On 8/30/2015 5:42 PM, Russell Standish wrote:
On Sun, Aug 30, 2015 at 07:46:33PM -0400, Jason Resch wrote:
There's roughly a 100x increase in number of neurons, scaling from the
nematode to the fruit fly, to the mouse, cat, and then human. If efficiency
and power of computers for a given cost continue to double, then what it
costs now to simulate a nematode brain, we will in 7 years (assuming
doubling per year) be able to simulate a fruit fly brain. Seven years
later, for the same cost, we will be able to simulate a mouse brain. Seven
years later, a cat brain, and then finally, 28 years later, we'll be able
to simulate neural networks with the same complexity of a human brain.
Historically, the rate of doubling is every 18 months, or 5 years for
an order of magnitude. So your numbers are 10 years to fruit fly, 20
to mouse and 30 to human brain resp.
But computational cost goes up faster than linearly with neuron count,
because the actual cost also depends on axon connectivity. IIRC, this
is something like xlog(x) complexity. So it might be a bit further out
for human complexity.
One intersting test I'd like to see is applying Tononi's integrated
information measure to these simple creatures to see if they're
producing any integrated information. I suspect Integrated Information
is a necessary requirement for conscious, but not so sure about
sufficiency.
Doesn't a NAND operation produce integrated information?
Really? Did you have a cite for this?
It seems we're in
danger of basing our answers to the binary question "is it conscious?" on
purely quantitative, rather than qualitative differences in computations.
How many neurons do you think are required to implement the algorithms you
consider necessary for consciousness?
I haven't fully grokked this, but I would have thought Tononi's
measure was at least normalised by the size of the system in question.
It's normalized by the size of the smallest subsystem. But I wouldn't put in
credence in IIT:
http://www.scottaaronson.com/blog/?p=1799
Bren
--
You received this message because you are subscribed to the Google Groups
"Everything List" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to everything-list+unsubscr...@googlegroups.com.
To post to this group, send email to everything-list@googlegroups.com.
Visit this group at http://groups.google.com/group/everything-list.
For more options, visit https://groups.google.com/d/optout.
