Comments on: Will We Ever… Simulate the Brain?

By: Roedy Green

Roedy Green — Sun, 24 Feb 2013 15:56:51 +0000

Circa 1990 I met Bernard C. Till, who for his masters thesis has simlated the nervous system of a nematode. He has mapped the 47 (149?) neurons of the beast, the coonections and the “S functions” (the key to the magic) of the various connections. What utterly blew my mind is the simlated worm exhibited EVERY known behaviour of the worm.

If I has tried to create such a simlation directly with a computer program it would have taken many megabytes of code.

Along simlar lines I learned that humans are specied by something like 18,000 genes. Computer programmers would require terabytes of design documents to specify a human, and its biochemstry, embryology etc. We have a lot to learn form nature’s terseness.

By: Stan Sandler

Stan Sandler — Sat, 23 Feb 2013 13:12:03 +0000

The worker bee brain has only about 1 billion neurons. With that it can control its body, interpret its senses, do all the tasks in a beehive (sequentially as it ages), function as a member of a complicated society (communicate…) and a great amount of learning (spatial mapping of nearly a thousand hectares, learn mazes in only 6 times the time of a rat….). So why start with something the size and complexity of the human brain.

By: Stephen Minhinnick

Stephen Minhinnick — Mon, 18 Feb 2013 10:41:12 +0000

“The project is impressive but might leave people baffled that someone would spend a lot of time and effort building something that doesn’t do anything,” says Chris Eliasmith.

What does the Large Hadron Collider “do”? Sounds like Henry Markram plans to build something similar to work out the underlying details of the working brain. You could call it the “Large Neuron Collider”.

By: Jeff

Jeff — Mon, 18 Feb 2013 08:57:42 +0000

The money for this grant was meant for a high-risk, potentially high-reward project. Someone with ambition is wanting to try to tackle a grand challenge in a bold way. Let them. We could sit around for several more centuries and debate about whether it’s going to help or not, or we can applaud those last few people in the world who actually TRY TO DO SOMETHING.

My god, I’m so glad our generation wasn’t tasked with exploring/settling the Americas, or fighting the world wars. Nothing would have gotten done, as everyone would want to debate it for a lifetime before getting started.

By: Roedy Green

Roedy Green — Mon, 18 Feb 2013 06:18:12 +0000

Back in the 70′s I wrote OPTOW, a computer program to design high voltage transmission lines. Engineers came by to laugh at its efforts. Every week I steadily improved it. One week it was as good as a human, the next week it was 10% better. 50 engineers with PhDs and masters degrees were out of a job. Psychologically we adjudicate capability even a bit below our own as hopeless. We also fail to take into account artificial intelligence constantly improves, unlike our own. I think AI will eventually take the world by surprise, just as OPTOW did, for the exact same reasons.

By: Roedy Green

Roedy Green — Mon, 18 Feb 2013 06:01:40 +0000

We are vain. Our large brain took a twinkling to evolve compared with the snail brain. It can’t be more than half a dozen tricks. There was not time for more subtlety.

By: Chris

Chris — Mon, 18 Feb 2013 03:56:25 +0000

I think it’s disingenuous to call the brain “the most powerful computer in the world”. It functions with great incoherence, noisyness and distraction. On a macroscopic level it is not that deterministic and works nothing like the Turing Machine – the accepted mathematical definition of a computer. No human can outperform a modern computer at speed in strictly following an algorithm to its termination. One must therefore not compare computer speed/memory with the brain, but instead compare neural matter’s capabilities with simulated neural nets. I reckon we are nearly capable of simulating smaller brains on current machines but we require better computer models more so than better hardware.

By: John Kubie

John Kubie — Sat, 16 Feb 2013 19:41:18 +0000

the notion of bottom up and top-down modeling isn’t new. People have been modeling single neurons and small networks of neurons for decades; and slowly adding specific cellular features, At the same time, people have been modeling larger networks of neurons with simpler neurons. And, at the same time, people have been making robotic devices using sensory systems, motor systems and computations somewhat similar to brain computations for awhile. Each of these approaches is advancing, and the approaches converging. Two quick points:
1. Modeling simple invertebrate nervous systems has advanced rapidily, but remains greatly simplified comared to the real thing. These will ‘work’ well before a human brain is simulated. I feel we have to understand how the brain evolved, before trying to build a complete one.
2. I feel trying to build a brain in a vat is wasted effort. A mature human brain gets that way through a develpmental sequence that requires extensive interaction with the environment. this can’t be short-circuited. My strong hunch is that the first close approximation to a functional brain will be a robot, not a computer.

By: Joonjeong Yi

Joonjeong Yi — Sat, 16 Feb 2013 15:05:01 +0000

The idea of Markram will be failed since any computer cannot simulate nuerons interconnected differently for each individual’s brain connnectome what Dr. Sebastian Seung explained in TED.

By: Johnny O

Johnny O — Fri, 15 Feb 2013 15:35:20 +0000

Why are they starting at “the top”? How about dealing with “the basics” first – like eating, sleeping and sh*tting? That was how the brain started: simply neurological functions had to be established before anything complex could happen.