State of Brain Emulation Report 2025 and how far the Sandberg Bostrom roadmap has moved
The State of Brain Emulation Report 2025, posted to arXiv on 17 October 2025 and updated in November, is the first systematic reassessment of the field since Anders Sandberg and Nick Bostrom published their Whole Brain Emulation roadmap in 2008. The report was authored by Niccolò Zanichelli, Maximilian Schons, Isaak Freeman, Philip Shiu, and Anton Arkhipov. It organizes the field around three capabilities, recording brain function through neural dynamics, mapping brain structure through connectomics, and emulating and embodying through computational neuroscience. The report’s central judgment is that progress has been uneven across the three, and that the bottleneck has shifted from measurement to computation.
The report matters for consciousness research because brain emulation is the strongest concrete test of the substrate independence claim. If consciousness is a property of causal organization, then a faithful emulation of a brain’s causal organization should, in principle, carry the same consciousness. The report does not take a position on that claim. What it does is report how close the field has come to being able to run the experiment.
The three capabilities
The report reorders the 2008 roadmap into three capability frontiers.
Recording brain function, which the report calls Neural Dynamics, is the capacity to capture the temporal activity of a nervous system in enough detail to reconstruct it. The report’s assessment is that this remains the hardest measurement problem, because the scale required, every neuron and every spike over a behaviorally meaningful window, is beyond current technology for any animal with a brain larger than a fly.
Mapping brain structure, which the report calls Connectomics, has advanced the furthest since 2008. The fruit fly connectome was completed, and the mouse cortex and other mammalian structures have made documented progress. FlyWire, the largest completed connectome, generated a wiring diagram of a fly brain, and the report treats it as the current proof that connectomics at full scale is achievable.
Emulating and embodying, the Computational Neuroscience capability, is where the report places the frontier. A wiring diagram alone does not initialize a functioning network. The emulation requires biophysical models, which require parameters, which require data the recording capability has not yet supplied.
How far the roadmap has moved
The 2008 roadmap estimated the requirements for whole brain emulation and ranked eleven levels of emulation detail, from a scanned wiring diagram to a full biophysical simulation. The 2025 report’s assessment is that no level of the human roadmap has been completed, and that the field has instead built credible prototypes at much smaller scale. The highest fidelity example the report and this site track is the Eon Systems fruit fly, 125,000 neurons in a physics simulated body.
The report’s structural conclusion is that the field has learned the roadmap’s cost estimates were too optimistic at the top end and too pessimistic about the value of small scale work. The mapping capability has proven tractable. The recording capability remains the gap, and the simulation capability is bounded by the first two. That is a hard result, and it is the reason the human clinical BCI work matters. Every cortical interface that records more channels narrows the recording gap the report identifies as the binding constraint.
The consciousness connection
The report does not claim to assess consciousness, and that silence is itself informative. Brain emulation as the report defines it is engineering, how to reproduce the brain’s function at a specified level of detail. Whether the reproduction is conscious is a separate question, and the report’s authors decline to speculate, which is the correct scientific posture. The site’s own brain emulation section separates the engineering question from the consciousness question in the same way, and the Sandberg-Bostrom emulation levels provide the vocabulary for doing so.
For the indicator checklist, the report’s value is that it supplies the measurement conditions. The indicators are only testable if the system under test reproduces the relevant causal organization, and the report states how far current emulation is from doing that for any mammalian brain. The distance is empirical, not metaphysical, which means it can shrink.
Comparison to The Consciousness AI
The Neutral Core project’s own emulation plans sit at the low end of the roadmap, and that is where an open source project can operate today. The project’s Neutral Core architecture simulates a multi-layer conscious architecture, and its spiking neural network work is exactly the kind of small scale simulation the report treats as credible. The project does not claim to be building toward a human emulation. It claims to be testing the indicators on a cheaper substrate, which is the method the report implicitly endorses by showing that full scale is not currently reachable.
The report’s most useful contribution to the site is its framing of the bottleneck. The site’s brain emulation roadmap and connectomics coverage each tracked part of the field. The 2025 report integrates them into a single assessment, and it gives the site a dated, citable benchmark for where the field stood. That is exactly what a research publication needs, a reference point that later posts can measure against.
*Niccolò Zanichelli, Maximilian Schons, Isaak Freeman, Philip Shiu, and Anton Arkhipov authored the State of Brain Emulation Report 2025, posted to arXiv as 2510.15745. Niccolò Zanichelli and Philip Shiu are independent researchers in the brain emulation community. Anton Arkhipov is a researcher working on neural simulation and the AIBRAIN program.
Researchers covered here
-
Anders SandbergFuture of Humanity Institute, Oxford UniversityCo-author of Whole Brain Emulation: A Roadmap (2008) with Nick Bostrom, computational neuroscience, and technological futures
-
Nick BostromOxford UniversitySuperintelligence (2014), the Simulation Argument (2003), and co-authoring Whole Brain Emulation: A Roadmap (2008)
- Anton ArkhipovAllen Institute for Brain ScienceLarge-scale neural simulation, biophysically detailed cortical models, AIBRAIN program, State of Brain Emulation Report 2025