Integrated Guide · Chapter 6

What to do now — guidance by position.

The current status in Chapter 3 and the dependency structure in Chapter 4 make it possible to identify the buttons that, when pushed, actually move the whole field. This chapter states the field-wide priorities first, then gives concrete suggestions for researchers, students, funders and individuals. It ends with something equally important: what not to do.

Four audiences16 references

In this chapter

6.1 Field-wide priorities
6.2 For researchers and engineers
6.3 For students and career changers
6.4 For funders and policymakers
6.5 For individuals
6.6 Regional notes
6.7 What not to do
References

6.1 Priorities

Five buttons that move the whole field.

#	Priority	Why it is rate-limiting	Who is working on it
P1	Fully automated proofreading — cutting AI segmentation error rates by orders of magnitude	The mouse connectome (M3) essentially depends on it. Straight extrapolation of manual proofreading gives thousands of person-years	Google, Princeton and E11 among others. The area with the highest demand for people¹
P2	The science of inferring dynamics from structure — recovering synaptic weights and modulatory state from EM morphology	Chapter 2's "greatest scientific uncertainty". A finished map will not run without it	MICrONS validation work and digital-twin approaches have reached the entrance²
P3	Validation benchmarks — extending the ZAPBench pattern to nematode, fly and mouse	Without a yardstick for "it worked", neither progress nor investment decisions are possible	ZAPBench is the first instance. Standardised behaviour and perturbation batteries are a blank space³
P4	Scientific validation of preservation — quantifying ASC quality in human post-mortem brains; demonstrating preserve-then-read-out	The only axis relevant to people alive now. Unvalidated adoption is an ethical risk; validation removes the time constraint	The Nectome preprint and BPF evaluations exist, but independent verification and standards are lacking⁴⁵
P5	Getting ahead on ethics and governance — consent design, emulation welfare, brain-data protection	Reactive handling risks halting research; proactive work turns social trust into an accelerant	Academic proposals exist (Sandberg 2014, Metzinger 2021) but institutionalisation is at zero⁶⁷

6.2 For Researchers

For researchers and engineers: open problems by axis.

Axis	Problems worth working on
1. Mapping	Self-correcting segmentation, including barcode-assisted approaches such as PRISM; integrating molecular annotation of transmitters and receptors at synapses; hybrid pipelines combining expansion microscopy, X-ray and EM
2. Readout	Stable long-term recording at 10⁵-neuron scale in behaving mammals; registering recorded and structural data from the same animal; BCI channel density and electrode lifetime
3. Modelling	Systematic validation of the mapping from EM morphology to physiological parameters (MICrONS is the test bed); state-space models of neuromodulation; identifying and implementing biological learning rules
4. Running	Memory and communication design for spiking simulators at 10⁹–10¹¹ neurons; standardising body and environment models (generalising the NeuroMechFly line); automatic mapping onto neuromorphic hardware
5. Preservation	Quantitative evaluation of ASC quality in human post-mortem brains; end-to-end demonstration of preserve → EM → reconstruct; intraoperative monitoring of perfusion quality
6. Validation	Activity-prediction benchmarks portable across species; protocols that eliminate degeneracy through perturbation response; a white paper defining emulation pass criteria

Because the problem is cross-disciplinary, there are entry points well outside neuroscience proper: machine learning (P1, P3), systems software and HPC (axis 4), microscopy, optics and materials (axes 1 and 5), control and robotics (axis 4), statistics and causal inference (axes 3 and 6), and bioethics and law (P5).

6.3 For Students

For students and career changers: a learning path.

Foundations, whichever route you take. A standard computational neuroscience text (Dayan & Abbott, Theoretical Neuroscience), the basics of machine learning, and a close reading of the two central documents in the resources: the 2008 roadmap⁸ and the State of Brain Emulation Report 2025¹. Neuromatch Academy is a high-quality free entry point⁹.
Data and tools you can touch today. FlyWire (explore and analyse the whole fly connectome in a browser)¹⁰, MICrONS Explorer (mouse structure plus function), ZAPBench (enter your own model for whole-brain activity prediction)³, and OpenWorm (contribute code to the nematode simulation)¹¹. Nearly all the field's major data is public — the barrier to entry is doing the work, not holding the degree.
Choosing a lab or organisation. Connectomics (Princeton, Harvard, Janelia, Allen, Google), modelling (Allen, Stanford, Baylor, the EPFL lineage), preservation (21st Century Medicine, Nectome, university brain banks), emulation execution (Eon, Open Brain Institute), theory and ethics (Oxford and philosophy departments generally). Chapter 3's player table doubles as a career map.
Community. Carboncopies' public workshops¹², the Foresight Institute's WBE group, and conferences such as COSYNE and SfN.

6.4 For Funders & Policymakers

For funders and policymakers.

The leverage is in P1–P3. A whole mouse connectome is a multi-hundred-million-dollar, decade-scale undertaking, but the rate limiter is proofreading automation and validation standards rather than instruments. Prize-based funding (the BPF is the working example⁵) and benchmark-based funding are highly efficient even at small scale.
Funding stability is itself a deliverable. The swing in the US BRAIN budget — halved between 2023 and 2025, restored in 2026¹³ — was maximally ill-suited to a decade-long data programme. Multi-year commitments and diversification across public, philanthropic and private sources are the structural prescription.
Public involvement in validating preservation. Preservation services already exist in the market. Establishing scientific verification and audit standards is neither promotion nor prohibition; it is consumer protection, and it is the most urgent policy gap.
Ethics ahead of time. Welfare guidelines for animal emulations, brain-data protection (neurorights), consent design around assisted death. Building institutions is cheaper than building technology, and the social cost of being late is higher.

6.5 For Individuals

For individuals: six realistic options.

1. Understand it accurately. Start with Chapters 2 and 3 and the introductory books in resources. If you can explain why both "certain by 2045" and "already possible today" are wrong, that is a sufficient starting point.
2. Track it. Follow the indicators in Chapter 3's watchlist — the zebrafish release, the 10 mm³ mouse work, Eon's mouse programme — on a yearly basis. Judging by indicators rather than headlines protects against both hype and despair.
3. Participate. FlyWire was completed partly through citizen-scientist proofreading¹⁰, and OpenWorm accepts code contributions¹¹. For donations, the BPF, OpenWorm and directed gifts to university labs are the most transparent options.
4. Keep records (a weak backup). Diaries, audio and video are not a substitute for preserving a brain, but they have limited value as future validation material, as a legacy, and as a behavioural trace. Commercial "digital clone" and "AI memorial" services replay records; they are not uploading, and should be used with that understood.
5. Evaluate brain preservation soberly. See the table and caveats below. This is not demonstrated life extension but a long-horizon bet in which only structural preservation is guaranteed. If you are considering it, satisfy four conditions first: (a) understand the scientific limits (Chapter 3, axis 4), (b) agreement with family, (c) after-death logistics and legal paperwork in order, and (d) a cost structure that does not damage your household finances.
6. Stay alive and healthy. Unglamorous, but in expectation the most reliable strategy is being around when the technology matures. Standard preventive medicine and health habits are a precondition of every upload argument ever made.

Preservation providers, August 2026 (for reference)

Organisation	Method	Region	Indicative cost	Notes
Alcor	Cryopreservation (vitrification)	United States	≈$80K brain / ≈$220K whole body	Founded 1972, the oldest provider. Commonly funded through life insurance
Cryonics Institute	Cryopreservation	United States	from ≈$28K whole body	Low-cost option; transport billed separately
Tomorrow Bio	Cryopreservation (vitrification)	Europe	≈€75K brain / ≈€200K whole body	Europe's first integrated standby and transport operation
Oregon Brain Preservation	Chemical fixation (ASC family)	US, Oregon area	Low cost; free places in-region	Non-profit, research-donation based. Operational use of ASC-family methods
Nectome	ASC (research stage)	—	No service offered	Research only. Assisted-death-compatible protocol preprint in 2026⁴

Costs are approximate figures from each organisation's published information as of 2026 and change over time; always check the official source. This table is informational and is not a recommendation to enter any contract.

6.6 Regional Notes

Regional notes.

United States. Home to most of the frontier — BRAIN CONNECTS, MICrONS, Janelia, Princeton, Harvard, Google — and to the only jurisdictions where assisted-death-compatible preservation protocols are legally conceivable. It is also where funding volatility poses the largest programme risk.
Europe. EBRAINS and the Open Brain Institute inherited the simulation infrastructure of HBP and Blue Brain, giving Europe a distinctive position in modelling and open platforms rather than in raw imaging scale. Tomorrow Bio provides the region's integrated preservation operation.
Japan. The strengths are primate (marmoset) brain mapping and work on consciousness and theory. Brain/MINDS 2.0 (AMED) is building a cross-species digital brain platform¹⁴. Masataka Watanabe at the University of Tokyo proposes a distinct route — connecting a living brain to a machine hemisphere and migrating consciousness and memory gradually — with the startup MinD in a Device pursuing related work¹⁵. The Whole Brain Architecture Initiative serves as a Japanese-language entry point to the computational theory of the brain¹⁶. There is effectively no domestic preservation provider, so that option means an overseas contract plus international transport, and the resulting delay directly degrades preservation quality.
Anywhere. Because FlyWire, MICrONS, H01 and ZAPBench are all open, analysis and contribution are not gated by geography. Accurate translation and explanation in under-served languages is itself a contribution.

6.7 Do NOT

What not to do.

Five things to avoid

1. Do not pay any current service claiming to upload consciousness. As of 2026 no provider can scan or upload a personality. Chatbot-style "digital clones" replay records; they are not a transfer of the person.

2. Do not let hopes about uploading accelerate irreversible medical or end-of-life decisions. Preservation guarantees structural retention only and remains an unproven bet. Nothing in current science justifies letting "sooner is better" distort decisions about the timing of death — which is precisely what makes ethical review of assisted-death-compatible protocols contentious⁴.

3. Do not spread confident dates. Neither "arriving in 2045" nor "impossible forever" has documentary support. Speak in distributions (Chapter 4).

4. Do not run experiments on systems that might have subjectivity without a validation and ethics framework. Welfare and stopping criteria for animal emulations should be designed before the experiment. Metzinger's moratorium argument⁷ is strong, but the precautionary principle behind it is increasingly shared.

5. Do not treat brain data casually. Connectomes and neural recordings are the ultimate personal information. Confirm the protection framework (neurorights) before letting them onto a default path toward sharing or commercialisation.

Chapter summary

The field's rate limiters are five: proofreading automation, structure-to-dynamics inference, validation standards, preservation validation, and ethics infrastructure. From any position, the highest-value action is one that connects to one of these.
Because nearly all the data is open, students and individuals can start today (FlyWire, ZAPBench, OpenWorm).
For an individual the realistic order is understand → track → participate, with any preservation contract treated as optional insurance taken out in full knowledge of the scientific limits.

References