# Nerves and thinking

**URL:** <http://discourse.iapct.org/t/nerves-and-thinking/7542>\
**Category:** CSG1994\
**Created:** [October 25, 1994, 5:47am UTC](http://discourse.iapct.org/t/nerves-and-thinking/7542 "1994-10-25T05:47:58Z")\
**Posts on this page:** 3\
**Page:** 1

<div class="post-metadata">

**Author:** ![Bruce\_Buchanan1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/ecccb3/32.png) [@Bruce\_Buchanan1](http://discourse.iapct.org/u/Bruce_Buchanan1)\
**Post date:** [October 25, 1994, 5:47am UTC](http://discourse.iapct.org/t/nerves-and-thinking/7542/1 "1994-10-25T05:47:58Z")

</div>

[From Bruce Buchanan 941025.00:50 EDT]

Bill Powers (941024.1105 MDT) writes:

> . . . you mentioned that neural activity involves "brief" events,  
> neural impulses and jolts of chemicals entering neural cell bodies. I  
> think this is looking too closely at the processes . . .

The context of my remarks was that of reporting an article in The  
Economist which I had found interesting. I posted a note about it to (the  
newsgroup) comp.ai.philosophy, where it is my impression many readers  
believe that the brain/mind may soon be modelled by digital computer  
networks and programs. My tacit message was probably that the tasks of  
modelling brain function have not yet been completed, to say the least. At  
the same time I also posted the note to csg-l, where such a message is  
perhaps somewhat irrelevant.

As I noted, the language was that of a magazine with a general readership,  
and it was not my intent to make any case in terms of the details. I  
disclaim responsibility for responses I may have evoked, yet Bill's  
comments as usual provide food for thought. Bill's account of neural  
activity, and analog functions are indeed what I understand from B:CP and  
make good sense to me.

> So what do the astrocytes do? Anything you can imagine, depending on how  
> they're connected to each other and to the main neural system. Could  
> they have to do with memory? With consciousness? With life support for  
> the neurons? With reorganization? Sure, any of the above and anything  
> else you can think up. By themselves, they probably don't "do" anything...

Agreed. (But then there is likely nothing in the nervous system that  
operates in isolation anyway.)

> "Thinking" is such a broad term that it would be hard to make any  
> general statement about its relation to nerve function. . . .

Agreed; (as I used it, this was part of a quote in "ordinary" not  
scientific language, as previously noted.)

Cheers!

Bruce B.

(P.S. I will be away on a trip for 3 weeks but hope to follow csgnet  
discussions when I return.)

---

<div class="post-metadata">

**Author:** ![William\_T\_Powers2](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/w/c67d28/32.png) [@William\_T\_Powers2](http://discourse.iapct.org/u/William_T_Powers2)\
**Post date:** [October 24, 1994, 7:03pm UTC](http://discourse.iapct.org/t/nerves-and-thinking/7542/2 "1994-10-24T19:03:10Z")

</div>

[From Bill Powers (941024.1105 MDT)]

Bruce Buchanan (941024.00.10 EDT)--

> My interpretation of the \*question\* is: how does activity of nervous  
> tissue relate to thinking, what are the structures involved and how do  
> they function? This may have a bearing upon the characteristics of  
> models of the processes involved. The point is that the new findings  
> introduce more possibilities for complex multidimensional functions.

I can give a general answer but not a specific one. In your comments of  
yesterday you mentioned that neural activity involves "brief" events,  
neural impulses and jolts of chemicals entering neural cell bodies. I  
think this is looking too closely at the processes, like trying to  
explain an electronic circuit in terms of individual electron  
trajectories.

I can make more sense of the system by thinking in terms of neural  
frequencies (impulses per second) and chemical concentrations that are  
adjusted upward and downward by changing the frequency with which  
neurotransmitters are sent across synaptic junctions. Inside the  
receiving cell body, concentrations rise and fall as a consequence of  
continuing parallel streams of incoming neurotransmitters, both  
excitatory and inhibitory, entering the cell in various locations and  
contributing to the net concentrations around the axon hillock, where  
outgoing impulses are generated. Most neurons do not respond to single  
impulses; rather, the chemical concentrations near the axon hillock  
adjust electric potentials and thus govern the rate of firing of the  
neuron. This doesn't work for all types of neurons, but it's a useful  
general picture.

With that basic picture, we can see neural activity as involving  
continuing streams of impulses going from one location to another, and  
continuously-variable chemical concentrations inside neurons which make  
the frequency of the output stream of impulses (one output per neuron)  
depend on the frequencies of multiple input streams. Each neuron thus  
acts as an analog computer, not a digital computer.

My mental model of the system identifies these ongoing activities as the  
objects of awareness. Since I don't know what awareness is or how it  
works, this is strictly a working hypothesis. My image comes from analog  
computing, a branch of computing that has unfortunately become almost a  
lost art. In an analog computer the computations are done without  
benefit of symbols. The signals themselves are the variables which show  
the results of the computations, and the computations are physical  
processes which have the same form as the physical processes being  
modeled (they are analogs of the modeled physical processes).

To model the response of a mass to a force, for example, we represent  
the magnitude of the force by one voltage and the velocity of the mass  
by another voltage. The first voltage enters a computing element where  
it is converted into a current running into a capacitor; the voltage on  
the capacitor is the integral of the input current in amperes, divided  
by the capacitance in farads. Thus the output voltage on the capacitor,  
the second voltage, is the time-integral of the input voltage. The  
voltages are related just as velocity and force are related, with  
capacitance corresponding to mass.

Now if the input voltage is made to vary as a force is observed to vary,  
the output voltage will at all times have a magnitude that corresponds  
to the velocity of the mass. We can predict what the velocity of the  
real mass will be if we apply a force to it that varies in any complex  
way with time, just by applying a corresponding voltage to the input of  
this little analog computer and recording the changes in output voltage.

When I use an analog computer, I must use a voltmeter or an oscilloscope  
to observe the inputs to the computation and the outputs from it. I must  
apply the proper conversion factors so I know, for example, that an  
output voltage of 1.5 volts means a velocity of 30 meters per second,  
and an input voltage of 2.0 volts means an applied force of 1.0 newtons  
of force, with the capacitance in farads corresponding to the mass in  
kilograms.

The important fact about using an analog computer is that the observer  
knows about the results of the computation only by receiving information  
about the behavior of the voltages. Even if the observer is not  
observing, the computation is still going on.

So this is my image of the relationship of the aware human observer to  
the operation of brain functions, at least at the lower levels. The  
analog computations go on continuously, input signals being converted  
into output signals according to the forms of the analog computing  
functions that apply. The output signals become inputs to other analog  
computations. And all these signals become the world that the Observer  
experiences, just like the user of an analog computer who is applying  
voltmeters to the wires to see what the computer is doing.

As to "The point is that the new findings introduce more possibilities  
for complex multidimensional functions", we don't have to look far to  
find the possibilities of as much complexity as we can grasp. Each  
neuron is an analog computer, and the individual computers can be hooked  
up in any possible way to do anything you can define. In a way we're  
like a kid who is poking around a huge electronics lab. The kid picks up  
a resistor and asks "What does this thing do?" The electronic designer  
is hard-pressed to answer. From one point of view it just determines how  
much current will flow when a given voltage is applied to it. From  
another, it fixes the gain of an operational amplifier, or the time-  
constant of a filter, or the output of a voltage divider, or limits the  
heat dissipation in another component -- the list is endless. What it  
does depends on what it's used for, and its relationship to all the  
other components in the whole system. By itself it doesn't do anything  
interesting.

So what do the astrocytes do? Anything you can imagine, depending on how  
they're connected to each other and to the main neural system. Could  
they have to do with memory? With consciousness? With life support for  
the neurons? With reorganization? Sure, any of the above and anything  
else you can think up. By themselves, they probably don't "do" anything,  
any more than a resistor does.

This reminds me, by the way, of how some people talk about  
neurotranmitters. You can tell that they have never had much to do with  
system design. A neurotransmitter is just a way of connecting one  
variable to another. The fact that neurotransmitters of a certain type  
are found in certain places in the brain doesn't tell us anything about  
function -- it just tells us that there is some neurotransmitter there.  
Instead of worrying about what the neurotransmitter "does", we should be  
looking at how it participates in the circuits in a certain place. It's  
like noticing that some parts of an electronic circuit use electrons to  
carry signals, and others use holes, except that in the brain there are  
a dozen or so types of message-carriers. The fact that you find a lot of  
electrons flowing in a given part of a circuit doesn't mean that the  
electrons explain what that part of the circuit does, or "cause" the  
kind of behavior we see. The electrons in an audio amplifier are  
certainly necessary for the amplifier to work, but the connections, the  
organizations of circuit relationships, tell us what the circuit \_does\_.  
The same circuit could be designed to use holes as the signal carriers,  
and it would still do the same thing. I imagine that the brain uses an  
assortment of neurotransmitters mainly as a substitute for insulation --  
to keep signals from one system from leaking into another system and  
interfering with its operation.

> **···**
>
> ---------------------------------------  
> "Thinking" is such a broad term that it would be hard to make any  
> general statement about its relation to nerve function. Do you mean  
> running sentences through your mind? Mathematical formulas? Estimating  
> how a swatch of color will go with the living-room decor? Wondering  
> whether your egg salad sandwich is swarming with Salmonella? Considering  
> the opening chords of Grieg's piano concerto? The term thinking covers  
> just about any covert activity in consciousness, and I don't think it  
> can be associated with any one kind of brain function. I've tried to  
> substitute semi-technical terms for general words like thinking, so we  
> can at least attempt to characterize the kind of functions that are  
> going on.  
> -----------------------------------------------------------------------  
> Best,
> 
> Bill P.

---

<div class="post-metadata">

**Author:** ![Bill\_Leach1](http://discourse.iapct.org/letter_avatar_proxy/v4/letter/b/848f3c/32.png) [@Bill\_Leach1](http://discourse.iapct.org/u/Bill_Leach1)\
**Post date:** [October 26, 1994, 3:14am UTC](http://discourse.iapct.org/t/nerves-and-thinking/7542/3 "1994-10-26T03:14:54Z")

</div>

\<[Bill Leach 941025.20:06 EST(EDT)]

> [Bill Powers (941024.1105 MDT)]

In your response to Bruce, I could not help but again lament the lousey  
timing with respect to the integrated circuit operational amplifier vs  
the integrated circuit digital logic gate.

I am afraid that the advent and rapid growth of the digital computer  
(with its supposedly statically verifiable performance) doomed the analog  
computer at just the time when the industry had finally produced circuits  
that worked in practice.

I was fortunate (in my opinion anyway) to experience an application where  
an analog computer was able to perform a task that was many orders of  
magnitude beyond even a super-computer (I suspect the situation would  
even be the same today).

The computer had to model the performance of a nuclear reactor core using  
real time reactivity parameters for a transient burst power reactor. A  
Cyber 205 required nearly a week's worth of processing time to estimate  
that the test would be safe... a test that ran for a total of  
approximately a third of a second.

-bill