Mouse Versus Supercomputer: No Contest
Fanny More  |  by blog.wired.com. All rights reserved. 12.07 | 1:21

IBM's BlueGene/L supercomputer contains 4,096 processors, each tricked out with 256MB of RAM. It crunches out 360 trillion floating point operations per second. But as University of Nevada researchers have discovered, it doesn't hold a digital candle to the mind of a mouse:

The mouse cortex has roughly 8x10⁶ (8m) neurons and 8,000 synapses per neuron, state researchers James Frye, Rajagopal Ananthanarayanan and Dharmendra S Modha.

Assuming an average firing rate of 1Hz, the entire memory must be refreshed every second, each neuron must be updated at every simulation time step, and each neuron communicates to each of its targets at least once a second. As the team says in a gloriously deadpan way: Modeling [sic] at this scale imposes tremendous constraints on computation, communication and memory capacity of any computing platform.
However, even this huge processing effort still only managed to run at a speed 10 times slower than real time, and only for 10 seconds - the equivalent of one second of mouse-thought.

Barely enough time for a mouse to register a hungry cat in the vicinity.

What's more, a mouse brain has just eight million neurons. A human brain, by comparison, has 100 billion, but that hasn't stopped Swiss researchers from trying to model one.


Would such a supercomputer develop consciousness? The Lausanne researchers are refreshingly candid on this: We really do not know, they say. If consciousness arises because of some critical mass of interactions, then it may be possible.


It is your line of thought that gave us religion. Oh, the vanity of Man! Consciousness is far too complex to understand, much less recreate.

Let's create a deity to explain it! Awesome call, dude.

What if consciousness is not in the brain.

.. but rather, the brain, and all else, is IN consciousness?


"It is a misconception that every neuron in the brain is active at all times. The brain can only focus on a few things at a time. If grandma comes into view, neurons associated with say, elephants or horses do not wake up.

"
1. All neurons have a baseline firing rate of action potentials, which is random and not due to stimulation.
2.

Saying that neurons are associated with certain objects is misleading. There is no 'grandma' or 'horse' neuron. Instead there is more of a chain reaction of many many different neurons in many different areas of the brain.


Basically...

the brain is very complicated, all the time.

Read more on by blog.wired.com. All rights reserved.
Keywords: In The
Related news
Post comments
Name
Place
3 + 4 =
Comments