thanks
[From Rick Marken (2017.08.22.1625)]
RobertsonGlinesPaper.pdf (310 KB)
PowersCommentonPlateau.pdf (58.9 KB)
···
On Tue, Aug 22, 2017 at 6:25 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:
thanks
RM: The best paper I know of on a PCT analysis of “operant
conditioning” during learning is a little gem by Robertson and Glines called “The Phantom Plateau Returns”. I’ve attached it along with Powers’ comment on it in the same journal.Â
Hi AlexÂ
RM: Another good source would be Bruce Abbott, an expert on operant conditioning and a PCT aficionado. Bruce is on CSGNet though he hasn’t posted in a while nor does he answer my personal emails. So I may have finally driven him off (the power law discussion possibly being the last straw). Or I just don’t have his correct email address. The one I have isÂ
bbabbott@frontier.com
Best regards
Rick
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.â?
                --Antoine de Saint-Exupery
Rick,
Hmm I donât recall having receivved any personal emails from you in the last several months. Did I miss them? You certainly have the right email address . . .
Bruce A.
···
From: Richard Marken [mailto:rsmarken@gmail.com]
Sent: Tuesday, August 22, 2017 7:25 PM
To: csgnet@lists.illinois.edu
Subject: Re: any paper or email on a PCT analysis of “operant conditioning” but during learning, namely, for re-organization (not in steady state)
[From Rick Marken (2017.08.22.1625)]
On Tue, Aug 22, 2017 at 6:25 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:
thanks
Hi Alex
RM: The best paper I know of on a PCT analysis of “operant conditioning” during learning is a little gem by Robertson and Glines called “The Phantom Plateau Returns”. I’ve attached it along with Powers’ comment on it in the same journal.
RM: Another good source would be Bruce Abbott, an expert on operant conditioning and a PCT aficionado. Bruce is on CSGNet though he hasn’t posted in a while nor does he answer my personal emails. So I may have finally driven him off (the power law discussion possibly being the last straw). Or I just don’t have his correct email address. The one I have is
bbabbott@frontier.com
Best regards
Rick
–
Richard S. Marken
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.â?
–Antoine de Saint-Exupery
Thanks, Richard. I could not see much explicit modeling there à la Powers (cf. 1971 “A feedback model for behavior: application to a rat experiment”), though… I am trying to build a similar model that would generate the behavior (sequences of lever presses) of a typical “fixed-ratio” schedule (if the mouse presses 8 times, it gets a food pellet) in a learning experiment, but not just when the animal has learned it and behaves as an ideal system, but as the animal is going through re-organization. I have a very simple one based on the E.Coli method, but I wonder if there is anything else explicitly done out there. Thanks again.
···
On Wed, Aug 23, 2017 at 1:25 AM, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2017.08.22.1625)]
On Tue, Aug 22, 2017 at 6:25 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:
thanks
RM: The best paper I know of on a PCT analysis of “operant
conditioning” during learning is a little gem by Robertson and Glines called “The Phantom Plateau Returns”. I’ve attached it along with Powers’ comment on it in the same journal.Â
Hi AlexÂ
RM: Another good source would be Bruce Abbott, an expert on operant conditioning and a PCT aficionado. Bruce is on CSGNet though he hasn’t posted in a while nor does he answer my personal emails. So I may have finally driven him off (the power law discussion possibly being the last straw). Or I just don’t have his correct email address. The one I have isÂ
bbabbott@frontier.com
Best regards
Rick
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
–
[From Rick Marken (2017.08.23.1115)]
···
On Wed, Aug 23, 2017 at 5:06 AM, Alex Gomez-Marin agomezmarin@gmail.com wrote:
AGM: Thanks, Richard. I could not see much explicit modeling there à la Powers (cf. 1971 “A feedback model for behavior: application to a rat experiment”), though… I am trying to build a similar model that would generate the behavior (sequences of lever presses) of a typical “fixed-ratio” schedule (if the mouse presses 8 times, it gets a food pellet) in a learning experiment, but not just when the animal has learned it and behaves as an ideal system, but as the animal is going through re-organization. I have a very simple one based on the E.Coli method, but I wonder if there is anything else explicitly done out there. Thanks again.
RM:  The modeling project sounds great!! I’m afraid I don’t know of any PCT modeling that includes the learning phase of operant behavior on a fixed ratio schedule. I believe Bill Powers and Bruce Abbott did do some work on this some years ago, and now that I see Bruce is still on CSGNet perhaps he could tell you a bit about that project, though I’m pretty sure it didn’t include modeling of the learning phase. And, of course, Henry Yin has some good fixed ratio data but I don’t think he did any modeling of it à la Powers (1971). But, again, if he did it’s unlikely that any modeling included the learning phase.Â
RM: I think it might be very difficult to model the learning phase of any operant behavior because PCT itself is a model of control so the model has mainly been applied to accounting for behaviors that involve good control. The operant conditioning modeling that has been done was less successful than Bill had hoped (although it was very successful by conventional standards, as in Powers (1971)) and that’s because organisms in model operant conditioning situations are constrained to not be able to control as well as they could. In the shock avoidance situation the rat could not control perfectly for avoiding shocks; it still got shocked occasionally. In food reinforcement studies the animal is usually not able to control for getting all the food it wants. The result is that there is apparently always some significant amount of error in the control system that are controlling for shock avoidance or food intake, resulting in some low level or reorganization and, hence, some randomness in the behavior.Â
RM: But please keep us posted on this project. It certainly sounds great!
BestÂ
Rick
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
[From Rick Marken (2017.08.23.1130)]
Powers power law relevant post.docx (28.3 KB)
···
On Tue, Aug 22, 2017 at 7:05 PM, Bruce Abbott bbabbott@frontier.com wrote:
BA: Rick,
Hmm I donât recall having received any personal emails from yyou in the last several months. Did I miss them? You certainly have the right email address . . .
RM: Yes, I sent an email to your frontier.net address on Aug 1. I asked if you have a copy of the “Little Man” program that includes the “physics” that Greg Williams gave to Bill. I was hoping that we could use it to reproduce the results Bill describes in this post of his that I posted earlier (attached). The relevant part is here:
BP: This paper was sent to me by Greg Williams as a source of data about
actual hand movements, for comparison with the hand movements generated by Little Man v. 2, the version using actual arm dynamics for the external part of the model. The model’s hand movements were, as Greg will attest, quite close to those shown in this paper, being slightly
curved lines connecting the end-points. Forward and reverse movements followed somewhat different paths, and by adjustment of model parameters this difference, too, could be reproduced.
RM: So I guess what I’m asking is whether you have a Delphi version of Little Man v.2 and, if so, could it be set up to make up and down movements like those made by subjects in the Atkeson and Hollerback (1985) paper that Bill discusses in that post so and then collect the movement data and analyze it to see if we find invariant velocity profiles like the authors did, showing that such profiles are a “side effect” of controlling perceptions of arm position when taking the physics of the real world into account. Â
Best regards
Rick
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.â?
                --Antoine de Saint-Exupery
[From Bruce Abbott (2017.08.23.1550 EDT)]
Rick Marken (2017.08.23.1130) –
BA: Rick,
Hmm I donât recall having received any personal emails from you in the last several months. Did I miss them? You certainly have the right email address . . .
RM: Yes, I sent an email to your frontier.net address on Aug 1. I asked if you have a copy of the “Little Man” program that includes the “physics” that Greg Williams gave to Bill. I was hoping that we could use it to reproduce the results Bill describes in this post of his that I posted earlier (attached). The relevant part is here:
BP: This paper was sent to me by Greg Williams as a source of data about
actual hand movements, for comparison with the hand movements generated by Little Man v. 2, the version using actual arm dynamics for the external part of the model. The model’s hand movements were, as Greg will attest, quite close to those shown in this paper, being slightly
curved lines connecting the end-points. Forward and reverse movements followed somewhat different paths, and by adjustment of model parameters this difference, too, could be reproduced.
RM: So I guess what I’m asking is whether you have a Delphi version of Little Man v.2 and, if so, could it be set up to make up and down movements like those made by subjects in theAtkeson and Hollerback (1985) paper that Bill discusses in that post so and then collect the movement data and analyze it to see if we find invariant velocity profiles like the authors did, showing that such profiles are a “side effect” of controlling perceptions of arm position when taking the physics of the real world into account.
BA: I checked my email archive and sure enough, there was your message. Somehow it was overlooked.
BA: I do have the Delphi Arm2 program; it includes a procedure created by Greg Williams that computes all the motion-generated forces on the arm segments, including any active Coriolis forces. A zip file containing sorce code and a Windows-executable program is available for download from my Google website, Perceptual Control Demos, at
https://sites.google.com/site/perceptualcontroldemos/home/other-demos
I suspect that modifying it to produce up-and-down hand movements would not be too difficult to do. But I suspect you would also want the ability to accurately record those movements as a time series.
Bruce
···
On Tue, Aug 22, 2017 at 7:05 PM, Bruce Abbott bbabbott@frontier.com wrote:
[From Rick Marken (2017.08.23.2240)]
BA: I do have the Delphi Arm2 program; it includes a procedure created by Greg Williams that computes all the motion-generated forces on the arm segments, including any active Coriolis forces. A zip file containing sorce code and a Windows-executable program is available for download from my Google website, Perceptual Control Demos, at
<https://urldefense.proofpoint.com/v2/url?u=https-3A__sites.google.com_site_perceptualcontroldemos_home_other-2Ddemos&d=DwMFaQ&c=8hUWFZcy2Z-Za5rBPlktOQ&r=-dJBNItYEMOLt6aj_KjGi2LMO_Q8QB-ZzxIZIF8DGyQ&m=PRIqkl9oo1DtYdX9_Vm3xg0528RRsw5SnyfJ11QBfH4&s=0hd806bBzUTv_3BKunF94M8tjZLndrAF36za6MWYoaw&e=>> https://sites.google.com/site/perceptualcontroldemos/home/other-demos
I suspect that modifying it to produce up-and-down hand movements would not be too difficult to do. But I suspect you would also want the ability to accurately record those movements as a time series.
RM; Great! Thanks. Here's a pointer to the article that Bill was referring to in his post:Â
<https://www.dropbox.com/s/qx27ow5je28qwk1/arm_movements.pdf?dl=0>https://www.dropbox.com/s/qx27ow5je28qwk1/arm_movements.pdf?dl=0\.Â
RM: It gives the details on the method used to obtain the arm trajectories. It would be great if you could create some code to have the Little Man's arm move back and forth between two target points as shown in the paper. It would just involve varying the references for the fingertip. Then we would have to collect the trajectories, normalize them and make plots corresponding to those shown in the paper.Â
RM: Whadaya think?
BestÂ
Rick
···
--
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
--
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
[From Bruce Abbott (2017.08.25.1050 EDT)]
Rick Marken (2017.08.23.2240) –
BA: I do have the Delphi Arm2 program; it includes a procedure created by Greg Williams that computes all the motion-generated forces on the arm segments, including any active Coriolis forces A zip file containing sorce code and a Windows-executable program is available for download fr
om my Google website, Perceptual Control Demos, at
https://sites.google.com/site/perceptualcontroldemos/home/other-demos
I suspect that modifying it to produce up-and-down hand movements would not be too difficult to do. But I suspect you would also want the ability to accurately record those movements as a time series.
RM; Great! Thanks. Here’s a pointer to the article that Bill was referring to in his post:
https://www.dropbox.com/s/qx27ow5je28qwk1/arm_movements.pdf?dl=0.
RM: It gives the details on the method used to obtain the arm trajectories. It would be great if you could create some code to have the Little Man’s arm move back and forth between two target points as shown in the paper. It would just involve varying the referenc
es for the fingertip. Then we would have to collect the trajectories, normalize them and make plots corresponding to those shown in the paper.
RM: Whadaya think?
BA: Are you thinking we might coauthor a paper on the results?
Bruce
[From Rick Marken (2017.08.25.1155)]
···
:
Bruce Abbott (2017.08.25.1050 EDT)–
Â
Rick Marken (2017.08.23.2240) –
Â
BA: I do have the Delphi Arm2 program…
RM; Great! Thanks.Â
Â
RM:… It would be great if you could create some code to have the Little Man’s arm move back and forth between two target points as shown in the paper. It would just involve varying the referenc
es for the fingertip. Then we would have to collect the trajectories, normalize them and make plots corresponding to those shown in the paper.Â
Â
RM: Whadaya think?
Â
BA: Are you thinking we might coauthor a paper on the results?
RM: Yes, indeed. If you would be willing I would love to do a joint paper with you on this!Â
BestÂ
Rick
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery
[From Bruce Abbott (2017.08.27.1300 EDT)]
[From Rick Marken (2017.08.25.1155)]
:
Bruce Abbott (2017.08.25.1050 EDT)–
Rick Marken (2017.08.23.2240) –
BA: I do have the Delphi Arm2 program…
RM; Great! Thanks.
RM:… It would be great if you could create some code to have the Little Man’s arm move back and forth between two target points as shown in the paper. It would just involve varying the referenc es for the fingertip. Then we would have to collect the trajectories, normalize them and make plots corresponding to those shown in the paper.
RM: Whadaya think?
BA: Are you thinking we might coauthor a paper on the results?
RM: Yes, indeed. If you would be willing I would love to do a joint paper with you on this!
BA: O.K., let’s do it!
Bruce
[From Bruce Abbott (2017.08.27.1300 EDT)]
···
[From Rick Marken (2017.08.25.1155)]
:
BruceAbbott (2017.08.25.1050 EDT)–
RickMarken (2017.08.23.2240) –
BA:I do have the Delphi Arm2 program…
RM;Great! Thanks.
RM:...It would be great if you could create some
code to have the Little Man’s arm move back
and forth between two target points as shown
in the paper. It would just involve varying
the referenc es for the fingertip. Then we
would have to collect the trajectories,
normalize them and make plots corresponding to
those shown in the paper.
RM:Whadaya think?
BA:Are you thinking we might coauthor a paper on
the results?
RM: Yes, indeed. If you would be
willing I would love to do a joint paper with you on
this!
BA: O.K., let’s do it!
Bruce
[From Rick Marken (2017.08.27.1900)]
···
Bruce Abbott (2017.08.27.1300 EDT)–
Â
Â
[From Rick Marken (2017.08.25.1155)]
:
Bruce Abbott (2017.08.25.1050 EDT)–
Â
RM; Great! Thanks.Â
RM:… It would be great if you could create some code to have the Little Man’s arm move back and forth between two target points as shown in the paper. It would just involve varying the referenc es for the fingertip. Then we would have to collect the trajectories, normalize them and make plots corresponding to those shown in the paper.Â
Â
RM: Whadaya think?
Â
BA: Are you thinking we might coauthor a paper on the results?
Â
RM: Yes, indeed. If you would be willing I would love to do a joint paper with you on this!Â
Â
BA: O.K., let’s do it!
RM: That’s great! Should we carry on the collaboration on or off CSGNet? I think we should start off line and give reports on line when we have something worthwhile.Â
Best
Rick
–
Richard S. MarkenÂ
"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery