Then a miracle occurs

[From Bill Powers (950618.1430 MDT)]

Dennis Delprato (950618) --

     If PCT offers a new fundamental unit of psychological behavior,
     then PCT is going "underneath" the operations of reinforcement. It
     is getting at what is more generally going on than what we observe
     on the surface. The question, then, is how to go beyond the
     surface of selection by consequences reinforcement.

What has me climbing the wall is the use of the term "selection by
consequences" when to me it is obvious that nothing of the sort is going
on.

I've been watching the U.S. Open golf tournament on TV. Shinnecock is a
tough course and club selection for short approach shots is critical. If
the club is too lofted the wind makes the ball drift, and if not lofted
enough the ball runs when it hits the hard green as well as falling
short. So what selects the club? The consequence of drifting in the wind
or falling short, or THE PLAYER? I contend that it is the player who
does the selecting, using information about previous experience to be
sure, but still using a process of selection that lies in the player and
not in the ball.

The behavior of a golf ball is just not the sort of thing that can
"select." Selection of behavior is a process of weighing many factors
and choosing an action that is estimated -- by a system capable of
making estimates -- to produce a desired result. The ball has no
preference for how it will behave and mechanism for making estimates or
choosing on that basis; only the player can have such a preference, or
draw conclusions from present-time data.

     One place to look might very well be the work on feedback
     functions. It seems to me that the molar behaviorists (where one
     finds feedback functions) have departed from Skinnerian "molecular"
     selection by consequences. They say, I think, "When responses are
     modified by contiguous relationships between responses and
     consequences, it is not what one sees directly (contiguous
     relationships) but what one does not see directly that is more
     generally important for describing what is going on."

Left out of this way of putting it is the actual mechanism that does the
modifying. That mechanism has to lie within the organism -- where else
could it possibly be? The mere fact that a response has a certain
consequence can't modify anything else. It's just a fact, something that
happened. I don't think that proposing that contiguous relationships can
modify responses is any better than saying that consequences can select
behavior. Neither one makes any sense to me. How on earth could a
contiguous relationship do anything but be a contiguous relationship?
Wherein lies its ability to modify a response?

Help me out, Dennis. Am I the only one in the world who sees statements
like this as invoking magic? Here is a consequence; here is a contiguous
relationship. Then a miracle occurs, and reponses are modified or
behavior is selected. Doesn't anyone else see something vital missing
from these statements?

···

-----------------------------------------------------------------------
Best,

Bill P.

[Dan Miller (950619)]

Bill Powers (950618)

Very nice post. The invokation of magic is everywhere around us.
I have to work hard to keep from doing it, with only modest success.

Regarding the selection of clubs in the U. S. Open Golf Tournament,
not only do the players select the clubs with what they perceive to
be the proper loft, but also, they have in mind the kind of shot they
are going to play. They "see" the flight of the ball, where, and how
it will land. The winner, Corey Pavin, is well-known as a "shot maker."
That is, he can hit a ball high, low, with draw, or with fade. It is
not surprising that he (or a player with these talents) won this
tournament. Also, his puts were firm and assured.

Golf balls may have no preference in how they fly, but the makers of
them do. It is possible to hit balls that will "tend" to take a high
trajectory, a low trajectory, or medium. Also, balls are being made
that will minimize hooks and slices. Alas, the players still must
select clubs, imagine shots, execute them, chase after the ball, and
hit it again.

I may take some time to think more of PCT and golf - thus combining
two of my favorite things.

Later,
Dan
millerd@udavxb.oca.udayton.edu

[FROM: Dennis Delprato (950619)]

Bill Powers (950618.1430 MDT)

What has me climbing the wall is the use of the term "selection by
consequences" when to me it is obvious that nothing of the sort is going
on.

I've been watching the U.S. Open golf tournament on TV. Shinnecock is a
tough course and club selection for short approach shots is critical. If
the club is too lofted the wind makes the ball drift, and if not lofted
enough the ball runs when it hits the hard green as well as falling
short. So what selects the club? The consequence of drifting in the wind
or falling short, or THE PLAYER? I contend that it is the player who
does the selecting, using information about previous experience to be
sure, but still using a process of selection that lies in the player and
not in the ball.

The behavior of a golf ball is just not the sort of thing that can
"select." Selection of behavior is a process of weighing many factors
and choosing an action that is estimated -- by a system capable of
making estimates -- to produce a desired result. The ball has no
preference for how it will behave and mechanism for making estimates or
choosing on that basis; only the player can have such a preference, or
draw conclusions from present-time data.

There appears to be a misunderstanding here; in reply,
I'll speak "operantly" to the best of my ability and with brevity
in view:
It seems like you are interested in understanding and explaining a
golfer's behavior of asking for and using a particular club on a
particular occasion. Young fellow, I assure you that the answer is
not a cognitive expectancy, a schemata, any sort of underlying
motive, a strong S-R bond, or any other purely hypothetical entity
that your elders may have taught you. The behavior also is not
a manifestation of an inherited trait.
The answer is in the golfer's
history of reinforcement. Consider a particular set of conditions
[wind conditions, lie (isn't this a golf descriptor?), distance from
hole, perhaps organic condition of golfer (e.g., degree of fatigue),
condition of green, slope of green, ...]. In the past, when the
golfer has used a particular club, they were more reinforced (positive
reinforcement: ball went closer to hole; negative reinforcement: ball
did not end up in bad place like 'the rough) than if they used other
clubs. In slightly more complete terms, we might say the golfer's
responses were 'differentially reinforced': Response-1 -->positive
reinforcers presented, negative reinforcers not presented;
Responses other than 1--> positive reinforcers less likely, negative
reinforcers more likely. (--> means followed by where the right
side of --> indicates what we call 'consequences (of responding).
The basic principle being obeyed here is that responses followed
by consequences that are reinforcers (presentation of positive
reinforcers, termination or withholding of negative reinforcers)
are selected. That is, they are more likely to occur in the future
under the same or similar conditions (ess Ds, discriminative
stimuli).

Thus, we see that you are basically correct: THE PLAYER
selects. But why does THE PLAYER select? Because of THE PLAYER'S
history of interactions with their environment. This history
changes THE PLAYER in ways that physiology ultimately will describe.
All we know now is that reinforcement histories change organisms
organically and in this way behavior is modified. It appears that
of past responses, along with current environmental (including
organic ones) to be of particular interest. You seem to looking
more than I into the question of what is happening organically
in conjunction with all this. Perhaps we could get together some
time.

  One place to look might very well be the work on feedback
  functions. It seems to me that the molar behaviorists (where one
  finds feedback functions) have departed from Skinnerian "molecular"
  selection by consequences. They say, I think, "When responses are
  modified by contiguous relationships between responses and
  consequences, it is not what one sees directly (contiguous
  relationships) but what one does not see directly that is more
  generally important for describing what is going on."

Left out of this way of putting it is the actual mechanism that does the
modifying. That mechanism has to lie within the organism -- where else
could it possibly be? The mere fact that a response has a certain
consequence can't modify anything else. It's just a fact, something that
happened. I don't think that proposing that contiguous relationships can
modify responses is any better than saying that consequences can select
behavior. Neither one makes any sense to me. How on earth could a
contiguous relationship do anything but be a contiguous relationship?
Wherein lies its ability to modify a response?

Help me out, Dennis. Am I the only one in the world who sees statements
like this as invoking magic? Here is a consequence; here is a contiguous
relationship. Then a miracle occurs, and reponses are modified or
behavior is selected. Doesn't anyone else see something vital missing
from these statements?

Is the operant view of the locus of the of the 'actual mechanism'
of modifying now clearer? They focus on reinforcement history, i.e.,
actual response-->consequence relationships in life of individual.
Skinner agrees that response-->consequence histories change organisms
oganically. He hold that despite this, the study of organism-environment
interactions is important in its own right and if done well will
provide fundamental material for those tracing what is happening to
the organism at the organic level as a function of particular
environmental histories.

On the miracle: I believe Skinner might ask if we believe
Darwin was invoking miracles with the principle of selection by
consequences (natural selection). K. U. and Tom Smith say neoDarwinians
are, but that is another (cybernetic) story. To Skinner, natural
selection is a basic law (causal mode, even) of biological evolution,
the ontogeny of individual behavior, and cultural practices. We are
not dealing with anything complicated with 'selection by consequences.'
The alternative is taken to be "the causality of classical mechanics"
or "'selection pressure,' which appears to convert selection into
something that forces change. ... [with a more serious example being]
the metaphor of storage" (Skinner, 'Selection by Consequences, _Science_,
1981, v. 213, p. 503).Basically, selection by consequences seems to
be more simplistic or incomplete, rather than mystical--what we find
now is found because there was something functional or adaptive about
it and that is why we find it.

Bill, you didn't comment on my suggestion that PCT modelers base
models on PCT units not on units from non-PCT areas such as operant
psychology. Perhaps when Bruce Abbott returns he will be able to
determine if I am making any sense at all, because he is well-schooled
at using the operant fundamental units (depending on one's view:
respondent, operant, discriminated operant). In other words, one
not practiced at using the operant units in technical ways could
be enticed to use them unknowingly in models and thus develop nothing
more than another operant model.

Dennis Delprato
psy_delprato@emuvax.emich.edu

···

from my operant perspective I find the environmental consequences

<[Bill Leach 950619.23:53 U.S. Eastern Time Zone]

[FROM: Dennis Delprato (950619)]

Perhaps when Bruce Abbott returns he will be able to determine if I am
making any sense at all, because he is well-schooled ...

Yes SIR! I for one will be very interested in what Bruce has to say
about this particular posting.

The "wording" might "tend to drive a PCTer up the wall" but the meaning
of your golf example sure sounds like "learned control" to me.

Your "reinforcement histories"; "more" "postitive reinforcers" and "less"
"negative reinforcers" (hell of a pair of terms there, those two!) is a
case of "control attempt with little or no error" and; "less" "postitive
reinforcers" and "more" "negative reinforcers" is a case of "control
attempt with high error possibly even control failure" (ta da!
Reorganization to the rescue?)

One place where there has not been a great deal of discussion with
respect to "reinforcement" is in terms of memory. There _seems_ to be
some correlation between repetitive experience and effectiveness of
memory AT TIMES (at other times, it appears that a single experience that
"was long forgotten" will in fact remain available.

Similarly (as a person that has recently just "half-heartedly" taken up
golf), many "mechanical skills" seem to require repeated performance and
"tuning" to eventually be performed well (yet again, occassionally there
is that example where "first try" is correct).

If this is the sort of thing that "reinforcement" actually refers to;
that is, the "consequences of a _control action attempt upon the
controlled variable_ AS PERCEIVED by the organism (ie: control error
does or does not exist) AND NOT just the consequences as observed by
others then maybe EAB is not as greatly different as it appears.

-bill