Cubic millimetre of brain mapped at nanoscale resolution

Reactions to the visualization

• Many find the interactive 3D viewer stunning and immersive; zooming and neuron highlighting are praised.
• Several report strong visceral reactions: awe, eeriness, “spiritual” feelings, or discomfort at seeing the physical substrate of mind.
• Some are unsettled by the idea that this tissue once belonged to a living person and that “this mess of fibers” implements consciousness.

Brain complexity and intelligence

• The density and tangled connectivity in just 1 mm³ reinforces a sense that brains are extraordinarily complex, perhaps the most complex known systems.
• Debate over whether a brain can ever fully understand itself; some argue we lack even a clear definition of “understanding.”
• Comparisons are made between human and bird brains; bird brains are noted as more neuron-dense and tightly packed, with implications for speed and efficiency but constraints on size.

Imaging, slicing, and reconstruction

• Sample prep: staining, embedding in resin, then ultra-thin slicing with diamond knives and automated tape collection (microtome/ATUM).
• EM resolution is discussed in the context of Feynman’s “plenty of room at the bottom”; progress is acknowledged but atomic-level detail is still challenging.
• Reconstructing 3D volumes and segmenting cells/synapses relies on ML; this is state-of-the-art but still requires laborious human proofreading, and only part of the dataset is fully checked.

Data volume, storage, and whole-brain emulation

• Back-of-the-envelope extrapolation: mapping a whole human brain at this resolution would require zettabytes of storage, far beyond any single system today.
• Some argue much of this data might be redundant for “mind uploading”; others respond we don’t yet know what can be discarded and even simple AI models are poorly understood.
• There is skepticism that whole-brain EM or uploads will arrive soon; even C. elegans remains poorly understood despite its tiny, mapped nervous system.

AI vs biological brains

• Rough parameter comparisons suggest human brains have orders of magnitude more “parameters” than current LLMs, even under very simplified assumptions.
• Others note neurons, dendrites, and even proteins perform complex local computation, so these estimates are likely still huge underestimates.
• Some see LLMs as surprisingly efficient per “connection”; others highlight their lack of embodiment, math rigor, and genuine understanding.

Philosophical and existential themes

• The dataset triggers reflections on consciousness as “just” physical wiring versus something more; some feel this diminishes human “magic,” others say the substrate doesn’t lessen the wonder.
• There is broad agreement that understanding intelligence will require advances in complexity theory and that both connectomics and AI alone are insufficient.