Color Bill Powers looking for a way out

[From Bruce Abbott (990404.1935 EST)]

Bill Powers (990404.1334 MDT) --

Bruce Abbott (990404.1335 EST)

I wrote a long reply and trashed it. Arguing with you about these
experiments is just as pointless now as it was the first time I gave up
doing it. The experiments you describe are so poorly conceived, the data
taken so badly analyzed, and your conclusions derived by such mysterious
processes, that I despair of explaining to you what is wrong with them. If
you could understand any explanation I would give, you wouldn't have done
the experiments that way in the first place, or admired whoever did them.

Ah, yes. Now the experiments are "poorly conceived," the data "badly
analyzed," the conclusions derived by "mysterious processes." A flurry of
unsupported assertions designed to pooh-pooh the results and the conclusions
I derive from them. Furthermore, I am now judged to be too stupid to
understand your explanations. Well, be that as it may, how about rendering
the explanation just for the benefit of any (if such exist) who may have the
intellectual horsepower to follow you. After all, I've taken the trouble to
lay out a coherent position, and all you have to do is explain why, from
your perspective, it falls short. Shouldn't be all that difficult.

I'll just give one example, the problem of measuring "rate of lever
pressing," a measure of a repetitive behavior. I thought you had understood
my original critique of the practice of using session averages for this
measure: total presses per session divided by session duration. If pressing
behavior were one simple kind of behavior that simply speeded up and slowed
down, a session average might (at a stretch) be a useful measure. But when
the animal's behavior changes _qualitatively_ during a session, so it
spends time away from the lever engaged in other activities, session
averages become completely invalid, giving true measures neither of
lever-pressing rates nor of any other actions that occur in the same session.

Apparently you are only on "send." I've already indicated that these
within-session changes do not occur on the schedule in question.
Nevertheless, you bring them up as an "example," as if they actually had
anything to do with the results I reported. Magnificent!

Now I do the same thing with another pendulum, but this time I notice, just
after the 50th swing, that the pendulum is squeaking. So I stop it, go get
an oil-can, oil it, go put the oil-can away, return to the pendulum, and
start it up again, resuming the swing count at 51. Then, after the 100th
swing I read the timer as before and compute the period in seconds per swing.

Do you understand why the measure of the second pendulum's period is
invalid? And if that measure is invalid how can it be valid to use a
session average to measure bar-pressing rate when the rat spends part of
the session sleeping?

How about if the pendulum's stopping is not an artifact imposed from outside
its system but a characteristic of its overall, unimpeded behavior? What if
it's one of those fancy "chaotic" pendulums that shows no regular period but
does manifest a long-term quasi-stability in the sense of giving evidence of
a strange attractor? The average period, taken over a run of sufficient
length, might be reasonably stable under given conditions, and change in a
regular way with changes in conditions. Perhaps this would provide a closer
analogy, as the tendency for the rat to do things other than lever-press
varies systematically with such parameters as the average schedule interval.

Do you understand why the measure of this pendulum's period is NOT invalid
as one measure of such a pendulum's performance?

Of course, the notion that the rat is sleeping during part of the session
assumes facts not in evidence. The rat does not spend time sleeping on VI
schedules. You are still confusing the VI results with those on CRF, where
the rat rapidly fills up on pellets and loses interest in earning them.

I thought we had reached agreement about this. But today, your post
indicates that you don't see why a session average doesn't give a true
measure of repetition rate. You haven't budged a millimeter.

It isn't that I haven't "budged." It's that the problem to which you refer
is not relevant to the data at hand.

Let's just drop the whole thing, Bruce. There are too many more problems
just like this, and you won't admit that any of them is a problem. Color me
fed up.

Or in other words, when you can't stand the heat, you get out of the kitchen.

Bill, your whole reply is nothing but a cheap cop-out. I wager that you
have no really convincing rational argument to make. These vague
hand-wavings are nothing but an obvious attempt to dismiss the observations,
and the conclusions based on them, without having to actually present any
argument. I call your bluff.

Regards,

Bruce

[From Bill Powers (990405.0817 MDT)]

Bruce Abbott (990404.1935 EST)--

Apparently you are only on "send." I've already indicated that these
within-session changes do not occur on the schedule in question.
Nevertheless, you bring them up as an "example," as if they actually had
anything to do with the results I reported. Magnificent!

But you said, I think, that the measures of pressing rate under the VI
schedule were obtained from whole-session averages, and for the
cyclic-ratio experiment from averages over the period for a given ratio.
Are you now telling me they were computed press by press?

Would the variations occur if the animal could "earn" as much food and in
the same period of time as in the FR-1 schedules? We are talking not only
about a change from FR-1 to FR-n, or FR to VI, but also about a change from
enough food to satisfy the animal temporarily to a far smaller amount of food.

What we are talking about, in PCT terms, is what happens when the CV is
brought by behavior nearly to its reference level. If a schedule prevents
this from happening we would expect the appropriate result, due not to the
schedule but to the difference in effort required to obtain a given amount
of food.

How about if the pendulum's stopping is not an artifact imposed from outside
its system but a characteristic of its overall, unimpeded behavior? What if
it's one of those fancy "chaotic" pendulums that shows no regular period but
does manifest a long-term quasi-stability in the sense of giving evidence of
a strange attractor? The average period, taken over a run of sufficient
length, might be reasonably stable under given conditions, and change in a
regular way with changes in conditions. Perhaps this would provide a closer
analogy, as the tendency for the rat to do things other than lever-press
varies systematically with such parameters as the average schedule interval.

In reverse order: I would claim that it varies with the error signal. At
very low error, other behaviors will occur because there is no error or
other errors are greater; at very high error, other behaviors will occur
because of reorganization or commencement of a search pattern. Between
these extremes, the animal will maintain its food intake where it wants it
by behaving as necessary, if the experimental design permits.

As to other fanciful kinds of pendulums, they do not affect what we mean by
"rate," or the way any physical scientist would measure it. Your attempts
to rationalize a really stupid way of trying to measure rates shows only
how strongly you defend the way you and your EAB colleagues reason (or
perhaps a better word would be rationalize).

Do you understand why the measure of this pendulum's period is NOT invalid
as one measure of such a pendulum's performance?

No. No matter how complex the waveform, a pendulum's period is the time
between repetitions of the same or nearly the same pattern. Except in
EAB-land, where it is anything that supports your conclusions.

Of course, the notion that the rat is sleeping during part of the session
assumes facts not in evidence. The rat does not spend time sleeping on VI
schedules. You are still confusing the VI results with those on CRF, where
the rat rapidly fills up on pellets and loses interest in earning them.

I agree -- on VI schedules, as you have pointed out several times (as well
as on the cyclic ratio schedule), the animals receive far less food than in
the FR -1 schedule with which I am somewhat familiar. I would not attribute
the changes in behavior to the changes in schedule until I had tried
equalizing the amount of food received across schedules.

In your field, changing schedules entails changes in many other important
conditions, yet the custom is to attribute any differences in behavior
entirely to the change in schedule. This doesn't seem to bother you. You
may recall that during our joint investigation I kept cautioning you about
the frequent changes in conditions (I stopped counting after the 17th major
change within 6 months in what was supposedly one continuous experiment).
Yet you kept coming up with still more variations as you pursued
ill-conceived notions one after another (for example, your switching the
home cages to the metabolic jars in the vain hope of measuring all intakes
and outflows despite not being able to measure evaporation, powder from
food, or gas exchanges through the lungs). Also, we started with juvenile
rats and ended with rats who had lived half their lifespans. I could not
understand why you persisted in doing this, or having done it, how you
could draw any conclusions from the data. As far as I am concerned, we
never did a usable experiment.

I thought we had reached agreement about this. But today, your post
indicates that you don't see why a session average doesn't give a true
measure of repetition rate. You haven't budged a millimeter.

It isn't that I haven't "budged." It's that the problem to which you refer
is not relevant to the data at hand.

Oh, if the problem were relevant you would agree with me? Today's post
belies that possibility.

I wish you would come right out and say that you measured the rates of
pressing on a second-by-second basis under all the schedules, and did NOT
use session averages in any of them. If long-term averages were used, I
dispute the validity of the results.

Let's just drop the whole thing, Bruce. There are too many more problems
just like this, and you won't admit that any of them is a problem. Color me
fed up.

Or in other words, when you can't stand the heat, you get out of the kitchen.

No. Whan I can't stand the intransigence I change to a more interesting
channel.

Best,

Bill P.

[From Bruce Abbott (990405.1255 EST)]

Bill Powers (990405.0817 MDT) --

Bruce Abbott (990404.1935 EST)

Apparently you are only on "send." I've already indicated that these
within-session changes do not occur on the schedule in question.
Nevertheless, you bring them up as an "example," as if they actually had
anything to do with the results I reported. Magnificent!

But you said, I think, that the measures of pressing rate under the VI
schedule were obtained from whole-session averages, and for the
cyclic-ratio experiment from averages over the period for a given ratio.
Are you now telling me they were computed press by press?

Are you telling me that you don't remember? As you well know, Bill, the
computer record of the data comprises the time of occurrence of every
lever-press. In addition there are video-tapes of the rat at work on these
schedules.

Would the variations occur if the animal could "earn" as much food and in
the same period of time as in the FR-1 schedules? We are talking not only
about a change from FR-1 to FR-n, or FR to VI, but also about a change from
enough food to satisfy the animal temporarily to a far smaller amount of food.

Yes. But as these session-variations are absent when the rate of pellet
delivery is low, they cannot explain why lever-pressing is less frequent
under long-interval VI schedules than under short-interval VI schedules.
The latter is a nice, systematic, repeatable finding.

What we are talking about, in PCT terms, is what happens when the CV is
brought by behavior nearly to its reference level. If a schedule prevents
this from happening we would expect the appropriate result, due not to the
schedule but to the difference in effort required to obtain a given amount
of food.

I have good evidence from other sources that effort is not much of a factor.

How about if the pendulum's stopping is not an artifact imposed from outside
its system but a characteristic of its overall, unimpeded behavior? What if
it's one of those fancy "chaotic" pendulums that shows no regular period but
does manifest a long-term quasi-stability in the sense of giving evidence of
a strange attractor? The average period, taken over a run of sufficient
length, might be reasonably stable under given conditions, and change in a
regular way with changes in conditions. Perhaps this would provide a closer
analogy, as the tendency for the rat to do things other than lever-press
varies systematically with such parameters as the average schedule interval.

In reverse order: I would claim that it varies with the error signal. At
very low error, other behaviors will occur because there is no error or
other errors are greater; at very high error, other behaviors will occur
because of reorganization or commencement of a search pattern. Between
these extremes, the animal will maintain its food intake where it wants it
by behaving as necessary, if the experimental design permits.

We don't disagree here; there has to be some kind of time-sharing of
behavioral resources going on -- the rat can't be pressing the lever _and_
exploring the corner of the chamber opposite the lever at the same time, for
example. The relative intensities of error signals would be one way for
such a system to prioritize its outputs.

As to other fanciful kinds of pendulums, they do not affect what we mean by
"rate," or the way any physical scientist would measure it. Your attempts
to rationalize a really stupid way of trying to measure rates shows only
how strongly you defend the way you and your EAB colleagues reason (or
perhaps a better word would be rationalize).

I don't believe I rationalized anything. If, for example, more and more
time is being spent away from the lever and food cup as the animal fills up
on pellets (as is observed on CRF), then this shows up in the data as a
highly systematic decline in rate, averaged over an appropriate time-period
( e.g., a minute or so). I never argued that it would be unimportant to
know what is the cause of this change, and in fact my own experiments were
carefully designed to collect the relevant data.

Do you understand why the measure of this pendulum's period is NOT invalid
as one measure of such a pendulum's performance?

No. No matter how complex the waveform, a pendulum's period is the time
between repetitions of the same or nearly the same pattern. Except in
EAB-land, where it is anything that supports your conclusions.

Of course, the notion that the rat is sleeping during part of the session
assumes facts not in evidence. The rat does not spend time sleeping on VI
schedules. You are still confusing the VI results with those on CRF, where
the rat rapidly fills up on pellets and loses interest in earning them.

I agree -- on VI schedules, as you have pointed out several times (as well
as on the cyclic ratio schedule), the animals receive far less food than in
the FR -1 schedule with which I am somewhat familiar. I would not attribute
the changes in behavior to the changes in schedule until I had tried
equalizing the amount of food received across schedules.

Which would automatically introduce a second confound -- time. In the ratio
study, I took the trouble to plot the data both as a function of time and
number of pellets consumed in order to separate these factors.

In your field, changing schedules entails changes in many other important
conditions, yet the custom is to attribute any differences in behavior
entirely to the change in schedule.

Not so.

This doesn't seem to bother you.

It would if it were true.

You
may recall that during our joint investigation I kept cautioning you about
the frequent changes in conditions (I stopped counting after the 17th major
change within 6 months in what was supposedly one continuous experiment).
Yet you kept coming up with still more variations as you pursued
ill-conceived notions one after another (for example, your switching the
home cages to the metabolic jars in the vain hope of measuring all intakes
and outflows despite not being able to measure evaporation, powder from
food, or gas exchanges through the lungs). Also, we started with juvenile
rats and ended with rats who had lived half their lifespans. I could not
understand why you persisted in doing this, or having done it, how you
could draw any conclusions from the data. As far as I am concerned, we
never did a usable experiment.

You have evidently forgotten that this was an exploratory study --
essentially an extended series of pilot studies. It wasn't working as
originally designed, so I launched into a series of changes to try to
understand what was going on and find a procedure that would generate better
data. To judge my abilities as an experimenter by this is simply unfair.
That said, I do not share in your opinion as to the effect of those
manipulations. In each case sufficient time was allowed for behavior to
stabilize under the new conditions that it is extremely doubtful that there
was any residual effect of the previous conditions. With the benefit of
hindsight I could of course now design and carry out a much cleaner study.

I thought we had reached agreement about this. But today, your post
indicates that you don't see why a session average doesn't give a true
measure of repetition rate. You haven't budged a millimeter.

It isn't that I haven't "budged." It's that the problem to which you refer
is not relevant to the data at hand.

Oh, if the problem were relevant you would agree with me? Today's post
belies that possibility.

This is too vague for me. What are you talking about?

I wish you would come right out and say that you measured the rates of
pressing on a second-by-second basis under all the schedules, and did NOT
use session averages in any of them. If long-term averages were used, I
dispute the validity of the results.

See above. I came right out and said it.

Let's just drop the whole thing, Bruce. There are too many more problems
just like this, and you won't admit that any of them is a problem. Color me
fed up.

Or in other words, when you can't stand the heat, you get out of the kitchen.

No. Whan I can't stand the intransigence I change to a more interesting
channel.

Translation: if you can't get the other party to agree with you, you give up
in frustration. It never occurs to you that your position might be the one
that should change.

Regards,

Bruce

P.S. I think it might be well to remind you that I have _not_ argued that
the VI data contradict PCT. I've argued that they contradict a specific PCT
model you once proposed.

[From Bill Powers (990406.0926 MDT)]

Bruce Abbott (990405.1255 EST) --

I need to regroup here. It's been too long since I had any data in front of
me and I'm misremembering things. Let's focus on the VI data you sent in
'97 and see if we can't at least talk about the same thing. I started
looking at it yesterday and it will take some time to figure out again what
it's about.

When I first recieved the data and your program, I wrote a modification of
your "cumulative record" display so I could look directly at the
interresponse times (the reciprocal of rate) versus time. Here's what I
read from the plots (eyeball estimates):

Run 97001.145 97001.149

Lower band 0.067-1 sec 0.067-0.7 sec
Upper band about 3-4 sec about 2 sec
total reinf. 210 87
total presses 4571 4997

The "lower band" and "upper band" are distributions of interresponse times
that remain essentially the same during the whole run. They are
distinguishable but not sharply separated. The 0.067-sec lowerf limit
reflects the 30/sec sampling rate. The upper band, as I interpret it,
reflects the collection time, which is clearly less when the total
reinforcements are smaller. This indicates to me that the rats are
consuming the pellets in more of a hurry when they are getting less food.
Also, there is an increase in mean rate of pressing, indicated by the
general shift to shorter interresponse times in the lower band.

On the other hand, I don't even know if these plots are for the same rat,
so all that may be spurious. How about reviewing for me the experimental
conditions and the meaning of the data?

My program "Cumrecp1.pas" is attached.

Best,

Bill P.
program CumRecP1;
{ Plots a cumulative record based on data files created by the FRATIO
  and RANDOMVI real-time experiment programs.
  Created 1/2/98 by B. Abbott

  Plots ln(interval) versus time. WTP 980103
}

uses dos, crt, frdata, graph, grutils;

var
  ch: char;
  Filename, Path: pathstr;
  R, P: dataptr;
  RTon, RToff, PTon, PToff, Event: timesptr;
  i, j, nR, nP: word;
  NWords, SubjYr, SubjNo, Year, Month, Day, Session, Ratio,
  Hour, Minute, Resp, Rft: word;

procedure InitScreen;
begin
  initgraphics;
  graphmode := getmaxmode;
  setgraphmode(graphmode);
  hsize := getmaxx + 1;
  vsize := getmaxy + 1;
  hcenter := hsize div 2;
  vcenter := vsize div 2;
end;

procedure WriteData(nD: word; On, Off: timesptr);
var
i: word;

begin
  writeln(nD:10);
  for i := 1 to nD do
    begin
      write(On^[i]:8, Off^[i]:8);
      if (i Mod 4 = 0) then writeln;
    end;
  writeln('Successfully wrote ', nD, ' values');
end;

function IRT(i: word; On, Off: timesptr): longint;
begin
  if i = 1 then IRT := On^[i] else IRT := On^[i] - On^[i-1];
end;

function DUR(i: word; On, Off: timesptr): longint;
begin
  DUR := Off^[i] - On^[i];
end;

procedure PlotIt;
var
  i, j, x, y: word;
  istr: string;
begin
  str(Session, istr);
  outtextxy(10,35, 'Sess = '+istr);
  str(nR, istr);
  outtextxy(10,50, 'Rsp = '+istr);
  str(nP, istr);
  outtextxy(10,65, 'Rft = '+istr);
  y := Vsize;
  j := 1;
  for i := 1 to nR do
    begin
      dec(y); if y = 0 then y := Vsize;
      x := (RTon^[i] * Hsize) div 72000;
      putpixel(x,y, white);
      if RTon^[i] = PTon^[j] then
        begin
          inc(j);
          line(x,y, x+5,y+5);
        end;
    end;
end;

procedure PlotlogInt;
var
  i, j, x, y,color: word;
  yr: real;
  istr: string;
begin
  str(Session, istr);
  outtextxy(10,35, 'Sess = '+istr);
  str(nR, istr);
  outtextxy(10,50, 'Rsp = '+istr);
  str(nP, istr);
  outtextxy(10,65, 'Rft = '+istr);
  y := Vsize;
  j := 1;
  for i := 1 to nR do
    begin
      y := vsize - round(50.0*ln(IRT(i,RTon,RToff)));
      x := (RTon^[i] * Hsize) div 72000;
      putpixel(x,y, white);
    end;
  j := 1;
  for i := 0 to 8
    do
    begin
      if i > 0 then j := j + j;
      y := j;
      str(y/30:4:3,istr);
      yr := ln(j);
       outtextxy(hsize - 100,vsize - 8 - round(50*yr),istr);
    end;
    outtextxy(hsize - 100, 100, 'IRT, sec');
    outtextxy(vcenter,0,'LOG IRT VERSUS TIME');
end;

begin
  ClrScr;
  Path := 'd:\tp\ratexp\vidata\';
  Filename := '97001.149';
  Filename := Path + Filename;
  if FileExists(Filename) then
    begin
      InitScreen;
      new(R); new(P);
      new(RTon); new(RToff);
      new(PTon); new(PToff);
      writeln('Reading File ', Filename, ' . . .');
      ReadData(Filename, SubjYr, SubjNo, Session,
            Ratio, Year, Month, Day, Hour, Minute, Nwords,
            Resp, Rft, R, P);
      writeln('Decoding data . . .');
      Decode(R, Nwords, RTon, RToff, nR);
      Decode(P, Nwords, PTon, PToff, nP);
      PlotLogInt;
      dispose(PToff); dispose(PTon);
      dispose(RToff); dispose(RTon);
      dispose(P); dispose(R);
      ch := readkey;
      CloseGraph;
    end
  else
    begin
      write('File Not Found . . .');
      ch := readkey;
    end;
end.