Watch the Third Part of Noah Hutton's 10-Year Film on a Mapless Quest To Simulate the Human Brain

What's the minimum amount of information needed to simulate a human brain ? Do we need the whole, complete map (known as a connectome) before we begin the (computer, obviously) simulation, or some lesser knowledge threshold from which everything else...

|
Nov 15 2012, 1:27pm

What’s the minimum amount of information needed to simulate a human brain ? Do we need the whole, complete map (known as a connectome) before we begin the (computer, obviously) simulation, or some lesser knowledge threshold from which everything else will follow? This the fundamental question of the Blue Brain Project (aka the Human Brain Project), which, manymanymany years before neuroscientists will ever have that map or close to it (there are around 100 trillion connections in the brain), is simulating away. The neurons are biologically correct within the project (e.g. might as well be real live neurons), but the totality of their connections is far from known; that part should happen naturally. The project’s progenitor Henry Markram has given the Human Brain Project 10 years to succeed.

The idea sounds like a waste. The theory’s leap, that the map will make itself, is too great. And testing that theory will take 10 years of continuous work, assuming it can maintain funding. From Scientific American, here’s the hopeful side.

Markram counters that one can build a realistic virtual brain without understanding the function of every last molecule or pinpointing every single connection in a living brain. In December 2006 Markram and his team simulated a rat’s neocortical column, a small functional unit of brain tissue containing about 10,000 neurons. A rat’s brain has perhaps 100,000 of these columns, whereas the human brain likely has more than two million of them, each of which comprises 100,000 neurons. Markram did not create a connectome for the rat’s cortical column. Instead, he gathered all the available information about how the many different types of neurons in a single column connect and programmed his virtual cells to link up in the same way. Instead of manually connecting his virtual neurons one by one, he’s hoping to get his supercomputers to do the linking on their own, following the same rules—genetic, geometric or otherwise—that govern neural connections in a fleshy brain.

I guess that sounds a bit like Kevin Kelly’s (What Technology Wants) idea of emergent technology, in which the rules themselves along with some initial momentum and sufficiently selective conditions are the tickets to big Earth-changing structures, like brains or internets. In both cases, it’s a super-optimistic view.

New York filmmaker Noah Hutton is making a movie about the project in 10 parts, one for every year. You can find the third part above. Apparently, Hutton got the idea after watching Markram give his idea at a TED thing. It’s nice to see he’s taking a balanced approach — allowing critics, at least — though plenty of neuroscientists would argue the Human Brain Project shouldn’t be getting attention at all. SA has a Q&A with Hutton online right now. A good sample:

Do you think Markram will achieve his ultimate goal of reverse engineering the human brain as a biologically realistic computer model? Do you think that reaching that goal matters?

I don’t have a prediction. I don’t feel I know enough yet to make one. Some do seem to make predictions with their tone of coverage of Blue Brain, and I’m somewhat skeptical of anyone in the press or elsewhere dooming it to fail or painting an excessively grim portrait at this stage without getting specific and staying open. Because Blue Brain is not off in the woods, impervious somehow to legitimate feedback—what I have learned is that they are periodically reviewed by their funders, by an outside panel of experts—they continue to do well and be funded, and continue to move forward. And from spending time there, I can say from my own perspective that the project is made up of very sharp scientists from around the world, and they are very detailed in their thinking—most of all, Henry Markram.

On the question of whether Blue Brain reaching its goal matters, I would say I’m not really in it to be rooting for one thing or another, and I’m not even sure how one would predict whether reaching their goal will eventually matter or not at this early stage. Though if evidence mounts that understanding the brain through Blue Brain’s methods will lead to medical advances in treating diseases or damage to the brain, I am rooting. From a more philosophical point of view, I can say for myself that in making this film, I’m equally interested in success as I am in failure. Because if we have mapped the brain, fully simulated it or otherwise grappled with it comprehensively, and if we’re still not yet understanding it or watching it come alive, what are we missing?

Likewise, I want some damn answers about consciousness and all that stuff too. But I also think there’s something a bit revealing about science itself in the project and its role in the greater scheme of understanding. Markram’s idea isn’t some out-of-thin-air sort of thing, but there’s a leap involved. How do we support leaps and long-shots in this day and age of science getting more expensive and science funding getting smaller? And then, what do we have without leaps?

Reach this writer at michaelb@motherboard.tv.