Friam

goal/function/robots

Classic

List

Threaded

3 messages Options

thompnickson2

goal/function/robots

Here’s the example I was trying to lay out for the zFriam Leftovers when my house got flooded. It concerns simple robots, “didabots”, that if unleashed in a room full of randomly placed Styrofoam cubes, will tidy up the room by herding the cubes into clusters.

http://www.verena-hafner.de/teaching/didabots.pdf

Here is what Louise Barrett’s BEYOND THE BRAIN says about the example:

It should be clear that this “clustering” is not the “goal” of the individual didabots … . Clustering is an emergent property (=consequence?) … this complex behavior is produced, not only by a very simple mechanism, but also by a mechanism that bears absolutely no relation to the behavioral outcome produced when that mechanism operates in the real world…. . Poking around inside a didabot to identify the nature of this mechanism won’t tell us anything about didabot [herding] behavior because it makes sense only after we have taken into account the interaction of the internal mechanisms with the physical structure of the didabot and the structure of the environment. (p49)

It seems to me that understanding the distinction between goal and function, we have to look at the same behavior, simultaneously from two different points of view. The first is what outcome the behavor is directed toward (avoiding obstacles on the side) and the outcome robot (organism) has been designed to produce (the herding of the blocks) within its
ecological” context. The seagull is a robot whose goal is to remove shiny things from the nest; natural selection has designed that seagull so that it avoids predation by vision-using predators. That this behavior constitutes a design is demonstrated by widening the lens still further and showing that shell removal is characteritic of surface nesting gulls but NOT of the closely related kitiwake gulls which nest on cliffs that ground searching predators cannot reach.

This is an old story that has no doubt been altered by time, but the logic of attribution of goal, selection mechanism, and design remains the same.

It is these sorts of examples that lead me to think that the decoupling between goal and function or design is an essential feature of control systems generally. Designers, human and natural, are all the same. As soon as the machine produces the outcome we desire, we stop designing.

Nick

Nicholas Thompson

Emeritus Professor of Ethology and Psychology

Clark University

[hidden email]

https://wordpress.clarku.edu/nthompson/

- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
FRIAM Applied Complexity Group listserv
Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam
un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/

Eric Charles-2

Re: goal/function/robots

" As soon as the machine produces the outcome we desire, we stop designing."

Oddly, in this crowd that is not true at all. Computer programming is largely an effort to increase efficiency in things that already do what we want them to do. Whole areas of computer science are dedicated to determining the theoretical limits of efficiency at tasks that are easy to program. Search optimization and list sorting might be the most obvious contexts for this, but a lot of agent based modeling is focused in this direction as well.

There are MANY projects I work on where I write a program that does a task in whatever way I think is easiest to write, and then spend 10 times longer rewriting to do the task in some other way (usually to run more quickly, but other times for the aesthetic pleasure of seeing the code get more "elegant").

-----------

Eric P. Charles, Ph.D.
Department of Justice - Personnel Psychologist

American University - Adjunct Instructor

[hidden email]

On Fri, Jul 24, 2020 at 6:06 PM <[hidden email]> wrote:

Here’s the example I was trying to lay out for the zFriam Leftovers when my house got flooded. It concerns simple robots, “didabots”, that if unleashed in a room full of randomly placed Styrofoam cubes, will tidy up the room by herding the cubes into clusters.

http://www.verena-hafner.de/teaching/didabots.pdf

Here is what Louise Barrett’s BEYOND THE BRAIN says about the example:

It should be clear that this “clustering” is not the “goal” of the individual didabots … . Clustering is an emergent property (=consequence?) … this complex behavior is produced, not only by a very simple mechanism, but also by a mechanism that bears absolutely no relation to the behavioral outcome produced when that mechanism operates in the real world…. . Poking around inside a didabot to identify the nature of this mechanism won’t tell us anything about didabot [herding] behavior because it makes sense only after we have taken into account the interaction of the internal mechanisms with the physical structure of the didabot and the structure of the environment. (p49)

It seems to me that understanding the distinction between goal and function, we have to look at the same behavior, simultaneously from two different points of view. The first is what outcome the behavor is directed toward (avoiding obstacles on the side) and the outcome robot (organism) has been designed to produce (the herding of the blocks) within its
ecological” context. The seagull is a robot whose goal is to remove shiny things from the nest; natural selection has designed that seagull so that it avoids predation by vision-using predators. That this behavior constitutes a design is demonstrated by widening the lens still further and showing that shell removal is characteritic of surface nesting gulls but NOT of the closely related kitiwake gulls which nest on cliffs that ground searching predators cannot reach.

This is an old story that has no doubt been altered by time, but the logic of attribution of goal, selection mechanism, and design remains the same.

It is these sorts of examples that lead me to think that the decoupling between goal and function or design is an essential feature of control systems generally. Designers, human and natural, are all the same. As soon as the machine produces the outcome we desire, we stop designing.

Nick

Nick

Nicholas Thompson
Emeritus Professor of Ethology and Psychology
Clark University
[hidden email]
https://wordpress.clarku.edu/nthompson/

- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
FRIAM Applied Complexity Group listserv
Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam
un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/

thompnickson2

Re: goal/function/robots

Well, I obviously have that wrong because once natural selection happens on a way to do something, it will go on to find ways to do it with less cost to other functions. Obviously, then, I have to make a more sophisticated argument – trapped in a local optima, or something – or I have to get off the field.

Maybe I will get off the field.

Nicholas Thompson

Emeritus Professor of Ethology and Psychology

Clark University

[hidden email]

https://wordpress.clarku.edu/nthompson/

From: Friam <[hidden email]> On Behalf Of Eric Charles
Sent: Friday, July 24, 2020 8:39 PM
To: The Friday Morning Applied Complexity Coffee Group <[hidden email]>
Subject: Re: [FRIAM] goal/function/robots

" As soon as the machine produces the outcome we desire, we stop designing."

-----------

Eric P. Charles, Ph.D.
Department of Justice - Personnel Psychologist

American University - Adjunct Instructor

On Fri, Jul 24, 2020 at 6:06 PM <[hidden email]> wrote:

Here’s the example I was trying to lay out for the zFriam Leftovers when my house got flooded. It concerns simple robots, “didabots”, that if unleashed in a room full of randomly placed Styrofoam cubes, will tidy up the room by herding the cubes into clusters.

http://www.verena-hafner.de/teaching/didabots.pdf

Here is what Louise Barrett’s BEYOND THE BRAIN says about the example:

It should be clear that this “clustering” is not the “goal” of the individual didabots … . Clustering is an emergent property (=consequence?) … this complex behavior is produced, not only by a very simple mechanism, but also by a mechanism that bears absolutely no relation to the behavioral outcome produced when that mechanism operates in the real world…. . Poking around inside a didabot to identify the nature of this mechanism won’t tell us anything about didabot [herding] behavior because it makes sense only after we have taken into account the interaction of the internal mechanisms with the physical structure of the didabot and the structure of the environment. (p49)

It seems to me that understanding the distinction between goal and function, we have to look at the same behavior, simultaneously from two different points of view. The first is what outcome the behavor is directed toward (avoiding obstacles on the side) and the outcome robot (organism) has been designed to produce (the herding of the blocks) within its
ecological” context. The seagull is a robot whose goal is to remove shiny things from the nest; natural selection has designed that seagull so that it avoids predation by vision-using predators. That this behavior constitutes a design is demonstrated by widening the lens still further and showing that shell removal is characteritic of surface nesting gulls but NOT of the closely related kitiwake gulls which nest on cliffs that ground searching predators cannot reach.

This is an old story that has no doubt been altered by time, but the logic of attribution of goal, selection mechanism, and design remains the same.

It is these sorts of examples that lead me to think that the decoupling between goal and function or design is an essential feature of control systems generally. Designers, human and natural, are all the same. As soon as the machine produces the outcome we desire, we stop designing.

Nick

Nick

Nicholas Thompson
Emeritus Professor of Ethology and Psychology
Clark University
[hidden email]
https://wordpress.clarku.edu/nthompson/

- .... . -..-. . -. -.. -..-. .. ... -..-. .... . .-. .
FRIAM Applied Complexity Group listserv
Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam
un/subscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/