[Bruce Nevin (2016.02.10.22.38 ET)]
[Bruce Nevin (2016.02.10.22.38 ET)]
[David Goldstein (2016.02.11.0329)]
Really interesting. A great example of reorganization.
What levels had to be unlearned and relearned.?
Age differences in this were very interesting.
If someone who never learned to ride a bike was given the task, they wouldn’t have to unlearn, just learn.
I can remember hours and days of p
racticing to ride a bike.
[From Rick Marken (2015.02.12.0950)]
···
Bruce Nevin (2016.02.10.22.38 ET)
BN: http://www.dailyliked.net/backwards-brain-bicycle/
David Goldstein (2016.02.11.0329)]
DG: Really interesting. A great example of reorganization.
RM: It sure is. But it’s something that was already famously demonstrated (at least to intro psychology students in my courses way back when I was a professor) by Austrian psychologist Erismann. He had his soon to be more well known student, Gestalt psychologist Ivo Kohler, wear inverting lenses to see if he could adapt to them. And indeed he did. There is a little You Tube of this at https://www.youtube.com/watch?v=jKUVpBJalNQ
Among the things Kohler learned to do was ride a bike. Somewhere in that 1950 Austrian video you can see him trying to ride it, just like the guy in the video that Bruce sent. And, indeed, he eventually adapts completely to the glasses and can ride the bike just fine. And just like the guy in Bruce’s video, when he takes off the glasses he as to relearn how to do things in an uninverted world.
RM: The point is 1) that these young guys like the one in Bruce’s video should study their history and 2) reorganization works
DG: What levels had to be unlearned and relearned.?
RM: I think it’s better to look at it as “what variables need to be learned to be controlled given the change in the feedback connection between output and the variable”? One variable is certainly “balance” (location of center of gravity) which is controlled by turning the front wheel, mainly (I think). What the inverting (or reversing) goggles do is change the feedback connection between wheel turn and the perceived location of your center of gravity. It’s like changing the polarity of the effect of direction of mouse movement on direction of cursor movement. The reorganization system has to reverse the polarity of the output functions that affect the center of gravity (in the bike riding case) and they have to do it in just the right way – getting the gain and time constants of the control system just right – so that control of balance is as good as it was without the goggles. And Kohler had just as much difficulty relearning how to ride the bike after the goggles were removed as did the fellow in Bruce’s video.
BN: Age differences in this were very interesting.
If someone who never learned to ride a bike was given the task, they wouldn’t have to unlearn, just learn.
RM: I didn’t see that part of Bruce’s video but I think people who haven’t learned to ride a bike would have more trouble learning to ride with reversing googles than without. That’s my prediction anyway because our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed (or inverted). Is these any evidence that non- bike rider learn to ride with reverse goggles slower than non bike rider who learn without reverse goggles? That would be my prediction. If it’s not true - if non bike riders learn to ride equally well with or without reversing googles then I think that would be a problem for the PCT idea that lower level control systems are used in the service of controlling many different higher level perceptions.
Best
Rick
I can remember hours and days of p
racticing to ride a bike.
–
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
[Bruce Nevin (2016.02.12.15:17 ET)]
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
https://www.youtube.com/watch?v=-kohUpQwZt8
And here’s a BBC clip, apparently in England:
https://www.youtube.com/watch?v=MHMvEMy7B9k
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
···
On Fri, Feb 12, 2016 at 12:50 PM, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2015.02.12.0950)]
Bruce Nevin (2016.02.10.22.38 ET)
BN: http://www.dailyliked.net/backwards-brain-bicycle/
David Goldstein (2016.02.11.0329)]
DG: Really interesting. A great example of reorganization.
RM: It sure is. But it’s something that was already famously demonstrated (at least to intro psychology students in my courses way back when I was a professor) by Austrian psychologist Erismann. He had his soon to be more well known student, Gestalt psychologist Ivo Kohler, wear inverting lenses to see if he could adapt to them. And indeed he did. There is a little You Tube of this at https://www.youtube.com/watch?v=jKUVpBJalNQ
Among the things Kohler learned to do was ride a bike. Somewhere in that 1950 Austrian video you can see him trying to ride it, just like the guy in the video that Bruce sent. And, indeed, he eventually adapts completely to the glasses and can ride the bike just fine. And just like the guy in Bruce’s video, when he takes off the glasses he as to relearn how to do things in an uninverted world.
RM: The point is 1) that these young guys like the one in Bruce’s video should study their history and 2) reorganization works
DG: What levels had to be unlearned and relearned.?
RM: I think it’s better to look at it as “what variables need to be learned to be controlled given the change in the feedback connection between output and the variable”? One variable is certainly “balance” (location of center of gravity) which is controlled by turning the front wheel, mainly (I think). What the inverting (or reversing) goggles do is change the feedback connection between wheel turn and the perceived location of your center of gravity. It’s like changing the polarity of the effect of direction of mouse movement on direction of cursor movement. The reorganization system has to reverse the polarity of the output functions that affect the center of gravity (in the bike riding case) and they have to do it in just the right way – getting the gain and time constants of the control system just right – so that control of balance is as good as it was without the goggles. And Kohler had just as much difficulty relearning how to ride the bike after the goggles were removed as did the fellow in Bruce’s video.
BN: Age differences in this were very interesting.
If someone who never learned to ride a bike was given the task, they wouldn’t have to unlearn, just learn.
RM: I didn’t see that part of Bruce’s video but I think people who haven’t learned to ride a bike would have more trouble learning to ride with reversing googles than without. That’s my prediction anyway because our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed (or inverted). Is these any evidence that non- bike rider learn to ride with reverse goggles slower than non bike rider who learn without reverse goggles? That would be my prediction. If it’s not true - if non bike riders learn to ride equally well with or without reversing googles then I think that would be a problem for the PCT idea that lower level control systems are used in the service of controlling many different higher level perceptions.
Best
Rick
I can remember hours and days of p
racticing to ride a bike.
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
–
[From Bruce Abbott (2016.02.12.1940 EST)]
(Bruce Nevin (2016.02.12.15:17 ET) –
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
https://www.youtube.com/watch?v=-kohUpQwZt8
And here’s a BBC clip, apparently in England:
https://www.youtube.com/watch?v=MHMvEMy7B9k
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backwardâ€? bicycle or wearing inverting prisms. Second, the reversion seemed just to “snapâ€? into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.Â
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Bruce A.
[From Bruce Abbott (2016.02.12.1940 EST)]
···
(Bruce Nevin (2016.02.12.15:17 ET) –
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
https://www.youtube.com/watch?v=-kohUpQwZt8
And here’s a BBC clip, apparently in England:
https://www.youtube.com/watch?v=MHMvEMy7B9k
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backward� bicycle or wearing inverting prisms. Second, the reversion seemed just to “snap� into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Bruce A.
[From Rick Marken (2016.02.13.1140)]
···
Bruce Nevin (2016.02.12.15:17 ET)–
BN: Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
RM: Yes, the relatively fast “recovery” when vision returns to normal is very interesting. I’ll give my ideas on it in a later reply to Bruce A. and Warren M. Right now I want to comment on this:
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
BN: …Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
RM: I think there is very good evidence that the motor systems – the systems that control the immediate effects of muscles forces – get references from higher level systems controlling perceptual aspects of the inputs to all modalities, including vision. The simple tracking task is an example; here a visual variable – distance between cursor and target – is controlled by varying the reference for the muscle force that moves the mouse – the mouse movements having effects on the cursor that protects the controlled variable --distance of cursor from target – from disturbances.
RM: Since I first got into PCT I have always wanted to develop a demo that shows these two levels of control – the level controlling cursor-target distance and the one controlling the force exerted on the mouse – operating simultaneously. I thought it would take a force joystick to do it since the force control system could only be revealed by resistance to force disturbances. But I recently realized, perhaps because of my close encounter with physics in the mass spring discussion, that the disturbance to force could be introduced by software by taking advantage of the fact that f = ma. The software could introduce the disturbance to acceleration – a – by changing the acceleration of the cursor resulting from mouse movement ; that would require a compensating change in how much the person has to accelerate the mouse.
RM: So I have developed a demonstration of two levels of control operating simultaneously in a tracking task. The demo still needs some work but I’m so excited about it I would like people to give it a try and let me know if it works for them. An initial version of the demo is up at:
www.mindreadings.com/TwoLevelControl.html
RM: It’s just a regular tracking task but it may take a little practice because the disturbance to the effect of the mouse on the cursor – which is a varying change in the effect of mouse velocity on the cursor – makes the tracking a tad noisy. But once you get the hang of it – and you are able to keep the cursor pretty close to the target – you should see results like this at the end of a test run:
RM: The top traces, labeled L2, are those for the higher level (Level 2) system that is controlling the cursor-target relationship. These traces show the usual mirror image relationship between Response (Resp – mouse movement) and Disturbance (Dist) and the correlation between then is the expected large negative one (-.961 in this case). The lower traces, labeled L1, are those for the lower level mouse movement control system. They show the relationship between the force disturbance to the mouse (L1 Dist – the disturbance to mouse velocity)) and the compensating force exerted on the mouse by the person doing the tracking task (L1 Resp). The relationship between them is again a mirror image indicating a negative relatioship between force disturbance and compensating force (as shown by the negative correlation between L1 Dist and L1 Resp (-.578 in this case). The negative correlation between Disturbance and Response at L1 is low and I am working on figuring out why. But the negative relationship clearly shows the operation of the lower level mouse position control system since the the L1 Resp curve is clearly opposed to the L1 Dist curve resulting in mouse movements (the L2 Resp curve) that are exactly what are needed to compensate for the L2 Dist and keep the cursor close to the target.
RM: Any comments or suggestions on this demo would be greatly appreciated.
Best
Rick
–
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
On Fri, Feb 12, 2016 at 12:50 PM, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2015.02.12.0950)]
Bruce Nevin (2016.02.10.22.38 ET)
BN: http://www.dailyliked.net/backwards-brain-bicycle/
David Goldstein (2016.02.11.0329)]
DG: Really interesting. A great example of reorganization.
RM: It sure is. But it’s something that was already famously demonstrated (at least to intro psychology students in my courses way back when I was a professor) by Austrian psychologist Erismann. He had his soon to be more well known student, Gestalt psychologist Ivo Kohler, wear inverting lenses to see if he could adapt to them. And indeed he did. There is a little You Tube of this at https://www.youtube.com/watch?v=jKUVpBJalNQ
Among the things Kohler learned to do was ride a bike. Somewhere in that 1950 Austrian video you can see him trying to ride it, just like the guy in the video that Bruce sent. And, indeed, he eventually adapts completely to the glasses and can ride the bike just fine. And just like the guy in Bruce’s video, when he takes off the glasses he as to relearn how to do things in an uninverted world.
RM: The point is 1) that these young guys like the one in Bruce’s video should study their history and 2) reorganization works
DG: What levels had to be unlearned and relearned.?
RM: I think it’s better to look at it as “what variables need to be learned to be controlled given the change in the feedback connection between output and the variable”? One variable is certainly “balance” (location of center of gravity) which is controlled by turning the front wheel, mainly (I think). What the inverting (or reversing) goggles do is change the feedback connection between wheel turn and the perceived location of your center of gravity. It’s like changing the polarity of the effect of direction of mouse movement on direction of cursor movement. The reorganization system has to reverse the polarity of the output functions that affect the center of gravity (in the bike riding case) and they have to do it in just the right way – getting the gain and time constants of the control system just right – so that control of balance is as good as it was without the goggles. And Kohler had just as much difficulty relearning how to ride the bike after the goggles were removed as did the fellow in Bruce’s video.
BN: Age differences in this were very interesting.
If someone who never learned to ride a bike was given the task, they wouldn’t have to unlearn, just learn.
RM: I didn’t see that part of Bruce’s video but I think people who haven’t learned to ride a bike would have more trouble learning to ride with reversing googles than without. That’s my prediction anyway because our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed (or inverted). Is these any evidence that non- bike rider learn to ride with reverse goggles slower than non bike rider who learn without reverse goggles? That would be my prediction. If it’s not true - if non bike riders learn to ride equally well with or without reversing googles then I think that would be a problem for the PCT idea that lower level control systems are used in the service of controlling many different higher level perceptions.
Best
Rick
I can remember hours and days of p
racticing to ride a bike.
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
–
[From Bruce Abbott (2016.02.14.0945 EST)]
Rick Marken (2016.02.13.1140) –
Bruce Nevin (2016.02.12.15:17 ET)–
BN: Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
RM: Yes, the relatively fast “recovery” when vision returns to normal is very interesting. I’ll give my ideas on it in a later reply to Bruce A. and Warren M. Right now I want to comment on this:
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
BN: …Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
RM: I think there is very good evidence that the motor systems – the systems that control the immediate effects of muscles forces – get references from higher level systems controlling perceptual aspects of the inputs to all modalities, including vision. The simple tracking task is an example; here a visual variable – distance between cursor and target – is controlled by varying the reference for the muscle force that moves the mouse – the mouse movements having effects on the cursor that protects the controlled variable --distance of cursor from target – from disturbances.
RM: Since I first got into PCT I have always wanted to develop a demo that shows these two levels of control – the level controlling cursor-target distance and the one controlling the force exerted on the mouse – operating simultaneously. I thought it would take a force joystick to do it since the force control system could only be revealed by resistance to force disturbances. But I recently realized, perhaps because of my close encounter with physics in the mass spring discussion, that the disturbance to force could be introduced by software by taking advantage of the fact that f = ma. The software could introduce the disturbance to acceleration – a – by changing the acceleration of the cursor resulting from mouse movement ; that would require a compensating change in how much the person has to accelerate the mouse.
RM: So I have developed a demonstration of two levels of control operating simultaneously in a tracking task. The demo still needs some work but I’m so excited about it I would like people to give it a try and let me know if it works for them. An initial version of the demo is up at:
www.mindreadings.com/TwoLevelControl.html
RM: It’s just a regular tracking task but it may take a little practice because the disturbance to the effect of the mouse on the cursor – which is a varying change in the effect of mouse velocity on the cursor – makes the tracking a tad noisy. But once you get the hang of it – and you are able to keep the cursor pretty close to the target – you should see results like this at the end of a test run:
RM: The top traces, labeled L2, are those for the higher level (Level 2) system that is controlling the cursor-target relationship. These traces show the usual mirror image relationship between Response (Resp – mouse movement) and Disturbance (Dist) and the correlation between then is the expected large negative one (-.961 in this case). The lower traces, labeled L1, are those for the lower level mouse movement control system. They show the relationship between the force disturbance to the mouse (L1 Dist – the disturbance to mouse velocity)) and the compensating force exerted on the mouse by the person doing the tracking task (L1 Resp). The relationship between them is again a mirror image indicating a negative relatioship between force disturbance and compensating force (as shown by the negative correlation between L1 Dist and L1 Resp (-.578 in this case). The negative correlation between Disturbance and Response at L1 is low and I am working on figuring out why. But the negative relationship clearly shows the operation of the lower level mouse position control system since the the L1 Resp curve is clearly opposed to the L1 Dist curve resulting in mouse movements (the L2 Resp curve) that are exactly what are needed to compensate for the L2 Dist and keep the cursor close to the target.
RM: Any comments or suggestions on this demo would be greatly appreciated.
BA: I tried the demo and have a few comments/suggestions:
-
The instruction “Press the “Basic Tracking Experiment" button below to open the window that runs the tracking taskâ€? is unnecessary – there is no such button and the screen iis already as if you had pressed such a button. -
There may be differences in speed of execution of the demo depending on what computer one is using. I found the cursor movements generally too fast to track with any degree of accuracy – well below the level of accuracy shown in the top graph from your own run. If fact, sometimes the cursor would just disappear from one location and reappear in another. It would be nice to know whether anyone else experiences this problem and, whether or not it occurs, what computer they were using. Mine is an older HP laptop running the Windows Vista operating system. (Come to think of it, the system I use to control the mouse is something of an antique as well!) -
It’s hard to tell, but perhaps the above problem is due to the effect of the “force� on the cursor. If the cursor suddenly accelerates it may reach a high rate before the participant has time to even begin to compensate for the change. In other words, the force disturbance may be causing effects that are outside the bandwidth for effective control. -
I don’t experience the disturbance as a force acting on the mouse, but rather as an acceleration of the cursor. To compensate for it, I of course must vary the force applied to the mouse so as to generate an opposing acceleration of the cursor. A true force disturbance to the mouse would be experienced as a push or pull on the mouse.
You should be able to implement a true force disturbance by switching from the mouse to one of those joysticks with force feedback.
What does the block diagram look like for your two-level system?
Bruce
Bruce A.
[From Bruce Abbott (2016.02.14.1140 EST)]
···
From: Warren Mansell [mailto:wmansell@gmail.com]
Sent: Friday, February 12, 2016 8:02 PM
To: csgnet@lists.illinois.edu
Subject: Re: reorganization at work
Couldn’t the process of reorganisation still be via the e.coli method, but just applied to a copy of the original hierarchical network, thereby keeping the old one as a back up, just like when we duplicate a file?
Warren
That’s possible, I suppose, but we really don’t know much about how the brain/nervous system creates or accesses a given organization. Bill Powers suggested that reference values may be stored in memory and accessed via associative addressing, for example, but exactly how that would work remains to be elucidated. This problem is extremely difficult because, unlike digital computers, the central nervous system has no store for memories separate from the elements that carry out the receipt, processing, and transmission of neural signals. Reorganization almost certainly involves changes in synaptic connections – whether creating new ones, eliminating old ones, merely changing the conductivity of existing synapses, or all of the above. But these are the same changes thought to be involved in the creation of new memories and making them accessible!
One of the oldest findings in psychology is the observation that it is easier to relearn something than to learn it in the first place, as evidenced by the amount of practice or rehearsal required. This has been interpreted as evidence that some trace of the original learning remains even after its “representation� in the nervous system has been degraded sufficiently that it is no longer accessible. But often it seems that the representation (organization) remains but has become unavailable because cues (inputs) that should produce recall (activate the network) are not present.  Provide the right cue and the “memory� is activated. This phenomenon demonstrates that, at least in those cases, the representation was still present.
In operant conditioning, one can train a rat to press a lever in order to receive a food pellet. Switching to extinction (no longer delivering the food) eventually leads to the near cessation of lever-pressing. But extinction does not produce forgetting, it produces new learning – learning that levver-pressing no longer produces food. Restore food delivery for lever-pressing and lever-pressing quickly resumes. In fact the reacquisition of lever pressing is much faster than the original acquisition of lever pressing.Â
In PCT terms, reorganization occurred during extinction, but this reorganization did not destroy the control system that used lever-pressing as the means to obtain food. Instead, the failure of lever-pressing to produce food during extinction led to a search (at random?) for a new means to obtain food and to the near abandonment of the previously effective means (lever-pressing). But it would appear that this particular output mechanism (involving a whole sequence of controlled perceptions and their associated actions, culminating in a lever-press and followed by approach to the food cup etc.) was not destroyed by reorganization, but instead was in effect disconnected (perhaps by setting the output gain to zero?) due to its lack of success in controlling the perception of food in the cup.
After extinction, lever-pressing does not go away entirely.  Such presses sometimes occur as a by-product of other activity but I suspect that there is another reason why such presses continue, albeit at a low rate.  Perhaps the control system does not completely “abandon� lever-pressing so long as no other solution to the pellet problem is found, but continues to “sample� lever-pressing with a low frequency. This sampling strategy would be adaptive given that lever-pressing is a solution that once worked, and may work again in the future.
So one way that reorganization might work is to vary the connections between error signal and output function. The discovery of another means to the same end (perhaps through trial and error) simply connects the error signal to a different output function, without destroying the previous output function. If later experience shows that one means works under Condition A and another under Condition B, then perceptions of Condition A and Condition B may become associatively connected to the control system in such a way as to cause Output Function X to be used under Condition A and Output Function Y to be used under Condition B, perhaps by setting the gains of the output functions appropriately.
All this is speculative, but it does not depart radically from Bill’s conception. Consider the arm-reorganization demo in LCS III. Here we have a number of joints being randomly connected to an equal number of joint-angle control systems through a set of output weights. Initially each control system may be connected to several joints at once, to varying degrees, through the matrix of weights. Reorganization eventually sorts it all out so that each control system ends up controlling a single joint (or nearly so), the weights to the other outputs being reduced to zero or near-zero. If the control-system outputs went to lower-level control systems instead of directly to the joints, this would match the scheme I described above.
Bruce
On 13 Feb 2016, at 00:40, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.12.1940 EST)]
(Bruce Nevin (2016.02.12.15:17 ET) –
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
And here’s a BBC clip, apparently in England:
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backward� bicycle or wearing inverting prisms. Second, the reversion seemed just to “snap� into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Bruce A.
[From Rick Marken (2016.02.14.1310)]
···
Bruce Abbott (2016.02.14.1140 EST)–
BA: So one way that reorganization might work is to vary the connections between error signal and output function. The discovery of another means to the same end (perhaps through trial and error) simply connects the error signal to a different output function, without destroying the previous output function. If later experience shows that one means works under Condition A and another under Condition B, then perceptions of Condition A and Condition B may become associatively connected to the control system in such a way as to cause Output Function X to be used under Condition A and Output Function Y to be used under Condition B, perhaps by setting the gains of the output functions appropriately.
RM: I was going to suggest something similar. The fact that you relearn riding the bike without the inverted lenses so quickly certainly suggests that reorganization does not reset control parameters directly by does something like what you suggest – learning to address the new parameters. The e. coli approach is still likely to be involved but the random variation would be of addresses and not the control parameters directly. Actually, it might be both, since at the very beginning of learning the control parameters themselves would have to be “built”. But perhaps once the control parameters have already been built (as they have for the control systems that control the direction of movement of the handle bars) then reorganization could just work on changing how those existing systems are “addressed” by the higher order systems (that are controlling for balance).Â
RM: Anyway, great observations. The fact relearning of old skills is an importnat observation to consider when developing models of reorganization.Â
BestÂ
Rick
Â
All this is speculative, but it does not depart radically from Bill’s conception. Consider the arm-reorganization demo in LCS III. Here we have a number of joints being randomly connected to an equal number of joint-angle control systems through a set of output weights. Initially each control system may be connected to several joints at once, to varying degrees, through the matrix of weights. Reorganization eventually sorts it all out so that each control system ends up controlling a single joint (or nearly so), the weights to the other outputs being reduced to zero or near-zero. If the control-system outputs went to lower-level control systems instead of directly to the joints, this would match the scheme I described above.
Bruce
Â
On 13 Feb 2016, at 00:40, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.12.1940 EST)]
Â
(Bruce Nevin (2016.02.12.15:17 ET) –
Â
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
And here’s a BBC clip, apparently in England:
It says that after adapting for a week, she reverts to normal in an hour.
Â
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
Â
RM:Â Â our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversedÂ
Â
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
Â
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backwardâ€? bicycle or wearing inverting prisms. Second, the reversion seemed just to “snapâ€? into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.Â
Â
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Â
Bruce A.
–
Richard S. MarkenÂ
Author, with Timothy A. Carey, of  Controlling People: The Paradoxical Nature of Being Human.Â
I think there is also the possibility that in a large hierarchy of control systems, where the lower levels are simply the means to achieve the control of a higher level perceptual variable, reorganisation needs only to tweak with the higher levels in that system (I.e. In the operant example maybe only at the level of ‘relationships’, and then only within the domain of the specific relationships utilised in the situation (between food, and perception of the lever position). Surely the rest of the hierarchy below would be intact anyway. It would also retain a whole load of other lower level systems that could be ‘tried out’ until higher level control is restored. So, I don’t see these examples as a weakness of PCT reorganisation per se, but as a limit of any of us to model this process in multi-layered, multiple-parallel-streamed, control systems. Until we do that, I think we will struggle to know what specifically needs adding, don’t you?
Warren
···
Bruce Abbott (2016.02.14.1140 EST)–
BA: So one way that reorganization might work is to vary the connections between error signal and output function. The discovery of another means to the same end (perhaps through trial and error) simply connects the error signal to a different output function, without destroying the previous output function. If later experience shows that one means works under Condition A and another under Condition B, then perceptions of Condition A and Condition B may become associatively connected to the control system in such a way as to cause Output Function X to be used under Condition A and Output Function Y to be used under Condition B, perhaps by setting the gains of the output functions appropriately.
RM: I was going to suggest something similar. The fact that you relearn riding the bike without the inverted lenses so quickly certainly suggests that reorganization does not reset control parameters directly by does something like what you suggest – learning to address the new parameters. The e. coli approach is still likely to be involved but the random variation would be of addresses and not the control parameters directly. Actually, it might be both, since at the very beginning of learning the control parameters themselves would have to be “built”. But perhaps once the control parameters have already been built (as they have for the control systems that control the direction of movement of the handle bars) then reorganization could just work on changing how those existing systems are “addressed” by the higher order systems (that are controlling for balance).
RM: Anyway, great observations. The fact relearning of old skills is an importnat observation to consider when developing models of reorganization.
Best
Rick
All this is speculative, but it does not depart radically from Bill’s conception. Consider the arm-reorganization demo in LCS III. Here we have a number of joints being randomly connected to an equal number of joint-angle control systems through a set of output weights. Initially each control system may be connected to several joints at once, to varying degrees, through the matrix of weights. Reorganization eventually sorts it all out so that each control system ends up controlling a single joint (or nearly so), the weights to the other outputs being reduced to zero or near-zero. If the control-system outputs went to lower-level control systems instead of directly to the joints, this would match the scheme I described above.
Bruce
On 13 Feb 2016, at 00:40, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.12.1940 EST)]
(Bruce Nevin (2016.02.12.15:17 ET) –
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
Youtube follows it with what looks like a replication by someone in Chicago.
And here’s a BBC clip, apparently in England:
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After reorganization, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backward� bicycle or wearing inverting prisms. Second, the reversion seemed just to “snap� into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Bruce A.
–
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
[From Bruce Abbott (2016.02.16.1020 EST)]
···
From: Warren Mansell [mailto:wmansell@gmail.com]
Sent: Tuesday, February 16, 2016 3:33 AM
To: csgnet@lists.illinois.edu
Subject: Re: reorganization at work
I think there is also the possibility that in a large hierarchy of control systems, where the lower levels are simply the means to achieve the control of a higher level perceptual variable, reorganisation needs only to tweak with the higher levels in that system (I.e. In the operant example maybe only at the level of ‘relationships’, and then only within the domain of the specific relationships utilised in the situation (between food, and perception of the lever position). Surely the rest of the hierarchy below would be intact anyway. It would also retain a whole load of other lower level systems that could be ‘tried out’ until higher level control is restored. So, I don’t see these examples as a weakness of PCT reorganisation per se, but as a limit of any of us to model this process in multi-layered, multiple-parallel-streamed, control systems. Until we do that, I think we will struggle to know what specifically needs adding, don’t you?
Warren
I agree that reorganization should operate only at the level where control is poor, but the reorganization process described in B:CP does not work like that. Supposedly, the reorganization system detects significant deviations from innate reference values in the so-called âintrinsicâ? or âessentialâ? (Ashbyâs term) variables and increases the rate at which changes are made, ecoli fashion, to the control hierarchy. To explain how changes might be restricted to take place only in those control systems that are failing to produce good control over their perceptual signals, it has been suggested that conscious attention is drawn to such malfunctioning systems and that reorganization is restricted to those systems to which attention has been drawn. That proposal seems to be consistent at least to some degree with subjective experience, but until someone develops a generative model that implements such a process, it remains little more than hand-waiving.
Bill saw his proposed reorganization system as a simplest-case model that demonstrates the process. In later work to be carried on by PCT researchers, the process would be elaborated to allow reorganization to be restricted to those systems that failed to adequately control their perceptual signals. Weâre not there yet, not by a long-shot, but the suggestions the Rick and I have offered may provide a starting point and at least call attention to a significant problem in need of a solution.
Bruce
On 14 Feb 2016, at 21:08, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2016.02.14.1310)]
Bruce Abbott (2016.02.14.1140 EST)–
BA: So one way that reorganization might work is to vary the connections between error signal and output function. The discovery of another means to the same end (perhaps through trial and error) simply connects the error signal to a different output function, without destroying the previous output function. If later experience shows that one means works under Condition A and another under Condition B, then perceptions of Condition A and Condition B may become associatively connected to the control system in such a way as to cause Output Function X to be used under Condition A and Output Function Y to be used under Condition B, perhaps by setting the gains of the output functions appropriately.
RM: I was going to suggest something similar. The fact that you relearn riding the bike without the inverted lenses so quickly certainly suggests that reorganization does not reset control parameters directly by does something like what you suggest – learning to address the new parameters. The e. coli approach is still likely to be involved but the random variation would be of addresses and not the control parameters directly. Actually, it might be both, since at the very beginning of learning the control parameters themselves would have to be “built”. But perhaps once the control parameters have already been built (as they have for the control systems that control the direction of movement of the handle bars) then reorganization could just work on changing how those existing systems are “addressed” by the higher order systems (that are controlling for balance).
RM: Anyway, great observations. The fact relearning of old skills is an importnat observation to consider when developing models of reorganization.
Best
Rick
All this is speculative, but it does not depart radically from Billâs conception. Consider the arm-reorganization demo in LCS III. Here we have a number of joints being randomly connected to an equal number of joint-angle control systems through a set of output weights. Initially each control system may be connected to several joints at once, to varying degrees, through the matrix of weights. Reorganization eventually sorts it all out so that each control system ends up controlling a single joint (or nearly so), the weights to the other outputs being reduced to zero or near-zero. If the control-system outputs went to lower-level control systems instead of directly to the joints, this would match the scheme I described above.
Bruce
On 13 Feb 2016, at 00:40, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.12.1940 EST)]
(Bruce Nevin (2016.02.12.15:17 ET) –
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
reorganization Youtube follows it with what looks like a replication by someone in Chicago.
And here’s a BBC clip, apparently in England:
It says that after adapting for a week, she reverts to normal in an hour.
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
RM: our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversed
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the âbackwardâ? bicycle or wearing inverting prisms. Second, the reversion seemed just to âsnapâ? into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.
Simple ecoli-style reorganization wouldnât be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Bruce A.
–
Richard S. Marken
Author, with Timothy A. Carey, of Controlling People: The Paradoxical Nature of Being Human.
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7442 / Virus Database: 4530/11630 - Release Date: 02/15/16
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7442 / Virus Database: 4530/11634 - Release Date: 02/16/16
[Bruce Nevin (2016.02.19.11:19 ET)]
The film seems not to have made it to the Internet. Martin reminds me that it was not Pribram, it was Papert working with J.G. Taylor. It may be that I never saw the clip, but only Martin’s description of it posted to CSG-net a decade or two ago.
With the correction, Martin just sent me his recollection of the experiments. I post it here without explicit permission because of its relevance to the speculations in this thread. Please don’t expect or ask him to elaborate or respond, he’s in the thick of a major writing project.
···
On Tue, Feb 16, 2016 at 10:22 AM, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.16.1020 EST)]
Â
From: Warren Mansell [mailto:wmansell@gmail.com]
Sent: Tuesday, February 16, 2016 3:33 AM
To: csgnet@lists.illinois.edu
Subject: Re: reorganization at work
Â
I think there is also the possibility that in a large hierarchy of control systems, where the lower levels are simply the means to achieve the control of a higher level perceptual variable, reorganisation needs only to tweak with the higher levels in that system (I.e. In the operant example maybe only at the level of ‘relationships’, and then only within the domain of the specific relationships utilised in the situation (between food, and perception of the lever position). Surely the rest of the hierarchy below would be intact anyway. It would also retain a whole load of other lower level systems that could be ‘tried out’ until higher level control is restored. So, I don’t see these examples as a weakness of PCT reorganisation per se, but as a limit of any of us to model this process in multi-layered, multiple-parallel-streamed, control systems. Until we do that, I think we will struggle to know what specifically needs adding, don’t you?
Warren
Â
I agree that reorganization should operate only at the level where control is poor, but the reorganization process described in B:CP does not work like that. Supposedly, the reorganization system detects significant deviations from innate reference values in the so-called “intrinsic� or “essential� (Ashby’s term) variables and increases the rate at which changes are made, ecoli fashion, to the control hierarchy. To explain how changes might be restricted to take place only in those control systems that are failing to produce good control over their perceptual signals, it has been suggested that conscious attention is drawn to such malfunctioning systems and that reorganization is restricted to those systems to which attention has been drawn. That proposal seems to be consistent at least to some degree with subjective experience, but until someone develops a generative model that implements such a process, it remains little more than hand-waiving.
Bill saw his proposed reorganization system as a simplest-case model that demonstrates the process. In later work to be carried on by PCT researchers, the process would be elaborated to allow reorganization to be restricted to those systems that failed to adequately control their perceptual signals. We’re not there yet, not by a long-shot, but the suggestions the Rick and I have offered may provide a starting point and at least call attention to a significant problem in need of a solution.
Bruce
On 14 Feb 2016, at 21:08, Richard Marken rsmarken@gmail.com wrote:
[From Rick Marken (2016.02.14.1310)]
Â
Bruce Abbott (2016.02.14.1140 EST)–
Â
BA: So one way that reorganization might work is to vary the connections between error signal and output function. The discovery of another means to the same end (perhaps through trial and error) simply connects the error signal to a different output function, without destroying the previous output function. If later experience shows that one means works under Condition A and another under Condition B, then perceptions of Condition A and Condition B may become associatively connected to the control system in such a way as to cause Output Function X to be used under Condition A and Output Function Y to be used under Condition B, perhaps by setting the gains of the output functions appropriately.
Â
RM: I was going to suggest something similar. The fact that you relearn riding the bike without the inverted lenses so quickly certainly suggests that reorganization does not reset control parameters directly by does something like what you suggest – learning to address the new parameters. The e. coli approach is still likely to be involved but the random variation would be of addresses and not the control parameters directly. Actually, it might be both, since at the very beginning of learning the control parameters themselves would have to be “built”. But perhaps once the control parameters have already been built (as they have for the control systems that control the direction of movement of the handle bars) then reorganization could just work on changing how those existing systems are “addressed” by the higher order systems (that are controlling for balance).Â
Â
RM: Anyway, great observations. The fact relearning of old skills is an importnat observation to consider when developing models of reorganization.Â
Â
BestÂ
Â
Rick
Â
Â
All this is speculative, but it does not depart radically from Bill’s conception. Consider the arm-reorganization demo in LCS III. Here we have a number of joints being randomly connected to an equal number of joint-angle control systems through a set of output weights. Initially each control system may be connected to several joints at once, to varying degrees, through the matrix of weights. Reorganization eventually sorts it all out so that each control system ends up controlling a single joint (or nearly so), the weights to the other outputs being reduced to zero or near-zero. If the control-system outputs went to lower-level control systems instead of directly to the joints, this would match the scheme I described above.
Bruce
Â
On 13 Feb 2016, at 00:40, Bruce Abbott bbabbott@frontier.com wrote:
[From Bruce Abbott (2016.02.12.1940 EST)]
Â
(Bruce Nevin (2016.02.12.15:17 ET) –
Â
Thanks for that. The last frame claims that with the ‘glasses’ off he reverted to normal vision in a few (“wenige”) minutes.
reorganization Youtube follows it with what looks like a replication by someone in Chicago.
And here’s a BBC clip, apparently in England:
It says that after adapting for a week, she reverts to normal in an hour.
Â
I was looking for the video that Bill posted on CSG-net. I thought it was Pribram riding the bike–tumbling and laughing about it. Couldn’t find it in a search associating his name with anything about prisms etc.
Â
RM:Â Â our lower level control systems – the one’s we use to do everything, including riding a bike – have been build to deal with a visual world that is not reversedÂ
Â
The visual hierarchy recognizes transitions and contributes to references at the top of the motor hierarchy, which is also the transition level. After, the visual system recognizes transitions again. Do lower levels of the motor hierarchy also get reference input from the visual modality? From my reading of Henry’s paper on the basal ganglia, I suspect not, but it’s possible.
Â
BA: A couple of observations suggest that something other than simple ecoli-style reorganization is going on. First, it takes less time to revert to the former organization than to create the new one, whether riding the “backwardâ€? bicycle or wearing inverting prisms. Second, the reversion seemed just to “snapâ€? into place (clearly in the bicycle case, at least) after a certain amount of experience trying to deal with the reverted relationships.Â
Â
Simple ecoli-style reorganization wouldn’t be expected to work like that. The old organization would be destroyed as reorganization blindly varied connections and parameters until control was reestablished. Reverting to the previous organization would then require just as must time and effort as learning the reversal.
Â
Bruce A.
Â
–
Richard S. MarkenÂ
Author, with Timothy A. Carey, of  Controlling People: The Paradoxical Nature of Being Human.Â
Â
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7442 / Virus Database: 4530/11630 - Release Date: 02/15/16
No virus found in this message.
Checked by AVG - www.avg.com
Version: 2016.0.7442 / Virus Database: 4530/11634 - Release Date: 02/16/16