State of
Brain Emulation
2025

Despite impressive progress in neuron recording capabilities, neuroscience has not yet achieved whole-brain recording (≥ 95% of neurons and brain volume) at single-neuron resolution in any organism. The closest achievements include larval zebrafish with approximately 80% brain coverage and C. elegans with roughly 50% of nervous system neurons recorded at single-cell resolution.

Even these figures come with substantial limitations: temporal resolution is typically well below neuronal firing rates (often 1-30 Hz for calcium imaging), recording durations remain short (minutes to hours), and the need for head-fixation severely constrains behavior repertoires. In larger organisms like mice, recordings focus on cortical regions or specific brain areas rather than whole-brain coverage.

Complete connectomes at synaptic resolution currently exist only for small organisms. C. elegans has multiple whole-nervous-system reconstructions from individual specimens, with approximately ten datasets available. Adult Drosophila has fully proofread connectomes for both the male central nervous system and the female brain.

For larger organisms, progress remains at the proof-of-concept stage. In mice, the largest densely reconstructed volume is a cubic millimeter of visual cortex, containing approximately 120,000 neurons and 523 million automatically detected synapses. In humans, the largest synaptic-resolution volume is approximately 1 mm³ of the temporal cortex (0.00007% of the whole brain).

Meaningful progress toward whole-brain emulation is currently confined to small organisms where comprehensive datasets are becoming available. In C. elegans, multi-scale, closed-loop simulations now reproduce basic behaviors by integrating neural dynamics, body mechanics, and environmental interaction.

For Drosophila, the adult connectome has enabled models spanning the entire brain, successfully predicting neural responses and circuit functions for behaviors like feeding and grooming. In larger organisms like mice and humans, comprehensive emulation remains at the proof-of-concept stage, demonstrating building biophysically detailed cortical circuits or running human-scale simulations on supercomputers as feasibility tests.

Brain emulation aims to computationally replicate neural dynamics, from perception to decision-making. This requires three integrated capabilities: recording brain function in vivo, mapping brain structure ex vivo through connectomics, and instantiating accurate models in silico. The report examines progress across five model organisms, focusing on electrical activity at single-neuron resolution, synaptic connectivity, and the computational frameworks needed to integrate this data.

Key terminological distinctions for the report. A simulation matches a target system's outputs without necessarily reproducing internal causal dynamics; an emulation matches outputs by implementing the same internal dynamics at a chosen biophysical level. A minimal brain emulation must cover approximately all neurons, be constrained by an accurate synaptic-level connectome, model cell type diversity, use at least point neurons, and operate at the timescale of neuronal spiking. "Whole brain" is defined as incorporating at least 95% of neurons across 95% of brain volume.

Methods for recording neural activity include EEG (millisecond temporal but poor spatial resolution), fMRI (whole-brain but indirect and slow), electrophysiology with Neuropixels probes (spike-resolution but sparse sampling), functional ultrasound (sub-100 μm in rodents), and optical methods like calcium/voltage imaging with two-photon or light-sheet microscopy (large-scale but depth-limited). Optogenetics enables precise perturbation experiments to map causal relationships. Neural data repositories are growing rapidly, with initiatives like BIDS and NWB promoting standardization.

Electron microscopy remains the primary modality for synaptic-resolution imaging, with multi-beam systems parallelizing acquisition and AI dramatically reducing segmentation costs. Alternative approaches include expansion microscopy (enabling molecular annotation) and X-ray microscopy (faster imaging of thicker sections). Dataset sizes now reach petabytes—a mouse brain at 10nm requires ~0.5 exabytes, a human brain ~1-1.4 zettabytes. Automated proofreading remains the cost bottleneck, though recent AI methods have improved error rates by an order of magnitude.

Models span multiple levels of abstraction, from simple integrate-and-fire neurons to biophysically detailed multi-compartment models. Connectome-constrained models use structural data to define network architecture, while data-driven approaches fit parameters to functional recordings. Simulating mammalian brains requires substantial resources: a mouse brain needs ~1-2 TB memory and ~5-10 PFLOP/s; human-scale requires ~1-3 PB memory and ~10 EFLOP/s. Memory capacity and interconnect bandwidth are now the primary bottlenecks.