A quick primer in uploading your brain

Amy Harmon has an article in the New York Times about cryonics and immortality that is heartrending and beautiful. Here is what the neuroscientists have to say:

“I can see within, say, 40 years that we would have a method to generate a digital replica of a person’s mind,” said Winfried Denk, a director at the Max Planck Institute of Neurobiology in Germany, who has invented one of several mapping techniques. “It’s not my primary motivation, but it is a logical outgrowth of our work.”

Other neuroscientists do not take that idea seriously, given the great gaps in knowledge about the workings of the brain. “We are nowhere close to brain emulation given our current level of understanding,” said Cori Bargmann, a neuroscientist at Rockefeller University in New York and one of the architects of the Obama administration’s initiative seeking a $4.5 billion investment in brain research over the next decade.

“Will it ever be possible?” she asked. “I don’t know. But this isn’t 50 years away.”

…The fundamental question of how the brain’s physical processes give rise to thoughts, feelings and behavior, much less how to simulate them, remains a mystery. So many neuroscientists see the possibility of reproducing an individual’s consciousness as unforeseeably far off.

“We have to recognize that there are many huge gaps that have to be leaped over,” said Stephen J. Smith, a neuroscientist at the Allen Institute for Brain Science in Seattle. “The brain is holding on to many of its secrets.”

Jeffrey Lichtman, a Harvard University neuroscientist, said, “Nothing happening now is close to a reality where a human patient might imagine that their brain could be turned into something that could be reproduced in silico.”

Count my on the Bargmann side of things (as I am in most things).

  1. Presumably we need to be able to reconstruct the 3D morphology of every cell body, axon, synaptic bouton, and vesicle
  2. Also, reconstruct all of those mysterious glia. What do they do, again?
  3. Is there a consistent learning rule between neurons? I cannot find my normal reference for this, but long-term potentiation, depression, STDP, etc is not the same in every anatomical region. There is a reason most LTP work was historically done in hippocampus.
  4. What about the extracellular matrix? Does it store our memories?
  5.  Don’t forget hormones. They can directly enter cells; think about what we need to know to simulate their diffusion throughout the brain.
  6. Are you planning on sequencing your microbiome as well? How much of the rest of your body contributes to your cognition?
  7. What genes, where, how are they transcribed, when are they transcribed, how do the enhancers interact with etc etc.
  8. Have you ever seen an image of the molecular pathways that contribute to neuronal function? Here is a (likely incomplete) description of the pathway for dopamine in a single cell. Remember, every cell is different and has how many modulatory pathways…?

In other words, we are probably not done solving biology very soon.

In neuroscience, there are more unknown unknowns than there are known unknowns. 40 years for ‘mind uploading’ is not even wrong.

(For the record, I would have my ‘mind’ uploaded the moment it is possible. I do think it is theoretically possible, to some approximation. But sadly, that won’t happen in my lifetime.)

One thought on “A quick primer in uploading your brain

  1. I agree entirely. Connectomics is one piece of the puzzle and there are so so so many more. To your list I’d add all the physiological properties that a given circuit/cell/piece of membrane can have.

    On the other hand, we both have the perfect model in mind when we think about this. Given how crazy perfect is, maybe the better question is how close could a model based only on a connectome get to real life? Is there Turing test equivalent here for determining when a model achieves something that from observations of its behavior is largely indistinguishable from real life? Do we even need a full connectome or could we extrapolate from, say, 10% of the connections to get the gist of the rest (assuming here tha you are building a general model, not a model of a specific individual)?

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s