Rickert on embedded computation (was re: science of consciousness.)

Neil Rickert rickert at cs.niu.edu
Wed Apr 29 10:32:48 EST 1998


andersw+ at pitt.edu (Anders N Weinstein) writes:
>In article <6i5fj2$a7c at ux.cs.niu.edu>, Neil Rickert <rickert at cs.niu.edu> wrote:

>>Surely, that the device is called a *computer* is no more than a
>>matter of social convention.

>But I didn't actually say anything about something being conventionally
>*called* a computer. I was talking about a stance you can take towards it
>in explaining its behavior, roughly a species of Dennett's "design stance".

>In general, I would suggest there are two sorts of computers, artifactual
>computers, whose computational description does depend on reference to social
>and historical factors, and -- possibly -- natural computers, whose 
>computational description depends on natural teleology, independent
>of any human conventions. (Of course if you do not believe in 
>natural teleology, e.g. you do not believe that the heart is a natural
>pump that *mal*functions when it does not pump well, then you would 
>not believe in natural computation). 

>>No, I disagree.  In fact this was the sort of thing that the
>>disagreement between Bill Modlin and me was about.  We can say that
>>something is a computation without having to map it into the action
>>of a formal Turing machine.  

>If you are including "analog" computation, that may be true.

There are things that are being done by the computer on my desk that
I could not map into a Turing machine.

I'm not talking 'analog'.  My concern is with interaction.  A Turing
machine is not interactive.  A person is, and the computer on my desk
is.

>                      But it has to be something to which the 
>theory of computation applies, doesn't it?

I don't see why.  The computer sends out a signal on one of its
output channels.  This causes something to happen in the world.
Whatever happens then results in further input to the computer.  In
effect the computer has coopted the world as part of the mechanism it
uses in its procedures.  If the computer on my desk is sufficiently
interactive, then you might not be able to apply the theory of
computation unless you can formalize the entire universe as a Turing
machine.

>>                             From my perspective, the Turing theory
>>is that of an idealized mathematical model of computation.  

>Ah, so you are treating it as a kind of physical theory, an idealized
>model of the actual (and counterfactual) state transitions in a
>physical system. As if theory of computation were a special branch of
>mathematical physics.

>On this view I ought to be able to look at any physical system at all, the
>tides or the goings on in stomach, and say: that's a computation, an 
>AND-gate, perhaps, and the rightness of what I say is to be judged
>solely by its predictive power.

Right.  Except you don't get much predictive power from treating
stomach actions as a computation.

>That seems like an OK concept, but I don't think it matches the use 
>of "computer" in actual use or in use in cognitive psychology.

Let's look at 'computation' in actual use:

When we use a lever, we might say that it multiplies the force by
three.  But we would not normally say it is a computation.

Early automobiles used a lever system for controlling the brakes.  We
could say that the lever system performed a computation, using the
pedal force as input, and the brake shoe force as output.  But we
would never actually say that.  More recent autos use a hydraulic
system in place of the levers.  Again, we could say that the
hydraulic system is performing computation.  But we don't say that.
Large semi-trailer trucks have long used air brakes.  Trains have
long used vacuum brakes.  If we can control the force with levers,
with hydraulics, with higher than atmospheric air pressure, with
lower then atmospheric air pressure, we might say that we have pretty
good evidence for the claim that the process is substrate neutral.
But still we would never call it computation.  But as soon as we use
electonics, we do call it computation, even if the electronics is
entirely analog (as it was in the earliest anti-skid braking
systems).

A digital computer mainly works by a process of setting and resetting
switches in accordance with inputs.  In an old fashioned telephone
exchange, operators manually did the setting and resetting of
switches, but we never called it computation.  When the switching was
based on mechanical relays and rotary selector switches, we never
called it computation.  But now that the telephone switching centers
are mostly electronic, we do call it computation.

Word processing used to be done by dictating to a stenographer, who
used a notepad and a typewriter.  It was never called computation.
Now that it is done in a digital computer, it is called computation.
Although an array of clerks sitting at typewriters transcribing from
their dictation pads were never called computation, a similar array
of clerks tabulating the account balances for a bank were considered
to be doing computation.

At one time the printing of address labels on magazines was
mechanized by a device known as an addressograph.  This operation was
never described as computation.  Now that it is done electronically,
it is called computation.

I'm suggesting that the use of the word 'computation' is rather
eratic, and is not in any obvious way tied to the ability to use
Turing machine models.  It might be more closely tied to the use of
electronics (so much for substrate neutrality), although that does
not completely work either.

Now that we consider telephone switching to be computation, I still
could not run it on my desktop computer.  Now that we consider the
control of a braking system to be computation, I still could not run
it on my desktop computer.

What I am suggesting is that the claim that X is computation is a
rather weak claim, made because we find that claim useful for some
purposes.  I'm not sure why philosophers consider the claim
"cognition is computation" to be a strong claim that they should
oppose.  I'm not sure why AI folk consider "cognition is computation"
a strong claim which they think has implications about cognition in a
desktop computer.  Perhaps we would do better to drop the word
"computation" from these discussion altogether, and start asking the
serious questions, such as "what does the brain do, and how does it
do it?"




More information about the Neur-sci mailing list