home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.ai.neural-nets
- Path: sparky!uunet!cis.ohio-state.edu!pacific.mps.ohio-state.edu!linac!uchinews!ellis!kps2
- From: kps2@ellis.uchicago.edu (kenneth paul scholz)
- Subject: Re: Neural Nets and Brains
- Message-ID: <1992Jul24.162237.15135@midway.uchicago.edu>
- Sender: news@uchinews.uchicago.edu (News System)
- Reply-To: kps2@midway.uchicago.edu
- Organization: University of Chicago Computing Organizations
- References: <arms.711935064@spedden>> <BILL.92Jul23224539@ca3.nsma.arizona.edu> <arms.711986585@spedden>
- Date: Fri, 24 Jul 1992 16:22:37 GMT
- Lines: 39
-
-
-
- Bill Armstrong wrote:
-
- >In sum: to study the brain, in my opinion, based on ignorance of how
- >the brain works, is that one should study a system that works on
- >logical signals. Although ALNs might help in this, they are not
- >designed as a brain model at all, and different learning algorithms
- >would have to be developed to capture the learning properties of
- >neurons.
- >
- >Are we getting closer?
-
- IMHO a system that works exclusively on logical signals would not
- approximate the function of the brain or even a relatively simple
- neuronal circuit. Let me use an example to stress my point, the
- monosynaptic reflexes in the spinal cord. When you touch a hot object,
- your sensory neurons fire and synapse directly onto motor neurons. The
- motor neurons activate muscle cells that cause withdrawal from the
- hot object. This form of reflex action could probably be modeled
- by a logical NN fairly effectively. However, you have now 'learned'
- that the object is (an can be in the future) hot. This learning
- might involve a number of modulatory actions (whatever that means) in
- the brain, which are often not logical processes. Yes, the end effect
- of these modulatory actions *might* be to increase or decrease firing
- rate of some neuron, but the 'memory' of the learned event is likely
- to be stored as a graded function at some synaptic connection.
-
- Someone earlier quoted a scientist as saying they would like to trace
- all the signals through the brain to examine how the signal is processed.
- If this was done just by recording action potential frequency, well, yes
- you might get something analogous to a Thevenin equivalent circuit, but
- you would miss alot. In particular, you would miss the underlying
- mechanisms that the brain uses to alter neuronal activity. Thus, you
- would probably miss many of the mechanisms that the brain uses to 'learn'
-
- --
- Ken Scholz Department of Pharm. and Physiology
- kps2@midway.uchicago.edu Univ. of Chicago
-