Science or Mush?

[From Dag Forssell (950629 1500)]

Bruce Abbott (950628.2020 EST)

Bill Powers (950628.0945 MDT) --

Bruce Abbott (950628.0900 EST)

However, these authors felt that these results might be

    explained by the complex interaction between running rate and
    postreinforcement pause observed in the data. The overall rate
    would result from the combination of these two factors.

This is a pretty feeble comment on an observation that goes

    directly against the fundamental assumptions behind
    reinforcement theory itself.

  Well, these guys aren't reinforcement theorists, they're just

    humble experimentalists reporting their data. I give them
    credit for perceiving and pointing out the apparent problem for
    reinforcement theory; apparently they felt less than
    comfortable going much further. Their objective was to
    discover what relationships occur in this situation, and that's
    what they did. They were willing to leave the theorizing to
    someone else.

I continue to appreciate and enjoy the discussion of reinforcement
theory. A great number of differences have been clarified. The
discussions are important since the basic notions of Reinforcement
Theory is engrained in the minds of not only psychologists, but
most lay people as well. You are playing the role of interpreter
and devil's advocate well, Bruce. So well, that I often have to
pinch myself to recall your stated objective, which as I understand
it is to merely offer arguments that are to be expected from those
with a reinforcement point of view.

Your statement above -- "Well, these guys aren't reinforcement
theorists, they're just humble experimentalists reporting their
data." -- offered with an apparently straight face is a sad comment
indeed. Surely these are psychologists with PhD's??? Was this
research report reviewed by a committee of expert peers before
publication? If so, did they not understand either?? Is
Reinforcement Theory so obscure and convoluted that a division of
labor as you indicate is called for? It comes across as a giant
cop-out in a way that I think unthinkable in the physical sciences.
The obvious implication is that the experimentalist accepts no
responsibility for the theory; the theoretician accepts no
responsibility for the experiment. The theoretician can pick and
choose what experiments he or she wishes to comment on, free to
select only those that support the theory for further comment.
Experiments not supportive of theory, in fact disproving it, die a
quiet death. This charade hardly qualifies Reinforcement Theory or
EAB as a science. It cannot be taken seriously. From a PCT
perspective we know that by misleading armies of well-meaning
students of psychology for many decades, it has actually done
mankind an enormous disservice.

Bruce, you have claimed that you are a PCTer, but claiming it does
not make you one. Please save me from having to pinch myself
bloody by following through on your promise to Rick two weeks back:

[Rick Marken (950616.1400)]
Maybe on your vacation you could try to think up some ways to

    show that consequences don't actually strengthen the responses
    that produce them. Then I wouldn't feel so bad about spending
    the same time developing the models that show that it can sure
    look like they can.

  [Bruce Abbott (950616.1720 EST)]
  O.K., I'll think about it. Th' check is in the mail, right? (;-

Bruce, you are obviously smart and witty. You have studied PCT,
spending time with DEMO1 and DEMO2. You know how to program, but
math and computer programs are not physical science and do not
necessarily represent physical mechanisms, especially when nebulous
magic like "probabilities" are part of the program.

You will be a tremendous asset to PCT and Humanity when you choose
to go beyond defending or explaining the non-science of
Reinforcement Theory in what seems like 100% of your posts to
actually supporting and promoting PCT and making constructive
suggestions as you indicated above (950616.1720 EST) in at least a
few of your posts. I will welcome some posts from you where you
show in-depth understanding of PCT, not just Reinforcement Theory.
I am very excited about the prospect of having another qualified
scientist on the PCT team and can hardly wait for you to
demonstrate that you are a PCTer by showing that you understand and
(as a consequence of that understanding) are making a personal
commitment to PCT.

Best, Dag

[From Dag Forssell (950909 1130)]

[Bruce Abbott (950908.2000 EST)]

   Yep! I'll take a good-ol' slimy, wiggley animal over an

     elementary particle any day -- snark over quark, as it were.
     What I want to know is, why do physicists call their field one
     of the "hard" sciences? It's damned _easy_, I'd say, compared
     to unraveling the mysteries of living, behaving organisms.
     Even then, they haven't got it right: relativity and quantum
     mechanics are fundamentally incompatible theories. And now
     what do they give us? _The Emperor's New Mind_ and
     fish-particle physics.

Bruce, back in July you indicated that you planned to answer my
posts on Science or Mush? I am back in the States, and note that
you never did. (I am very busy with a translating assignment which
puts bread on our table, so I limit myself to lurking.)

Unraveling the mysteries of behaving organisms IS easy, once you
shift your perspective on that Necker cube away from the hopeless,
unproven, perspective of Reinforcement Theory to the solid ground
of PCT. PCT lays the foundation for a "hard" science of life. I agree
with your apparent assessment that this makes things MUCH easier.

Now that you have confessed to having studied physics, I am even
more anxious to have your comments on Science or Mush? (and
PCT in my bones).

Best, Dag

[From Bruce Abbott (951030.1715 EST)]

    "The time has come," the Walrus said, "to speak of many things,
     Of shoes and ships and sealing wax, of cabbages and kings,
     And why the sea is boiling hot, and whether pigs have wings."

From Dag Forssell (951029 1445) --

Bruce Abbott (951028.1335 EST)] PCT in His Bones

   Dag has reminded me several times now that I have not replied

    to his posts of last July on "Science or Mush" and "PCT in My
    Bones." The general thrust of both posts was that it seemed
    to Dag that I really don't quite "get" PCT, that somehow I
    seem still to be clinging to many of my old beliefs, beliefs
    which, to a true PCT convert, are obviously wrong.

The general thrust was that Reinforcement Theory is mush, not
science. You are avoiding the basic issue in your superficial
treatment of "PCT in My Bones", which I made an effort to clarify
without invoking magic or religious faith.

O.K., my mistake. I thought there was another issue beneath this one that
was driving your questions, and that it would be simpler and easier for me
to respond to that issue rather than to the one lying on the surface. So
let us discuss the issue you raised. In order to respond clearly, I will
have to do a little setting up.

Back in 1898, Edward L. Thorndike performed a classic experiment in which a
cat was placed in a wooden box having a door on one side whose latch could
be released from within the box, if only the cat knew what to do. The idea
was to determine how the cat learned to solve the problem. To make sure
that the cat would be motivated to get out of the box, Thorndike arranged to
perform the experiment immediately before the cat's dinner-time, and placed
a fish or a bowl of milk just outside the box. Thorkdike put the cat in the
box, closed the latch, and started his stopwatch. When the cat emerged from
the box, Thorndike stopped the watch and recorded the time. After allowing
the cat a bite of fish or a few laps of milk, Thorndike put the cat back in
the box. This procedure was repeated until the cat was consistently
escaping immediately, upon being returned to the box.

Thorndike observed that, when the cat was first placed in the box, it did a
lot of different things: try to squeeze between the slats of the box, lick
its fur, investigate the corners and ceiling. Eventually, apparently quite
by accident, the cat happened to make the required movement (e.g., push
against a pole sticking up in the center of the box) and the door flopped
open. It would seem that the cat had at last discovered the solution to the
puzzle, and would hereafter emerge immediately from the box whenever placed
there. However, this did not prove to be the case. Instead, on Trial 2,
the cat went back to doing the kinds of things it had been observed to do on
Trial 1. Eventually it again brushed against the pole, the door opened, and
the cat escaped. Something similar happened on Trial 3, on Trial 4, and so
on; only by observing the whole series could one see that a subtle trend was
gradually developing: with a lot of ups and downs to obscure this trend, the
cat was slowly coming to emit the required behavior a little sooner, and a
little sooner still, until, after many trials, the cat was finally emerging
quickly from the box.

To explain this gradual and subtle change in performance across trials,
Thorndike appealed to a process of "selecting and connecting." Little was
known about brain processes, so Thorndike stuck fairly close to his
observations when trying to account for the change in behavior. He began by
noting that the different behaviors observed when the cat was first placed
in the box occurred with different frequencies, and that some tended to
occur earlier than others. This observation could be accounted for if it
were assumed that, when confronted with this particular combination of
conditions (which Thorndike called the "situation") different sorts of
behavior tended to occur with differing probabilities, which could be
arranged into a hierarchy from most probable (those occuring often and,
usually, early) to least probable (those occuring rarely and, usually, late.
He assumed that some of these activities may have been instinctive reactions
to the situation (e.g., confinement may innately potentiate certain classes
of behavior, such as exploration), whereas others may have been the products
of the cat's experience with similar situations in the past (these were
street-wise alley cats he was testing), but the source of the activities was
of less importance than their liklihood within the box.

Of course, it would be much better to have a mechanism that would account in
purely physical terms for the observed behavior patterns and their differing
frequencies, so as to derive the observed patterns from first principles,
but such a model was not available. Fortunately, it is possible to make at
least some progress without one (just as our PCT models are able to work
well enough without requiring a detailed knowledge of how the perceptual
system identifies and locates an object within the visual field; we just
assume that it can and go on from there).

Thorndike's latch mechanism required a behavior that was not very likely.
This action was thus near the bottom of the probablility hierarchy. Whether
the cat happened to nudge the rod at any given moment was to all appearances
a matter of chance. Under the circumstances, the probability of this result
would seem to be a reasonable variable to infer from response latency. An
increase in probability would tend to reduce the time required for the
response to appear, but would guarentee such a result only over the long run.

Thorndike reasoned that a particular event--getting out of the box and into
the food--had an effect on the behavior it reliably followed (bumping the
rod), and this effect was to increase the probability of that behavior in
the situation in which it had occurred. He envisioned that this consequence
of the cat's behavior automatically and mechanically strengthened an
associative connection encoded somewhere in the cat's brain, so that the
stimulus conditions identifying the situation now were able to potentiate
this particular behavior a bit more strongly than before.

But what sort of consequences would have this effect? Thorndike suggested
that the consequence must qualify as a "satisfying state of affairs" TO THE
CAT. And how would one know that a given event qualifies as a satisfying
state of affairs? By observing how the cat behaved with respect to it. If
the cat voluntarily approached the fish (it did) and resisted being
separated from the fist (it did), then the fish was a satisfying state of
affairs, and, when made an immediate consequence of a given bit of behavior,
such a consequence would increase the liklihood that, when placed into the
same situation, the cat would produce the same bit of behavior. Thorndike
proposed that such a change constituted a law of behavior. He called it the
"law of effect."

The law of effect explained why the cat did not immediately begin to produce
the action required to free itself from the box once it had succeeded for
the first time: the probability of the act was only slightly increased by
its being followed by the satisfying state of affairs. With many
higher-probability behaviors going on in the box, it was unlikely that the
still-low-probability behavior of pushing the rod would be seen again any
time soon. Only after many "strengthenings" would the probability of the
behavior be elevated to the point where it would be the most likly thing the
cat would do, and thus most likely the thing the cat would do first.

The details concerning how all this is accomplished physiologically in the
nervous sytem are lacking, but Thorndike's explanation is clearly a
mechanistic one and is consistent with the empirical evidence. Mostly it is
a description of what is observed to happen. The changes in response
probability are inferred directly from the changes in the observed
frequencies of various activities. As an empirical statement of
relationships, the law of effect has a status similar to that of Boyle's law
in physics: a generalized statement of observed relationships. It is,
however, somewhat weaker than Boyle's law in that Thorndike did not attempt
a quantitative analysis of the observed changes.

As I recall, Newton was quite vexed by the problem inherent in his view of
gravity, which is action-at-a-distance. A lot of other people of Newton's
day were unhappy with it, too. They said it appealed to magic. How does
the sun prevent the earth from flying off into space? Calling it gravity
does not explain it. The effect on the probability of a response Thorndike
observed when the response was followed by his "satisfying state of affairs"
was similarly mysterious to Thorndike, although he allowed himself to
speculate that it might have something to do with "associative connections"
in the brain. Magical? Certainly. But if a bit of magic turns science
into mush, then not only is Thorndike's law of effect mush, so is Newton's
law of gravity.

If it is quantitative analysis that makes the difference, then Newton has
the advantage, but only because the law of gravity turned out to be
independent of so many other things. Only mass and distance were important.
If Thorndike's cats had decreased their escape-times in as orderly a fashion
as marbles rolling down an inclined plane increase their speeds, a
quantitative law of effect would have emerged easily from the observations.
They did not, for there were too many other variables at work that Thorndike
could not see and measure. Newton was very bright and an excellent
mathematician. But his main advantage was that the relationships he
observed were simple and easy to reproduce.

At this point, Dag, I will stop to ask you what you think about what I've
said so far in this little exposition of reinforcement theory. Do you think
that Thorndike's approach was a sensible one? Was it, in your opinion,
science or mush? Explain.

Regards,

Bruce

Science or Mush?

[From Dag Forssell (951031 0945)]

[Bruce Abbott (951030.1715 EST)]

   O.K., my mistake. I thought there was another issue beneath

     this one that was driving your questions, and that it would be
     simpler and easier for me to respond to that issue rather than
     to the one lying on the surface. So let us discuss the issue
     you raised. In order to respond clearly, I will have to do a
     little setting up.

Whatever other issue you suspected, I hold no malice toward you.
You have made great contributions in the past year, often drawing
out the best in Bill P. I hear from Rick that you are a very
pleasant person. What I have wanted to challenge are some of your
systems concepts, so carefully constructed by your mind over a long
time, still resisting disturbances.

···

-------------------

   At this point, Dag, I will stop to ask you what you think

     about what I've said so far in this little exposition of
     reinforcement theory. Do you think that Thorndike's approach
     was a sensible one? Was it, in your opinion, science or mush?
     Explain.

I think this is a good idea. If indeed you take Thorndike seriously
and have written your post and explanation to me in all sincerity,
it will be very much worth my while to give you a careful answer.
I have no reason to doubt that this is the case. There are glaring
errors in the Thorndike scenario, and Rick has already given you
the short answer. But you deserve a thorough answer from me. For
example, Thorndike does not suggest an explanation that is a
functional mechanism, but suggests a flow chart of suggestive words
which you fill with meaning from your imagination. But reviewing
this will require more time than I have now.

It is now Tuesday morning. I have made a commitment to deliver two
translation assignments Thursday, before I leave Friday morning for
a wedding celebration in Pittsburgh. Back Monday evening. I will
not promise to give you my analysis until next week.

This is an open book assignment, and CSG-L is an open forum. If
anyone else butts in with comments, I will simply incorporate,
ignore or comment on them in my response. I expect to learn a
little about Thorndike from this exercise, both from my friends on
CSG-L and from reading about him myself.

I am glad you have joined my challenge to distinguish science from
mush. I have come to think that this is a very important issue, one
that the overwhelming majority of people never give any thought,
and one that is at the root of many fruitless discussions on CSG-L.
I expect to deal with this issue up front in my teaching of PCT as
it evolves. By the way, have you read my book? Send me your snail
mail address and I will send you a courtesy copy.

Regards, Dag

[From francisco arocha, 95/10/31-14.14]

[From Dag Forssell (951031 0945)]

It is now Tuesday morning. I have made a commitment to deliver two
translation assignments Thursday, before I leave Friday morning for
a wedding celebration in Pittsburgh. Back Monday evening. I will
not promise to give you my analysis until next week.

This is an open book assignment, and CSG-L is an open forum. If
anyone else butts in with comments, I will simply incorporate,
ignore or comment on them in my response. I expect to learn a
little about Thorndike from this exercise, both from my friends on
CSG-L and from reading about him myself.

Dag, before trying to discuss Thorndike's case, I think it would be useful
to define what is that you are trying to do. In your post on science or
mush, you make the claim that behaviourism, is not scientific. I won't try
to answer that; instead I'll make a short comment about a distinction I
find useful and which I think is sometimes at the root of some discussions
of metatheoretical nature on the net. It may be useful to clarify the
question by listing a set of criteria by which Thorndike's "law", or any
other law, explanation, or theory should be evaluated. As a start, I'll
propose a distinction between

a) an non-scientific theory
b) a scientific and factually correct theory
c) a scientic, but factually wrong theory

By a non-scientific theory, I mean one that does not meet the criteria for
scientificity. For instance, a theory that is not testable or that
introduces concepts that refer to non material things or processes. So here
we have two criteria: a scientific theory should be testable and refer only
to material things or processes. There are of course many others (of
course, there are people who would disagree on the criteria or simply say
that there are no criteria). But any way, the distinction is the following:

A scientific and factually correct theory is one that meets all the
criteria for scientificity, has been tested, and found to be supported by
the facts.

A scientic, but factually wrong theory is one that although meeting the
scientificity criteria has been tested and found contrary to fact (that is,
being wrong does not make a theory unscientific).

By starting this way I think the discussion will be more productive, while
avoiding misinterpretations.

Cheers,

francisco

j. francisco arocha Tel: (514) 398-4985
1110 Pine Avenue W. Fax (514) 398-7246
Cognitive Studies in Medicine
Centre for Medical Education E-mail: francisco@medcor.mcgill.ca
McGill University (alt) email: cybn@musica.mcgill.ca
Montreal, QC H3A 1A3
Canada

[From Dag Forssell (951031 1630)]

Bruce, 95% of this message belonged on CSG-L, so I'll put it there.
I am not altogether sure what to make of your qualification. The
one glaring error I mentioned already had nothing to do with
Thorndike's analysis. I am at a loss to see what unresolved issues
you are alluding to. I'll think it over til next week.

Best, Dag

Enclosed direct post:

···

-------------------------------------
Subj: Re: Book, issues
Date: 95-10-31 17:27:25 EST
From: abbott@cvax.ipfw.indiana.edu (Bruce Abbott)
To: DForssell@AOL.COM

Hi Dag,

Whatever other issue you suspected, I hold no malice toward you.

I'm glad to hear that, although I didn't think that you did; nor do
I hold any malice toward you. Rather, I thought that perhaps you
misunderstood my motives for arguing for the various positions I
have taken in opposition to accepted in-group views; often I sound
more like one of PCT's critics than one of its supporters, and you
must sometimes wonder just whose side I am on. My response to this
perception was to point out that even supporters of a view need to
vigorously challenge it, if they are to act as scientists. At any
rate, I seem to have guessed wrong.

   What I have wanted to challenge are some of your systems

     concepts, so carefully constructed by your mind over a long
     time, still resisting disturbances.

Fair enough. That is precisely what I have been doing with regard
to certain systems concepts of yours.

   If indeed you take Thorndike seriously and have written your

     post and explanation to me in all sincerity, it will be very
     much worth my while to give you a careful answer. I have no
     reason to doubt that this is the case. There are glaring
     errors in the Thorndike scenario, and Rick has already given
     you the short answer.

I take Thorndike seriously, but perhaps not in the way you imagine.
I offer Thorndike's experiment and his analysis in order to bring
into focus what sorts of activity you (and I) believe qualify as
legitimate science. The fact that Thorndike's analysis may have
"glaring errors" (from our perspective nearly 100 years later) is
another issue. I take Thorndike seriously, not because I think his
analysis was correct, but because I think his experiment and its
analysis point to unresolved difficulties for HPCT. But this,
again, is another issue.

   I am glad you have joined my challenge to distinguish science

     from mush. I have come to think that this is a very important
     issue, one that the overwhelming majority of people never give
     any thought, and one that is at the root of many fruitless
     discussions on CSG-L. I expect to deal with this issue up
     front in my teaching of PCT as it evolves. By the way, have
     you read my book? Send me your snail mail address and I will
     send you a courtesy copy.

Yes, I think this is a good issue to discuss, and look forward to
hearing your views on it.

I haven't read your book, but I would like to. Thanks for the
offer! My address appears below:

Bruce Abbott
Psychological Sciences
Indiana U.-Purdue U. Fort Wayne
2101 Coliseum Bvd. East
Fort Wayne, IN 46805-1499

Any chance you could stop by on your way back from Pittsburg? It
would be great to meet you and talk about these issues, and you'll
practically have to fly over my house to get home.

Regards, Bruce
--------------------------------
End enclosed post.

[From Bruce Abbott (951106.1055 EST)]

Shannon Williams (951105) --

Bruce Abbott (951102.1420 EST)

Not what I said. Explain your interpretation.

You said: "To make sure that the cat would be motivated...". The
ramifications of this point of view is that motivation is not studied. It
is assumed. If you had asked Thorndike why he thought the cat wanted to
escape, what would he have answered?

I needed to be clear about the scope of your statement. You meant that
Thorndike did not manipulate motivation in this study, which is correct.
But Thorndike had already established the need for some kind of incentive
during pilot work. It's like wanting to study the effect of load on the
rotational speed of an electric motor: you have to apply sufficient voltage
to the motor first. But during that study you are not investigating the
relationship of speed to voltage.

What constitutes an explantion?

An explanation explains the process by which you start at state 'A'
and end up at state 'C'. This process, which we often call a thing (a
mechanism), must be well enough described so that we can implement/verify
the steps of the process.

Are there different types of explanation, or only one type ?

There is only one type of explanation. An explanation describes steps in a
process. An explanation details how things happen.

Now I know what you are calling an explanation, although I do not agree that
there is only one type of explanation. You are looking for a mechanistic
explanation, as opposed to, say, a functional one.

If you do not know how to physically proceed from state A towards
state C, then do you know how to get from state A to C? Just because you
know what needs to be done, does that mean that you know how to do it?
Does an explanation _explain_ anything, if after understanding it, you
still do not know how to get from A to C?

Assume you are just beginning this process, and do not know anything about
the requirements of the mechanism you wish to discover. Would it help to
know what this mechanism _does_ and the conditions that must be satisfied if
the mechanism is to do its job?

Did Kepler's discovery that the planets move around the sun in eliptical
orbits in any sense explain the observations? Did Newton's formulation of
gravity explain anything?

Are there different levels of explanation, or only one level?

There are different levels of explanation. The steps in each
process/mechanism, are themselves processes/mechanisms.

I was thinking of "levels" in a different sense, but now that you brought it
up, can one construct an explanation at one level (as you are using the
concept) while still lacking explanations at a lower level? If so, how? If
not, why not?

Might Thorndike have been offering one kind of explanation, one that
then sets the stage for asking another question, aimed at another level
of explanation?

No. Thorndike offered a black box description of a process. He said:

       "If we hold these inputs constant, these outputs appear. At
       first the outputs are random, but over time positive results are
       reinforced so that something selects for the most effective output."

This is a description of the characteristics of the box. If you were to
buy the box, or I were to build the box, these external characteristics
would be part of the box's functional specifications. They would not
explain how the box is to be built.

I believe that this is the locus of our apparent disagreement. I would hold
that specifying the operating requirements of the box provides a kind of
explanation. Why is the motor turning? Because I've connected it to the
battery, and the motor needs a source of current to work. It's a looser
sort of explanation, but it can provide the basis for the mechanistic one
you seek. I think we basically agree, but that you have simply chosen
different words to express this. You said that Thorndike's formulation
defined the problem. Yes, once you can explain how the mechanism works from
a functional perspective, you are in a better position to discover the
mechanism itself.

The definition is the description of the problem (mechanism in our case).
It is the functional specifications of the mechanism.

The solution to the problem is the specifications which explain how to
build the mechanism.

Yes.

Regards,

Bruce

[From Shannon Williams (951108)]

Bruce Abbott (951106.1055 EST)--

I needed to be clear about the scope of your statement. You meant that
Thorndike did not manipulate motivation in this study, which is correct.
But Thorndike had already established the need for some kind of incentive
during pilot work.

But Thorndike wanted to figure out how the learned to excape. He needed
to study what motivated the cat to make each of its movements. Did he
determine any of the cat's motivations?

Assume you are just beginning this process, and do not know anything about
the requirements of the mechanism you wish to discover. Would it help to
know what this mechanism _does_ and the conditions that must be satisfied if
the mechanism is to do its job?

Yes. You must have a functional spec.

Did Kepler's discovery that the planets move around the sun in eliptical
orbits in any sense explain the observations? Did Newton's formulation of
gravity explain anything?

Kepler's did not. Newton's did. Newton defined a mechanism that did what
Kepler's laws required.

I was thinking of "levels" in a different sense, but now that you brought it
up, can one construct an explanation at one level (as you are using the
concept) while still lacking explanations at a lower level? If so, how? If
not, why not?

This is a strange question to ask. I cannot even begin to visualize what you
are asking.

I think that you would understand the answer to this question everytime you
explain to someone how to buy groceries or use a computer program. You
'explain' a process which dependes upon other processes that you cannot
explain.

Why is the motor turning? Because I've connected it to the battery, and
the motor needs a source of current to work. It's a looser sort of
explanation, but it can provide the basis for the mechanistic one you seek.

This explanation is based upon knowing the internal mechnism of the motor.
What if I explained that the motor turns because a switch causes it to
respond by turning? Would this explanation provide a clue to the internal
mechanisms?

-Shannon

[From Shannon Williams (951101)]

Bruce Abbott (951030.1715 EST)--

   "The time has come," the Walrus said, "to speak of many things,
    Of shoes and ships and sealing wax, of cabbages and kings,
    And why the sea is boiling hot, and whether pigs have wings."

And not a word was heard when the story was done,
Because he had deceived them all, and then eaten every one.

(or something like that)

Back in 1898, Edward L. Thorndike performed a classic experiment in which a
cat was placed in a wooden box having a door on one side whose latch could
be released from within the box, if only the cat knew what to do. The idea
was to determine how the cat learned to solve the problem.

Ok. Thorndike studied the evolution of cat behavior.

To make sure that the cat would be motivated to get out of the box,
Thorndike arranged to perform the experiment immediately before the cat's
dinner-time...

Thorndike did not study the motivation behind the evolution of cat behavior.

To explain this gradual and subtle change in performance across trials,
Thorndike appealed to a process of "selecting and connecting."

Thorndike was not 'explaining the change' when he 'appealed to a process of
selecting and connecting'. He was describing the change. He was
describing what needed to be explained by an explanation for the change.

Of course, it would be much better to have a mechanism that would account
in purely physical terms for the observed behavior patterns and their
differing frequencies, so as to derive the observed patterns from first
principles, but such a model was not available. Fortunately, it is
possible to make at least some progress without one

The progress that you describe in this post is towards defining the problem.

How do we distinguish between definition of a problem, and solution to the
problem?

-Shannon