LETTER TO/FROM (DELETED) WITH ENCLOSURE TITLED "OPERANT REINFORCEMENT OF A MEDIATED AUTONOMIC RESPONSE" RE LEARNING; BEHAVIORAL MODIFICATION

• Document Type: SPECIALCOLLECTION
• Collection: General CIA Records
• Document Number (FOIA) /ESDN (CREST): 00173702
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 9
• Document Creation Date: January 21, 2025
• Document Release Date: January 15, 1983
• Publication Date: June 21, 1967

-:;�
21 June 1967
Dear Dr."
_Enclosed is o_ Li or the raper
which is to be slthni-tted for publication in the journal Psychartrysioloszr.
The article as edited has been fcamally cleared for publication in
---:open-litcrature with the proviso that no Agency affiliation will be
revealed. . �
��� .
-1-
OPERANT REIWFORCEMEN: OF A, =DIA= AUTONOMIC RESPONSE
Backrround:
Nowrer and some other psychologists have argued that there are two
distinct methods of conditioning or learning (roughly, classical con-
ditioning and instrumental learning). Furthermore they have insisted
that instrumental learning cannot be reduced to classical conditioning,
and that classical conditioning cannot be explained by the laws of
instrumental learning. As evidence, Mowrer points to the fact that
autonomic responses and skeletal responses are qualitatively different
as they are given in nature. "Here we are assuming that behavioral
responses are categorically different from emotional responses: the
former are 'voluntary' and subject to influence through reward and
punishment (and not conditionable, strictly speaking), whereas the
latter are involuntary and conditionable and not subject to control
through reward and punishment, or a; least not in the same way as are
the overt behavioral responses." (Aowrer, 1960.)
Tnere are everyday inatances of apparent learning of an autonomic
response through reward; Skinner cites the child who cried "real tears"
because tears had been followed by attention and candy in the past, and
recently some experimental evidence.
1
!has shown that
auIonoc responses can be directly modified by response contingent
reinforcement, however, it is not possible to rule out the effect of
skeletal mediating responses in any of these cases. If the autonomic
response in question is elicited by an unobserved skeletal response, it
-2-
is possible that all of the evidence for apparent autonomic learning
merely reflects the parasitic reinforcement of the underlying skeletal
tesponse.
Some authors (Kendler, 1962) have insisted that even the possibility
of skeletal mediators is sufficient to reject the hypotnesis of autonomic
learning through direct reinforcement;
The mediation explanation asserts that the mediator is a skeletal
response which precedes and elicits the autonomic response. As a result
of rewarding the autonomic response, the skeletal response is parasitically
-- reinforced and learned. Thus it is implied that a typical negatively
accelerated learning curve would develop for the skeletal response. It
is also implied that the variability in the autonomic response can be
- fully explained by the changes in the frequency of the underlying skeletal
response.
In the present experiment, it is proposed to set up an explicit medi-
ating skeletal response which elicits a drop in galvanic skin resistance,
and then reinforce the elicited sutoncmic response.
If the mediation explanation is correct, it should be possible to
discover both the negatively accelerated curve of the skeletal response
and to explain the variability of the elicited autonomic response. It
may also be possible to compare these findings with the results of the
earlier studies where an autonomic response was apparently directly
reinforced. If the learning carves are very similar, It would sugeest
that an unknown skeletal mediator was responsible for the changes
observed in the earlier work.
-3-
It is possible that results from this study will give us some kind
of model of operant learning of autonomic responses as it occurs in
stressful situations.
Method:
The method will be a fairly direct extension of methodology used in
- I
earlier studies /College age female volunteers will be
used; this vill permit a reasonable comparison with the old data. Ten
experimental and 10 control subjects will be run, they will be matched
for frequency of GER nonspecific deflections on the first day of the
experiment and yoked for schedules of reinforcement for the remainder of
the experiment.
.In the earlier studies it was observed that very large respirations
occurred intermittently in what vs otherwise a very regular respiration
rate. These large respirations were often followed by a GER deflection.
For example, on the last day of the 1963 experiment, the eighteen subjects
emitted 0 to 26 gross irregularities of respirations during the twenty
minute session with an average of 7 per subject. These respirations
often elicited GER deflections. .Ti:e rate of GER elicitation varied from
A
121,for one of the subjects to 1004.", for four of the subjects; the median
WW1 564,
in that experiment a apeeial effort was made to avoid reiuforcing
respiration-elicited GZ,R deflectioes. In the present study, it is pro-
pored that these gross irregularities of respirution be utilised as
a.:eletel mediators; and that their associated GER deflections be systemati-
cally reinforced. Thiv response WA the aevantage cf being relatively
subtle and not immediately discernible to the subject.
It is suggested that the responses be monitored by the experimenter
at this point. It would probably be too complicated to set up an
automatic reinforcer for what should be a relatively short, contained
study.
Otherwise, methodology will follow the earlier studies. All subjects
will be given. 2 days of adaptation, five days of reinforcement (contingent
or yoked) and three days of extinction. We will follow the policy of
havine the subjects sit quietly for twenty-five minutes before each day's
session of twenty minutes. All girls will be screened for medical
problems and run in between menstrual periods. The experiment should
be conducted in a sound-proof constant temperature room.
Equipment:
Sanborn GS R with Wenger electrodes, plethysmograph (Kenelco Corp.)
pneumanometer.
Data Analysis:
Do you think we can get direct taping of thq output of all three
variables? In GM, all we need is a frequency count�amplitude of the
deflections does not seem to be in:hrt,nt. That should simplify the
proble, too.
However, minute by zinute basal resistance counts should
be obtained. The fact that basal resistance shifts over the twenty
minute period may present special difficulties. On the plethysmosrsph
record, I'd like to take off heart rate (in twenty second intervals) and
amplitude (about once every twenty seconds). The respiration record may
be the most difficult to automate, if our experience in recording by
hand is any indicator. / have ample samples of all these, plus a fairly
specific outline of how they were counted by hand.
Once the data are in digital form, we will need individual and group
curves on al�hree variables. in the past I have used some rather
simple nonparametric statistics for looking at differences. It ia quite
clear that the major variable will be an intra-individual Change fran
adaptation (days 1 and 2) to extinction (days 9 and 10). It has been
suggested to me that there is some way of looking at clusters of all
three variables at once as a measure of learning (Hoteling's T). However,
/ am not sure that with this small sample, non-normal data, etc., that
that is feasible. Any suggestions?
It seems to me that at this early stage of experimenting, the auto-
mation of all these things should be kept simple--to leave roam for
changes in procedure and to maintain quite a bit of flexibility in the
system. We are handicapped by the fact that our background knowledge of
ongoing basal rates and individual differences in GSR nonspecifics,
plethyamegraph amplitudes, etc. is still �inite limited.
Data Parameters:
1. Basal Resistance
In general, all subjects show an increase in basal resistance fram
day one to ten in this kind of stu4. The controls tend to show a
-6-
relatively greater increase than the experimental subjects.
Mere are side individual differences in basal resistance. In my
1964 study, I found that it varied from 61.9 K (MT, day 1) to 444.6 K
(Ms day 5)�
Within an individual a maximum change of about 150-200 K could be
seen from day one of the experiment to day ten. Median readings for
each day were used as estimates of each subject's daily level.
Basal resistance can show rather narked change within a daily session
for sane individuals. In the 1964 study, at least one subject had an
increase of over .200 K from beainning to end of the twenty minute session
on day 9 (CJ). Other subjects tend to remain quite stable within a day's
run and from day to day.
In ceneral, subjects with low basal resistance pit out a high fre-
quency of nonspecifics, while subjects with high basal resistance proauce
few nonspecifics. For instance, in the 19G4 study, one subject (MT) had
an initial basal resistance of en1: 61 K and her output of nonspecifice
was over 6 per minute. another subject with a high basal resistance
(331 K) had a nonspecific rate of .4 per minute.
2. Nonspecifics
In the 1964 study, experimentals tended to maintain their initial rate
of nonspecifics while controls declined.
There are large individual differences in frequency of nonspecificc.
In the 1964 study, the lowest reading on day I was .4 per minute the
high was over 6 per minute. Both experimentals and controls show some
-7-
decline in rate over the ten dais, so the range on day 10 is .05 per
minute to about 5 per minute.
I do not have any current data on amplitude of deflections. The
smallest we counted was .4 K and the largest was 30 It, but there were
unquestionably larger ones. Data from the 1962 study suggests that size
tends to increaue as basal resistance increases from day one to day ten;
median amplitude increased at madh as three times the day one amplitude
(conductance readings).
3. Heart Rate .
In the /964 study, experimentals maintained their initial level of
heart rate while controls showed a small but consistent decline.
Again, there are fairly large individual differences. In the 1964
study, the eighteen subjects ranged fram 61.8 to 104.7 with a standard
deviation of 12.8.
Under non-strescful repeated trials, the standard deviation within
an individual is about 6 beats per minute. The maximum within individual
variation that we saw over days was 15 beats per minute.
i� �
4. Plethyamograph
This data is purely rankea data. On the first day of the 1964 study
we found that subjects varied from a minimum amplitude reading to twice
that reading. Over the ten day session* controls tended to increase in
amplitude while experimentals reuain relatively constant. Some subjects
doubled their amplitude over the ten days* while at least one subject
showed a reading which was half a.. large as his initial amplitude,
The heaviest expenses for this study will be tape and computer
time. Each subject is run for 3 1/3 hours. We made some estimates at
\for this kind of study. I.don't know whether these costa
are comparable, but for what they're worth:
Apparatus batteries, etc.
.Analos Recorder tape at
Digital tape at
Computer time. Analog' to digital con-
version at
Computer time. Digital analysis at
Subjects at
This does not include the cost of a computer programmer to set up the
, program. We estimated that might take as long as three months and
-coat about