Re: Why do we need definitions?
From: Just Playing (gms2004_at_lycos.com)
Date: 06/22/04
- Next message: zerkanX: "Re: Why do we need definitions?"
- Previous message: Markus Triska: "Re: Why do we need definitions?"
- In reply to: Wolf Kirchmeir: "Re: Why do we need definitions?"
- Next in thread: Acme Diagnostics: "Re: Why do we need definitions?"
- Reply: Acme Diagnostics: "Re: Why do we need definitions?"
- Messages sorted by: [ date ] [ thread ]
Date: 22 Jun 2004 07:52:13 -0700
Just Playing
Sorry, I guess I did not express myself clearly.
I was wondering more about the amount of information we can process
while taking a written test, like in school or for taking a driver
license.
At the same time I was looking at this amount of information, maybe
expressed in bytes per unit of time, and its connection to the
intellectual concepts we use.
And next I was interested how all these limitations reflect on our
daily life, in our discussions.
Just Playing
>
> A neuron responds in the 100s of milliseconds. IOW, a train of pulses
> can be repeated about 10 times per second. (Eg, the neurons (light
> receptors) in the eye have a cycle time of about 1/10th of a second; the
> illusion of motion in movies and videos depends on this.) The question
> is how to interpret the neural response as a bit rate. AFAIK, the only
> variation in neural response is the rapidity of the pulses within a
> train, which suggests that a train carries two bits (fast/slow). Thus a
> neuron operates at a bit rate of about 20.
>
> However, the connection topology of the neurons guarantees that the
> processing rate of the brain as a whole, or of subsets of it, will be
> many times higher. Consider, for example, that small clusters of neurons
> in the visual cortex respond to specific features of the visual field
> (sometimes, a single neuron does so.) EG, most of us can pick out a
> sought for face in a crowd in a matter of a few seconds at most. This
> corresponds to a bit rate of several megapixels a second, as measured by
> the (linear) processors used in computers. The usual phrase used to
> describe this aspect of the brain is "massively parallel," an
> unfortunate term IMO, since it implies a two dimensional array or tree.
> In fact, the connection topology of neurons is multidimensional, since
> many neurons are connected to many neurons. (See the concept of
> fractional dimensions to understand how a network embedded in 3-space
> can have a dimensionality higher than 3.)
>
> As for what all this has to do with the human capacity for symbolism: I
> don't think bit-rates have anything to do with it. It's the topology of
> the network that has something to do with it. IMO, at the level of brain
> function symbols are specific firing patterns of specific networks,
> whose activity is stimulated by inputs from elsewhere in the brain (and
> ultimately, in some but not all cases, by input from sensors.) If this
> guess is valid, we ought to find symbolic behaviour in any system with
> a sufficiently complex neural topology. Some people claim to have
> observed evidence symbolic processing in apes (as distinct from
> conditioned responses to arbitrary shapes and colours); if true, such
> observations imply that apes have brains roughly as complex as humans.
> OTOH, it seems no artificial system is complex enough to exhibit
> symbolic behaviour.
>
> BTW, mere "connectivity", as measured in average number of connections
> per neuron, doesn't tell us much about the topology of the network.
- Next message: zerkanX: "Re: Why do we need definitions?"
- Previous message: Markus Triska: "Re: Why do we need definitions?"
- In reply to: Wolf Kirchmeir: "Re: Why do we need definitions?"
- Next in thread: Acme Diagnostics: "Re: Why do we need definitions?"
- Reply: Acme Diagnostics: "Re: Why do we need definitions?"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
|
