In 2008, Anders Sandberg and Nick Bostrom of the Future of Humanity Institute published Whole Brain Emulation: A Roadmap. This document served as the first serious technical evaluation of mind uploading, moving the concept from science fiction into the realm of theoretical engineering. It established the “scan-and-copy” approach as the most probable path to substrate-independent minds.

The Three Pillars of Emulation

The roadmap breaks the problem down into three sequential technological capabilities, all of which must mature to achieve success.

1. Scanning (Data Acquisition)

The requirement is to capture the connectivity and functional properties of a brain at sufficient resolution. The roadmap identifies three potential scales:

• Synaptic resolution: Mapping every synapse.
• Ultrastructural resolution: Mapping ion channel distributions and receptor types.
• Molecular resolution: Mapping protein states (likely unnecessary for neural emulation).

Current connectomics efforts, such as the FlyWire project (2025) and ongoing mouse brain mapping, operate at the synaptic resolution. The 2008 roadmap correctly identified electron microscopy (SEM/TEM) as the workhorse for this stage, a prediction that holds true today.

2. Translation (Image Processing)

Raw scan data must be converted into a symbolic representation—a connectome. This involves:

• Tracing axons and dendrites through 3D stacks of images.
• Identifying synapses.
• Classifying cell types.

Sandberg and Bostrom anticipated that AI would be crucial here. In 2026, we see this fulfilled; manual tracing is impossible at scale. Automated segmentation algorithms are now the standard, though error correction remains a significant hurdle.

3. Simulation (Emulation)

The final step is running the reconstructed network on non-biological hardware. This requires:

• Mathematical models of neurons (e.g., Hodgkin-Huxley, Izhikevich).
• Hardware capable of handling massive parallel processing.

Technology Readiness Levels (TRL)

The roadmap introduced the concept of assessing emulation feasibility via TRLs.

• Scale separation: The assumption that biological details below a certain scale (e.g., quantum effects) can be abstracted away.
• Computability: The brain functions as a physical system governed by computable laws.

The Bottleneck Shift

In 2008, the primary concern was often computational power. However, looking at the trajectory from 2008 to 2026, the bottleneck has shifted. We now possess Exascale supercomputers capable of simulating neuron counts exceeding the human brain (in simplified models). The critical constraint remains scanning throughput and translation accuracy. We can run the code, but we cannot yet extract the source code from a physical brain fast enough or cheaply enough to make human emulation practical in the near term.

Official Sources

• Sandberg, A., & Bostrom, N. (2008). Whole Brain Emulation: A Roadmap. Technical Report #2008-3, Future of Humanity Institute, Oxford University.
• Future of Humanity Institute Archive