Looking for tracking fit parameter values

[Martin Taylor 2017.03.23.23.23]

To check out some ideas for a different kind of dynamic control analysis, I would like to collect parameter values from reasonably well fitted simulations of tracking studies. The nature of the controlled variable and its level in the Powers hierarchy are irrelevant, because I would like to have some variety. If you have done such a study (and I know lots of you have), and you used a leaky integrator as your output function in a single loop model, I would like to know your results if the parameters include transport lag, integrator gain rate, and slowing factor (or leak rate) (both rates either per sample with sample rate, or per second), plus any other parameters you may have included in the model. I must have the three named parameters. Any others would be gravy.

I don't need the raw data, just the parameters that produced a good fit to the human performance for a task with enough difficulty that tracking wasn't too near perfection.

I have some of my own, but they aren't by themselves what I am looking for. Any help would be appreciated, no matter what perception was being tracked.

Martin

[Martin Taylor 2017.03.24.13.07]

I forgot to mention that if you have data, I would need to know the quality of control achieved by the human, in the form of the control ratio -- the ratio of effect that the disturbance would have had in the absence of control to the controlled variation of the controlled perception. Usually this is the ratio of RMS values, but sometimes variance is used. Let me know which you do use. If you could also describe the way the disturbance was generated, that would also be useful,

I thank all contributors (as yet none).

Martin

···

On 2017/03/23 11:37 PM, Martin Taylor wrote:

[Martin Taylor 2017.03.23.23.23]

To check out some ideas for a different kind of dynamic control analysis, I would like to collect parameter values from reasonably well fitted simulations of tracking studies. The nature of the controlled variable and its level in the Powers hierarchy are irrelevant, because I would like to have some variety. If you have done such a study (and I know lots of you have), and you used a leaky integrator as your output function in a single loop model, I would like to know your results if the parameters include transport lag, integrator gain rate, and slowing factor (or leak rate) (both rates either per sample with sample rate, or per second), plus any other parameters you may have included in the model. I must have the three named parameters. Any others would be gravy.

I don't need the raw data, just the parameters that produced a good fit to the human performance for a task with enough difficulty that tracking wasn't too near perfection.

I have some of my own, but they aren't by themselves what I am looking for. Any help would be appreciated, no matter what perception was being tracked.

Martin

[From Rick Marken (2017.03.25.1100)]

···

[Martin Taylor 2017.03.23.23.23]

MT: To check out some ideas for a different kind of dynamic control analysis, I would like to collect parameter values from reasonably well fitted simulations of tracking studies. The nature of the controlled variable and its level in the Powers hierarchy are irrelevant, because I would like to have some variety. If you have done such a study (and I know lots of you have), and you used a leaky integrator as your output function in a single loop model, I would like to know your results if the parameters include transport lag, integrator gain rate, and slowing factor (or leak rate) (both rates either per sample with sample rate, or per second), plus any other parameters you may have included in the model. I must have the three named parameters. Any others would be gravy.

RM: I guess that let’s me off the hook because I can’t think of any model fitting I did where I included transport lag as a parameter. I know I have used transport lag as a parameter in some work. I think I did try it at some point when modeling some of the object interception data but I don’t recall it making much of a difference. Anyway, I could send you the object interception model parameters if you like – just gain and slowing. And I could get measures of stability of control but that may be a tad difficult given the non-linear controlled variables.Â

Best

Rick

Â

I don’t need the raw data, just the parameters that produced a good fit to the human performance for a task with enough difficulty that tracking wasn’t too near perfection.

I have some of my own, but they aren’t by themselves what I am looking for. Any help would be appreciated, no matter what perception was being tracked.

Martin


Richard S. MarkenÂ

"Perfection is achieved not when you have nothing more to add, but when you
have nothing left to take away.�
                --Antoine de Saint-Exupery

[Martin Taylor 2017.03.25.14.10]

[From Rick Marken (2017.03.25.1100)]

Thanks. I find I didn't use transport lag as a parameter either when

I fit my sleep data. That’s why I said that my own data don’t help.
But TrackAnalyze (LCS III Demo 4-1) does include lag, Maybe someone
has run it and got some results.

Martin
···
          [Martin

Taylor 2017.03.23.23.23]

          MT: To check out some ideas for a different kind of

dynamic control analysis, I would like to collect
parameter values from reasonably well fitted simulations
of tracking studies. The nature of the controlled variable
and its level in the Powers hierarchy are irrelevant,
because I would like to have some variety. If you have
done such a study (and I know lots of you have), and you
used a leaky integrator as your output function in a
single loop model, I would like to know your results if
the parameters include transport lag, integrator gain
rate, and slowing factor (or leak rate) (both rates either
per sample with sample rate, or per second), plus any
other parameters you may have included in the model. I
must have the three named parameters. Any others would be
gravy.

          RM: I guess that let's me off the hook because I can't

think of any model fitting I did where I included
transport lag as a parameter. I know I have used transport
lag as a parameter in some work. I think I did try it at
some point when modeling some of the object interception
data but I don’t recall it making much of a difference.
Anyway, I could send you the object interception model
parameters if you like – just gain and slowing. And I
could get measures of stability of control but that may be
a tad difficult given the non-linear controlled
variables.