X-Message-Number: 21412
Date: Sun, 16 Mar 2003 13:25:15 -0800
From: Pat Clancy <>
Subject: Re: CryoNet #21399 Conciousness
References: <>

> In that view, a thermostat has no conciousness, a computer neither, an 
> ordinary neural network is not concious, but it could be built with multi 
> reset frequencies and then be trained to simulate other similar networks. 
> That system would be conscious.
> 

A "neural network" is just another type of algorithm, basically a kind
of fancy pattern matching (I've implemented them). It has nothing
whatsoever to do with consciousness, nor is there any reason to think
that a nn-based algorithm would be any more suitable to "implement"
consciousness on a computer than any other algorithm. The term "neural
network" is in fact quite misleading, as it leads many people to think
that it is somehow "brain-like".

When you say it could be trained to simulate "similar" networks, you
presumably mean other "networks" that are conscious (otherwise what's
the point?), of which the only known instance is the brain. However, to 
train a neural net algorithm you need a (multi-dimensional)
point of output data to correspond with every point of input data,
covering all possible inputs. I.e. even assuming the inputs and outputs
of neurons could be simulated digitally (which they can't), you'd need
data covering all possible inputs to every neuron and the corresponding
outputs. So, let's see, the human brain has on the order of 10^15
synapses. Even if each synapse could have only 2 states (which is not
the case), that's 2^(10^15) states, which is slightly larger than
10^10^14 (that's 10 following by 10^14 zeroes).  So the size of the
training data set is on the order of this size. A neural net will need
on the order of thousands of training iterations; let's assume just
1000. Now assuming each training iteration took just a nanosecond
(highly unrealistic), that's (10^3)*(10^(10^14))*(10^(-9)) =
10^((10^14)-6) seconds, still approx. 10^10^14 seconds. Assuming the age
of the universe is 10^10 years, the time to train the network is about
10^12 times the age of the universe. And even if you could build and 
train it, my guess is that all you'd have is a computer-based digital 
simulation of the neuron state-space, and it wouldn't be remotely conscious.

> To conclude: Our machines are not conscious but we could build conscious 
> system without too much problem. Conciousness is not an emergent process 
> here, it is an evolution product selected because of this survival value.
> 

Well the obvious response when someone says something is easy to build
is to ask them why they haven't done so.

Pat Clancy

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=21412