whole large-mammal brain preserved with synaptic connectivity intact across the entire organ, verified by an independent prize committee using three-dimensional electron microscopy — and it dates from 2018, resting on a method paper published three years before it and now more than a decade old. Established What does not exist is any agreed answer to whether the preserved structure contains the person, or any imaging capability within six orders of magnitude of reading a human brain out.

Frontier That gap is the spine of this brief. Preservation is the solved half of the problem. Readout is not, and the constraint on readout is industrial rather than scientific: nothing in physics forbids imaging a human brain at electron-microscope resolution, and everything in throughput does. Speculative Underneath both sits a question no experiment has yet been designed to answer in any organism, though the organisms in which it could be answered cheaply have had complete connectomes for years. Frontier A reader who takes away one thing should take away the shape of the asymmetry: the preservation step has a demonstrated procedure with an independently judged result behind it, and the readout step has a demonstrated unit cost that, multiplied out, produces numbers with no precedent in any scientific undertaking ever attempted.

Established The method is a decade old, has a paper, and has two independently judged demonstrations behind it. Aldehyde-stabilised cryopreservation (ASC) was published in Cryobiology in 2015 by Song A — who published that paper as McIntyre RL and now publishes under the first name — with Gregory Fahy as senior author. Established The procedure perfuses glutaraldehyde through the vasculature, cross-linking proteins within minutes and arresting decay outright; a cryoprotectant is then perfused and the brain cooled to a solid, storable state, reported at -135 degC for the prize demonstrations. Established Two properties of that sequence do all the work and both follow from the same choice: fixation is what makes cryoprotectant tolerable at concentrations that would otherwise be lethal, and fixation is what makes the result irreversible for any biological purpose. Established In February 2016 the Brain Preservation Foundation awarded its Small Mammal Brain Preservation Prize to 21st Century Medicine for an entire rabbit brain in which, on independent 3D electron-microscopy evaluation, “synaptic connectivity was preserved across the whole brain to the standard required by the prize.” Established In March 2018 the same laboratory took the Large Mammal Prize — $80,000, a whole pig brain, evaluated by the foundation’s committee with 3D EM after rewarming from cold storage. Established The awarding body’s own conclusion was a call for argument rather than for adoption: “The neuroscience and medical community should begin an open debate regarding ASC’s ability to preserve the information content of the brain.”

Established The prize committee attached a limit to its own award that the field has largely ignored. The BPF stated in the same announcement that a “single prize-winning laboratory demonstration is insufficient to address the quality-control measures that should be expected of any procedure applied to humans.” Frontier That is the body which funds and judges the prize explicitly declining to endorse clinical use of the thing it had just paid for, and it remains the most conservative published position held by anyone inside the preservation community. Established The sharpest technical objection, meanwhile, comes from a commercial rival. Max More, then president of Alcor, published a position statement on 13 March 2018 conceding that “ASC produces clearer images than current methods of vitrification without fixation,” and objecting on four grounds: that fixation wreaks havoc on a molecular scale, with chemical changes that are “extreme and difficult to evaluate”; that “fixation is likely to be much harder to reverse so as to restore biological viability as compared to vitrification without fixation”; that fixation amplifies cryoprotectant damage under the non-ideal field conditions typical of real cases; and that ASC is “a research dead end” for reversible preservation. Established This is an interested-party document twice over — a vendor criticising a rival method that won a prize it did not — and the molecular-damage objection is nevertheless technically substantive, which is why it is quoted at length here rather than waved away.

Established Two paradigms exist, they optimise for incompatible things, and no published method does both. ASC maximises structural information retention and abandons viability by design. Vitrification — Alcor’s practice since 2001, using the M22 cryoprotectant line developed by the very company that won both ASC prizes — maximises potential viability and accepts worse structural imaging, which Alcor itself concedes. Frontier No measurement currently available adjudicates between them, because the question is not technical: it depends on whether the person is the structure or the living tissue. Established The strongest evidence that the viability branch is not hopeless is Han and colleagues’ 2023 result — rat kidneys vitrified, stored at -150 degC for up to 100 days, then rewarmed volumetrically by iron-oxide nanoparticles in an alternating magnetic field, with all five nanowarmed recipients surviving the full 30-day study and terminal creatinine matching fresh controls. Frontier That organ weighs about a gram, has a forgiving vascular tree, and is not a brain; scaling to roughly 1,300 g of the least thermally and osmotically tolerant tissue in the body is entirely untouched.

Established The best connectome ever published covers about one cubic millimetre and cost six months of automated microscopy and two petabytes. The MICrONS Consortium’s April 2025 Nature paper reconstructed a volume of roughly 1.3 × 0.87 × 0.82 mm of mouse visual cortex containing over 200,000 cells, about 75,000 neurons with matched functional calcium imaging, and approximately 0.5 billion synapses. Established Raw electron-microscopy data: two petabytes. Imaging: about six months on automated microscopes. Established Manual proofreading of the automated segmentation: 1,046,656 edits as of 16 September 2024, described by the consortium as the largest single expense in person-hours, at a rate of 400 to 600 edits per full-time proofreader per week. Established This is a genuine landmark, and it is also the number that prices the entire field, because it is the only measured throughput from which a statement about reading a human brain can honestly be derived. Section 4 does that arithmetic and shows its working.

Established A connectome is a wiring diagram, and the only organisms in which connectomes can be compared between individuals say the diagram is neither identical nor decisive. Witvliet and colleagues reconstructed eight isogenic C. elegans across developmental stages and reported in Nature in 2021 that while “the overall geometry of the brain is preserved from birth to adulthood,” comparison between individuals “reveals substantial differences in connectivity that make each brain partly unique.” Frontier The result cuts both ways at once and this brief declines to pick one. It establishes that the connectome carries individual-specific information, which is a reason to preserve it. Frontier It also establishes that genetically identical animals with measurably different wiring behave alike, which means connectome differences at some scale are not identity-bearing — the mapping from connectome to organism is not even known to be injective, and structure cannot be assumed sufficient merely because it is plausibly necessary. Speculative No comparable within-species, between-individual comparison exists for any vertebrate, and none is planned at a scale that would settle the question.

Established Nothing has ever been revived and nothing has ever been read out. No preserved mammalian brain has been restored to function; no preserved brain of any species has been reconstructed as a whole connectome. The demonstrated capability is preservation plus verification by electron microscopy of sampled regions, and nothing beyond that. Established No party to the dispute contests this. Frontier The living-tissue branch is separate, real, and recent: a 2024 Cell Reports Methods paper reports a cryoprotectant regime for cryopreserving human brain tissue and neural organoids with functional recovery. The research behind this brief could not retrieve that paper, so no storage duration, electrophysiological endpoint or recovery fraction is printed here — it is named because a brief on brain preservation that omits living-tissue cryopreservation is incomplete, not because its figures are being leaned on.

Established Every result in this subject was obtained under laboratory conditions that no human case can reproduce. The prize brains were perfused promptly, by the people who developed the method, on healthy animals under controlled conditions. Established A human case cannot begin until legal death has been pronounced, which guarantees an interval of warm ischaemia before any fixative reaches the tissue — minutes in a planned, standby-attended case, days in an unattended death, an autopsy jurisdiction or an international transport. Frontier No published dataset quantifies the distribution of those intervals across the several hundred human preservation cases that exist in the adjacent cryonics industry, and no human ASC case exists at all, which makes the single largest determinant of case quality also the field’s least reported number. Frontier Alcor’s own objection that fixation amplifies cryoprotectant damage under non-ideal field conditions is precisely a claim that laboratory performance does not transfer, and it is the part of that objection least answered by the prize results.

Established The only company to have offered the procedure commercially proposed to kill the patient in order to perform it, and said so out loud. Nectome, covered by MIT Technology Review on 13 March 2018, described a procedure in which a patient under anaesthesia is connected to a heart-lung machine that perfuses fixative through the carotid arteries over approximately six hours. Established The reported finances: a $10,000 fully refundable deposit, 25 people on the waitlist, $120,000 from Y Combinator, and a $960,000 federal NIH grant for whole-brain preservation and imaging, roughly a third of which funded Edward Boyden’s expansion-microscopy laboratory at MIT. Established The two quotations that framed the coverage remain the two poles of the argument: the neuroscientist Michael Hendricks called brain uploading “abjectly false hope” and “a joke,” while the BPF’s Kenneth Hayworth said that “if the brain is dead, it’s like your computer is off, but that doesn’t mean the information isn’t there.” Established MIT severed the research relationship the following month, in April 2018. Frontier Both halves of that sequence are load-bearing and they point opposite ways: a federal science agency funded the work, and a university walked away from it, within weeks of each other.

Frontier The bottleneck has moved from biology to instruments, and that is the most important change of the last decade. Between the 2016 rabbit prize and the 2025 MICrONS release the field went from asking whether a brain can be preserved to reporting that a cubic millimetre can be imaged in six months for two petabytes. Frontier Whole-brain electron microscopy is now an industrial process with a unit cost rather than a research question with an unknown answer. Frontier The residual scientific uncertainty in readout is concentrated in one place: proofreading. MICrONS required over a million manual corrections to its automated segmentation for one cubic millimetre, and the published rate of 400 to 600 edits per proofreader-week is a labour figure, not a compute figure. Speculative Automated segmentation is improving, and every workback plan in the field assumes machine learning eliminates this link; no published result demonstrates it, so that assumption is an assertion with a good prior rather than a finding.

Frontier ASC has never been independently replicated at large-mammal scale. Both prizes were won by the same laboratory, using the same method, evaluated by the same committee. Established That is not an accusation; the arrangement was published openly and nobody has alleged otherwise. It is simply the ordinary standard the rest of biology applies, and the preservation literature does not meet it here. Frontier A replication would be cheap by the standards of the money already in the field: a second laboratory performs ASC on a pig brain and a third, unconnected to either, evaluates by 3D EM against the published prize criteria. Speculative The absence of such a replication eight years after the prize is better explained by the small number of laboratories with any interest in the question than by any suppressed negative result.

Frontier The field’s founding hypothesis has never been tested in the cheapest organism where it could be. The structural hypothesis — Hayworth’s powered-off computer — requires that long-term memory and personal identity be encoded wholly in features that survive fixation: synaptic connectivity, synaptic weights inferable from morphology, and cell-type identity. Speculative That is an assumption, and the experiment that would test it is small. Train an animal on a learned association, preserve it by ASC, image it, simulate the recovered connectome, and ask whether the trained behaviour is recoverable and distinguishable from a naive individual’s. Established Complete connectomes exist for C. elegans and for Drosophila. Frontier The demonstration has not been performed in either. It is the single most decisive missing experiment in the subject and it is tractable today.

Frontier The molecular objection is a live research question, not a philosophical one, and it has an assay. Memory may reside partly in molecular state — receptor phosphorylation, methylation, CaMKII autophosphorylation, local translation state — much of which glutaraldehyde cross-links, denatures, or leaves invisible at electron-microscope contrast. Frontier Alcor’s objection is this argument stated in the preservation community’s own vocabulary. Speculative The measurement that would settle how much is lost is a matched fixed-versus-unfixed assay of synaptic post-translational modification state by mass spectrometry or expansion microscopy — which is very likely why Boyden’s expansion-microscopy laboratory received part of the Nectome grant. Speculative A further and more radical version of the objection holds that memory-associated information can be carried by RNA species and transferred between animals. This brief could not obtain a primary source for that line of work and therefore attaches no claim, number or citation to it; it is named because if any part of memory is carried by molecules that fixation destroys or that EM cannot read, structural preservation is insufficient in principle rather than in practice.

Speculative Hibernation supplies an awkward natural experiment that the preservation literature has not absorbed. Hibernating mammals undergo massive synaptic pruning and tau hyperphosphorylation during torpor, reversed on arousal, and emerge with memory apparently intact. Speculative If that is right, the instantaneous synaptic wiring diagram is not the memory — the memory is recoverable from something that survives the diagram being taken apart and put back together. Handwave Read one way this is encouraging for preservation, because it implies redundancy; read the other way it is alarming, because it implies the identity-bearing quantity is not the thing electron microscopy measures. Nobody has run the behavioural experiment that would separate the readings. See Human Hibernation.

Established The binding constraint is imaging throughput and proofreading labour, and the arithmetic is worth doing in public. Take the three published MICrONS figures for one cubic millimetre of mouse cortex: 2 petabytes of raw EM data, about six months of automated microscope time, and 1,046,656 manual proofreading edits. Established A human brain is approximately 1.2 × 10^6 mm³. Frontier Straight-line scaling gives roughly 2.4 × 10^6 petabytes — about 2.4 zettabytes of raw data; on the order of 6 × 10^5 microscope-years of imaging at the demonstrated rate of about 2 mm³ per instrument-year; and on the order of 1.3 × 10^12 proofreading edits, which at 400 to 600 edits per proofreader-week is something like 5 × 10^7 proofreader-years.

Frontier Every one of those three numbers is this brief’s own arithmetic on a published rate, not a measured total, and nobody should cite them as a result. The projection assumes no improvement in throughput, which is certainly wrong as a forecast and exactly right as a baseline: it prices the problem at today’s demonstrated capability, which is the only capability anyone has demonstrated. Established No archive of 2.4 zettabytes exists. Frontier The honest summary is that the gap between “a brain can be preserved” and “a brain can be read” is about six orders of magnitude, and that it sits in instrumentation and labour rather than in biology or physics. Speculative Microscope-years is a capital number rather than a waiting time, which is the one piece of good news in the arithmetic: 6 × 10^5 microscope-years is 600,000 instruments running for a year, or 6,000 instruments running for a century, and electron microscopes are manufactured objects rather than scarce natural phenomena. Handwave What no one has costed in public is what 6,000 purpose-built multi-beam instruments and their operators would cost, or who would own the result.

Frontier Two published quantities track progress and a reader can follow them for the next twenty years. The first is cubic millimetres per instrument-year, against the MICrONS baseline of roughly two. Frontier The second is proofreading edits per cubic millimetre — or, better, per millimetre of reconstructed neurite — against a baseline of about a million per cubic millimetre. Speculative Closing the gap requires the first to rise by something like 10^5 and the second to fall towards zero; multi-beam electron microscope arrays address the first and self-correcting segmentation the second, and neither has yet been shown to deliver more than a factor of tens. Handwave A claim that either metric will improve by five orders of magnitude on any stated schedule is currently a forecast about semiconductor-style scaling in an instrument class that has never exhibited it.

Frontier This subject waits on four results, none of which it produces itself. The first is a method for inferring synaptic weight from fixed morphology. Speculative Electron microscopy sees the geometry of a synapse, not its strength; if weights matter and morphology does not determine them, a perfect connectome is an underdetermined description of the network it came from, and every downstream simulation inherits free parameters nobody can constrain. Frontier The second is a quantitative account of what glutaraldehyde fixation does to identity-relevant molecular state, which is a mass-spectrometry and expansion-microscopy problem rather than a preservation one. Speculative A related and usually unstated dependency is cell-type identification from fixed morphology: a connectome without cell types is a graph without node labels, and the parameters that turn a graph into a dynamical system are carried mostly by the labels.

Frontier The third is segmentation that does not need a human. Connectomics is currently rate-limited by a labour input, and the machine-learning result that removes it would be a result in computer vision, not in neuroscience. Speculative The fourth is whole-brain simulation from a connectome plus inferred parameters — a capability that is downstream of the first and that no existing project claims at vertebrate scale. See Mind Uploading and Digital Minds. Frontier The viability branch depends instead on volumetric rewarming at organ scale, which is the same dependency Cryonics has and which nanowarming partly satisfies at 1 g. Speculative Underneath all of it sits a dependency no experiment supplies: a defensible account of what memory physically is, which is Memory Engineering’s problem and Consciousness Research’s. Frontier None of these four dependencies is being pursued by anyone whose primary interest is preservation, which is the structural weakness of the field: it needs results from communities that have no reason to produce them on its schedule.

Frontier The workback chain has nine links and the ordering matters more than the list. (1) Independent replication of ASC at large-mammal scale — a second laboratory preserves a pig brain and a third evaluates by 3D EM against the published prize criteria. Not done. Speculative (2) Show that a learned behaviour is recoverable from a preserved, imaged individual — train a C. elegans or a Drosophila, preserve by ASC, image, simulate, and test whether the trained state is distinguishable from a naive control. Both connectomes already exist; the experiment costs a fraction of one prize purse, and its endpoints are behavioural rather than interpretive, so it can return a clean negative. Speculative A negative would be the most valuable result the field could produce, because it would tell everyone currently paying for preservation that the thing being preserved is not the thing that matters. Frontier (3) Measure what fixation destroys — matched fixed and unfixed assay of synaptic post-translational modification state.

Frontier (4) Establish quality-control standards for a human procedure, which the BPF states do not exist. (5) Raise EM throughput by about 10^5, tracked as mm³ per instrument-year. (6) Eliminate manual proofreading, tracked as edits per millimetre of neurite. Speculative (7) Storage that outlives institutions — either passively stable room-temperature preservation or an endowment structure with a century of demonstrated survival. Neither exists. (8) Simulate at vertebrate scale. (9) A legal framework for pre-mortem preservation.

Frontier Which link binds depends entirely on which question is being asked, and this brief separates the three. For is this worth doing at all, the binding link is 2: a cheap, decisive, currently unperformed invertebrate experiment whose absence is the most striking gap in the subject. Frontier For can it be completed, the binding links are 5 and 6 — industrial imaging throughput, six orders of magnitude short, with a labour component that has not been shown to scale sub-linearly. Established For can it be offered to humans, the binding link is 9, and it is absolute rather than merely difficult: a fatal procedure performed on a living person is homicide in every jurisdiction absent a specific legal framework, and no jurisdiction has one.

Frontier The preservation side is an organ-perfusion engineering problem and it is largely solved at the demonstrated scale. Perfusing 1,300 g of tissue with glutaraldehyde and then with cryoprotectant requires an intact enough vasculature to reach every capillary bed before fixation stiffens the tissue against further flow, which is why case quality in the field depends overwhelmingly on how quickly perfusion begins. Established Storage at -135 degC is a materials and refrigeration problem with commercial answers; storage below the glass transition means the specimen is a brittle solid under thermal stress and can fracture. Speculative A fracture plane is arguably the most recoverable form of damage there is — a clean break rather than diffuse degradation — but no published dataset quantifies fracture incidence at brain scale.

Frontier The readout side is an instrument-building problem of a kind nobody has attempted. It needs multi-beam electron microscope arrays operating in parallel at production yield, automated ultrathin sectioning of an entire brain without losing or distorting a section, terabit-per-second data capture, and exabyte-class storage with random access for proofreading. Speculative The section-handling step is the one most often skipped in cost estimates: a human brain sectioned at connectomic thickness is a number of physical sections with no precedent in laboratory practice, and every one of them has to survive handling. Handwave Estimates that reduce the readout problem to petabytes and GPU-hours are pricing the easy half.

Speculative A third requirement is a metrology nobody has built: a way to grade a preserved brain without destroying it. The prize evaluations worked by cutting samples and imaging them, which is acceptable for a demonstration and unacceptable for a stored patient. Speculative A non-destructive quality assay — some combination of imaging, spectroscopy and mechanical measurement that returns a defensible score for how well a given brain was preserved — is what would turn preservation from a craft into a process with a specification. Frontier Its absence is why the prize-awarding foundation says quality-control measures for human procedures do not exist, and it is an engineering gap rather than a scientific one.

Frontier Four adjacent technologies would each change this subject if they matured, and only one of them is a preservation technology. Expansion microscopy — physically swelling fixed tissue so that optical microscopy resolves structures below the diffraction limit — is the obvious tool for measuring what fixation destroys, and it is the reason a third of the Nectome federal grant went to an MIT laboratory rather than to a preservation company. Frontier Functional imaging is the second: MICrONS matched calcium imaging to 75,000 of its reconstructed neurons, which is the only existing template for pairing a wiring diagram with what the wiring was doing.

Frontier Volumetric rewarming is the third and it belongs to the rival paradigm. Nanoparticle-driven rewarming in an alternating magnetic field is what made the 2023 kidney result possible, and if it scales it makes the viability branch a live programme rather than a promissory note — which would reduce the case for irreversible fixation rather than strengthen it. Speculative The fourth is machine segmentation, which is a computer-vision capability that the whole subject is waiting on and none of its practitioners are building. Speculative Further out, Brain-Computer Interfaces supply the only route by which a living brain’s functional state could ever be measured densely enough to validate a structural readout against a ground truth.

Speculative The adjacent technology that would change the economics most is chemical preservation that needs no cold at all. If aldehyde fixation plus a resin embedding or a stabilised dehydrated state held ultrastructure indefinitely at room temperature, the entire cost model collapses: no liquid nitrogen, no dewars, no century-long refrigeration liability, and no failure mode in which a specimen is lost because an organisation stopped paying an electricity bill. Handwave This brief could not verify the current state of room-temperature or fluid-preservation programmes and prints no result, no organisation and no figure for them; the direction is named because it is being pursued and because it removes the institutional failure mode that has historically killed preservation organisations, which is a larger effect than any plausible improvement in fixation chemistry.

Established The binding institutional requirement is custodial continuity over an interval no organisation of this type has ever survived. Preservation is worthless if the custodian fails, and the industry adjacent to this one supplies the base rate. Established Alcor publishes three mutually inconsistent patient counts across its own materials — an itemised census of 181 patients and 1,338 members as of 31 January 2021, an undated fact sheet claiming “over 250” patients and “over 1,500” members, and an encyclopaedic entry citing October 2023 for 222 preserved of 1,927 members. Frontier An organisation asking members to trust it with century-scale custody has a data-hygiene problem worth naming, without inflating it into an accusation. Established The Cryonics Institute reports 268 patients and 1,995 members as of March 2025. Frontier Fewer than 600 preserved humans exist worldwide across all providers, which means the entire custodial track record is a few decades long and a few hundred cases deep.

Established The legal frame is the second requirement and it currently forbids the procedure outright in its most defensible form. Cryonics operates in the United States under the Uniform Anatomical Gift Act: patients are received as anatomical donations, and preservation begins only after pronouncement of legal death. Established That is what makes it lawful, and it is also what guarantees an ischaemic interval before perfusion — the single largest uncontrolled variable in preservation quality. Established ASC applied optimally would begin before death, which is why Nectome’s procedure was fatal by design and why no jurisdiction permits it. Speculative A legal framework for pre-mortem preservation would collapse the ischaemic interval to zero and improve every downstream step, and it is the change least likely to be made. Frontier The third requirement is a quality-control standard for human application, whose absence is asserted not by critics but by the prize-awarding foundation itself. Speculative The fourth is a financial instrument nobody has designed: an endowment whose spending rule, governance and legal form can be shown to survive the death of everyone who set it up, through inflation regimes, jurisdictional changes and at least one institutional bankruptcy. Handwave Perpetual trusts and cathedral fabric funds are the nearest analogues, and neither carries an obligation whose beneficiary cannot complain.

Established The central ethical fact is that the best version of this procedure kills the patient. Nectome’s protocol was described by its own founders as one hundred per cent fatal, performed under anaesthesia on a living person over roughly six hours, and marketed to a waitlist of 25 people who had each placed a refundable $10,000 deposit. Frontier That configuration — a terminal patient, an irreversible procedure, an unfalsifiable benefit, and a company holding deposits — is the hardest consent problem in the whole longevity cluster, and it is harder than the equivalent in assisted dying because the claimed benefit cannot be assessed by anyone now living. Speculative A defensible framework would require the benefit to be described as an option on a future capability with a stated probability, rather than as a service, and no provider does this.

Frontier The second problem is what a preserved brain is, legally and morally. It is currently property or an anatomical donation. Speculative If readout ever became possible, the same object would become either a person or the sole evidence of one, and no legal system has a category that survives the transition. Speculative A third problem is distributive: neuropreservation costs $80,000 at the most expensive US provider and EUR 75,000 in Europe, so the population that would be preserved is selected by wealth and by the willingness to hold a life-insurance policy for a century. Handwave The deepest objection is that preservation exports a cost to the future: every stored brain is a claim on the attention, energy and moral seriousness of people who did not consent to receive it. Speculative The reply available to the field is that this is true of every archive, cemetery and cathedral, and that societies have generally judged the burden worth carrying — but no archive has ever asked its custodians to decide whether its contents are people.

Speculative If the structural hypothesis holds, death becomes a data-retention problem, and civilisations are historically bad at data retention. The change would not be that people stop dying; it would be that dying becomes conditional on an archive surviving, which shifts the locus of mortality from medicine to institutions, endowments and formats. Handwave A society organised that way would have to invent custodial law that operates on the scale of cathedral maintenance, and would discover that its bottleneck is not cryogenics but trusteeship.

Speculative If the hypothesis fails, the accumulated inventory is the problem. A few hundred preserved brains is a curiosity; a few million, held against a readout that never arrives, is a permanent liability with a constituency that will not permit disposal. Frontier The asymmetry is the interesting part: the cost of being wrong in the optimistic direction is storage, and the cost of being wrong in the pessimistic direction is the irreversible destruction of whatever the structure contained. Speculative That asymmetry is the entire practical argument for preservation and it is a decision-theoretic argument, not a neuroscientific one. Handwave It is also the argument most vulnerable to the institutional failure mode, because an option that expires when its custodian does is worth much less than its holders believe.

Handwave The third-order effect is legal rather than technical. A society that could read a preserved brain would have to decide whether the readout inherits, votes, is married, or can be tried — and would have to decide it about a class of entity that exists in the thousands before any court has heard a case. Speculative The precedent most often reached for is corporate personhood, which is a poor guide because a corporation has no prior instance of itself to be compared against. Handwave The likeliest outcome is that the question is settled administratively and badly, decades before it is settled philosophically, by whichever jurisdiction happens to hold the storage facility.

These horizons track two independent clocks — the scientific question of whether structure suffices, which could be answered cheaply next year, and the industrial question of readout throughput, which cannot.

• 10 yr: Frontier Independent replication of ASC at large-mammal scale is achievable and overdue. Speculative The invertebrate preserve-image-simulate-behave experiment is performed by somebody, and it is the result most likely to change the subject in either direction. Frontier EM throughput improves by tens, not by orders of magnitude, and proofreading remains a labour line item that shows up in grant budgets rather than in press releases. Speculative A non-destructive quality assay for preserved tissue is the most likely piece of genuine engineering progress in this window.
• 25 yr: Speculative A complete mouse brain becomes the field’s stated milestone and is plausibly reached, which requires throughput gains of two to three orders of magnitude over MICrONS. Speculative A quantitative account of what fixation destroys exists, and it either rules ASC in or forces a modified protocol. Handwave A jurisdiction somewhere considers a legal framework for pre-mortem preservation and does not pass it. Speculative The custodial question becomes the field’s public problem rather than its private one, because by then the earliest preserved specimens are approaching the age at which the founding organisations’ original trustees are all dead.
• 50 yr: Handwave Human whole-brain electron microscopy becomes conceivable only if the two tracking metrics have each moved by four to five orders of magnitude, which no instrument class has done on that schedule. Speculative If it happens, the first readout is a dataset nobody can interpret rather than a mind, because simulation from a connectome plus inferred parameters remains unsolved. Handwave Everything about this row is a projection from a single published rate, and a reader is entitled to treat the 50-year horizon as the point at which the honest answer becomes that nobody knows.
• 100 / 250+ yr: Handwave The horizon on which the whole enterprise is actually staked, and the one where the institutional risk dominates every technical risk. Handwave Whether anything preserved today survives to be read depends less on chemistry than on whether any organisation founded in the twenty-first century is still discharging its obligations in the twenty-third. Speculative On this horizon the two most consequential developments would be room-temperature preservation, which removes the refrigeration liability entirely, and a decisive negative from the invertebrate experiment, which would end the programme within a decade of being run.

• Depends on Depends on results this subject does not produce: a method for inferring synaptic weight from fixed morphology; a quantitative assay of what aldehyde fixation does to identity-relevant molecular state; automated segmentation that does not require human proofreading; and, for the rival paradigm, volumetric rewarming at organ scale. The first two are neuroscience and analytical chemistry; the third is computer vision; the fourth belongs to organ cryopreservation and was partly demonstrated in a 1 g rat kidney in 2023. None of the four is being pursued by anyone whose primary interest is preservation, which is why progress here has historically arrived as a side effect of other people’s programmes rather than as the output of a plan.
• Requires (not on this map) Two constraints sit outside this map and neither is a scientific result. The first is an organisation that can hold a specimen at cryogenic temperature, solvent, uncontested and unbroken, for a hundred years or more — a governance and endowment problem no preservation organisation has yet demonstrated, and one whose failure mode is documented in the industry’s own history. The second is electron-microscope imaging capacity roughly five orders of magnitude above the demonstrated rate of about two cubic millimetres per instrument-year, together with a segmentation pipeline that removes the million manual proofreading edits per cubic millimetre the published record still requires. Neither token is a discovery waiting to be made: both name capacity the world could choose to build, which is why they are recorded as institutional and industrial rather than scientific. The first is the one that has already failed once in the adjacent industry’s history, and it is the one on which nothing preserved today depends more completely.
• Enables Enables Mind Uploading in the only sense in which that subject has a physical substrate to work from, and supplies Digital Minds with the one input it cannot generate for itself. It also enables a decisive test for Memory Engineering: a preserved and imaged trained animal is the cleanest available probe of whether memory is structural. What it enables is conditional in a way the enabled subjects rarely acknowledge: if fixation destroys identity-bearing molecular state, then brain preservation enables mind uploading only in the sense that a photograph of a library enables reading the books.
• Adjacent Adjacent to Cryonics, which shares the storage and rewarming problems while rejecting the fixation choice; to Consciousness Research, which owns the question of whether a readout is anyone; to Human Hibernation, which asks whether memory survives massive reversible synaptic remodelling; and to Biological Immortality, which is testing the same proposition about what a person is made of from the opposite end — persistence by continuous replacement against persistence by frozen structure. Adjacent also, and uncomfortably, to connectomics itself, a field with no interest in preservation whose published throughput figures nonetheless determine whether preservation is ever worth anything.

Frontier “A pig brain was successfully preserved, so brain preservation works.” What was demonstrated in March 2018 is that synaptic connectivity was visible under 3D electron microscopy across a whole pig brain after cold storage. No function was preserved, nothing was revived, nothing was read out, and the awarding foundation itself stated that one laboratory demonstration is insufficient for human application. Established This is an enthusiast-side error and the foundation is on record contradicting it.

Established “ASC is a form of cryonics.” ASC deliberately abandons viability: glutaraldehyde is chosen precisely because it stabilises structure, and the public objection from Alcor — the cryonics provider whose own method ASC competes with — is that fixation makes biological revival harder. Established The two are competing paradigms with incompatible objectives, and the 2018 exchange between the BPF and Alcor is the clearest public statement of the split.

Established “Connectomics is nearly able to read a human brain.” The state of the art is one cubic millimetre, two petabytes, six months of imaging and 1,046,656 manual proofreading edits. Frontier A human brain is about 1.2 million times that volume, which puts straight-line readout at roughly 2.4 zettabytes and 6 × 10^5 microscope-years. Established The scale gap here is routinely understated by around six orders of magnitude, and it is the largest single misjudgement in the subject. Frontier It is worth repeating that the zettabyte and microscope-year figures are this brief’s own scaling of a published rate rather than a published result — the error being corrected is a mis-statement of magnitude, and replacing it with a false precision would be the same failure in the other direction.

Frontier “The connectome is the person.” This is the field’s founding assumption, not a finding, and it has never been tested in any organism — including ones whose complete connectomes have been available for decades. Established Genetically identical C. elegans have “partly unique” connectomes and behave alike, so connectome differences at some scale carry no identity. Speculative A defender can answer that the identity-bearing scale is finer than the between-individual variation measured, which is coherent and currently unfalsified; it is also an additional assumption rather than a reply.

Established “Fixation destroys everything, so ASC is worthless.” Independent 3D EM evaluation found synaptic connectivity preserved across a whole brain in two separate prize demonstrations, and a commercial competitor concedes in writing that ASC gives clearer images than unfixed vitrification. Frontier The real objection is about molecular information and about reversibility, not about structural preservation. This one is a sceptic-side error.

Established “Nectome was a fringe operation with no scientific backing.” It held $120,000 from Y Combinator and a $960,000 NIH grant, roughly a third of the federal money funding a laboratory at MIT. Established It is equally true that MIT terminated the relationship weeks later. Both halves belong on the page; quoting either alone is a sceptic-side or an enthusiast-side error depending on which half is dropped.

Frontier “Brain preservation is a solved storage problem; only revival is hard.” Storage is a century-scale institutional problem, and the historical record of preservation organisations is poor enough that the industry’s own past contains a facility failure in which patients thawed. Speculative This brief could not obtain a primary source for that episode and prints no date, no organisation detail and no patient number for it; the structural point stands regardless, which is that whether an organisation survives two centuries is not a technical question and has never been demonstrated by any organisation of this kind.

Handwave “Preserving the brain preserves you.” Even under the most optimistic technical assumptions, a readout produces a copy, and whether that copy is you is a question about personal identity that no experiment resolves. Speculative The optimistic reading requires a functionalist theory of personal identity to be correct, which is a philosophical commitment and not a neuroscientific finding. Frontier A brief that lets neuroscience answer that question is making the field’s characteristic error, and so is one that treats the question as settled in the sceptical direction.

Speculative One number in circulation should not be repeated. A figure of 150 trillion synaptic connections for the preserved pig brain appears in at least one retrieval of the prize announcement. It is close to standard estimates for a human brain, it could not be confirmed in the source, and it is therefore not printed as a fact anywhere in this brief. Handwave Where an enthusiast and a sceptic both have reasons to want a number to be true, the number is usually the first thing to check.

Established The strongest case that structural preservation preserves nothing that matters deserves stating at full strength. No preserved brain has ever been revived or read. Both prizes went to a single laboratory using a single method with no independent replication. The readout technology is six orders of magnitude away and the shortfall is in labour and instruments, which improve slowly. Frontier The sufficiency of structure for memory is unproven and there is a published mechanism by which it would fail. Genetically identical animals have measurably different connectomes and indistinguishable behaviour. The leading commercial actor’s procedure is, by its own description, fatal. Established Michael Hendricks’ phrase for the whole enterprise — “abjectly false hope” — is the compressed version of that case, and it is not a fringe view. Speculative The reply that survives it is not a scientific one: preservation dominates cremation under uncertainty for anyone assigning non-zero probability to future readout, at a cost far below the value of the option. Frontier That is a genuine tie between an evidential argument and a decision-theoretic one, and this brief declares it rather than resolving it.