EP Model -- Delphi version, revised -- again!

[From Bruce Abbott (2014.02.09.1420 EST)]

···

From: Control Systems Group Network (CSGnet) [mailto:CSGNET@LISTSERV.ILLINOIS.EDU] On Behalf Of Adam Matic
Sent: Sunday, February 09, 2014 1:26 PM
To: CSGNET@LISTSERV.ILLINOIS.EDU
Subject: Re: EP Model – Delphi version, revised – again!

On Sun, Feb 9, 2014 at 6:56 PM, Warren Mansell wmansell@gmail.com wrote:

To what degree are there different labels for the same thing (e.g. Lambas and reference values).

AM: I’m interested in this too. Are there counterparts to R and C in some of PCT models?

BA: I assume so – for example, Bill’s 1999 model could increase joint stiffness by raising the references for muscle force on both sides of a joint. But I need to review Bill’s model before I say more.

Bruce

[From Rick Marken (2014.02.09.1600)]

Martin Taylor (2014.02.08.16.46)–

If you did the simulation in a spreadsheet, could you send it to me?

RM: OK, I’ve attached the spreadsheet. The model is run as a macro (by hitting ctrl-r) but you don’t have to enable the macros if you don’t want to (or can’t, as is the case in some versions of Excel). I have plotted the results of the model runs for you.

The top graph gives the main results and the bottom one shows a comparison of the behavior of the P-control and EP models in greater detail. Both graphs show a variable corresponding to limb angle (Y axis) changing over time (the X axis). The top graph shows a step disturbance (the red line labeled d) to limb angle; this disturbance is equivalent to stepping through the weights added to the limb in Bruce’s simulation, where the weights can be increased in steps of 1 kg.

The blue line, labeled cv, is the behavior of the proportional (P) control model controlling limb angle as the CV. The loop gain of the model is .1; a P control model is only stable when its gain is <1.0. The green line is the behavior of a comparable EP model (comparable in terms of how much of the disturbance is prevented from having an effect on limb angle) as observed in Bruce’s “EP Model” simulation. The main comparison is between these two functions.

It was suggested that the behavior of the EP model could be that of a P control system. This spreadsheet shows that the behavior of a P control and EP model are quite different. This is most easily seen in the lower graph which shows a more detailed picture (looking at only 10 time sample points) of the behavior of limb angle in response to a disturbance for the P control model (cv) and EP model(ep).

Note that for the EP model limb angle moves continuously toward the new equilibrium point after disturbance; there is no evidence of “push back” against the disturbance. (This is the behavior that is seen in the continuous graph of limb angle in Bruce’s EP model each time the disturbance is increased by 1 kg). The response of the P control model is quite different; limb angle is initially increased by the disturbance but the model “pushes back” to get the limb back as far as it can toward the reference angle (of zero). The gain of the P control system is so low that it doesn’t get the limb angle back very close to the reference but it is clearly working toward that end; there is push back.

I’ve set the simulation up so that the P control system brings the cv (limb angle) back to a limb angle that is equivalent to the equilibrium point limb angle achieved by the EP model (ep). This makes it easier to see that the behavior of a P control model is quite different than that of an EP model, which suggests strongly that the EP model is not a control model at all.

The fact that the EP model is not a control model is even more evident when one compares the behavior of the EP model to that of a control model that better represents what actually happens when increasing step disturbances of weight are applied to a limb. This is shown by the yellow line (labelled icv to indicate that these are the variations in limb angle that result when limb angle is controlled by an integral control system). Except for the brief “jerks” that occur at the points where the step disturbance increases, the control system keeps the limb angle right at the reference angle (0 in this case) protected from the increasing step disturbances. This corresponds to the behavior you would actually observe in a human. You could see this by by having someone hold a bag in their hand at a fixed angle from their body and then drop one pound weights one at a time into the bag. I think you will find that the behavior of the person’s arm angle over time will looks a lot more like the yellow plot (icv) than the green one (ep).

This, by the way, is the correct way to test the models to see if they match real behavior.What you want to do is see if the model (EP or control) controls like the human does. So far the EP model has not been tested to see if it controls like humans. Indeed, the tests that have been conducted on the EP model have asked the participants to try not to control. As Bruce Abbot said in a recent post regarding the tests of the EP model in the Lan/Zhu paper:

BA: The participant has been instructed not to attempt to hold the initial joint

angle, but to allow the joint angle to change as it will. The angle at which
the forearm stops is a measure of the equilibrium point.

I read this in the Lan/Zhu paper myself and I was gobsmacked; I couldn’t believe what I was reading. They are not testing to see if the model controls; they are testing to see if the model acts like a mass on springs whose spring constants are set by a machine at a control panel – a control panel that is presumably inside the participants who would otherwise be inclined to control the position of their limb. So the methodology used to test the EP model confirms (without my having had to have done all this testing) that it is not a model of control (of how people “hold the initial joint angle”, for example). The EP model is a model of the behavior of purposeless (non-control) systems. So the only mistake the EP theorists are making is in applying their model to the behavior of living systems. An easy mistake to make,apparently;-)

Best

Rick

Control vs EP.xlsm (26.8 KB)

···


Richard S. Marken PhD
www.mindreadings.com

The only thing that will redeem mankind is cooperation.

                                               -- Bertrand Russella

[From Rick Marken (2014.02.09.2015)]

···

On Sun, Feb 9, 2014 at 2:24 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi Rick, I disagree. I think the PCT model should be built from first principles of PCT, not inferred from the mistaken workings of a completely different theory…

RM: I suggested using the version of the EP model that controls as the PCT model because the cachet of the EP model seems to come from its apparent fidelity to the underlying neurophysiology. But really any model that controls can be considered a PCT alternative to the EP model because the EP model is not a control model; so its just not a model of the behavior of living systems.

PCT is not really an alternative to EP because EP doesn’t even explain what PCT explains: the behavior of living systems, which of course, we know is a control process. So really any model that controls is a better model of behavior than EP because EP doesn’t control. That’s really the only message necessary; we don’t have to show that PCT provides a better model of limb movement than does the EP model. All we have to show is that EP can’t possibly be even close to the right model of limb movement, even with all the important sounding physiological language.

EP strikes me as being kind of like the (possibly mythical) model of the bumblebee – based on all the best physical principles – that proved that a bumblebee can’t fly. EP is a model of limb movement – based on all the best physiological principles regarding the properties of nerves and muscles – that proves that the position of a limb can’t be controlled. I think Bruce’s modeling demonstrates the failure of the EP quite clearly. It’s a tremendously important piece of work. And I hope it becomes one of the prominent papers in the book you’re putting together.

Best

Rick

Warren

Sent from my iPhone

On 8 Feb 2014, at 18:48, Richard Marken rsmarken@GMAIL.COM wrote:

[From Rick Marken (2014.02.08.1050)]

On Sat, Feb 8, 2014 at 10:29 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi Rick, that’s what I was thinking. I guess the plan would be, after programming the next threshold control model, to reproduce their anatomy exactly but use a PCT model instead and compare?

RM: Yes, I agree. But I think the PCT model was already written by

Bruce. It was the second (or third, I’ve lost track) version of

Bruce’s attempts to reproduce the EP model; it’s the version of the

model that clearly controlled perceptions of muscle length (the

lambda*'s) relative to references for these lengths (the lambda’s). If

Bruce can reproduce that model we can look and see why it controls and

the “correct” EP model doesn’t.

Another interesting result of this modeling exercise is that it shows

that the muscle length control version of the EP model (the PCT

version) and the non-control (current) version of the model behave

exactly the same and can’t be distinguished without doing the test for

the controlled variable (ie. applying varying weight – torque–

disturbances to limb). Without disturbances, both models change their

angles smoothly in response to smooth changes in R. But in the control

version of the model, R determines the values of the lambda references

that “command” input (desired muscle length); in the non-control

version these lambda “references” command outputs (flexor/extensor

torque).

Best

Rick

Warren

Sent from my iPhone

On 8 Feb 2014, at 17:17, Richard Marken rsmarken@GMAIL.COM wrote:

[From Rick Marken (2014.02.08.0920)]

On Sat, Feb 8, 2014 at 8:05 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi both, I may be barking up the wrong tree here, but when there is a weight in hand, why should co-activating both muscles be necessary - why doesn’t just the flexor muscle get the force it needs correct to keep the weight still?

RM: That’s one question. Another (possibly related) is "why is the

joint angle not moving back toward the reference angle after the

disturbance is applied"? You will notice that even when the weight is

just 1 kg, this weight disturbance causes the arm to move to a new

angle and remain there; there is not even the slightest sign of

resistance to the disturbance; no sign that the angle is being

returned to a reference state.

While the size of the deviation produced by a 1 kg weight disturbance

is smaller as “gain” (the value of C) increases, there is no evidence

that there is a system working to counter this small disturbance. The

decreased size of the deviation produced by a weight of 1 kg is

exactly what is expected if the same torque were applied to an

increasingly “stiff” joint. Compare this to what happens when limb

position is controlled; the 1 kg weight disturbance is almost

instantly countered, even when the gain of the control system is

relatively low.

What you see with a low gain (poor control) system after a 1 kg weight

is applied is the limb returning toward but not quite reaching the

original angle. There is no sign of this at all in the EP model at any

setting of “gain” (C). When you apply the weight the limb moves to a

new position and stays there; with higher “gain” the deviation

produced by the weight is smaller but it is still a fixed deviation.

This is exactly what is expected if one were applying a constant 1 kg

force to a stick anchored in the ground. If the stick is thin the

deviation produced by the constant force will be large but constant;

if the stick is thick, the deviation will be small but still constant.

This is what I think is going on with the EP model; increasing C is

equivalent to increasing the thickness of a stick in the ground. No

control at all.

So I retract my judgement that this “final” version of the EP model

controls. There is no behavioral evidence that the system controls;

the decreasing effect of a 1 kg weight disturbance to the limb is

exactly what is expected if the weight disturbance is being applied

to a joint that is increasingly stiff.

I had said to Bruce:

RM: So if this is the right version of the EP model, I think you can say

that it is a control model, that perceptions of muscle length are the

controlled variables and that it controls much more poorly that the real

system; and we have to find out why.

And Bruce replied:

BA?: I think we’ve finally got it nailed.

I guess not. I believe I was wrong in concluding that the model is a

control model. This “final” version of the EP model is not controlling

at all. I was fooled by looking at the inner workings of the model,

which suggests that the flexor is increasing its upward torque in

response to the downward torque produced by the weight. It may be

doing that but this doesn’t seem to be compensating for the

disturbance at all.

I think this may have to do with your point, Warren.There’s something

funny going on with the flexor/extensor torques. First, both are

always positive. What’s that about? And the increase in C (“gain”)

leads to an increase in both torques, suggesting that what C is

doing is not increasing the gain of a control system but increasing

the stiffness of the joint, in terms of the strength of the offsetting

torques. In other words, increasing C looks like it does something

equivalent to tensing your arm muscles.

But just based on the behavior of this model I think it’s clear that

it doesn’t control at all, in the sense the the observed effect of

disturbances are exactly what are expected on physical grounds; a

disturbance is completely effective; there is no resistance to the

disturbance at all.

So I think we can conclude that the EP model can move the limb to

reference positions but it doesn’t do this in a controlled manner. So

the EP model can’t be considered a serious model of real limb

movement.

Best

Rick

It seems to be reproducing the

data in Lan and Zhu (2007), Figure 6 and handles the effect of the added

weight correctly.

Richard S. Marken PhD

www.mindreadings.com

The only thing that will redeem mankind is cooperation.

                                             -- Bertrand Russell

Richard S. Marken PhD

www.mindreadings.com

The only thing that will redeem mankind is cooperation.

                                              -- Bertrand Russell


Richard S. Marken PhD
www.mindreadings.com
The only thing that will redeem mankind is cooperation.
– Bertrand Russell

Hi Rick, thanks for that. I agree that the test of whether the EP controls is critical and publishable and that PCT models advance on that. I think it would still be desirable to then test a PCT model, either making the same or different assumptions about the physiology.

Thanks Bruce for the further details of the EP model and a summary of your work finding/developing a PCT model.

Warren

···

On Sun, Feb 9, 2014 at 2:24 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi Rick, I disagree. I think the PCT model should be built from first principles of PCT, not inferred from the mistaken workings of a completely different theory…

RM: I suggested using the version of the EP model that controls as the PCT model because the cachet of the EP model seems to come from its apparent fidelity to the underlying neurophysiology. But really any model that controls can be considered a PCT alternative to the EP model because the EP model is not a control model; so its just not a model of the behavior of living systems.

PCT is not really an alternative to EP because EP doesn’t even explain what PCT explains: the behavior of living systems, which of course, we know is a control process. So really any model that controls is a better model of behavior than EP because EP doesn’t control. That’s really the only message necessary; we don’t have to show that PCT provides a better model of limb movement than does the EP model. All we have to show is that EP can’t possibly be even close to the right model of limb movement, even with all the important sounding physiological language.

EP strikes me as being kind of like the (possibly mythical) model of the bumblebee – based on all the best physical principles – that proved that a bumblebee can’t fly. EP is a model of limb movement – based on all the best physiological principles regarding the properties of nerves and muscles – that proves that the position of a limb can’t be controlled. I think Bruce’s modeling demonstrates the failure of the EP quite clearly. It’s a tremendously important piece of work. And I hope it becomes one of the prominent papers in the book you’re putting together.

Best

Rick

Warren

Sent from my iPhone

On 8 Feb 2014, at 18:48, Richard Marken rsmarken@GMAIL.COM wrote:

[From Rick Marken (2014.02.08.1050)]

On Sat, Feb 8, 2014 at 10:29 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi Rick, that’s what I was thinking. I guess the plan would be, after programming the next threshold control model, to reproduce their anatomy exactly but use a PCT model instead and compare?

RM: Yes, I agree. But I think the PCT model was already written by

Bruce. It was the second (or third, I’ve lost track) version of

Bruce’s attempts to reproduce the EP model; it’s the version of the

model that clearly controlled perceptions of muscle length (the

lambda*'s) relative to references for these lengths (the lambda’s). If

Bruce can reproduce that model we can look and see why it controls and

the “correct” EP model doesn’t.

Another interesting result of this modeling exercise is that it shows

that the muscle length control version of the EP model (the PCT

version) and the non-control (current) version of the model behave

exactly the same and can’t be distinguished without doing the test for

the controlled variable (ie. applying varying weight – torque–

disturbances to limb). Without disturbances, both models change their

angles smoothly in response to smooth changes in R. But in the control

version of the model, R determines the values of the lambda references

that “command” input (desired muscle length); in the non-control

version these lambda “references” command outputs (flexor/extensor

torque).

Best

Rick

Warren

Sent from my iPhone

On 8 Feb 2014, at 17:17, Richard Marken rsmarken@GMAIL.COM wrote:

[From Rick Marken (2014.02.08.0920)]

On Sat, Feb 8, 2014 at 8:05 AM, Warren Mansell wmansell@gmail.com wrote:

WM: Hi both, I may be barking up the wrong tree here, but when there is a weight in hand, why should co-activating both muscles be necessary - why doesn’t just the flexor muscle get the force it needs correct to keep the weight still?

RM: That’s one question. Another (possibly related) is "why is the

joint angle not moving back toward the reference angle after the

disturbance is applied"? You will notice that even when the weight is

just 1 kg, this weight disturbance causes the arm to move to a new

angle and remain there; there is not even the slightest sign of

resistance to the disturbance; no sign that the angle is being

returned to a reference state.

While the size of the deviation produced by a 1 kg weight disturbance

is smaller as “gain” (the value of C) increases, there is no evidence

that there is a system working to counter this small disturbance. The

decreased size of the deviation produced by a weight of 1 kg is

exactly what is expected if the same torque were applied to an

increasingly “stiff” joint. Compare this to what happens when limb

position is controlled; the 1 kg weight disturbance is almost

instantly countered, even when the gain of the control system is

relatively low.

What you see with a low gain (poor control) system after a 1 kg weight

is applied is the limb returning toward but not quite reaching the

original angle. There is no sign of this at all in the EP model at any

setting of “gain” (C). When you apply the weight the limb moves to a

new position and stays there; with higher “gain” the deviation

produced by the weight is smaller but it is still a fixed deviation.

This is exactly what is expected if one were applying a constant 1 kg

force to a stick anchored in the ground. If the stick is thin the

deviation produced by the constant force will be large but constant;

if the stick is thick, the deviation will be small but still constant.

This is what I think is going on with the EP model; increasing C is

equivalent to increasing the thickness of a stick in the ground. No

control at all.

So I retract my judgement that this “final” version of the EP model

controls. There is no behavioral evidence that the system controls;

the decreasing effect of a 1 kg weight disturbance to the limb is

exactly what is expected if the weight disturbance is being applied

to a joint that is increasingly stiff.

I had said to Bruce:

RM: So if this is the right version of the EP model, I think you can say

that it is a control model, that perceptions of muscle length are the

controlled variables and that it controls much more poorly that the real

system; and we have to find out why.

And Bruce replied:

BA?: I think we’ve finally got it nailed.

I guess not. I believe I was wrong in concluding that the model is a

control model. This “final” version of the EP model is not controlling

at all. I was fooled by looking at the inner workings of the model,

which suggests that the flexor is increasing its upward torque in

response to the downward torque produced by the weight. It may be

doing that but this doesn’t seem to be compensating for the

disturbance at all.

I think this may have to do with your point, Warren.There’s something

funny going on with the flexor/extensor torques. First, both are

always positive. What’s that about? And the increase in C (“gain”)

leads to an increase in both torques, suggesting that what C is

doing is not increasing the gain of a control system but increasing

the stiffness of the joint, in terms of the strength of the offsetting

torques. In other words, increasing C looks like it does something

equivalent to tensing your arm muscles.

But just based on the behavior of this model I think it’s clear that

it doesn’t control at all, in the sense the the observed effect of

disturbances are exactly what are expected on physical grounds; a

disturbance is completely effective; there is no resistance to the

disturbance at all.

So I think we can conclude that the EP model can move the limb to

reference positions but it doesn’t do this in a controlled manner. So

the EP model can’t be considered a serious model of real limb

movement.

Best

Rick

It seems to be reproducing the

data in Lan and Zhu (2007), Figure 6 and handles the effect of the added

weight correctly.

Richard S. Marken PhD

www.mindreadings.com

The only thing that will redeem mankind is cooperation.

                                             -- Bertrand Russell

Richard S. Marken PhD

www.mindreadings.com

The only thing that will redeem mankind is cooperation.

                                              -- Bertrand Russell


Richard S. Marken PhD
www.mindreadings.com
The only thing that will redeem mankind is cooperation.
– Bertrand Russell

[Martin Taylor 2014.02.13.10.47]

[From Rick Marken (2014.02.08.1205)]

Martin Taylor (2014.02.08.14.10) --
MT: You see the return after the initial move only if the output stage of the
controller is an integrating stage.

RM: Not according to my simulations. After the onset of a step
disturbance, the variable controlled by a proportional (P) control
system immediately starts moving back to the reference state; how far
back towards the reference state it moves depends on gain but even a
low gain P controller will start moving back toward the reference
state immediately after the onset of the disturbance.

You are quite right. I was thinking of a P system without transport lag, which doesn't exist in real life or in a simulation. With lag, the system with gain <1.0, and a fixed reference value of zero, oscillates with declining amplitude around d*G/(1+G) after a step of the disturbance from zero to d.

The problem with simulating a pure P controller is that somewhere in a real-life loop the effect of an input impulse (disturbance or reference) must spread out over time. If it doesn't, what you get is simply an impulse (or a step) that cycles around endlessly, either exponentially diminishing (if loop gain <1) or increasing (if loop gain >1). At a "critical gain" of 1.0, the oscillation amplitude neither increases nor diminishes. If, say, the environmental feedback path smears over time the effect of output, then the critical gain is higher. But a pure P controller still oscillates around its asymptotic target value of d*G/(1+G).

Since muscles apply forces and forces accelerate masses or keep them moving against friction and viscosity, the environmental feedback path often provides integration appropriate for controlling perception of velocity or position. If it does, then the biological output stage of the control loop need not.

  If the EP model
were a P controller of arm angle we would see the trace for arm angle
start to move back toward the reference state immediately after
applying the weight disturbance. We don't see that at all. Indeed, the
response of the EP model to a step disturbance of weight is exactly
what is expected of a P controller with 0 gain, ie. the behavior of a
non-control system.

That may be so. I haven't looked at it for myself, so I'll take your word for it. The real question is whether the EP model correctly models how the biological systems it purports to emulate behave. Not all the component legs of a control loop are themselves control loops.

Martin

[From Bruce Abbott (2014.02.13.1450 EST]

Rick Marken (2014.02.09.1600) –

RM: Note that for the EP model limb angle moves continuously toward the new equilibrium point after disturbance; there is no evidence of “push back” against the disturbance. (This is the behavior that is seen in the continuous graph of limb angle in Bruce’s EP model each time the disturbance is increased by 1 kg). The response of the P control model is quite different; limb angle is initially increased by the disturbance but the model “pushes back” to get the limb back as far as it can toward the reference angle (of zero). The gain of the P control system is so low that it doesn’t get the limb angle back very close to the reference but it is clearly working toward that end; there is push back.

RM: I’ve set the simulation up so that the P control system brings the cv (limb angle) back to a limb angle that is equivalent to the equilibrium point limb angle achieved by the EP model (ep). This makes it easier to see that the behavior of a P control model is quite different than that of an EP model, which suggests strongly that the EP model is not a control model at all.

RM: The fact that the EP model is not a control model is even more evident when one compares the behavior of the EP model to that of a control model that better represents what actually happens when increasing step disturbances of weight are applied to a limb. This is shown by the yellow line (labelled icv to indicate that these are the variations in limb angle that result when limb angle is controlled by an integral control system). Except for the brief “jerks” that occur at the points where the step disturbance increases, the control system keeps the limb angle right at the reference angle (0 in this case) protected from the increasing step disturbances. This corresponds to the behavior you would actually observe in a human. You could see this by by having someone hold a bag in their hand at a fixed angle from their body and then drop one pound weights one at a time into the bag. I think you will find that the behavior of the person’s arm angle over time will looks a lot more like the yellow plot (icv) than the green one (ep).

RM: This, by the way, is the correct way to test the models to see if they match real behavior.What you want to do is see if the model (EP or control) controls like the human does. So far the EP model has not been tested to see if it controls like humans. Indeed, the tests that have been conducted on the EP model have asked the participants to try not to control. As Bruce Abbot said in a recent post regarding the tests of the EP model in the Lan/Zhu paper:

BA: The participant has been instructed not to attempt to hold the initial joint
angle, but to allow the joint angle to change as it will. The angle at which
the forearm stops is a measure of the equilibrium point.

RM: I read this in the Lan/Zhu paper myself and I was gobsmacked; I couldn’t believe what I was reading. They are not testing to see if the model controls; they are testing to see if the model acts like a mass on springs whose spring constants are set by a machine at a control panel – a control panel that is presumably inside the participants who would otherwise be inclined to control the position of their limb. So the methodology used to test the EP model confirms (without my having had to have done all this testing) that it is not a model of control (of how people “hold the initial joint angle”, for example). The EP model is a model of the behavior of purposeless (non-control) systems. So the only mistake the EP theorists are making is in applying their model to the behavior of living systems. An easy mistake to make,apparently;-)

When I suggested that the EP model may be behaving like a low-gain proportional controller, the proportional controller I had in mind was one that acts through the same environment that the EP model does. In the EP model, a change in R produces a torque on the joint. This torque, divided by the moment of inertia, produces an angular acceleration, which is integrated with each time-step to get the velocity, which is integrated to get the position. In your spreadsheet simulation, you could have simulated the effect of these integrations on the stability of the system by employing our usual leaky integrator output. This would permit the controller to have a gain higher than 1.0.

Be that as it may, your simulation does show that the proportional controller begins to resist the initial effect of the disturbance in the next time-step and does “push back,” as you say, although not by much due to the low gain. The muscles of the EP model just stretch under the increased load – but not entirely passively (see below).

The EP model does have a feedback loop; in essence it determines something equivalent to the spring constant of the muscle. Without feedback the muscle would simply stretch passively when an external torque was applied to the limb. Feedback from the muscle spindles increases the activity of the alpha motor neurons, increasing muscle tension. The muscle will stretch under the external torque until the increasing counterforce developed by the muscle is enough to prevent further stretching.

In the EP model, changing R alters the lambda value of each muscle, the threshold length at which the muscle will begin to resist further stretching. When no external torques are present, the torques developed by the muscles will move the limb to the specified angular position if there is no co-contraction. Changing lambda effectively changes the zero-point of the spring, the point at which no counterforce is generated by the muscle, equivalent to the resting position of a spring that not being stretched.

I haven’t gotten far enough along in my research to know anything about proposals for the systems that set R and C. What if the physiologists have it right and the muscles are in fact contracted by changing alpha motor neuron thresholds as the EP model proposes? If that were the case, then a PCT model would have to act through the same mechanism. I can imagine a level 2 system that would set the R and C values while receiving its own proprioceptive input from the muscles, tendons, and joints. Such a system would automatically compensate for the springiness of the muscles by employing R and C values that brought the joint to the specified angle against external loads, so long as the loads were not overpowering.

With respect to the methods used to determine equilibrium points, those were specifically designed to determine whether the changes in joint angle following various degrees of unloading could be understood in terms of simple changes in alpha motor neuron thresholds, changes necessitated by the requirement to maintain position against the external torque prior to partial unloading. The change in joint angle after partial unloading is supposed to reflect only the involuntary reflex adjustments, which is why participants were told not to attempt to voluntarily resist the joint movement. Voluntary movement presumably involves participants changing R and C values via higher levels in the nervous system.

Until I learn more about how the EP model has been applied (e.g., to deal with voluntary control of joint position), I can’t say whether it is a mistaken model or a reasonable portrayal of how things work at the physiological actuator level. If the latter, then any model – command and compute, PCT, or something else, will have to deal with exerting control over joint angle via this EP mechanism.*

*That said, there are other models out there that propose somewhat different mechanisms at this level. The EP model may not be the right one to capture the actual physiology.

Bruce

[From Bruce Abbott (2014.02.13.1450 EST]

···

Rick Marken (2014.02.09.1600) –

RM: Note that for the EP model limb angle moves continuously toward the new equilibrium point after disturbance; there is no evidence of “push back” against the disturbance. (This is the behavior that is seen in the continuous graph of limb angle in Bruce’s EP model each time the disturbance is increased by 1 kg). The response of the P control model is quite different; limb angle is initially increased by the disturbance but the model “pushes back” to get the limb back as far as it can toward the reference angle (of zero). The gain of the P control system is so low that it doesn’t get the limb angle back very close to the reference but it is clearly working toward that end; there is push back.

RM: I’ve set the simulation up so that the P control system brings the cv (limb angle) back to a limb angle that is equivalent to the equilibrium point limb angle achieved by the EP model (ep). This makes it easier to see that the behavior of a P control model is quite different than that of an EP model, which suggests strongly that the EP model is not a control model at all.

RM: The fact that the EP model is not a control model is even more evident when one compares the behavior of the EP model to that of a control model that better represents what actually happens when increasing step disturbances of weight are applied to a limb. This is shown by the yellow line (labelled icv to indicate that these are the variations in limb angle that result when limb angle is controlled by an integral control system). Except for the brief “jerks” that occur at the points where the step disturbance increases, the control system keeps the limb angle right at the reference angle (0 in this case) protected from the increasing step disturbances. This corresponds to the behavior you would actually observe in a human. You could see this by by having someone hold a bag in their hand at a fixed angle from their body and then drop one pound weights one at a time into the bag. I think you will find that the behavior of the person’s arm angle over time will looks a lot more like the yellow plot (icv) than the green one (ep).

RM: This, by the way, is the correct way to test the models to see if they match real behavior.What you want to do is see if the model (EP or control) controls like the human does. So far the EP model has not been tested to see if it controls like humans. Indeed, the tests that have been conducted on the EP model have asked the participants to try not to control. As Bruce Abbot said in a recent post regarding the tests of the EP model in the Lan/Zhu paper:

BA: The participant has been instructed not to attempt to hold the initial joint
angle, but to allow the joint angle to change as it will. The angle at which
the forearm stops is a measure of the equilibrium point.

RM: I read this in the Lan/Zhu paper myself and I was gobsmacked; I couldn’t believe what I was reading. They are not testing to see if the model controls; they are testing to see if the model acts like a mass on springs whose spring constants are set by a machine at a control panel – a control panel that is presumably inside the participants who would otherwise be inclined to control the position of their limb. So the methodology used to test the EP model confirms (without my having had to have done all this testing) that it is not a model of control (of how people “hold the initial joint angle”, for example). The EP model is a model of the behavior of purposeless (non-control) systems. So the only mistake the EP theorists are making is in applying their model to the behavior of living systems. An easy mistake to make,apparently;-)

When I suggested that the EP model may be behaving like a low-gain proportional controller, the proportional controller I had in mind was one that acts through the same environment that the EP model does. In the EP model, a change in R produces a torque on the joint. This torque, divided by the moment of inertia, produces an angular acceleration, which is integrated with each time-step to get the velocity, which is integrated to get the position. In your spreadsheet simulation, you could have simulated the effect of these integrations on the stability of the system by employing our usual leaky integrator output. This would permit the controller to have a gain higher than 1.0.

Be that as it may, your simulation does show that the proportional controller begins to resist the initial effect of the disturbance in the next time-step and does “push back,â€? as you say, although not by much due to the low gain. The muscles of the EP model just stretch under the increased load – but not entirely passivelyy (see below).

The EP model does have a feedback loop; in essence it determines something equivalent to the spring constant of the muscle. Without feedback the muscle would simply stretch passively when an external torque was applied to the limb. Feedback from the muscle spindles increases the activity of the alpha motor neurons, increasing muscle tension. The muscle will stretch under the external torque until the increasing counterforce developed by the muscle is enough to prevent further stretching.

In the EP model, changing R alters the lambda value of each muscle, the threshold length at which the muscle will begin to resist further stretching. When no external torques are present, the torques developed by the muscles will move the limb to the specified angular position if there is no co-contraction. Changing lambda effectively changes the zero-point of the spring, the point at which no counterforce is generated by the muscle, equivalent to the resting position of a spring that not being stretched.

I haven’t gotten far enough along in my research to know anything about proposals for the systems that set R and C. What if the physiologists have it right and the muscles are in fact contracted by changing alpha motor neuron thresholds as the EP model proposes? If that were the case, then a PCT model would have to act through the same mechanism. I can imagine a level 2 system that would set the R and C values while receiving its own proprioceptive input from the muscles, tendons, and joints. Such a system would automatically compensate for the springiness of the muscles by employing R and C values that brought the joint to the specified angle against external loads, so long as the loads were not overpowering.

With respect to the methods used to determine equilibrium points, those were specifically designed to determine whether the changes in joint angle following various degrees of unloading could be understood in terms of simple changes in alpha motor neuron thresholds, changes necessitated by the requirement to maintain position against the external torque prior to partial unloading. The change in joint angle after partial unloading is supposed to reflect only the involuntary reflex adjustments, which is why participants were told not to attempt to voluntarily resist the joint movement. Voluntary movement presumably involves participants changing R and C values via higher levels in the nervous system.

Until I learn more about how the EP model has been applied (e.g., to deal with voluntary control of joint position), I can’t say whether it is a mistaken model or a reasonable portrayal of how things work at the physiological actuator level. If the latter, then any model – command and compute, PCT, or something else, wwill have to deal with exerting control over joint angle via this EP mechanism.*

*That said, there are other models out there that propose somewhat different mechanisms at this level. The EP model may not be the right one to capture the actual physiology.

Bruce