Mind Uploading integrated guide
EN 日本語
Integrated Guide · Chapter 3 · Running assessment

Current status — where the field stands in August 2026.

This is the core chapter of the site, and the one that is rewritten with each update. Following the pipeline of Chapter 2, it establishes six axes and assesses each as done / underway / not started, with primary citations.

As of: 2026-08-08Six axes39 references
The August 2026 verdict in three lines
  • Structural mapping is complete for insects (139,000–167,000 neurons) and has begun scaling organisationally in mammals. For humans, roughly one millionth of the volume is mapped.
  • Running a brain was demonstrated for the first time in the fly: a whole-brain spiking model matched experiment in 2024, and in 2026 natural behaviour emerged with a simulated body attached. Plasticity and neuromodulation are not included.
  • Human WBE has not entered implementation on any axis. A 2025 researcher review put it at "decades away even optimistically", and the prediction-market median sits around 2070.

3.1 Six Axes

Axis-by-axis assessment.

Axis 1 · Structural mapping (connectomics) insects complete mammals starting

StatusDetail
DoneComplete nematode wiring diagrams including both sexes and developmental stages (1986 → 2019 → 2021)12. Fly larva, 3,016 neurons (2023)3. Whole adult fly brain, 139,255 neurons and roughly 50 million synapses (FlyWire, 2024)4. Male fly brain plus ventral nerve cord, 166,000 neurons — the first wiring diagram spanning a whole central nervous system with the neck connective intact (2025)5. One cubic millimetre of human cortex: 1.4 petabytes, 57,000 cells, 150 million synapses (H01, 2024)6. One cubic millimetre of mouse visual cortex: about 200,000 cells and 523 million synapses, plus activity from over 70,000 neurons (MICrONS, 2025)7. Whole-brain cell-type atlases for mouse and human (2023)8.
UnderwayLarval zebrafish whole brain: EM acquisition is complete for an animal whose ~70,000-neuron activity was already recorded, and reconstruction is in progress (Google, Janelia, Harvard)9. Ten cubic millimetres of mouse hippocampus (Harvard-led, roughly $33 million, with Google and the Allen Institute)10 and 10 mm³ of the cortico-basal ganglia-thalamic loop (Allen)11. E11 Bio released hippocampal pilot data from PRISM, its protein-barcoding plus expansion microscopy approach aimed at cutting proofreading cost by orders of magnitude12.
Not startedWhole mouse brain (about 70 million neurons, an estimated 500 petabytes to 1 exabyte). Straight-line extrapolation of current technology puts imaging and proofreading beyond ten years13. Whole human brain (an estimated 1–2 zettabytes) has no established acquisition method at all.

For scale: proofreading the fly brain still required tens of person-years and millions of manual edits even after automatic segmentation4. Naive extrapolation to the mouse gives thousands of person-years, so full automation of proofreading is a precondition rather than an optimisation.

Axis 2 · Functional readout (recording and BCI) advancing rapidly

StatusDetail
DoneNeuropixels probes made simultaneous recording from thousands of sites standard14. In larval zebrafish, about 70,000 neurons — effectively the whole brain — were recorded during behavioural tasks (ZAPBench)9. In humans, intracortical speech reconstruction reached 60–78 words per minute with single-digit word error rates1516, and near-real-time streaming voice synthesis was demonstrated in 202517. Walking was restored via a brain–spine interface (2023)18. Implanted BCI clinical work expanded: Neuralink reached 26 patients by June 202619, with Synchron, Paradromics and Precision Neuroscience also in trials20.
UnderwayHigher electrode density and manufacturability (Neuralink's automated surgery and volume production plans19), thousands-to-10,000-channel human cortical recording, and bandwidth improvements in non-invasive methods.
Not startedSimultaneous recording of every neuron in a mammalian brain. Stable long-term recording at 100,000-channel scale in humans. No established route by which readout alone reaches the information content required for uploading (Chapter 2).

Axis 3 · Running it (simulation / emulation) first demonstration in the fly

StatusDetail
DoneFly: a leaky integrate-and-fire model built on the whole-brain connectome reproduced circuit responses — taste to feeding motor output, mechanosensation to grooming — consistently with experiment (2024)21. In March 2026, Eon Systems coupled that model to a physics-simulated body (NeuroMechFly v2) and demonstrated the first embodied whole-brain emulation, in which walking and grooming emerged without behaviour being programmed2223. Nematode: MetaWorm closed the loop between brain, body and environment and reproduced chemotaxis (2024)24. Mammals: biophysical simulation of cortical microcircuitry (2015)25 and a foundation-model digital twin predicting visual cortex responses to novel stimuli (2025)26.
UnderwayEon's mouse whole-brain emulation programme, about 560 times the fly in neuron count22. Competition to improve whole-brain activity prediction on ZAPBench9. Simulation infrastructure at the Open Brain Institute and EBRAINS, which inherited Blue Brain's assets27.
Not startedEmulation including plasticity (learning), neuromodulation and development. A nematode emulation that passes validation criteria. Any mammalian whole-brain emulation.
How to place the current headline result

The 2026 embodied fly emulation is the first whole-brain-scale partial affirmative answer to Chapter 2's central question: how much function follows from structure alone. But what was reproduced is reflexive, stereotyped behaviour. Learning, motivation and internal state are absent. It is not "a fly's life reproduced".

Axis 4 · Preservation (fixation and storage) clinical use now the question

StatusDetail
DoneAldehyde-stabilized cryopreservation demonstrated retention of synaptic ultrastructure across whole rabbit brains (BPF Small Mammal Prize, 2016) and whole pig brains (Large Mammal Prize, 2018) under third-party evaluation2829. In March 2026, Nectome released a preprint reporting whole-pig-brain ultrastructural preservation using a procedure compatible with physician-assisted death, which is legal in Oregon and elsewhere30. In practice, Oregon Brain Preservation offers low-cost chemical fixation while Tomorrow Bio (Europe), Alcor and the Cryonics Institute offer conventional cryopreservation31.
UnderwayQuantifying ASC quality in human post-mortem brains, demonstrating large-scale connectome extraction from a preserved brain, and building protocol standards and third-party audit frameworks.
Not startedDemonstrating that memory and personality information actually survive in a preserved brain and can be read out. This is the foundation of the entire preservation argument, but identifying the physical substrate of memory (synaptic strength plus something else) is itself an active research problem. Current preservation guarantees the retention of structure, and nothing beyond that.

Axis 5 · Compute lower-bound order of magnitude reached

Sustained supercomputer performance passed 10¹⁸ FLOPS32, entering the same order of magnitude the 2008 roadmap estimated for a spiking-level human brain (≈10¹⁸ FLOPS). Low-precision arithmetic in AI datacentres runs two to three orders above that. Neuromorphic machines — Intel's Hala Point at 1.15 billion neurons, DeepSouth at 228 trillion synaptic operations per second — are maturing as execution environments for whole-brain-scale spiking. What remains is not arithmetic but the memory and communication architecture for holding, updating and routing the state of 10¹⁴ synapses, and processing zettabyte-scale scan data (see Chapter 2.5).

Axis 6 · Validation and theory the furthest behind

StatusDetail
DoneThe first standard benchmark: ZAPBench, which asks models to predict whole-brain activity in the same animal and compares them quantitatively9. In the fly, a loop of model prediction followed by optogenetic verification has begun to turn21. The field also updated its self-assessment: the State of Brain Emulation Report 2025 organises progress around recording, mapping and emulation and judges human WBE to be decades away even in optimistic scenarios33.
UnderwayDesigning pass criteria from behaviour, perturbation and held-out data. Discussion of standardised cross-species validation protocols.
Not startedProcedures for validating personal identity — what would establish that an emulation is "the same person". Consciousness assessment: the 2025 adversarial test of the leading theories produced results at odds with predictions from both, leaving theory selection open34. This axis is a scientific and philosophical problem rather than a technical one, and continues in Chapter 5.

3.2 The Ladder

Species by species.

Organism (neurons)Structural mapDynamical dataEmulationOne-line verdict
Nematode (302)Complete both sexes, developmentPartial (whole-brain calcium imaging possible; electrophysiology scarce)Partial (MetaWorm reproduces behaviour; validation incomplete)The origin of the "map without dynamics" lesson
Fly larva (3,016)Complete 2023LimitedNoneUsed mainly as a teaching case for circuit analysis
Adult fly (139,255)Complete female 2024, male CNS 2025Rich (optogenetics, behaviour)First demonstration 2024 model → 2026 embodiedThe current frontier
Larval zebrafish (≈10⁵)In reconstructionWhole-brain, ~70,000 neuronsPrediction models competing (ZAPBench)The only species proceeding activity-first, structure-second
Mouse (≈7×10⁷)0.2% (1 mm³) plus a 10 mm³ programmeLarge-scale partial recording (thousands to tens of thousands)Regional digital twin (visual cortex)The main battleground of the next decade
Marmoset (≈6×10⁸)Mesoscale maps (Brain/MINDS)PartialNoneThe primate foothold, led from Japan
Human (8.6×10¹⁰)≈10⁻⁶ (1 mm³ of cortex)Point measurements via BCI onlyNoneEven the means of structural acquisition is unsettled

3.3 Projects & Players

The major projects and players.

OrganisationTypeWhat they doLatest milestone
Janelia FlyEM (HHMI)InstituteIndustrial EM connectomics in the flyMale CNS, 166,000 neurons (2025)5
FlyWire / Seung lab (Princeton)University + citizen scienceWhole-brain reconstruction combining AI with community proofreadingAdult whole-brain connectome (2024)4
Google Research ConnectomicsIndustryAutomated segmentation, H01, ZAPBench, mouse programmeWhole-brain zebrafish benchmark (2025)9
MICrONS consortium (Allen, Princeton, Baylor)Public consortiumCombined structural and functional connectomicsNature collection on 1 mm³ of mouse visual cortex (2025)7
BRAIN CONNECTS (NIH)PublicTechnology development toward a whole mouse brain ($150M, 11 projects)10 mm³ hippocampus and cortico-basal ganglia efforts underway1011
Lichtman lab (Harvard)UniversityHuman cortical EM (H01), hippocampus programmeH01 released (2024)6
E11 BioNon-profit institutePRISM: barcoding plus expansion microscopy, targeting a 100× cut in proofreading costHippocampal pilot data released (2025)12
Eon SystemsStartupEmulation execution; states "upload the human mind" as its goalEmbodied whole fly brain emulation (2026)22
OpenWorm / MetaWormOpen sourceIntegrated nematode simulationClosed-loop brain–body–environment model (2024)24
EBRAINS / Open Brain InstitutePublic / non-profitSuccessor simulation infrastructure to HBP and Blue BrainBlue Brain assets open-sourced (2024–25)27
Cortical Labs / FinalSparkIndustryCultured-neuron computation (a wet control condition for the field)CL1 on sale and available via cloud (2025)35
Neuralink · Synchron · Paradromics · PrecisionIndustryClinical translation of invasive human BCINeuralink at 26 patients (June 2026)1920
NectomePrivate researchHuman-compatible ASC preservation protocolsWhole pig brain, assisted-death-compatible protocol preprint (2026)30
BPF · CarboncopiesNon-profitThird-party evaluation of preservation; WBE roadmap outreachContinuing evaluation criteria and educational material29
Brain/MINDS 2.0 (Japan, AMED)PublicMarmoset brain mapping and digital brain developmentMarmoset standard brain atlas v2 (2025)36
MinD in a Device / Watanabe lab (Japan)Industry + universityGradual migration of consciousness to machine via BCI (hemisphere connection)Concept and component research37

3.4 Funding

Money and institutions: the centre of gravity moves private.

3.5 Watchlist

What to watch over the next twelve months.

Mapping

Zebrafish whole-brain connectome release

On release this becomes the first dataset combining whole-brain activity and whole-brain structure in the same animal — a step change for modelling research.

Mapping

Progress on 10 mm³ of mouse hippocampus

The technology proof for a whole mouse brain (over 500 mm³). Acquisition rate and proofreading automation will decide whether the mouse arrives in the early or late 2030s.

Running

First interim result from Eon's mouse programme

Fly to mouse is a 560-fold scale-up. Even a validated partial circuit would put a concrete timeline on mammalian emulation.

Readout

BCI bandwidth and patient numbers

Neuralink's volume production and automated surgery; scale-up at Paradromics and Precision. Annual implants reaching three digits would mark the transition to a clinical standard.

Preservation

Peer review of the Nectome preprint, and the regulatory response

Whether the assisted-death-compatible protocol survives review, and how states and professional bodies respond. The test of whether preservation medicalises or stalls.

Validation

ZAPBench score trajectory

Whether models using connectome information beat purely statistical ones. Quantifying that structure helps predict dynamics would be evidence for the field's whole premise.

3.6 Summary

Chapter summary.

Key points
  • Of the six axes, mapping, readout and compute advance exponentially while modelling and validation set the pace. Preservation is technically maturing, and its open questions have moved to deployment.
  • The 2024–2026 turn: the field shifted its subject from building maps to running maps and validating them (the fly demonstrations, ZAPBench, digital twins).
  • An honest summary for human WBE: six orders of magnitude short on structure, one species demonstrated, and zero agreed validation standards. The grounds for optimism (the speed of automation) and for pessimism (the unknowns in modelling) both genuinely exist.

References

Chapter 3 references (39).

  1. Cook, S. J., et al. (2019). Whole-animal connectomes of both C. elegans sexes. Nature, 571, 63–71.
  2. Witvliet, D., et al. (2021). Connectomes across development reveal principles of brain maturation. Nature, 596, 257–261.
  3. Winding, M., et al. (2023). The connectome of an insect brain. Science, 379, eadd9330.
  4. Dorkenwald, S., et al. (FlyWire) (2024). Neuronal wiring diagram of an adult brain. Nature, 634, 124–138.
  5. Janelia FlyEM (2025). Male CNS Connectome (brain + ventral nerve cord, 166,000 neurons).
  6. Shapson-Coe, A., et al. (2024). A petavoxel fragment of human cerebral cortex reconstructed at nanoscale resolution. Science, 384, eadk4858.
  7. The MICrONS Consortium (2025). Functional connectomics spanning multiple areas of mouse visual cortex. Nature, 640.
  8. Yao, Z., et al. (2023). A high-resolution transcriptomic and spatial atlas of cell types in the whole mouse brain. Nature, 624, 317–332.
  9. Google Research, Janelia & Harvard (2025). ZAPBench: whole-brain activity prediction benchmark in a single larval zebrafish (connectome in progress).
  10. Google Research (2023–). Harvard-led 10 mm³ mouse hippocampus connectome project (BRAIN CONNECTS).
  11. Allen Institute (2023–). Mouse cortico-basal ganglia loop and macaque connectivity projects.
  12. E11 Bio (2024–25). PRISM technology and hippocampal pilot dataset (released on AWS).
  13. BioTechniques (2024). Building a comprehensive mouse brain connectome (includes duration estimates).
  14. Steinmetz, N. A., et al. (2021). Neuropixels 2.0. Science, 372, eabf4588.
  15. Willett, F. R., et al. (2023). A high-performance speech neuroprosthesis. Nature, 620, 1031–1036.
  16. Card, N. S., et al. (2024). An accurate and rapidly calibrating speech neuroprosthesis. N. Engl. J. Med., 391, 609–618.
  17. Littlejohn, K. T., et al. (2025). A streaming brain-to-voice neuroprosthesis. Nat. Neurosci., 28.
  18. Lorach, H., et al. (2023). Walking naturally after spinal cord injury using a brain–spine interface. Nature, 618, 126–133.
  19. The Debrief (2026). Neuralink patient counts and volume production plans (26 patients as of June 2026).
  20. Paradromics (first human implant, 2025), Synchron and Precision Neuroscience — company sources.
  21. Shiu, P. K., et al. (2024). A Drosophila computational brain model reveals sensorimotor processing. Nature, 634, 210–219.
  22. Eon Systems (7 March 2026). How the Eon team produced a virtual embodied fly.
  23. The Register (2026). Digital fruit fly brain model walks and cleans its feelers (independent coverage).
  24. Zhao, M., et al. (2024). An integrative data-driven model simulating C. elegans brain, body and environment interactions. Nat. Comput. Sci.
  25. Markram, H., et al. (2015). Reconstruction and simulation of neocortical microcircuitry. Cell, 163, 456–492.
  26. Wang, E. Y., et al. (2025). Foundation model of neural activity predicts response to new stimulus types. Nature, 640.
  27. Open Brain Institute (2025–) / EBRAINS. Successor simulation infrastructure to Blue Brain.
  28. McIntyre, R. L., & Fahy, G. M. (2015). Aldehyde-stabilized cryopreservation. Cryobiology, 71, 448–458.
  29. Brain Preservation Foundation. Small Mammal Prize (2016) and Large Mammal Prize (2018) evaluation records.
  30. Nectome (2026). Ultrastructural preservation of a whole large mammal brain… (preprint and research page).
  31. Oregon Brain Preservation / Tomorrow Bio / Alcor / Cryonics Institute (services currently offered).
  32. TOP500. El Capitan (1.74 exaFLOPS sustained) and others.
  33. Zanichelli, N., et al. (2025). State of Brain Emulation Report 2025. arXiv:2510.15745.
  34. Cogitate Consortium (2025). Adversarial testing of global neuronal workspace and integrated information theories. Nature.
  35. Cortical Labs (2025). CL1: a cultured human neuron computer (about 200,000 neurons).
  36. Brain/MINDS 2.0 (AMED, 2024–). Marmoset brain mapping and digital brain development.
  37. NeurotechJP (2023). Interview with Masataka Watanabe: the science of consciousness and consciousness uploading.
  38. The Transmitter (2025). $278 million cut in BRAIN Initiative funding.
  39. The Transmitter (2026). BRAIN Initiative gains a 33% budget increase in the FY2026 bill.