X-Message-Number: 12700
Date: Tue, 02 Nov 1999 21:30:02 -0700
From: Mike Perry <>
Subject: Feelings, Robot Onboard Computer

Thomas Donaldson, Message #12685 says:

>How is a robot and a virtual robot different?
>
>A real robot (ie. such as the machines that wander about looking for 
>electrical outlets) will actually find outlets and use them. This is not
>the same as a program which imitates such a robot in a computer. The
>difference comes from the difference between reality (and all the things
>that can really get in the way of a real robot: obstacles of various 
>kinds, things falling on it etc) and a fictional story.
>
I see no necessary difference for the "brain" of the robot, its onboard
computer that controls its actions and interprets its sensory data. In my
message #12679 I was trying to make this point but perhaps wasn't clear.

>Whenever we deal with computers, WE use or write a program for them. This
>is essentially the same as telling them what to do on a more primitive
>level --- it's just that what we tell them is much more complex.

But a program can easily have provisions for modifying its own code in the
course of its operation. (When that improves the performance we call it
"learning.")

> To tell
>a computer to act like a robot (all inside the computer) creates a 
>situation which isn't fundamentally different from that of writing a book
>of fiction.

To me it's different because the fiction is static, while the program is
active, and for other reasons too.

>Another possibility that Mike Perry seems not to discuss is that of a 
>halfway robot: the robot still exists, but it is controlled by a computer
>program in a computer (either in the robot or elsewhere doesn't really
>matter).

I did consider this, but I think I wasn't clear. Here is what I said in
message #12679, with additional comments in square brackets [].

>Let's say we have 3 scenarios. In #1
>a robot has an onboard computer, call it an ROC ("robot onboard computer").
[This onboard computer should be understood as *controlling* the robot, via
a *program* that it, the ROC, is running. The program was originally created
extraneous to the robot's operation, but it is possible that it will itself
undergo some modification as the robot does its things, which is easy enough
to make a program do.] >It [the robot, here] does various goal-oriented
things, including >looking for electrical outlets when its batteries are
low, and some of us (self >included) conclude it [the robot, again, but
really the ROC and its running >program, which is where the robot "lives"]
has feelings, at some low level.

Thomas, again:
> Yes, here we have a shading over. I would say that the same
>problem exists as when EVERYTHING is virtual: the robot no longer is
>doing what it wants, it's merely doing what we ask it to do by writing
>the program we've written.

I don't see it that way. In my example the ROC is running (yes, running a
program). There were 3 scenarios. In #1 the ROC is inside the robot and
physically moves around with it and controls its motions. In #2 the ROC is
stationary and sitting on a desk, but communicates with the (now "brainless"
but tuned in) robot via radio signals, and again controls its motions. In #3
the ROC communicates with another computer which generates a virtual robot
so that the ROC "thinks" it is controlling a real robot, but really isn't.
In all 3 cases, I postulate, the actual computations of the ROC (including
possibly modifying the code in the program it is running) are identical. So
I claim the ROC is expressing the same subjective "feeling" in all 3 cases.
Do you disagree?

Mike Perry

Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=12700