University of Southern California Dissertations and Theses

Human Gsr Classical Conditioning And Awareness Of The Cs-Ucs Relation

(USC Thesis Other)

Human Gsr Classical Conditioning And Awareness Of The Cs-Ucs Relation

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content This dissertation has been
microfilmed exactly as received 6 7 -1 3 ,0 1 9
DAWSON, Michael Edward, 1940-
HUMAN GSR CLASSICAL CONDITIONING AND AWARE
NESS OF THE CS-UCS RELATION.
University of Southern California, Ph.D., 1967
Psychology, experimental
University Microfilms, Inc., Ann Arbor, Michigan
HUMAN GSR CLASSICAL CONDITIONING AND AWARENESS
OF THE CS-UCS RELATION
by
Michael Edward Dawson
A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Psychology)
June 1967
UNIVERSITY O F SO U TH ER N CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 9 0 0 0 7
This dissertationj written by
Mi ckae 1. _ .Edwar d _ _ Daw_son..............
under the direction of hL3...Dissertation Com
mittee, and approved by all its members, has
been presented to and accepted by the Graduate
School, in partial fulfillment of requirements
for the degree of
D O C T O R OF P H I L O S O P H Y
g / z & d z f i z S .....
Dean
Date..................... June _ . 8 ^ . 19 67.....................
DISSERTATION COIVPMTTEE ,
Chairmt
ACKNOWLEDGMENTS
The present research was supported by a National
Institute of Mental Health predoctoral fellowship (MH 10554-
01). The author wishes to thank the government for provid
ing and Dr. William Grings for acquiring this support.
Equipment used in this study were generously supplied by
Dr. Grings from his research grant (MH 3916).
Appreciation is expressed to my guidance and disser
tation committees for their advise. Special appreciation
is expressed to my chairman, Dr. William Grings, for four
years of truly excellent training and guidance. I also wish
to thank my contemporary graduate students for helping to
make graduate training more stimulating and exciting.
And to the U.S.C. Computer Science Laboratory, the
U.C.L.A. Health Sciences Computing Facility and Ann Schell
are expressed gratitude for invaluable help in data reduc
tion and analysis.
I also wish to take this opportunity to thank my
parents for their continual help. And to Lavina, without
whose enthusiastic and loving support this paper would not
be possible, this work is dedicated.
ii
TABLE OF CONTENTS
Page
ACKNOWLEDGMENTS ...................................... ii
LIST OF T A B L E S ...................................... V
LIST OF FIGURES...................................... vii
Chapter
I. INTRODUCTION ............................... 1
II. CR ACQUISITION AND R L ...................... 5
The Measurement of RL
III. CR ACQUISITION AND THE INTERSTIMULUS
INTERVAL................................. 23
IV. EXPERIMENTAL HYPOTHESES .................... 28
V. M E T H O D ...................................... 32
Subjects
Apparatus
Design and Procedure
Response Measurement
VI. RESULTS...................................... 43
Experimental Presuppositions
Equality of groups
Requirements for classical conditioning
Experimental Hypothesis; RL Defined
by Instructions
CS response
Pre-UCS response
Post-UCS response
Questionnaire
iii
Chapter Page
Experimental Hypotheses: RL Defined by
ITI Reports
Experimental Subhypotheses: RL Defined
by Post-conditioning Interview
Absolute RL
Relative RL
VII. DISCUSSION................................. 88
VIII. SUMMARY..................................... 99
REFERENCES......................................... 103
APPENDIX A. Instructions ........................... 110
APPENDIX B. Trial Sequence ......................... 115
APPENDIX C. Post-conditioning Interview ......... 119
LIST OF TABLES
Table Page
1. Schematic Representation of the First Two
Acquisition Trials ........................ 39
2. Analysis of Variance of the CS Response Data
during Familiarization Trials ........... 45
3. List of Number of CS Identification Errors
for Each S in Each Experimental Group . . 47
4. List of Number of Perceptual Task Errors for
Each S in Each Group..................... 47
5. Analysis of Variance of the CS Response Data
during Acquisition ........................ 52
6. Analysis of Variance of the CS Response
Acquisition Data for the II and 18 Groups 55
7. Analysis of Variance of the Pre-UCS Response
during Acquisition ........................ 56
8. Analysis of Variance of Post-UCS Response
during Acquisition ........................ 62
9. CS Response Magnitude (/AC) Data for the
Four Experimental Groups during the
Questionnaire
10. Trial Numbers and Conditions on which ITI
Report Revealed Expectancy of Shock and
Adjacent Trials where No Expectancy was
Reported................................... 67
11. Mean CS Response Magnitude (/AC) for ITI+
and ITI- for the Nl G r o u p ............... 69
v
Table Page
12. Mean GSR Magnitude (/AC) for ITI+ and ITI-
for the Three Response Measure for the N8
G r o u p .................................... 69
13. Absolute and Relative RL Scores for Ss in
the II Group............................. 71
14. Absolute and Relative RL Scores for Ss in
the 18 Group............................. 72
15. Absolute and Relative RL Scores for Ss in
the Nl Group........................... . 73
16. Absolute and Relative RL Scores for Ss in
the N8 Group............................. 74
17. CS Response Magnitude (/AC) to CS+ during
Questionnaire ............................ 82
18. CS Response Discrimination (/AC) Scores
during Questionnaire ..................... 87
LIST OF FIGURES
Figure Page
1. Mean Magnitude of CS Response during
Familiarization Trials of the Four
Experimental Groups ...................... 44
2. Mean CS Response Magnitude Data during
Acquisition for the II and Nl Groups
Mean CS Response Magnitude Data during
Acquisition for the 18 and N8 Groups . . 51
3. Discrimination Scores of the CS Response
during Acquisition of the Four Experi
mental Groups............................. 54
4. Mean Pre-UCS Response Magnitude during
Acquisition for the 18 and N8 Groups . . 57
5. Discrimination Scores of the Pre-UCS
Response during Acquisition of the N8 and
18 Groups......................... 58
6. Mean Post-UCS Response Magnitude during
Acquisition of the N8 and 18 Groups . . . 60
7. Discrimination Scores of the Post-UCS
Response during Acquisition of the N8 and
18 Groups............................. .. . 61
8. Mean CS Response Magnitude to CS+ for Ss
with Absolute RL and Ss without Absolute
RL in the Nl G r o u p .............. 77
9. Mean CS Response Magnitude to CS+ for Ss
with Absolute RL and Ss without Absolute
RL in the N8 G r o u p .............. 78
vii
Figure Page
10. Mean Pre-UCS Response Magnitude to CS+ for
Ss with Absolute RL and Ss without
Absolute RL in the N8 Group............. 79
11. Mean Post-UCS Response Magnitude to CS+ for
Ss with Absolute RL and Ss without
Absolute RL in the N8 Group............. 80
12. Discrimination Scores of the CS Response
for Ss with Relative RL and £s without
Relative RL in the N1 Group.............. 83
13. Discrimination Scores of the CS Response
for Ss with Absolute RL and Ss without
Relative RL in the N8 Group............. 84
14. Discrimination Scores of the Pre-UCS
Response for Ss with Relative RL and Ss
without Relative RL in the N8 Group . . . 85
15. Discrimination Scores of the Post-UCS
Response for Ss with Relative RL and Ss
without Relative RL in the N8 Group . . . 86
CHAPTER I
INTRODUCTION
Classical conditioning is generally defined as
response modification (called CR acquisition) attributable
solely to the repetitive temporal pairing of a relatively
neutral stimulus (CS) with a potent, response-producing
stimulus (UCS) (Grings, 1963; Hilgard and Marquis, 1940;
Kimble, 1961). However, the repetitive temporal pairing of
two stimuli may also produce— at least with normal human
subjects— the ability to verbalize the relation between the
two stimuli (called relational learning or RL) (Dawson and
Grings, in press; Razran, 1955a). Thus, the operation of
administering CS-UCS pairings to normal humans may produce
two measurable phenomena: (1) CR acquisition and (2) RL.
The primary purpose of the present investigation was to
determine whether or not human GSR CR acquisition can occur
in the absence of RL. As there is evidence to suggest that
different learning processes may occur with short and long
CS-UCS intervals (Lockhart, 1966a), the experimental ques
tion was studied with a 1 sec. and a 8 sec. interval. An
1
additional reason for employing two intervals is that more
than one GSR "CR" may be measured with a long interval
while only one may be measured with a short interval
(Grings, Lockhart, and Dameron, 1962).
A currently prevalent opinion is that human CR
acquisition can occur in either (1) its "pure" and primi
tive form which is independent of the higher RL processes
or (2) its "mixed" form due to its interaction with the
higher RL processes (Bridger and Mandel, 1965; Grant, 1964;
Grings, 1965; Kimble, 1962; Razran, 1955a; Spence, 1966).
According to this interpretation, CR acquisition is a
process of learning which is different and separable from
that of RL. Kimble summarized this interpretation, "It is
quite obvious that the verbalization of the sequence of
stimuli and the acquisition of the conditioned response are
two very different forms of learning, both of which occur
in the conditioning situation" (1962, p. 35).
Another current opinion, although not as explicitly
stated and perhaps not as prevalent, is that CR acquisition
can occur only when RL is present (Tolman, 1959; Woodworth,
1958). According to this interpretation, CR acquisition is
a peripheral reflection of an underlying RL process. Thus,
CR acquisition is not considered a process of learning
3
which is separable from RL. Woodworth suggested this
interpretation when he stated that Pavlov's dogs did not
learn to salivate to the bell but instead learned that food
followed the bell and salivation was merely a peripheral
concomitant of this learning (1958, p. 229).
It can be seen that both of the above interpreta
tions hypothesize that CR acquisition may be influenced by
RL, the difference between the interpretations is whether
CR acquisition is a learning process independent of RL.
This difference has vast theoretical and methodological
implications.
In order to appreciate the theoretical implications,
assume that CR acquisition can occur in the absence of RL.
It then becomes clear that we have a method of making the
"unconscious observable" (Razran, 1961), and at least one
author (White, 1964) has relied on this implication in
developing a theory of neurosis. Another theoretical
implication of this assumption is that a complete theory of
CR acquisition must describe how RL and CR acquisition
interact. On the other hand, assume that RL is a central
process necessary for CR acquisition. It then becomes
clear that human classical conditioning, sensory precondi
tioning, paired-associate learning, serial learning, and
possibly other types of learning may share the same basic
process ("sequence learning" according to Woodworth, 195 8).
If this assumption is true, then we have a much broader
base from which to study the learning process than is now
thought and, ultimately, one theoretical system may account
for all of the above-listed phenomena.
In terms of methodological implications, the type
of independent variable manipulated (e.g., variables de
signed to affect central-cognitive processes versus vari
ables designed to affect peripheral-response processes) and
the type of control procedure employed (e.g., attempting to
allow individual development of RL versus attempting to
equate RL between Ss) depends largely on which assumption
is accepted, either explicitly or implicitly.
CHAPTER II
CR ACQUISITION AND RL
Experimental evidence cited in support of the hy
pothesis that CR acquisition can occur in the absence of RL
may be divided into five categories (Razran, 1955a, pp. 87-
89): (1) CR acquisition occurs with animals low in the
phyletic scale, (2) CR acquisition occurs with decorticate
or even spinal animals, (3) CR acquisition occurs when the
CS is below the threshold of verbally reportable perception
(subliminal conditioning), (4) CR acquisition occurs when
the CS and/or UCS is of internal origin (interoceptive con
ditioning) , and (5) CR acquisition occurs when the CS-UCS
relation is masked by a distracting task.
If RL is defined as the ability to verbalize the
CS-UCS relation then the first two categories of evidence
constitute proof that CR acquisition can occur in the
absence of RL. This conclusion is true because neither
animals low in the phyletic scale nor decorticate animals
can verbalize the CS-UCS relation and yet they can acquire
a CR. There are two possible counter-arguments to these
6
lines evidence: (1) redefine RL in such a way that its
presence or absence can be inferred with nonverbal organ
isms or (2) limit the present definition of RL to organisms
who have the ability to verbalize. In theory, the first
approach is to be preferred because of its greater general
ity and there are some signs of success in this direction
(Egger and Miller, 1962 and 1963). However, due to our
present lack of understanding of the RL process, its defi
nition is limited in the present paper to the verbalization
of the CS-UCS relation by subjects who have the ability to
verbalize (i.e., normal adult humans). Therefore, the first
two categories of evidence, and other data with nonverbal
izing Ss (e.g., Grings, Lowell, and Honnard, 1961), are not
considered relevant to the present definition. The latter
three categories of evidence are considered relevant to the
present definition and will now be discussed.
First, it is asserted that CR acquisition can occur
when the CS is below the threshold of verbalizable percep
tion (subliminal conditioning). Certainly, if the CS is
not perceived the CS-UCS relation cannot be perceived.
Experimental evidence relevant to subliminal conditioning
was reviewed by Eriksen (19 60) and he concluded, "The evi
dence for subliminal conditioning is for the most part
negative and at best controversial" (p. 283). The situation
does not seem to have changed substantially since the time
of that evaluation. For example, Fishbein and Engle (1966)
could find no evidence of eyelid conditioning to a tone
which was below the threshold of verbalizable perception.
Block and Reiser (1962) found discrimination conditioning
of the GSR using CS intensities which were verbally percep
tible only 50% of the time. However, as Eriksen (1960) has
pointed out, the 50% correct identification criterion of
threshold allows S to sometimes perceive the stimulus and,
thus, to demonstrate CR acquisition with such a stimulus
does not constitute evidence that CR acquisition can occur
with a CS which is below the threshold of perception. In
fact, additional data from the same study by Block and
Reiser indicated that discrimination accompanied a confi
dence rating of "not sure" but did not accompany a confi
dence rating of "guess." This finding supports the
interpretation that CR acquisition occurred to the percep
tible presentations of the CS and not the unperceptible
(subliminal) presentations.
A report by Barratt and Beh (1964) seems to have
avoided or at least reduced this interpretational problem
by defining the "subliminal" CS as a tone 2 db below the
8
intensity at which 20% correct identification occurred.
Once this value was determined it was used as a CS in a
discrimination conditioning paradigm in which £ 3 was asked
to report whether the tone ever "accidently" occurred.
Successful conditioning of the EEG was obtained and, con
cerning the reporting of CS perception, the authors state,
"No such report was ever given, nor did subsequent question
ing elicit a positive response" (p. Ill) . Although the
instructions that the stimulus occurrence would be "acci
dental" may have caused hesitancy in reporting and despite
the fact that "subsequent questioning" was never specified,
this experiment probably constitutes the strongest evidence
to date that subliminal conditioning is a real phenomenon.
Whether or not this study will be replicable when these
deficiencies are remedied remains to be seen for, as
Eriksen has pointed out, the failure to replicate positive
results remains a major problem in the area.
Secondly, it is asserted that CR acquisition can
occur when the CS and/or UCS is of internal origin (intero
ceptive conditioning). Actually, the reason that intero
ceptive conditioning is used as evidence that conditioning
can occur without awareness of the CS-UCS relation is that
internal stimuli are often considered to be unperceptible
9
(Bykov, 1954; Razran, 1961). In this sense, interoceptive
conditioning can be considered a subclass of subliminal
conditioning.
Razran (1961) reviewed the interoceptive condition
ing literature, which is primarily of Soviet origin, and
concluded that "there should be absolutely no doubt" that
interoceptive conditioning "is by its very nature largely
unconscious in character" (p. 97). He went on to state
that the existence of interoceptive conditioning constitutes
evidence contrary to the interpretation that CR acquisition
is due solely to "cognitive-perceptual expectancies and
anticipations" (p. 9 8). The present author would agree
with the last conclusion if interoceptive conditioning were
shown to be necessarily independent of RL. Unfortunately,
Razran did not report any measures of RL and, thus, whether
or not interoceptive conditioning necessarily excludes
RL remains an unanswered empirical question.
A more recent review of the interoceptive condi
tioning literature (Uno, 19 66) arrived at a different
conclusion:
Impulses originating from interoceptors cannot be
categorically classified as being unconscious without
taking into consideration the nature of the stimulus
conditions. The general conclusions specifying that
interoceptive conditioning is unconscious are based on
10
misinterpretations and over-generalizations of empirical
observations of Soviet investigators. Soviet studies
have generally used animals and, in some instances,
human subjects, without including the methodology for
determining various levels of awareness for both
animals and human beings. (pp. 16-17)
In a recent interoceptive conditioning study, Uno
(1966) employed water of different temperatures in an
esophageal balloon as interoceptive CSs and found that the
majority of Ss could differentiate among them. This find
ing demonstrates the falsity of the statement that inter
oceptive stimuli are necessarily below the threshold of
perception.
Uno also found FIR (First Interval Response is the
response which occurs after the CS but before the UCS) and
SIR (Second Interval Response is the response which occurs
after the UCS point on test trials where the UCS is not
presented) GSR discrimination by Ss who were aware of the
CS-UCS relation. However, for Ss who were unaware of the
CS-UCS relation only SIR discrimination was reported. The
latter finding would seem to indicate that at least a
limited form of CR acquisition can occur in the absence of
RL. In the words of the author, "Awareness is necessary
for FIR conditioning, but not for SIR conditioning" (p. 137).
However, as Uno pointed out, the method of measuring RL was
11
somewhat incomplete. The various methods of measuring RL,
including the one employed by Uno, will be discussed later.
Thirdly, it is asserted that CR acquisition can
occur when S is made unaware of the CS-UCS relation by a
masking task. Studies which used the masking task method
will now be reviewed.
Razran (1955b) claims to have established salivary
CR acquisition in the absence of RL by misinforming £ of
the purpose of the experiment and embedding the CS-UCS rela
tion in a masking task. However, the only evidence pre
sented to support the claim that RL was absent was the
statement, "At no time was there any evidence that any
'misinformed' S turned 'informed'— partly or fully— in any
task by himself" (p. 650) . Unfortunately, the method of
assessing awareness and the criterion of awareness were not
specified and the reader is left to wonder whether RL could
have been a central process mediating the salivary CR.
Spence and Spence (1966) reported a series of "experiments
in which conditioning was masked by the probability-learning
task" (p. 139). This quote would seem to indicate that CR
acquisition was studied without the influence of RL. Yet,
like Razran's, at least two of the cited experiments failed
to report how RL was measured (Goldstein, 1962; Homzie and
12
Weiss, 1965).
The above studies were found to be inconclusive
concerning the question of whether CR acquisition can occur
in the absence of RL because of the failure to specify the
methods used to assess RL. A series of experiments by
Lacey and Smith (1954), Chatterjee and Eriksen (1962),
and Golin (1961) illustrate inconclusive results even when
the method of measuring RL is specified. All three sets of
authors employed an improved version of Diven's (1937)
design which required to free-associate to a list of
words while shock was administered 15 sec. after one par
ticular word. Lacey and Smith (1954) reported unaware
heart rate conditioning and generalization using this
method. They reported assessing RL by intensively ques
tioning S^ as to whether he could verbally identify the
particular word that was followed by shock (p. 1045).
Chatterjee and Eriksen (1960) later speculated that
although S might not be able to identify the shocked word
he might be able to state varying degrees of expectancy of
shock for each word and that these expectancies might be
directly related to CR acquisition. To test this specula
tion, Chatterjee and Eriksen essentially replicated the
Lacey and Smith experiment except that (1) GSR, rather than
13
heart rate, was recorded and (2) S's expectancies of shock
were more extensively explored. They found that the CRs to
the words incorrectly identified as being followed by shock
were as large as the CRs to the actually shocked words and
concluded, contrary to Lacey and Smith, that CR acquisition
is no more specific nor precise than verbally assessed RL.
Golin (1961) used essentially the same design as
the above two studies and assessed awareness by having S
identify from a checklist the relation of shock to each
word. The checklist contained the following choices:
(1) I am certain that this word was not followed by shock,
(2) I think that this word was not followed by shock but I
am not certain, (3) I don't remember whether this word was
followed by shock, (4) I think that this word was followed
by shock but I am not certain, and (5) I am certain that
this word was followed by shock. Ss who selected category
5 were excluded from the data analyses and the remaining Ss
demonstrated GSR conditioning— i.e., they responded more to
the shocked word than to a nonshocked word. Further analy
sis revealed that the responses to the shocked word did not
significantly differ between those who selected category 4
and those who selected the remaining categories.
The present author feels that all three of the above
14
studies (Chatterjee and Eriksen, 1960; Golin, 1961; Lacey
and Smith, 1954) made significant advances on the measure
ment of RL from what had been used up to that time. How
ever, it is felt that the question of whether or not CR
acquisition can occur in the absence of RL was not answered
by these studies due to logical deficiencies in the analyses
of the RL measures. These deficiencies will be specified
and elaborated upon in the next section of this paper.
To summarize the five categories of evidence pre
sented at the beginning of this chapter: (1) and (2) data
derived from nonverbal organisms (e.g., phyletically simple
and decorticate animals) are not considered relevant to the
issue because RL is presently defined so as to apply only
to verbal organisms, (3) subliminal conditioning has not
been demonstrated convincingly (Eriksen, 19 60), (4) intero
ceptive conditioning has not been demonstrated to necessar
ily involve subliminal conditioning (Uno, 1966), and (5)
experiments using masking tasks with human Ss have not
employed adequated methods of measuring and analyzing RL
(Chatterjee and Eriksen, 1962; Golin, 1961; Lacey and Smith,
1954; Uno, 1966). In view of the ubiquitousness of the
criticism of the questionnaire techniques, it is appropri
ate to examine at this time the problems associated with
15
measuring RL— especially in reference to the above studies.
Before examining the questionnaire techniques, how
ever, one additional line of evidence that CR acquisition
can occur without RL should be mentioned. Kimble (1962)
has presented data which reveal that Ss who possess RL do
not always exhibit CR acquisition and that in the typical
eyelid conditioning situation RL occurs several trials
before the CR appears. These data show that RL is not al
ways a sufficient condition for CR acquisition but they
shed no light on the question of whether RL is a necessary
condition for CR acquisition. Therefore, these data are not
considered relevant to the experimental question of whether
or not CR acquisition can occur in the absence of RL.
The Measurement of RL
The goal of a questionnaire designed to measure RL
should be to elicit all of the information (correct and
incorrect) which S possess about the predictability of the
UCS. From this information, awareness of the CS-UCS rela
tion and any correlated hypotheses (Adams, 1957) can be
evaluated. One direct method of measuring RL is to simply
ask S if he was aware of any stimulus being associated with
the UCS (Uno, 1966). There are several shortcomings to
16
this method: (1) S may interpret "associated" to mean
"perfectly correlated" and there is no association in this
sense if partial reinforcement and special test trials are
employed, as they were by Uno, (2) the degree of estimated
association and degree of certainty of the stated associa
tion are not evaluated, (3) differential association between
the reinforced CS and the nonreinforced CS is not evaluated.
A method of measuring RL which circumvents the
first two shortcomings is to have £ select from a series of
statements the one which best describes the association
between the reinforced CS and the UCS (Golin, 1961). How
ever, this method does not eliminate the third shortcoming.
More specifically, the third shortcoming is that CR acqui
sition is defined as a difference in responding to the
reinforced CS and the nonreinforced CS while RL is measured
as only the response to the reinforced CS. The present
author feels that if differential conditioning is going to
be studied then differential RL should be measured. One
way of measuring differential RL would be to record the
verbal expectancies of the UCS for the reinforced CS and
the nonreinforced CS and then compute a difference score
between the two expectancy ratings.
Chatterjee and Eriksen (1960) stated somewhat the
17
same idea when they criticized Lacey and Smith's question
naire on the grounds that it did not elicit all of the
information about shock expectancy available from S's
verbal report. Unfortunately, although Chatterjee and Erik-
sen implied the need for equating the expectancy ratings
between reinforced and nonreinforced stimuli, they did not
satisfy it. Their method of determining the relation
between RL and CR acquisition can be divided into two
stages. First, expectancy of shock ratings were obtained
for the experimental (reinforced) and for the control (non
reinforced) stimuli. Second, the GSR to the experimental
stimulus, regardless of its expectancy rating, was compared
with the GSR to control stimuli which were rated as being
followed by shock. No significant differences were found
between the GSRs and it was concluded that conditioning is
no more specific nor precise than verbally assessed expect
ancy ratings (RL). However, 5 of the 14 Ss actually rated
the experimental stimulus as not being followed by shock
and their responses were included in the above comparison.
In effect, then, Chatterjee and Eriksen compared the GSRs
to stimuli unanimously rated as being followed by shock (in
reality, the nonreinforced control stimuli) with the GSRs
to stimuli only partially rated as being followed by shock
18
(in reality, the reinforced experimental stimuli). Thus,
because Chatterjee and Eriksen failed to equate expectancy
ratings, their conclusion about conditioning and RL is
presently judged to be not valid. The present author feels
that, to make the conclusion that conditioning is no more
specific nor precise than RL, the ratings of the experi
mental and control stimuli should have been equated. Thus,
the 5 Ss mentioned above who rated the shocked stimulus as
not being followed by shock should have had a control
stimulus which was also rated as not being followed by
shock.
A subtle distinction between two types of unaware
ness is implied in the above discussion. The first type is
defined as ! 3 being unable to verbalize an expectancy of the
UCS following the reinforced CS and is called "absolute
unawareness" while its converse is called "absolute RL."
The second type is defined as S being unable to verbalize
differential expectancy of the UCS following the reinforced
and nonreinforced stimuli and is called "relative unaware
ness" while its converse is called "relative RL." Thus,
for example, in Golin's experiment (page 13 of this paper)
Ss who selected categories 1, 2, or 3 for the reinforced CS
would fulfill the criterion of absolute unawareness while
19
Ss who selected the same categories (be they 1, 2, 3, 4, or
5) for the reinforced and nonreinforced CSs would qualify
as relatively unaware. It is the present contention that
if CR acquisition is defined as differential responding
then RL must also be defined as differential responding in
order to legitimately determine, whether one can occur with
out the other.
Another aspect of the problem of measuring RL con
cerns the task of logically inferring the relationship
between RL and CR acquisition. In order to appreciate this
aspect, assume for a moment that an experiment did distin
guish between the two types of RL and obtained the follow
ing results: Ss who expressed relative and absolute RL
demonstrated good CR acquisition, Ss with only relative RL
demonstrated fair CR acquisition and Ss with only absolute
RL demonstrated no CR acquisition. These hypothetical data
reveal a positive correlation between two types of responses
— verbalized RL and CR acquisition. At least three conclu
sions can be drawn from these data: (1) relative RL
mediated CR acquisition, (2) CR acquisition mediated rela
tive RL, and (3) relative RL and CR acquisition were
mediated by a systematically correlated subject variable.
Any of the three conclusions could be correct therefore
20
none of the three can be legitimately inferred.
One way to avoid this inferential problem is to
define RL in terms of stimulus manipulations. This method
eliminates the practice of selectively rejecting Ss, which
the method of post-conditioning verbalizations involves,
and allows a causal inference regarding the relation between
RL and CR acquisition. This inference can be made because
it is based on a stimulus-response relation rather than a
response-response relation (Underwood, 1957, pp. 34-35).
Still another way to avoid the inferential problems
associated with the post-conditioning verbalizations is to
introduce the concept of time (DeNike, 1964; Fuhrer and
Baer, 1965 and 1966). For example, DeNike demonstrated that
only aware Ss acquired an instrumental CR and that they did
so only after they expressed awareness during intertrial
interval reports. By demonstrating this temporal relation
between awareness and conditioning he was able to separate
the three conclusions described above, namely, to support
the conclusion that awareness mediated CR acquisition.
In light of the above discussion it was decided to
employ four different operational definitions of RL:
1. RL is defined as the pre-conditioning presenta
tion of instructions which inform S of the
impending CS-UCS relation. This is a stimulus
definition of RL and allows an inference of a
causal relation between RL and CR acquisition.
2. RL is defined as an intertrial interval report
which expresses the CS-UCS relation or some
correlated relation. This is a response defi
nition of RL and allows an inference of a tem
poral relation between RL and CR acquisition.
3. RL is defined as the post-conditioning selec
tion in a multiple-choice situation of a state
ment which expresses the fact that CS+ was
followed by the UCS. This is a response defi
nition of RL and allows an inference of a
relation between absolute RL and CR acquisition.
4. RL is defined as the post-conditioning selec
tion in a multiple-choice situation of a state
ment which expresses the fact that CS+ was
followed by the UCS more often than was CS-.
This is a response definition of RL and allows
an inference of a relation between relative RL
and CR acquisition.
It is the purpose of this study to determine the
relation between human CR acquisition of the GSR and each
22
of the operational definitions of RL. The specific hypoth
eses are presented in Chapter IV of this paper.
CHAPTER III
CR ACQUISITION AND THE INTERSTIMULUS INTERVAL
The purpose of the present investigation was to
determine whether human GSR CR acquisition can occur in the
absence of RL. This question was studied with a short and
a long CS-UCS interval (called interstimulus interval or
ISI) for two reasons: (1) different learning processes may
occur with short and long ISIs and (2) only one GSR may be
measured with the short ISI while more than one may be
measured with a long ISI.
The first reason for using two ISIs is that avail
able data suggest that fundamentally different learning
processes occur with short and long ISIs. For example,
short ISIs (in the neighborhood of .5") have almost uni
versally been found to produce optimal conditioning of
skeletal responses (Fitzwater and Thrush, 1956, with human
finger withdrawal; McAdam, Knott, and Chiornia, 1965, with
cat leg flexion; McAllister, 1953a & b, with human eye-
blink; Schneiderman, 1966, with rabbit nictitating membrane).
These data suggest that there is some unique property of
23
24
the short ISI, at least for skeletal responses.
The search for an optimal ISI with autonomic, rather
than skeletal, conditioning has not yielded as clear a pic
ture. Early experiments using this variable reported
optimal autonomic conditioning with short ISIs (again
approximating .5") (Moeller, 1954; White and Schlosberg,
19 52). More recent experiments, however, have reported
equally good CR acquisition of autonomic responses with
long ISIs as with short ones (Bierbaum, 1958; Grings,
Lockhart, and Dameron, 1962; Silverman, 1960). However, a
close inspection of the data of those who reported no
difference in CR acquisition between short and long ISIs
still reveals a certain uniqueness of the short ISI. For
example, Silverman (1960) used a simple conditioning para
digm (i.e., only one CS was employed) with .5" and 6" ISIs.
He reported that the conditioned GSR of the 6" ISI group
rapidly decreased to its adaptation level during extinction
while the conditioned GSR of the .5" ISI group remained
significantly above its adaptation level during extinction.
This difference between ISI groups suggested to Silverman
"that the .5 sec. is conditioned, while the 6 sec. group is
either weakly conditioned or merely is sensitized to the
impending shocks" (p. 124). In another example, Grings,
25
Lockhart and Dameron (1962) compared .5" and 5" ISIs with a
discrimination conditioning paradigm (i.e., two CSs, one
reinforced and the other nonreinforced, were employed).
They, like Silverman, measured the GSR and found the con
ventional increasing learning curve only for the .5" ISI
group. Learning was also demonstrated for the 5" ISI group
but it was not of the conventional acquisition pattern.
To summarize the first reason, two ISIs were used
in the present study because data exist which suggest that
fundamentally different learning processes occur with short
and long ISIs. If this is true, then it may be that CR
acquisition can occur without RL with one ISI but not with
the other.
The second reason for using two ISIs is that a long
interval (8") permits an analysis of at least three dis
tinct GSRs while a short interval (1") permits only one
analyzable GSR (Grings, Lockhart and Dameron, 1962). The
approximate time periods of the three distinct GSRs are:
(1) the response(s) immediately following CS onset (from
1 to 4 seconds after CS onset), (2) the response(s) prior to
the UCR point in time (from 4 to 9 seconds after CS onset)
and, (3) the response(s) after the UCR point in time on
test trials where the UCS is omitted (from 9 to 13 seconds
26
after CS onset). A great deal of interest has been gener
ated recently concerning these distinct responses but the
labelling of them has proved a source of confusion. Lock
hart (1966b) has proposed the following terminology for
each of the three responses respectively: (1) CS response,
(2) pre-UCS response and, (3) post-UCS response. For want
of a better terminology, Lockhart's terms will be adopted
for the present study although no theoretical interpreta
tion of the response mechanism is implied. It is meant as
a purely descriptive system.
There is evidence to suggest that the three re
sponses reflect different processes. For example, Grings,
Lockhart and Dameron (1962) reported that the post-UCS
response conditioned faster and to a greater degree than
the largest of either the CS response or the pre-UCS
response. This finding so impressed the authors that they
stated that an explanation of the post-UCS response "will
require a reinterpretation of the nature of autonomic
classical conditioning" (p. 33). In a subsequent article,
Lockhart and Grings (1964) speculated that the post-UCS
response represents a more basic process and may be more
independent of RL than either the CS response or the pre-
UCS response. Additional evidence that the different
27
responses reflect different processes comes from a recent
study by Prokasy and Ebel (1967). Upon comparing the CS
response with the pre-UCS response, which they call the
first response and the second response, they concluded:
They are essentially independent responses. The
evidence takes a variety of forms. First, CS intensity
affects the first response, but not the second response.
Second, the CS-UCS pairing resulted in an amplitude
change in the first response but not in the second
response. Third, the first response tends to decrease
in frequency over trials while the second response
increases slightly with CS-UCS pairing. Fourth, des
pite the opportunity for scoring biases, no conditional
relationships between the first and second responses
were found. (p. 254)
To summarize the second reason, two ISIs were used
in the present study because data exist which demonstrate
that with long ISIs several responses may occur while with
short ISIs only one may occur and each of these responses
may reflect fundamentally different processes. If this is
true, then it may be that CR acquisition may occur without
RL with one or more of these responses but not with the
others.
CHAPTER IV
EXPERIMENTAL HYPOTHESES
The principal experimental question can be phrased
quite simply: can human discrimination GSR CR acquisition
occur in the absence of RL. Those theorists who hypothe
size that CR acquisition is a simple direct association
(e.g., Razran, 1955a) predict the occurrence of CR acquisi
tion in the absence of RL. Those theorists who hypothesize
that CR acquisition is due to a cognitive association
(e.g., Woodworth, 1958) predict the lack of CR acquisition
in the absence of RL. For the remainder of the present
paper the former will be called the "direct-association"
interpretation and the latter will be called the "cognitive-
association" interpretation.
In order to test the principal experimental ques
tion, one group was presented a paradigm which contained
all of the generally accepted requirements for CR acquisi
tion and yet excluded RL (noninformed group). So as to
insure that this paradigm did not contain elements which
would preclude GSR discrimination, another group was
28
29
presented the identical paradigm except that RL was induced
by instructions of the CS-UCS relation (informed group).
Tests of whether the paradigm did contain all of the
generally accepted requirements for CR acquisition and
whether the two groups did have functionally identical
tasks will be presented in the Results section.
From the point of view of the cognitive-association
interpretation, to which the present investigator ascribes,
the following are the major experimental hypotheses.
1. Repetitive CS-UCS pairings embedded in a mask
ing situation will not be sufficient to estab
lish CR acquisition if S is not informed of the
CS-UCS relation. Yet, the identical paradigm
will be sufficient to establish CR acquisition
if S^ is informed of the CS-UCS relation.
2. CR acquisition will occur only after relative
RL is expressed during the intertrial interval
reports.
The above are called the major experimental hypoth
eses because they are capable of answering the principal
experimental question of whether CR acquisition can occur
in the absence of RL and because the direct-association and
the cognitive-association interpretations make conflicting
predictions.
Additional hypotheses, called subhypotheses, are
also made. They are called subhypotheses because they do
30
not bear directly upon the principal experimental question
and because the two theoretical interpretations do not
necessarily make conflicting predictions. The following
are the subhypotheses:
1. Ss who express absolute RL (post-conditioning
selection of a statement indicating that the
reinforced CS was followed by shock) will
respond more to the reinforced CS than Ss who
do not express absolute RL.
2. Ss who express relative RL (post-conditioning
selection of a statement which indicates that
the reinforced CS was followed by shock more
often than the nonreinforced CS) will discrimi
nate better than Ss who do not express relative
RL.
As stated above, the direct-association and the
cognitive-association interpretations do not necessarily
make conflicting predictions concerning the subhypotheses
because both interpretations admit that cognitions may
influence autonomic behavior. The only real conflict,
then, between the two interpretations concerns whether CR
acquisition can occur in the absence of RL— i.e., the major
experimental hypotheses.
The logic of the major experimental hypotheses rests
partly upon the phrase "generally accepted requirements for
classical conditioning.1 1 What are these requirements?
Although there is not unanimous agreement, a reading of
31
current literature seems to indicate that there are three
generally accepted requirements for classical conditioning.
The three are the following:
1. A relatively neutral stimulus which is attended
to by S must serve as the CS.
2. A potent (significant or emotional) stimulus
which reliably elicits an unconditioned response
must serve as the UCS.
3. The CS must regularly precede the UCS a large
percentage of the time on a number of occasions.
In order to test the principal experimental ques
tion of whether human CR acquisition can occur in the
absence of RL it is necessary to satisfy these three re
quirements and yet exclude the development of RL.
CHAPTER V
METHOD
Subjects
Fifty students from introductory psychology classes
at the University of Southern California served as subjects
(Ss). The data of 10 Ss were discarded, 8 due to _E error
and 2 due to base levels exceeding 100,000 ohms. Of the
remaining 40 Ss, 10 were assigned to each of the four
experimental groups employed in the present design. The
sex ratio in each group was kept approximately equal, being
either 6 males and 4 females or 5 males and 5 females.
The assignment of Ss to the informed or the non
informed conditions was not random. Noninformed conditions
were more likely to be filled early in the experiment while
informed conditions were more likely to be filled later.
More specifically, 1 of the first 10 £s was informed, 5 of
the second 10 Ss were informed, 7 of the third 10 Ss were
informed and, 7 of the fourth 10 Ss were informed. The
reason for this systematic assignment of Ss to the experi
mental conditions was to decrease the probability of an
32
33
informed S telling a subsequently noninformed classmate of
the CS-UCS relation. The chance of this communication
among Ss was deemed a more likely and a more undesirable
source of error than that due to the systematic assignment
of Ss. Besides, as will be shown in the Result section,
initial group differences were assessed by comparing the
initial orientating response of the groups.
Apparatus
The two auditory CSs were presented among three
additional tones which were part of a masking task. The
frequencies of the CSs were 800 and 1200 cps while the fre
quencies of the masking task tones were 950, 1000, and 1050
cps. All of the tones were originally produced by a
Hewlett-Packard audio-oscillator and were recorded on a
Wollensak tape recorder. During the experimental session,
tones were presented from the tape recorder through a loud
speaker three feet to the right-front of S. Measured at
S's location, tone intensities were 5-7 db. above a 50 db.
constant masking noise.
The UCS, a dc shock from a Grass Model S-5 stimula
tor, was administered to the volar surface of the right
forearm through 1.7 cm. circular silver electrodes (spaced
34
2.5 cm. from center to center). UCS intensity was arrived
at by increasing shock voltage in 2-3 volt steps until S
reported "uncomfortableness."
In the original taping, tone durations of .5 sec.
were controlled by a Hunter electronic timer while the
intervals between the tones of 1 sec., 3 sec., 5 sec., or
8 sec. were manually controlled. The intertrial intervals
of 30 sec., 40 sec., and 50 sec. (average 40 sec.) were
also manually controlled. The intertrial interval is
defined as the time from the CS offset of one trial to the
onset of the first tone of the next trial. While the CSs
were automatically presented once they were on tape, the
UCSs were manually presented on each trial. This was done
by E closing a relay switch at the onset of the CS. The
relay switch initiated two timers: one timer controlled
the ISI of either 1 or 8 sec. and the other controlled the
UCS duration of .5 sec.
A Bell and Howell tape recorder was on continually
throughout the experimental session and was used to record
the ITI reports and the oral segment of the post
conditioning interview. A Grason-Stadler noise generator
was used to produce a white noise "answer signal" which
informed S when to begin the ITI report.
35
The GSR was obtained as a dc resistance change
through 1/2 by 5/8 in. silver electrodes, coated with elec
trode paste, taped firmly to S's first and third finger
tips of the left hand. The GSR was recorded on a Type R
Offner Dynograph which received input from a Darrow-type
bridge circuit. A constant 45 yA passed through the GSR
electrodes.
Design and Procedure
Two independent variables, each varied at two lev
els, produced a 2 X 2 factorial design with four different
experimental groups. The two independent variables were:
(1) instruction— half of the Ss were informed of the CS-UCS
relation while the other half were not informed and
(2) interstimulus interval (ISI)— half of the informed and
half of the noninformed Ss were administered a 1 sec. ISI
while the other halfs were administered an 8 sec. ISI. The
four resultant experimental groups were: informed 1 sec.
(II), informed 8 sec. (18), noninformed 1 sec. (Nl), and
noninformed 8 sec. (N8).
The experimental session can be divided into 7
stages.
1. Application of electrodes. After being led
36
into the experimental room, which was separated from the
equipment room by a wall and a one-way mirror, S was seated
in a comfortable chair and GSR and shock electrodes were
applied. At this time all Ss were told that the purpose of
the experiment was to determine whether "physiological
activity," as measured by the GSR, would facilitate percep
tion. Shock, it was explained, would be periodically
administered in order to create the "physiological activ
ity." The specific instructions are presented in Appendix
A.
2. Shock work-up. Shock was increased in 2-3 volt
steps with S instructed to say aloud when it was first
felt and when it became "uncomfortable." The voltage at
which £ reported "uncomfortableness" was the intensity of
the UCS employed throughout the remainder of the session.
The specific instructions are presented in Appendix A.
3. Perceptual task instructions I. All Ss were
told that, first, one tone (called the reference tone) and
then four more tones (called choice tones) would be pre
sented. Their task was to select one tone from the four
choice tones which sounded most like the reference tone and
then to make their selection known by saying aloud the
ordinal number (1, 2, 3, or 4) of their selection. They
37
were to do this only after they heard an "answer signal"
(white noise) from a loud speaker in front of them. They
were also instructed to report, after they had stated their
answer, their "observations, thoughts, and feelings" that
had occurred since the last time they reported. The inter
trial interval definition of RL and the subsequent test of
experimental hypothesis number 2 (p. 29) were based on the
content of these reports. The specific instructions are
presented in Appendix A.
4. Familiarization trials. All Ss were told that
a total of 5 different tones would be presented during the
entire experiment. They were told that the 5 tones would
now be presented so as to acquaint them with the tones.
The specific instructions are presented in Appendix A. All
Ss were then administered one presentation of each of the
5 tones (2 CSs and 3 perception task tones) in counter
balanced order. The time intervals between the tones
during the familiarization trials were approximately 5 sec.
The specific orders of tone presentations are presented in
Appendix B.
5. Perceptual task instructions II. All Ss were
told that one additional perceptual task would be required.
This task was to state whether the last of the four choice
38
tones was the highest or the lowest of all the tones and
they were assured that the last tone would always be one of
these two. At this time the first independent variable was
employed. The informed Ss were told that shock would
usually but not always be presented after the highest (low
est) tone and never after the lowest (highest) tone while
the noninformed Ss were merely reminded that shock would be
presented periodically. This difference in instructions
was the only procedural difference between the informed
and the noninformed conditions. The instructional defini
tion of RL and the subsequent test of experimental hypothe
sis number 1 (p. 29) were based on the difference between
these two conditions. All Ss were then asked to repeat
what was going to happen and what they were supposed to do.
The repetition of instructions was required in order to
insure that the task was clearly understood. The specific
instructions are presented in Appendix A.
6. Acquisition trials. All Ss were administered
30 trials of a discrimination conditioning paradigm embedded
in a masking task. Each trial consisted of one reference
tone and four choice tones with the last of the choice
tones being either the highest or the lowest of all the
tones and serving as either CS+ or CS-. A schematic
39
representation of the first two acquisition trials is shown
in Table 1. The frequency of tones serving as CS+ and CS-
were counterbalanced in each group. There were a total of
15 presentations of CS+ and 15 presentations of CS- with
shock (UCS) usually following CS+ (67% reinforcement) and
never following CS-. Half of the informed group and half
of the noninformed group were administered a 1 sec. ISI
while the other halfs were administered an 8 sec. ISI.
TABLE 1
SCHEMATIC REPRESENTATION OF THE FIRST TWO
ACQUISITION TRIALS
Trial
Reference
Tone
Choice Tones
1
x2
X3 Xi X2 CS+ UCS ans. signal
2
x3
xx x3 X2 CS- — ans. signal
Acquisition was administered in three blocks of
10 trials so that there were 5 presentations of each CS in
each block. Test trials (nonreinforced presentations of
CS+ and adjacent presentations of CS-) occurred on the
following trials: 5 (CS-), and 6 (CS+); 11 (CS+) and 12
(CS-); 19 (CS-) and 20 (CS+); 25 (CS+) and 26 (CS-); and
29 (CS-) and 30 (CS+). All Ss received identical
40
acquisition trials except for the counterbalancing of fre
quencies of CS+ and CS- and except for the ISI variation.
The specific sequence of acquisition trials along with the
intertrial intervals and the time intervals between tones
are presented in Appendix B.
It should be mentioned that the use of a .5 sec.
duration CS and a 1 sec. or 8 sec. ISI produces a trace
conditioning paradigm with a .5 sec. or a 7.5 sec. trace
interval.
7. Questionnaire. The post-conditioning interview
consisted of a tape recorded oral segment and a written
part. The oral segment consisted of: asking S to review
what he had observed, thought, and felt during the experi
ment, asking S to estimate the number and uncomfortableness
of the shocks and, asking J 3 to state whether he could, on
any basis, predict the occurrence of shock. An additional
question was asked of the informed Ss? namely, whether the
instructions about shock occurrence were correct. The
specific questions are presented in Appendix C.
The written part of the questionnaire consisted
simply of repeating the familiarization trials, while
recording the GSR, and asking £ to select one of the follow
ing 7 categories for each of the 5 tones.
41
1. I am certain that this stimulus was never followed
by shock.
2. I don't think this stimulus was ever followed by
shock.
3. I don't remember whether this stimulus was ever
followed by shock.
4. I think this stimulus was sometimes followed by
shock.
5. I am certain that this stimulus was sometimes
followed by shock.
6. I think that this stimulus was always or almost
always followed by shock.
7. I am certain that this stimulus was always or
almost always followed by shock.
The definition of absolute and relative RL and the
subsequent tests of experimental subhypotheses 1 and 2
(p. 30) were based on these questionnaire data. Absolute
RL is defined as the selection of categories 4, 5, 6, or 7
for the CS+ stimulus. Relative RL is defined as the selec
tion of a higher category for CS+ than for CS-. The post
conditioning interview terminated with the selection of
ISIs for those stimuli which S had demonstrated absolute
RL. The specific written questions are presented in Appen
dix C.
In summary, all Ss were presented the same percep
tual masking task. At the same time, all Ss were adminis
tered identical paradigms which contained all of the
generally accepted requirements for classical conditioning.
Two groups (Nl and N8) were not informed of the CS-UCS
42
relation while two additional groups (II and 18) were
informed of the CS-UCS relation. All £s were instructed to
state their "observations, thoughts, and feelings" during
the intertrial intervals. Finally, all Ss were administered
a post-conditioning interview.
Response Measurement
For reasons noted previously, the three time peri
ods of from 1.0 to 3.9 sec., from 4.0 to 8.9 sec., and from
9.0 to 12.9 sec. were analyzed separately for the 8 sec.
ISI groups while only the first time period was analyzed
for the 1 sec. ISI groups. Statistical analyses were com
puted on the largest response beginning in each of the
above time periods after the response had been transformed
to a square root of change in conductance score. All
statistical tests were two-tailed.
CHAPTER VI
RESULTS
Experimental Presuppositions
Equality of groups. As described earlier, Ss were
not assigned randomly to the informed and noninformed
groups. Therefore, it is especially relevant in the pres
ent study to test for possible initial group differences
and the method of doing this was to analyze the data of the
familiarization trials. Figure 1 presents the mean magni
tude of the CS response (GSRs with latencies ranging from
1.0 to 3.9 sec.) to each of the five familiarization tones
of each of the four experimental groups. An analysis of
variance was computed and its summary is shown in Table 2.
The absence of significant effects supports the experimen
tal suppositions of (1) no initial difference in GSR
responsivity among the groups and (2 ) no initial difference
in GSR elicitation properties among the stimuli.
A more subtle type of group difference also might
occur. It is possible that differences in the experimental
43
Mean GSR Magnitude (/AC)
44
. 0
. 9
• - -------4
. 8
.7
.6
.5
.4
. 3
# • 11
o— O 1 8
•- -# N1
O - O N8
.2
.1
o
1000 1050 900
Stimuli (cps)
1200
800
Fig. 1— Mean magnitude of CS response during
familiarization trials of the four experimental groups.
45
tasks may have been created by the differential instruc
tions regarding the CS-UCS relation. For example, the
informed group may have been more attentive to the CSs
because they were told that the CSs signaled shock and con
versely, been less attentive to the perceptual discrimina
tion task because they were told that it was irrelevant to
shock. To test for this possibility, the four experimental
groups were compared on how well they identified the CSs
and how well they performed the perception task.
TABLE 2
ANALYSIS OF VARIANCE OF THE CS RESPONSE
DATA DURING FAMILIARIZATION TRIALS
Source df
SS
MS F
Between subjects ( 39) (38.60)
Information (I) 1 .17 .17 —
ISI 1 .33 .33 —
ISI X I 1 .32 .32 —
Error (b) 36 37.78 1.05
Within subjects (160) (35.38)
Tones (T) 4 1. 38 .34 1.55
I X T 4 . 1 2 .03
—
ISI X T 4 1.03 .26
—
ISI X I X T 4 1 . 2 1 .30 1.36
Error (W) 144 31.64 . 2 2
Total (199) (73.98)
46
The number of errors made by each group in the
identification of CSs is shown in Table 3. A Kruskal-
Wallis analysis of variance-*- (Siegel, 1956, pp. 184-193)
was computed and, despite a tendency for the noninformed
groups to make more errors, the difference among groups was
not significant [H(3)=3.50]. The number of perceptual task
errors made by each group is shown in Table 4. Another
Kruskal-Wallis analysis of variance was performed and again,
despite a tendency for the noninformed groups to make more
errors, the difference among groups was not significant
[H(3)=3.21].
Requirements for classical conditioning. As noted
previously (p. 31), there appear to be three generally
accepted requirements for the establishment of classical
conditioning. The question is, did the present experimen
tal design satisfy each of these requirements.
The first requirement is that a neutral stimulus
which is attended to by S must serve as the CS. The
neutrality of a CS is difficult to demonstrate with the GSR
because any perceptible novel stimulus will elicit an
■ * " A nonparametric analysis was employed because the
data were not normally distributed and because they were
dimensioned along an ordinal scale.
47
TABLE 3
LIST OP NUMBER OF CS IDENTIFICATION ERRORS FOR EACH S
IN EACH EXPERIMENTAL GROUP
1 1 18 N1 N8
s# Errors S# Errors S# Errors S# Errors
4 0 14 0 1 0 2 0
16 0 22 0 8 0 7 0
2 0 0 25 0 15 0 9 0
2 1 0 27 0 34 0 18 0
26 0 29 0 17 1 23 0
30 0 37 0 24 1 31 0
35 0 38 0 28 1 36 1
39 0 32 1 5 4 1 0 1
40 1 33 2 6 4 13 2
1 1 2 19 2 1 2 4 3 6
X .3 .5 1.5 1 . 0
LIST OF NUMBER OF
TABLE 4
PERCEPTUAL TASK
IN EACH GROUP
: ERRORS FOR EACH S
1 1 18 N1 N8
S# Errors S# Errors S# Errors CO
Errors
4 0 38 0 15 1 23 0
2 0 0 14 1 1 2 2 3
35 1 37 1 28 2 7 3
39 1 27 3 8 3 13 3
2 1 2 32 3 24 4 31 6
30 3 2 2 4 34 4 18 7
26 4 29 5 12 10 9 8
40 5 25 6 17 10 10 8
16 6 33 6 5 13 36 9
1 1 1 2 19 13 6 15 3 1 0
X 3.4 4.2 6.4 5.7
orienting response (OR), one component of which is the GSR.
However, it can be said that the CSs in the present experi
ment were relatively neutral in the sense that the responses
elicited by the UCS were much larger than the responses
elicited by the CSs. It should also be pointed out that
classical conditioning has been reported with approximately
the same CS intensities (Grings and Dawson, 1965) and with
apparently much more intense CSs (Lockhart, 1966a) than
those used in the present study.
The second aspect of the first requirement is that
the CS must be attended to by S. This aspect poses a genu
ine interpretational problem for studies employing a masking
task. For example, Dawson and Grings (in press) reported
the failure of CR acquisition when a masking task was
employed. However, the masking task may have prevented the
perception of the individual CSs and prevented the learning
of the CS-UCS relation, therefore the authors were unable
to determine which effect was responsible for the lack of
CR acquisition. In the present study, the perception task
required S to attend to the last of the choice tones (CSs)
and to verbally report whether they were the highest or the
lowest of all the tones (CS+or CS-). This requirement
insured that S would attend to the CS and also permitted an
49
analysis of the correctness of S/s perception. Table 3
reveals that the majority of Ss correctly perceived (i.e.,
verbally reported) all of the CSs and that only 10% incor
rectly reported more than 2 of the 30 CSs.
In summary, it may be concluded that relatively
neutral stimuli which were attended to by S served as CSs.
The second requirement is that a potent, response-
producing stimulus must serve as the UCS. The UCS in the
present study was potent in the sense that most of the pre
conditioning (shock work-up) and post-conditioning ratings
agreed that the shock was "uncomfortable." An examination
of the UCR data revealed that 380 of the 400 UCS presenta
tions (i.e., 95%) elicited UCRs (where a UCR is defined as
at least a 1 0 0 0 ohm resistance change occurring between 1 . 0
and 3.5 sec. after UCS onset). In summary, it may be con
cluded that a potent, response-producing stimulus served
as the UCS.
The third requirement is that the CS must regularly
precede the UCS on a number of occasions. In the present
experiment, the CS+ preceded the UCS by 1 sec. or 8 sec.
on 67% (i.e., 10 of 15) of its presentations. The UCS was
never presented without being preceded by the CS+. Whether
this is a sufficient percentage of reinforcement or a
50
sufficient number of acquisition trials is not known with
certainty. However, in defense of the present parameters,
it should be noted that other studies have reported dis
crimination GSR classical conditioning with similar para
meters (e.g., Grings, Lockhart, and Dameron, 1962; Mandel,
1965; Uno, 19 6 6 ).
In summary of the experimental presuppositions,
the four experimental groups did not differ in GSR respons-
ivity nor in experimental tasks, the two CSs did not differ
in GSR eliciting properties, and the generally accepted
requirements for classical conditioning were present.
Experimental Hypothesis;
RL Defined by Instructions
CS response. The mean CS response magnitude data
during acquisition for the four experimental groups are
shown in Figures 2A and 2B. An analysis of variance was
performed and its summary is presented in Table 5. The
only statistically significant effects were the discrimina
tion [F (1/36)=21.00, £ <.01] and the discrimination by
information interaction [F(1/36)=18.37, £ <.01] effects.
The significant main effect indicates that the over-all
responding to CS+ was greater than that to CS-. The
Mean GSR Magnitude (/AC)
1.6r—
Short ISI
II, CS+
II, CS-
# ---# Nl, CS+
o -o Nl, CS-
Test Trials
Fig. 2A— Mean CS response
magnitude data during acquisition for the
II and Nl group.
# — •
18, CS+
O — — 0
18, cs-
m -
N8 , cs+
o - O
N8 , cs-
\
i i i \ i
1 2 3 4 5
Test Trials
Fig. 2B— Mean CS response
magnitude data during acquisition for m
the 18 and N8 groups.
52
TABLE 5
ANALYSIS OF VARIANCE OF THE CS RESPONSE
DATA DURING ACQUISITION
Source df SS MS F
Between subjects ( 39) (66.80)
Information (I) 1 4.52 4. 52 2 . 8 6
ISI 1 4.99 4.99 3.16
ISI X I 1 .25 .25 —
Error (b) 36 57.04 1.58
Within subjects (360) (115.33)
Discrimination (D) 1 9.03 9.03 2 1 .0 0 a
Trials (N) 4 1.35 .34 —
I X D 1 7.90 7.90 18.37a
ISI X D 1 1.49 1.49 3.47
I X N 4 .41 . 1 0 —
ISI X N 4 .72 .18 —
D X N 4 .32 .08 —
I X ISI X D 1 1.18 1.18 2.74
I X ISI X N 4 .54 .15
—
I X D X N 4 .98 . 25 1.19
ISI X D X N 4 1 . 82 .46 2.19
I X ISI X D X N 4 1 . 81 . 45 2.14
Error (w) (324) (87.78)
Ei (w) 144 41.84 .29
E2 (w) 36 15.37 .43
E3 (w) 144 30.57 . 2 1
Total 399 182.13
ap < .0 1 .
53
significant interaction effect indicates that discrimina
tion performance differed between the informed and nonin
formed groups.
Figure 3 presents the discrimination scores (defined
as the CS+ response minus the CS- response) for the four
experimental groups. The significant effects can be seen
clearly in this figure. Individual t-tests revealed that
the II and the 18 groups significantly discriminated
between CS+ and CS- [t(9)=4.33, p <.01 and t(9)=2.34, p
<.05, respectively] while the Nl and N8 groups did not.
These data suggest that the generally accepted requirements
for the establishment of classical conditioning are not
sufficient to establish CS response CR acquisition unless
RL is also present.
An interesting trend is also apparent from inspect
ing Figure 3. Whereas discrimination of the II group
increased as a function of trials that of the 18 group
decreased. In order to test whether this trend is more
than a chance phenomenon, a separate analysis of variance
was computed on the CS response acquisition data of the II
and 18 groups. The summary of this analysis is available
in Table 6 . The trend noted above is reflected by the ISI
by discrimination by trials interaction. This triple
GSR Discrimination (CS+ Minus CS-
• ---• II
54
+1.0
+ .9
+ .8
+ .7
+ .6
+. 5
+ .4
+. 3
+. 2
+. 1
0
-.1
- .2
-.3
-.4
-.5
O- -O N 8
\
/ \
\
\ /
1
1 2 3 4 5
Test Trials
Fig. 3.--Discrimination scores of the CS response
during acquisition of the four experimental groups.
55
interaction failed to reach the conventionally accepted
level of significance [F(4/72)=2.44, .10>p> .05]. Thus,
despite a suggestive trend, it cannot be reliably concluded
that trials had a differential effect on the discrimination
of the 1 1 and 18 groups.
TABLE 6
ANALYSIS OF VARIANCE OF THE CS RESPONSE ACQUISITION
DATA FOR THE II AND 18 GROUPS
Source df SS MS F
Between subjects ( 19) (33.36)
ISI 1 1.51 1.51 —
Error (b) 18 31.85 1.77
Within subjects (180) (61.96)
Discrimination (D) 1 16. 91 16.91 23.82a
Trials (N) 4 .41 . 1 0
—
D X N 4 .18 .05 —
ISI X D 1 2.67 2.67 3.76
ISI X N X D 4 1.46 . 37 2 .44
Error (w) (162) (39.73)
Ei (w) 72 16.15 . 2 2
E2 (w) 18 12.78 .71
E3 (w ) 72 1 0 . 80 .15
Total 199 95.32
ap < .0 1 .
b.10 > p > .05.
56
Pre-UCS response. The pre-UCS response can be
separately analyzed only for the 8 sec. ISI groups. The
magnitude data of this response for the 18 and N8 groups
are presented in Figure 4 and the discrimination scores are
presented in Figure 5. An analysis of variance, whose
summary is shown in Table 7, was computed and none of the
effects proved to be significant. This means that, despite
a suggestive trend in the 18 group, there was no pre-UCS
response discrimination for either the informed or the non
informed conditions.
TABLE 7
ANALYSIS OF VARIANCE OF THE PRE-UCS RESPONSE
DURING ACQUISITION
Source df SS MS F
Between subjects ( 19) (24.49)
Information (I) 1 1.16 1.16
—
Error (b) 18 23.33 1.30
Within subjects (180) (45.94)
Discrimination (D) 1 .60 .60 1.76
Trials (N) 4 .75 .19
—
I X D 1 . 1 2 . 12 —
I X N 4 1.38 .35 1.40
D X N 4 1.07 .27 1.17
I X D X N 4 1.48 .37 1.61
Error (b) (162) (40.54)
Ei (w) 72 18.14 .25
E2 (w) 18 6.07 .34
E3 (w) 72 16.33 .23
Total 199 70.38
Mean GSR Magnitude (/AC)
57
1 .
18, CS+
18, CS-
CS+
cs-
Test Trials
Fig. 4.— Mean pre-UCS response magnitude during
acquisition for the 18 and N8 groups.
GSR Discrimination (CS+ Minus CS-
58
+1.0
# • 18
Q ~ ~ Q N8
4 3 5 1 2
Test Trials
Fig. 5.— Discrimination scores of the pre-UCS
response during acquisition of the N8 and 18 groups.
59
Post-UCS response. As with the pre-UCS response,
the post-UCS response can be analyzed separately only for
the 8 sec. ISI conditions. The magnitude data of this
response for the 18 and N8 groups are presented in Figure 6
and the discrimination scores are shown in Figure 7. A
summary of the analysis of variance performed on these data
is shown in Table 8 and reveals that only the discrimina
tion by information interaction effect was significant
[F (1/18) =5. 84 , £<.05]. Individual t-tests indicated that
the discrimination of the 18 group was significant [t(9)=
3.50, p<.01] while the discrimination of the N8 group was
not significant. These data suggest that the generally
accepted requirements for classical conditioning are not
sufficient to establish post-UCS response CR acquisition
unless RL is also present.
Questionnaire. In addition to the familiarization
and acquisition trials, the GSR was also recorded during
the portion of the questionnaire when the stimuli were
being presented to S. The interval between the stimulus
onset and £ writing his answer was approximately 5 sec. so
only the CS response data were available for analysis.
Mean GSR Magnitude (/AC)
60
# ---# 18, CS+
O O 18, CS-
0 - 0 N8 , CS+
1 .
o~ ~o N8 , CS-
-a
i 2 4 3 5
Test Trials
Fig. 6.--Mean post-UCS response magnitude during
acquisition of the N8 and 18 groups.
GSR Discrimination (CS+ Minus CS-
61
+1.0
9 • 18
O” ~ 0 N8 + .9
+ .7
+ • 6
+ .5
+ .4
+ • 2
+.i
Test Trials
Fig. 7.— Discrimination scores of the post-UCS
response during acquisition of the N8 and 18 groups.
62
TABLE 8
ANALYSIS OF VARIANCE OF POST-UCS
RESPONSE DURING ACQUISITION
Source df SS MS F
Between subjects ( 19) (28.57)
Information (I) 1 1.13 1.‘ 13
—
Error (b) 18 27.44 1.52
Within subjects (180) (42.53)
Discrimination (D) 1 1.05 1.05 4.20
Trials (N) 4 1.71 .43 1.65
I X D 1 1.46 1.46 5. 84a
I X N 4 .55 .14
—
D X N 4 1.57 .39 2.29
I X D X N 4 .35 .09 —
Error (w) (162) (35.84)
Ei (w) 72 18.81 .26
E2 (w ) 18 4.52 .25
E3 (w ) 72 12.51 .17
Total 199 71.10
ap < .05.
63
Table 9 presents the mean CS response magnitude
data of the CS+, CS-, the difference between CS+ and CS-,
and the corresponding t-score for each group. There is no
sign of discrimination by any group during the question
naire except for the II group whose t-score barely missed
the conventionally accepted level of significance [t(9)=
1.97/ .10< p< .05]. These results reveal essentially the
same pattern as was present on the last acquisition trial—
i.e., CS response discrimination occurred only for the II
group (see Figure 3).
TABLE 9
CS RESPONSE MAGNITUDE (/AC) DATA FOR THE FOUR
EXPERIMENTAL GROUPS DURING THE QUESTIONNAIRE
Group CS+ CS- Dif f. t
1 1 .936 .483 .453 1. 97a
Nl .724 .578 .146 .91
18 .927 .878 .049 . 2 1
N8 .892 .529 .362 1.65
a.10 > p > .05.
64
Experimental Hypotheses;
RL Defined by ITI Reports
The experimental hypothesis concerning the ITI
reports and CR acquisition (hypothesis number 2, p. 29)
presupposed that at least some Ss would fail to exhibit RL
in the beginning of acquisition and yet would exhibit RL at
the end of acquisition. This presupposition was not con
firmed. That is, the noninformed Ss never expressed RL
during the ITI reports. Therefore, the experimental hy
pothesis number 2 is not testable in its original form
because there are no pre-RL and post-RL conditions during
which to compare CR acquisition.
It should not be concluded, however, that the ITI
reports are completely useless. It is still possible to
analyze the ITI reports on a trial by trial basis and then
to compare the "CRs" on trials where S^ reported expecting
shock with the "CRs" on trials where no such expectancy was
reported. The cognitive-association interpretation would
predict greater responding on trials where the expectancy
was reported than on trials where no such expectancy was
reported. However, the direct-association interpretation
may also make the same prediction. The direct-association
theorists do not necessarily deny that verbalized
65
expectancies may initiate CR-like behavior; they just deny
that this is the mechanism of "true" CR acquisition. Thus,
the revised form of hypothesis number 2 is reduced to the
level of a subhypothesis because it will not directly help
to answer the principal experimental question.
If GSR magnitude is found to be positively corre
lated with the ITI reports of expectancy of shock then
neither the direct-association nor the cognitive-association
interpretation will be differentially supported. However,
if no relation or a negative relation is found between GSR
magnitude and the ITI reports then at least one form of the
cognitive-association will be called into question, namely,
that expectancy of shock necessarily elicits autonomic CRs.
The trial by trial analysis described above is not
appropriate for the informed groups (II and 18) because
these Ss never expressed expectancy of shock for any stimu
lus except for CS+. Thus, such an analysis for these
groups would confound the effects of ITI reported expect
ancy with the effects of RL instructions. In light of this
confounding, the ITI reports are related to GSR magnitude
only for the Nl and N8 groups.
Several precautions are required with this type of
analysis. First, the pitch (frequency) of the tone should
66
be the same for the trials with expectancy as for the trials
without it. Second, the two trials being compared should
occur in close succession. Third, the two trials should be
nonreinforced so that the CRs can be analyzed independently
of the UCRs. Fourth, the two trials should occur after
approximately the same number of acquisition trials. And
fifth, each S should contribute approximately the same
amount toward the over-all mean.
In order to satisfy the above requirements, the
following procedure was employed. The acquisition trials
were divided into two blocks of 15 trials each. The first
nonreinforced trial within each block on which S reported
expecting shock was designated ITI+. The nearest nonrein
forced trial of the same tone-frequency on which S made no
report of expecting shock was designated ITI-. The order
of ITI+ and ITI- was counterbalanced.
Blocks of 15 trials were chosen so that the major
ity of Ss would be represented once in each block. The
first trial of the block was chosen so as to approximate
the number of prior acquisition trials. Nonreinforced
trials were chosen so that the CRs could be analyzed with
out interference by the UCRs. The specific trials which
met these criteria are in Table 10. A horizontal line in
s#
1
5
6
8
12
15
17
24
28
34
2
3
7
9
10
13
18
23
31
36
67
TABLE 10
TRIAL NUMBERS AND CONDITIONS ON WHICH ITI REPORT
REVEALED EXPECTANCY OF SHOCK AND ADJACENT
TRIALS WHERE NO EXPECTANCY WAS REPORTED
First Block Second Block
ITI+ ITI- ITI+ ITI-
Nl Group
2 (CS-) 3 (CS-) 29 (CS-) - 28 (CS-)
6 (CS-) - 1 1 (CS-)
26 (CS-) - 28 (CS-)
10 (CS-) 7 (CS-) 2 1 (CS-) - 23 (CS-)
2 (CS-) 5 (CS-) 18 (CS-) - 15 (CS-)
19 (CS-) - 18 (CS-)
3 (CS-) 2 (CS-) 18 (CS-) - 23 (CS-)
3 (CS-) 2 (CS-) 18 (CS-) - 15 (CS-)
N8 Group
10 (CS-) - 7 (CS-)
2 (CS-) - 5 (CS-)
3 (CS-) - 2 (CS-)
10 (CS-) - 12 (CS-)
2 (CS-) - 14 (CS-)
3 (CS-) - 2 (CS-)
19 (CS-) - 21 (CS-)
30 (CS+T) - 25 (CS+T)
18 (CS-) - 29 (CS-)
21 (CS-) - 19 (CS-)
19 (CS-) - 18 (CS-)
68
Table 10 indicates that a pair of trials meeting the above
criteria could not be found in that block for that S.
The mean CS response data for the ITI+ and the ITI-
trials for the N1 group are shown in Table 11. The differ
ence is large and in the direction predicted by the
cognitive-association interpretation but barely failed to
reach the conventionally accepted level of significance
[t(7)=2.13, .10 >p> .05] due to the small N and the large
degree of between-subject variability. The magnitude data
of the three response measures for the N8 group are pre
sented in Table 12. The CS response discrimination is in
the predicted direction and significant [t(6)=3.15, p< .05],
the pre-UCS response discrimination is also in the pre
dicted direction but not significant, and the post-UCS
response is in the opposite direction predicted and is not
significant. In summary, these data indicate that ITI
reports of expectancy of shock are positively related to CS
response discrimination but are unrelated to pre-UCS and
post-UCS response discrimination.
Experimental Subhypotheses: RL Defined
by Post-conditioning Interview
The primary purpose of the post-conditioning inter
view was to distinguish between and evaluate absolute RL
69
TABLE 11
MEAN CS RESPONSE MAGNITUDE (/AC) FOR ITI+ AND ITI-
FOR THE Nl GROUP
ITI+ ITI- Diff. t
.881 .412 .469 2.13a
a.10 > p > .05.
TABLE 12
MEAN GSR MAGNITUDE (/AC) FOR
THREE RESPONSE MEASURE
ITI+ AND
FOR THE
ITI- FOR
N8 GROUP
THE
Response ITI+ ITI- Diff. t
CS response . 8 8 6 .476 .410 3.15a
Pre-UCS Response .714 .595 .119 1.08
Post-UCS Response .468 .548 -.080 .36
ap < .05.
70
and relative RL. Tables 13 through 16 present the absolute
and relative RL scores for each S in each group. The rat
ing of the probability of shock following CS+ is dimen
sioned along the vertical axis and a score of 4 or more
(see Appendix C) indicates the presence of absolute RL.
The rating of CS+ minus the rating of CS- is dimensioned
along the horizontal axis and a score of + 1 or more indi
cates the presence of relative RL.
Table 13 shows that all Ss in the II group exhib
ited both absolute and relative RL. Table 14 shows that,
with the exception of subject number 32, all Ss in the 18
group exhibited absolute RL and all but one, subject num
ber 37, expressed relative RL. An inspection of the
interview data of subject number 32 revealed that this £ 3
expressed both absolute and relative RL in the oral portion
of the interview and then reversed CS+ and CS- in the writ
ten portion. This discrepancy suggests that the interview
data are dependent upon at least two processes: (1) the RL
state during acquisition and (2) S's memory of that state.
Tables 15 and 16 reveal that there are large indi
vidual differences in the absolute and relative scores of
the Nl and N8 groups. These large individual differences
are necessary in order to test whether absolute and
CS+ Rating (Absolute Awareness)
TABLE 13
ABSOLUTE AND RELATIVE RL SCORES FOR Ss IN THE II GROUP
(The numbers in the various RL categories refer to subject numbers.)
CS+ Minus CS- Ratings (Relative Awareness)
+6 +5 +4 +3 +2 +1 0 -1 -2 -3 -4 -5 -6 -7
(1 1 )
(21) (39)
(26)(40)
(35)
(2 0 )
(4)
(16)
(30)
CS+ Rating (Absolute Awareness)
TABLE 14
ABSOLUTE AND RELATIVE RL SCORES FOR Ss IN THE 18 GROUP
(The numbers in the various RL categories refer to subject numbers.)
CS+ Minus CS- Rating (Relative Awareness)
+6 +5 +4 +3 +2 + 1 0 -1 -2 -3 -4 -5 -6 -7
1
_
2
3
__ _
4
(14B) (22B)
5 (19)
(25)
6
_____
7 (33)
(38B)1 1 1 1 1 1 1 1 1 1 1 1 1 ^
to
+6 +5 +4 +3 +2 +1 0 -1 -2 -3 -4 -5 -6 -7
(32)
(29B) (37)
(14B)
(19)
(22B)
(25)
(27)
(33)
(38B)
CS+ Rating (Absolute Awareness)
TABLE 15
ABSOLUTE AND RELATIVE RL SCORES FOR Ss IN THE Nl GROUP
(The numbers in the various RL categories refer to subject numbers.)
CS+ Minus CS- Rating (Relative Awareness)
+6 +5 +4 +3 +2 +1 0 -1 -2 -3 -4 -5 -6 -7
(15B)
(5)
(1 )
(12B) (6)
(8) (34)
(17)
(24B) (28)
CS+ Rating (Absolute Awareness)
TABLE 16
ABSOLUTE AND RELATIVE RL SCORES FOR Ss IN THE N8 GROUP
(The numbers in the various RL categories refer to subject numbers.)
CS+ Minus CS- Rating (Relative Awareness)
+6 +5 +4 +3 +2 +1 0 -1 -2 -3 -4 -5 -6 -7
(3)
(1 0 ) (2 ) (23)
(7)
(13B)
(18)
(31)
(9)
(36)
75
relative RL are related to GSR performance. That is, to
test the experimental subhypotheses it is necessary to have
Ss who possess each kind of RL and Ss who do not possess
them. Thus, the experimental subhypotheses can be tested
only with the Nl and N8 groups. Even with these groups,
only trends, not exact probabilities, can be demonstrated
because of the small size of the subgroups.
Absolute RL. Ideally, the method of determining
the relation between absolute RL and GSR magnitude to CS+
(subhypothesis number 1, p. 30) should consist of holding
relative RL constant while examining various levels of abso
lute RL. For example, Table 15 shows that £s 5, 15, 24,
and 2 8 have similar relative RL scores while the former two
Ss failed to express absolute RL and the latter two Ss did
express absolute RL. Similarly, Table 16 shows that Ss 3,
10, 31, and 36 have identical relative RL scores and yet
the former two Ss failed to achieve absolute RL while the
latter two Ss did achieve absolute RL.
In practice, the method of holding relative RL
constant while comparing Ss with and without absolute RL is
not workable because the subgroup samples are too small
(N=2) from which to discern even trends. In light of this
76
fact, the method chosen consisted of comparing the GSRs of
Ss who exhibited absolute RL with those of Ss who did not
exhibit absolute RL regardless of their relative RL scores.
This method was chosen with the realization that if differ
ences in GSR magnitudes were found then it would be impos
sible to separate the influences of the two kinds of RL.
In the Nl group (Table 15), two Ss (5 and 15)
failed to express absolute RL while the remaining eight Ss
(17, 8 , 1, 24, 12, 28, 6 , and 34) did not possess absolute
RL. Figure 8 presents the CS response magnitude data to
CS+ during acquisition for these two subgroups. There does
appear to be a trend for Ss who exhibited absolute RL to
respond more to CS+, at least in the latter part of acqui
sition. As for the N8 group (Table 16), five Ss (3, 10, 7,
2, and 2 3) failed to achieve absolute RL while the remain
ing five (9, 13, 18, 31, and 36) did achieve it. The mean
magnitude of the three response measures to CS+ for these
two subgroups are shown in Figures 9, 10, and 11. There is
no apparent trend for Ss who expressed absolute RL to
respond more to CS+. In fact, a slight trend in the oppo
site direction can be seen at the end of acquisition for
the three response measures.
Mean GSR Magnitude (/AC)
77
1.0 —
-o
With absolute RL
o -o Without absolute RL
1 2 3 4 5
Test Trials
Fig. 8 .— Mean CS response magnitude to CS+ for
Ss with absolute RL and Ss without absolute RL in the
Nl group.
Mean GSR Magnitude (/AC)
78
1 . With absolute RL
O" ~o Without absolute RL
cx
Test Trials
Fig. 9.— Mean CS response magnitude to CS+ for
Ss with absolute RL and Ss without absolute RL in the
N8 group.
Mean GSR Magnitude (/AC)
79
With absolute RL
o -o Without absolute RL
9
3
0
Test Trials
Fig. 10.— Mean pre-UCS response magnitude to CS+
for Ss with absolute RL and Ss without absolute RL in
the N8 group.
Mean GSR Magnitude (/AC)
80
1.
absolute RL
o ~ ~ o Without
absolute RL
1 .
1 .
1 .
1 .
1 3 4 2 5
Test Trials
Fig. 11.--Mean post-UCS response magnitude to CS+
for Ss with absolute RL and Ss without absolute RL in
the N8 group.
81
Since a trend toward a relation between absolute RL
and GSR magnitude during acquisition was not found a new
approach was initiated. The new approach was to determine
whether there is a relation between absolute RL and GSR
magnitude during the post-conditioning interview. It
should be remembered that the written portion of the ques
tionnaire consisted of presenting the familiarization tri
als while recording the GSR and having S estimate whether
each stimulus was followed by shock. If verbalization of
absolute RL is an important factor in controlling concur
rent GSR activity then absolute RL should be related to GSR
magnitude during the questionnaire.
Table 17 presents the mean CS response magnitude
data to CS+ during the questionnaire. The differences are
in the predicted direction, that is, Ss who expressed
absolute RL did give larger GSRs to CS+ than Ss who did not
express absolute RL. However, the sample sizes are too
small from which to draw statistical conclusions.
Relative RL. In the Nl group (Table 15), five Ss
exhibited relative RL (17, 8 , 5, 1, and 24) while the re
maining five Ss (15, 12, 28, 6 , and 34) did not. The dis
crimination scores of these subgroups are shown in
82
TABLE 17
CS RESPONSE MAGNITUDE (/AC) TO CS+
DURING QUESTIONNAIRE
Nl N8
With Without With Without
Absolute RL Absolute RL Absolute RL Absolute RL
(N=8 ) (N=2) (N=5) (N=5)
.905 . 0 0 0 .946 .837
Figure 12. No trend toward greater discrimination for Ss
who achieved relative RL was found. As for the N8 group
(Table 16), three Ss (9, 13, and 18) expressed relative RL
and the remaining seven Ss (3, 10, 31, 36, 7, 2, and 23)
did not. The discrimination scores of the three response
measures for these two subgroups are shown in Figures 13,
14 and 15. Again, there does not appear to be any consist
ent trend toward greater discrimination for Ss who ex
pressed relative RL.
Because a relation between relative RL and acquisi
tion discrimination was not found, it was decided to
examine the relation between relative RL and questionnaire
discrimination. Table 18 presents the discrimination
scores of Ss who expressed relative RL and Ss who did not
for the Nl and N8 groups. The N8 discrimination is in the
GSR Discrimination (CS+ Minus CS-
83
0 ----0 With relative RL
o - o Without relative RL
+ .7
+ .5
+ .4
+. 3
+ .2
+ .1
- 1.0
Test Trials
Fig. 12.— Discrimination scores of the CS re
sponse for Ss with relative RL and Ss without relative
RL in the Nl group.
GSR Discrimination (CS+ Minus CS-
84
+ • 6
9 W With relative RL
O- O Without relativ^RL
+ .5
+ .2
+ .1
4 1 2 3 5
Test Trials
Fig. 13.— Discrimination scores of the CS
response for Ss with absolute RL and Ss without relative
RL in the N8 group.
GSR Discrimination (CS+ Minus CS-
85
^ 9 With relative RL
o- - -o Without relative RL
+ .7
+ .5
+. 3
+ • 2
+. 1
Test Trials
Fig. 14.— Discrimination scores of the pre-UCS
response for Ss with relative RL and Ss without relative
RL in the N8 group.
GSR Discrimination (CS+ Minus CS-
86
+ .7
+ .6
+ .5
+ .4
+. 3
+. 2
+ .1
0
-.1
-.2
-.3
-.4
-.5
"•6
-.7
-.8
-.9
■1.0
0 ----- 0 With relative RL
o - -o Without relative RL
1 2 3 4 5
Test Trials
Pig. 15.— Discrimination scores of the post-UCS
response for Ss with relative RL and Ss without relative
RL in the N8 group.
87
predicted direction while the Nl discrimination is the
opposite direction.
TABLE 18
CS RESPONSE DISCRIMINATION (/AC) SCORES
DURING QUESTIONNAIRE
Nl N8
With
Relative RL
(N=5)
Without
Relative RL
(N=5)
With
Relative RL
(N=3)
Without
Relative RL
(N=7)
.005 .287 .455 .322
In summary, the absolute and relative RL scores
obtained from the post-conditioning interview failed to
reveal any consistent trends of a relation with GSR per
formance. This lack of a relation was found whether the
GSR was recorded during acquisition or during the question
naire.
CHAPTER VII
DISCUSSION
The primary purpose of the present investigation
was to determine whether human GSR CR acquisition can occur
in the absence of RL. The principal finding was that the
noninformed groups, who received all of the generally
accepted requirements for classical conditioning, did not
acquire a CR while the informed groups, who received the
identical paradigm except that they were informed of the
CS-UCS relation, did acquire a CR. This finding suggests
that the generally accepted requirements for classical con
ditioning are not sufficient, by themselves, to produce CR
acquisition. It further suggests that RL is a necessary
requirement for human CR acquisition. In other words, the
principal finding of the present investigation suggests that
human GSR CR acquisition cannot occur in the absence of RL.
The lack of CR acquisition in the absence of RL is
in general opposition to the findings of Diven (19 37),
Golin (1961) , Lacey and Smith (1954) , and Uno (1966) who
reported successful unaware human autonomic conditioning.
88
89
However, as pointed out previously, the conclusion of the
above studies is suspect because relative RL was not mea
sured. That is, the above studies measured only whether
could name the reinforced CS (absolute RL) and not whether
S had differential expectancy of shock for CS+ and CS-
(relative RL). Therefore, it is quite possible that the
"unaware" Ss of the above studies, in reality, possessed
relative RL. A similar criticism of these studies was made
by Chatterjee and Eriksen (1960) but, unfortunately,
Chatterjee and Eriksen failed to adequately remedy the
defect in their own study.
It is possible, of course, that CR acquisition can
occur in the absence of RL with the designs used by Diven,
Golin, Lacey and Smith, and Uno and not occur with the
present design. Indeed, future research may demonstrate
unaware conditioning in a variety of paradigms which differ
from the present one (e.g., with shorter ISIs, longer peri
ods of acquisition, delay rather than trace conditioning,
or with different masking tasks). The point remains, how
ever, that before one can conclude that unaware condition
ing can occur the above criticism must be remedied, that
is, relative RL must be measured. Until the criticism is
remedied, it must be concluded that the evidence to date
90
supports the view that RL is a necessary condition for
human GSR CR acquisition.
The finding of CR acquisition among Ss who were
instructed of the CS-UCS relation generally confirms the
findings of several previous reports (Bridger and Mandel,
1964; Chatterjee and Eriksen, 1962; Cook and Harris, 1937;
Dawson and Grings, in press; Fuhrer, 1964). There is, how
ever, a unique importance of this finding in the present
context. This finding demonstrates that there was nothing
about the perception task per se which inhibited discrimi
nation performance with the GSR. In other words, this
finding clearly eliminates the objection that the nonin
formed groups did not discriminate because of some response-
interference property of the perception task. It does not
eliminate, however, the objection that the perception task
inhibited CR acquisition for some reason other than its
masking properties. For example, it may be argued, and
with some historical precedent (Pavlov, 1927, p. 28), that
the perception task suppressed discrimination because CR
acquisition can not be established with an activated organ
ism. This possibility may be tested in one of two ways:
(1) use a method of eliminating RL which does not cause
activation (e.g., hypnosis or drugs) and (2 ) use a method
91
which activates the organism but does not eliminate RL.
It should be pointed out that the instructions of
the CS-UCS relation failed to produce pre-UCS response
discrimination. This finding is contrary to the report of
Fuhrer (19 64). However, Fuhrer used a frequency measure
while the present study employed a magnitude measure of the
GSR. In light of this dependent variable difference, the
frequency data of the pre-UCS response in the present study
were analyzed with a sign-test (Siegel, 1956, pp. 68-75).
As with magnitude, the frequency data were in the direction
of conditioning but were not significant (N=7, x=6 , p< .12).
There are several other procedural differences between the
present investigation and Fuhrer's study which may account
for the different pre-UCS response data— e.g., Ss in
Fuhrer's experiment did not have a masking task and were
merely told to pay attention to the CSs and UCS.
In addition to the principal finding, three auxili
ary observations were made. The first such observation was
that ITI reported expectancy of shock accompanied CS re
sponse but not pre-UCS nor post-UCS response discrimination.
The only other studies which concurrently measured autonom
ic conditioning and ITI reports were done by Fuhrer and
Baer (1965 and 19 6 6 ) . However, unlike the present study,
92
Fuhrer and Baer did not perform a trial by trial analysis
of ITI reports and GSR performance and therefore their data
are not comparable to the present data.
The critical aspect of the first auxiliary finding
is that RL defined by the ITI reports correlated with only
CS response discrimination while the RL defined by instruc
tions accompanied both CS response and post-UCS response
discrimination. This difference suggests that the two
definitions have quantitatively or qualitatively different
processes as referents. There are several possible reasons
why the two definitions may have different referent pro
cesses. One possible reason is that the source of the RL
is an authority figure (E) with the instruction definition
while the source is S_-himself with the ITI definition.
McAllister and McAllister have also noted, "knowledge about
conditioning obtained from different sources may affect
conditioning differently" (1958, p. 582). Another differ
ence which should not be ignored is that a true CS-UCS
relation is known with the instruction definition while
only an amorphous feeling that shock is somehow "due" is
present with the ITI definition. However, all that may be
concluded from the present data is that RL defined by
instructions differs in some unknown way from RL defined by
93
ITI reports and that this difference is directly related to
post-UCS response discrimination.
Another implication of the first auxiliary observa
tion is that CS response discrimination can occur independ
ently of post-UCS response discrimination. This finding
adds to the already mounting evidence which suggests that
the GSR prior to the UCS point obeys different laws than
the GSR after the UCS point (on test trials where the UCS
is omitted) (Lockhart, 19 6 6b).
The second auxiliary observation was that absolute
RL, as measured by the post-conditioning interview, did not
accompany greater responding to CS+. This finding is in
general agreement with Golin (1961) who reported that Ss
who selected category 4 (p. 13 of this paper) did not
respond more to CS+ than Ss who selected categories 1, 2,
or 3.
The second auxiliary observation indicates a lack
of correspondence between verbalization and autonomic
response and hence would seem to pose a major problem for
cognitive-association interpretations of human classical
conditioning. However, it does not pose a problem if three
simple assumptions are made: (1) RL differs in degree,
(2) there exists a threshold above which RL will initiate
94
autonomic responses and below which it will not, and (3) RL
at the recall level is above that threshold while RL at the
recognition level is below the threshold. The reason that
these three assumptions would predict the second auxiliary
observation is that absolute RL was measured at the recog
nition level. That is, absolute RL was defined by having S
select a statement from a multiple-choice situation (recog
nition level) and not by answering an open-ended question
(recall level).
The three assumptions suggested above receive
modest support from Dawson and Grings (in press) who found
that RL at the recognition level did not tend to accompany
GSR conditioning but that RL at the recall level did tend
to accompany conditioning. Unfortunately, there were no
noninformed Ss in the present study who expressed RL at the
recall level. Therefore, it is not possible to directly
test the three assumptions with the present set of data.
The third auxiliary observation was that relative
RL, as measured by the post-conditioning interview, did not
accompany greater discrimination performance. It is pos
sible to interpret this finding in a similar way as the
second observation. That is, relative RL did not accompany
greater GSR discrimination because it was measured at the
95
recognition level which is below the threshold of initiat
ing autonomic responses. There were no noninformed Ss who
expressed relative RL at the recall level, therefore, it is
not possible to determine whether RL at the recall level is
above the assumed threshold.
The three assumptions employed to explain the
second and third auxiliary observations receive support
from an apparent discrepancy between the present results
and those of Branca (1957), Chatterjee and Eriksen (1960)
and Fuhrer and Baer (196 5). The latter three studies
reported that RL expressed during a post-conditioning
interview did accompany greater GSR discrimination which
would seem to contradict the present results. However,
Branca, Chatterjee and Eriksen, and Fuhrer and Baer mea
sured RL at the recall level while the present investiga
tion measured it at the recognition level. Therefore, this
apparent discrepancy is quite consistent with the three
assumptions suggested above.
It should be pointed out that, although the three
assumptions suggested above can account for the present
data and do receive some support from previous findings,
they remain ad hoc assumptions. In fact, other forms of
these assumptions, e.g., that the recall and recognition
96
levels differ qualitatively rather than just quantitatively,
can also account for the present results and receive sup
port from the previous studies. Separate experiments
designed to directly test these assumptions are needed.
In order to summarize, the two experimental hypoth
eses and the two experimental subhypotheses will now be
briefly reviewed.
1. It was hypothesized that a paradigm which con
tained all of the generally accepted requirements for
classical conditioning would establish human GSR CR acqui
sition only if RL were present. This hypothesis was con
firmed with the CS response of the 1 sec. ISI groups and
with both the CS response and post-UCS response of the 8
sec. ISI groups. It was concluded that, within the limit
ing conditions of the present investigation, human GSR CR
acquisition cannot occur in the absence of RL.
2. It was hypothesized that CR acquisition would
occur for the noninformed groups only after RL was expressed
during the ITI reports. This hypothesis was not testable
because the noninformed groups never expressed RL during
the ITI reports. A revised form of the hypothesis involved
the prediction that responses would be larger on trials
where ITI reported expectancy of shock occurred than on
97
adjacent trials without the reported expectancy. This hy
pothesis tended to be confirmed with the CS response of the
1 sec. ISI group (p< .10) and the 8 sec. ISI group (p< .05).
However, it was not confirmed with the pre-UCS nor the
post-UCS responses. It was concluded that RL defined by
instructions differs, quantitatively or qualitatively, from
RL defined by ITI reports and that this difference is
directly related to post-UCS response discrimination.
3. It was hypothesized that absolute RL would
accompany greater responding to CS+. This hypothesis was
not confirmed. In order to account for this finding, it
was suggested that absolute RL at the recognition level is
below the threshold of initiating autonomic responses
while absolute RL at the recall level is above the thresh
old. This suggestion received modest support from the
findings of Dawson and Grings (in press).
4. It was hypothesized that relative RL would
accompany greater discrimination. This hypothesis was not
supported. It was suggested that this finding may be
interpreted in a similar way as the last-noted observation.
That is, relative RL at the recognition level is below the
threshold of initiating CR-like GSRs while relative RL at
the recall level is above the threshold. This suggestion
was supported by differences in the results of previous
studies which measured RL at the recall level and results
of the present investigation which measured RL at the
recognition level.
CHAPTER VIII
SUMMARY
Classical conditioning is generally defined as
response modification attributable solely to the repetitive
temporal pairing of a CS with a UCS. However, this repeti
tive temporal pairing may also produce— at least with
normal adult human subjects— the ability to verbalize the
CS-UCS relation (called relational learning or RL). The
primary purpose of the present investigation was to deter
mine whether human GSR classical conditioning can occur in
the absence of RL.
All subjects, college students, were told that the
purpose of the experiment was to correlate "physiological
activity" with perception. It was explained that the
physiological activity would be produced by a periodic
shock and would be recorded by the GSR. Thirty trials of a
perception task were then presented with each trial com
posed of five tones. Subjects were instructed that the
task was twofold: (1) to identify which of the last four
tones sounded most like the first tone, and (2) to identify
99
100
whether the last tone was the highest or the lowest of all
the tones.
Actually, the perception task was employed merely
to mask the true CS-UCS relation. Shock, the UCS, followed
the highest (or lowest) tone 67% of the time and never
followed the lowest (or highest) tone. In effect then, all
subjects were administered a discrimination classical con
ditioning paradigm embedded in a masking task.
Two methods of determining whether the subject
became aware of the CS-UCS relation were employed: (1) in
tertrial interval (ITI) reports— subjects were asked to
report their "observations, thoughts, and feelings" during
the ITI, and (2) post-conditioning interview— subjects
selected statements which indicated whether they thought
the reinforced CS was followed by shock (absolute RL) and
whether they thought it was followed by shock more often
than the nonreinforced CS (relative RL).
Two independent variables were employed: (1) in
structions— half of the subjects were informed of the CS-
UCS relation (informed group) while the other half were not
informed (noninformed group) and (2) interstimulus interval
(ISI)— half of the informed and the noninformed groups were
administered a 1 sec. ISI while the other halfs were
101
administered an 8 sec. ISI. The two variables were ortho
gonal, thus establishing a 2 by 2 factorial design with 10
subjects in each of the four experimental groups.
The GSR records of the 8 sec. ISI groups were
divided into three time periods: first, from 1.0 to 3.9
sec. after CS onset; second, from 4.0 to 8.9 sec. after CS
onset; and third, from 9.0 to 12.9 sec. after CS onset (on
test trials where the UCS is omitted). Only the first time
period was employed with the 1 sec. ISI groups.
The principal finding of the present investigation
was that the informed groups demonstrated GSR discrimination
during the first and third time periods while the nonin
formed groups did not. This finding suggests that CS-UCS
pairings, by themselves, are not sufficient to establish
human GSR conditioning and also that RL is a necessary pre
requisite for such conditioning. That is, within the
limiting conditions of the present design, it was concluded
that human GSR classical conditioning cannot occur in the
absence of RL.
In addition to the principal finding, three auxil
iary observations were made. The first such observation
was that ITI reports of expectancy of shock accompanied
first period but not second or third period discrimination.
102
This finding suggests that first period responses are rela
tively independent of second and third period responses.
It further suggests that RL defined by instructions differs
from RL defined by ITI reports.
The second and third auxiliary observations were
that absolute and relative RL did not influence GSR per
formance. Based on these observations and those of previ
ous studies, it was suggested that RL at the recall level
is above threshold to initiate autonomic responses while RL
at the recognition level is below threshold.
REFERENCES
103
REFERENCES
Adams, J. K. Laboratory studies of behavior without aware
ness. Psychological Review, 1957, 5_4, 383-405.
Barratt, P. E. H., and Beh, Helen C. Subliminal discrimi
nation and the concept of vigilance. Australian
Journal of Psychology, 1964, IL6, 107-119.
Bierbaum, W. B. The temporal gradient in GSR conditioning.
Journal of General Psychology, 1958, 59_, 97-103.
Block, J. D., and Reiser, M. F. Discrimination and recog
nition of weak stimuli I. Psychological and physiolog
ical relationships. Archives of General Psychiatry,
1962, 6, 25-36.
Branca, A. A. Semantic generalization at the level of the
conditioning experiment. American Journal of Psych
ology, 1957, 70, 541-549.
Bridger, W. H., and Mandel, I. J. A comparison of GSR fear
responses produced by threat and electric shock.
Journal of Psychiatric Research, 1964, 2_, 31-40.
Bridger, W. H., and Mandel, I. J. Abolition of the PRE by
instructions in GSR conditioning. Journal of Experi
mental Psychology, 1965, 6J3, 476-482.
Bykov, K. M. The cerebral cortex and the internal organs.
Moscow: Foreign Language Publishing House, 1959;
Medgiz, 1954.
Chatterjee, B. B., and Eriksen, C. W. Conditioning and
generalization of GSR as a function of awareness.
Journal of Abnormal and Social Psychology, 1960, 60,
396-403.
104
105
Chatterjee, B. B., and Eriksen, C. W. Cognitive factors
in heart rate conditioning. Journal of Experimental
Psychology, 1962, 64, 272-279.
Cook, S. W., and Harris, R. E. The verbal conditioning of
the galvanic skin response. Journal of Experimental
Psychology, 1937, 21, 202-210.
Dawson, M. E., and Grings, W. W. Comparison of classical
conditioning and relational learning. Journal of
Experimental Psychology, In press.
DeNike, L. D. The temporal relationship between awareness
and performance in verbal conditioning. Journal of
Experimental Psychology, 1964, 6_8, 521 - 529.
Diven, K. Certain determiners in the conditioning of
anxiety reactions. Journal of Psychology, 1937, 3_,
291-308.
Egger, D., and Miller, N. Secondary reinforcement in rats
as a function of information value and reliability of
the stimulus. Journal of Experimental Psychology,
1962, 64, 97-104.
Egger, D., and Miller, N. When is a reward reinforcing?:
An experimental study of the information hypothesis.
Journal of Comparative and Physiological Psychology,
1963, 56, 132-137.
Eriksen, C. W. Discrimination and learning without aware
ness. Psychological Review, 1960, 67_, 279-300.
Fishbein, H. D., and Engel, S. Human eyelid conditioning
at detection threshold. Psychonomic Science, 1966, £,
291-292.
Fitzwater, M. E., and Thrush, R. S. Acquisition of a con
ditioned response as a function of forward temporal
contiguity. Journal of Experimental Psychology, 1956,
51, 59-61.
106
Fuhrer, M. J. Differential verbal conditioning of heart
rate with minimization of changes in respiratory rate.
Journal of Comparative and Physiological Psychology,
1964, 58, 283-289.
Fuhrer, M. J., and Baer, P. E. Differential classical con
ditioning: Verbalization of stimulus contingencies.
Science, 1965, 150 (3702) , 1479-1481.
Fuhrer, M. J., and Baer, P. E. Effects of intertrial
interval reports on cognitive and GSR differentiation
of conditional stimuli. Proceedings of the 74th Annual
Convention of the American Psychological Association,
1966, 53-54.
Goldstein, H. The effects of set factors and reinforcement
schedules on the acquisition and extinction of the con
ditioned eyelid response. Unpublished Ph.D. disserta
tion. State University of Iowa, 1962.
Golin, S. Incubation effect: Role of awareness in an
immediate versus delayed test of conditioned emotional
ity. Journal of Abnormal and Social Psychology, 1961,
63, 534-539.
Grant, D. A. Classical and operant conditioning. In A. W.
Melton (Ed.) Categories of human learning. New York:
Academic Press, 1964.
Grings, W. W. Classical conditioning. In M. Marx (Ed.)
Theories in contemporary psychology. New York: Mac
millan, 1963.
Grings, W. W. Verbal-perceptual factors in the condition
ing of autonomic responses. In W. F. Prokasy (Ed.)
Classical conditioning. New York: Appleton-Century-
Crofts, 1965.
Grings, W. W., and Dawson, M. E. Multiple response compar
isons in classical autonomic discrimination condition
ing. Paper presented at The Psychonomic Society,
Chicago, October, 1965.
107
Grings, W. W. , Lockhart, R. A., and Dameron, L. E. Condi
tioning autonomic responses of mentally subnormal
individuals Psychological Monographs, 1962, 7^6, No. 39
(Whole No. 558) .
Grings, W. W., Lowell, E. L., and Honnard, R. R. GSR con
ditioning with pre-school age deaf children. Journal
of Comparative and Physiological Psychology, 1961, 54_,
143-148.
Hilgard, E. R., and Marquis, D. G. Conditioning and learn
ing . New York: Appleton-Century-Crofts, 1940.
Homzie, M. J., and Weiss, G. UCS intensity and the associ
ative strength of the eyelid CR with a masked condi
tioning procedure. Journal of Experimental Psychology,
1965, 69, 101-103.
Kimble, G. A. Hilgard and Marquis1 conditioning and
learning. New York: Appleton-Century-Crofts, 1961.
Kimble, G. A. Classical conditioning and the problem of
awareness. In C. W. Eriksen (Ed.), Behavior and aware
ness . Durham, N. C.: Duke University Press, 1962.
Lacey, J. J., and Smith, R. L. Conditioning and general
ization of unconscious anxiety. Science, 1954, 150,
120, 1045-1052.
Lockhart, R. A. Dominance and contiguity as interactive
determinants of autonomic conditioning. Unpublished
doctoral dissertation, University of Southern Cali
fornia, 1966a.
Lockhart, R. A. Comments regarding multiple response
phenomena in long interstimulus interval conditioning.
Psychophysiology, 1966b, 3_, 108-114.
Lockhart, R. A., and Grings, W. W. Inter-stimulus interval
effects in GSR discrimination conditioning. Journal of
Experimental Psychology, 1964, 67_, 209-214.
McAdam, D., Knott, J. R., and Chivorini, J. Classical
conditioning in the cat as a function of the CS-UCS
interval. Psychonomic Science, 1965, 3^, 89-90.
108
McAllister, W. R. Eyelid conditioning as a function of the
CS-UCS interval. Journal of Experimental Psychology,
1953a, 45, 417-422.
McAllister, W. R. The effect on eyelid conditioning of
shifting the CS-UCS interval. Journal of Experimental
Psychology, 1953b, A5_, 423-428.
McAllister, W. R., and McAllister, D. E. Effect of
knowledge of conditioning upon eyelid conditioning.
Journal of Experimental Psychology, 1958, E>5, 579-583.
Mandel, I. J. Two levels of GSR conditioning: Interstimu-
lus interval effects. Unpublished doctoral disserta
tion, Yeshiva University, 19 65.
Moeller, G. The CS-UCS interval in GSR conditioning.
Journal of Experimental Psychology, 1954, 4J3, 162-166.
Pavlov, I. P. Conditioned reflexes (Trans, by G. V.
Anrep). New York: Oxford University Press, 1927.
Prokasy, W. F., and Ebel, H. C. Three distinct components
of the classically conditioned GSR. Journal of Experi
mental Psychology, 1967, 7_3, 247-256.
Razran, G. H. S. Conditioning and perception. Psycholog
ical Review, 1955a, 6_2, 83-95.
Razran, G. H. S. A direct laboratory comparison of
Pavlovian conditioning and traditional associative
learning. Journal of Abnormal and Social Psychology,
1955b, 51, 649-652.
Razran, G. H. S. The observable unconscious and the
inferable conscious in current Soviet psychophysiology:
Interoceptive conditioning, semantic conditioning, and
the orienting reflex. Psychological Review, 1961, 68,
81-147.
Schneiderman, N. Interstimulus interval function of the
nictitating membrane response of the rabbit under delay
versus trace conditioning. Journal of Comparative and
Physiological Psychology, 1966, 6_2, 397-402.
109
Siegel, S. Nonparametric statistics for the behavioral
sciences. New York: McGraw Hill, 1956.
Silverman, R. E. Eliminating a conditioned GSR by the
reduction of experimental anxiety. Journal of Experi
mental Psychology, 1960, 5£, 122-125.
Spence, K. W. Cognitive and drive factors in the extinc
tion of the conditioned eye blink in human subjects.
Psychological Review, 1966, 7j3 445-458.
Spence, K. W., and Spence, Janet T. Sex and anxiety differ
ences in eyelid conditioning. Psychological Bulletin,
1966, 65, 137-42.
Tolman, E. C. Principles of purposive behavior. In S.
Koch (Ed.) Psychology: A study of a science. Vol. 2,
New York: McGraw Hill Co., 1959.
Underwood, B. J. Psychological research. New York:
Appleton-Century-Crofts, Inc., 1957.
Uno, T. Inhibitory-excitatory interaction of conditioned
autonomic responses through interoceptive and extero
ceptive conditioning. Unpublished Ph.D. dissertation,
University of Southern California, 1966.
White, R. W. The abnormal personality. New York: Ronald
Press Co., 1964.
White, C. T. and Schlosberg, H. Degree of conditioning of
the GSR as a function of the period of delay. Journal
of Experimental Psychology, 1952, 43_, 357, 362.
Woodworth, R. S. Dynamics of behavior. New York: Holt,
1958.
APPENDIX A
INSTRUCTIONS
1. Application of Electrodes Instructions
2. Shock Work-up Instructions
3. Perceptual Task Instructions I
4. Familiarization Trials Instructions
5. Perceptual Task Instructions II
110
Ill
APPENDIX A
INSTRUCTIONS
1. Application of Electrodes Instructions
We are going to measure the Galvanic Skin Response
or GSR while you perform a perceptual discrimination task.
The perceptual task will involve you listening to a bunch
of tones coming from this speaker (E points to the loud
speaker) but I will go into more detail about that later.
By the way, have you talked to anyone who has been
in this experiment? OK, I will explain what I am doing as
we go along. (While applying the GSR electrodes) First, I
am going to put these electrodes on the finger tips of your
left hand. You won't feel anything from these electrodes,
they just measure the electrical resistance of your skin.
The reason that I am going to measure your GSR
while you perform a perceptual discrimination task is that
there is a theory in perception which states that when you
are physiologically active, what you might call tense, you
can perceive things better. This is the theory I am test
ing. I am going to measure your physiological activity by
the GSR and I am going to measure your perception by having
you listen to and discriminate among tones coming from that
speaker. I am going to see if you discriminate better when
your GSR is more active.
(While applying shock electrodes) I am going to
have to create a little artificial physiological activity
by presenting shock through these electrodes. But you will
set the intensity of the shock so you don't have to worry
about it being too strong.
Now, the first thing I want to do is to measure the
GSR just to be sure that everything is working all right.
So you just sit back and relax now while I go into the
other room to measure the GSR. When I come back we will
set the intensity of the shock.
112
2. Shock Work-up Instructions
OK, now we are ready to set the intensity of the
shock. We will do this by turning on the shock so low you
won't feel it. Then I will gradually increase it and you
say when you first feel it and then you tell me when its
uncomfortable. I will stop when you say its uncomfortable
and that will be the amount we will use later in the exper
iment .
Now, the shock doesn't have to be at a point where
it lifts you out of your chair but it does have to be at a
point where you would rather not have it in order to cause
this physiological activity I talked about before.
3. Perceptual Task Instructions I
Now, I will tell you about the perception task.
You will hear first one tone then four more tones from this
loudspeaker (E points to the loudspeaker) and its your task
to select the one tone from the four that sounded most like
the first one. You will make your selection known by saying
aloud the number of your answer into the microphone on your
left. But you won't say that number until you hear a blast
of noise from that speaker in front of you (E points to
another speaker). So, the sequence goes like this: you
hear one tone and then four more, you have your answer of
which of the four sounded most like the first and wait for
the blast of noise from that speaker and then you say your
answer out loud into that microphone.
Now, what your thinking and feeling while you do
this can influence how you do on the perception task and it
can influence the GSR. So, I want you to give me a running
account of what you are observing, thinking, and feeling on
each trial. You will do this by briefly saying what you
observed, thought, and felt on the last trial after you
have given your answer. For example, you might say that
you were just concentrating on the tones, or that you were
daydreaming, or that you were worried about the shock. You
don't have to be detailed but do try to be honest.
113
4. Familiarization Trials Instructions
We are going to use only five different tones
throughout the whole experiment. Of course, the order of
the tones and the times between them will vary on each
trial. Now I want to present you the five tones so as to
acquaint you with them before you have to discriminate
between them. So just listen to the five tones and see if
you can tell the difference between them.
5. Perceptual Task Instructions II
OK, could you tell the difference between them?
Now, so far we have only talked about one kind of
perception. This is the kind where you tell which of the
last four tones was the same as the first tone. In this
kind of perception all you have to do is decide whether
each tone was the same as or different from the first tone.
There is also another kind of perception. Instead
of just deciding whether a tone is the same as or different
from some other tone this other kind of perception requires
that you perceive the actual quality of that individual
tone, e.g., whether it is high or low. I want to measure
this other kind of perception too. In order to do this I
want you to tell me whether the last of the four tones is
the highest of all the tones or the lowest of all the
tones. And its been arranged so that it is always one of
those two— the highest of all the tones or the lowest of
all the tones.
So, now you have three things to do. First, decide
which of the four tones is most like the first tone,
second to decide whether the last of the four tones is the
highest of all the tones or the lowest of all the tones,
and third, to report what you were observing, thinking, and
feeling since the last time you reported. Now, to be sure
you understand, I want you to tell me what's going to
happen and what your supposed to do right from the very
beginning. (E listens to S review the procedure and makes
any necessary corrections.)
(This is what is told to the noninformed groups)
OK, that's fine. Now, remember you are going to get the
shock once in a while in order to change your activity
level. The rest of the session takes about 35 minutes.
114
(This is what is told to the informed groups) 0Kf
that's fine. Now, the shocks I told you about are not
going to be completely random. They will come only after
the fourth tone in the series of four and then only when
the fourth tone is the highest (lowest) and never after
the fourth tone when it is the lowest (highest). But not
every high tone will be followed by shock, only most of
them. In other words, shock will usually but not always
follow the high tone and will never follow the low tone.
But don't let this shock interfere with your perception
task. Now, will you tell me when you're going to get the
shock? (E listens to S and makes any necessary correc
tions) OK, that's fine. The rest of the session takes
about 35 minutes.
APPENDIX B
TRIAL SEQUENCE
1. Familiarization Trials
2. Acquisition Trials
115
116
APPENDIX B
TRIAL SEQUENCE
1. Familiari
Order No.
1 950 1050
2 1050 800
3 800 1200
4 1200 1000
5 1000 950
ation Trials
Stimuli (cps)
800 1200 1000
1200 1000 950
1000 950 1050
950 1050 800
1050 800 1200
Acquisition Trials
ITI Trial No. Ref. Stim. Choice Stimuli
1 ISI 2 ISI 3 ISI 4 UCS
50 1 950 (5) 1000
(1)
950 (5) 1050 (3) 800 +
40 2 950 (8) 950 (3) 1000
(1)
1050 (5) 1200
-
30 3 1000 (8) 1050 (5) 950 (3) 1000
(1)
1200
-
50 4 1050 (5) 1050 (3) 1000 (5) 950
(1)
800 +
30 5 1000 (8) 950
(1)
1050 (3) 1000 (5) 1200 -test
30 6 950 (8) 1000
(1)
1050 (3) 950 (5) 800 +test
40 7 1000 (5) 1000 (5) 950
(1)
1050 (3) 1200
-
30 8 1050 (5) 1050 (3) 1000
(1)
950 (5) 800 +
40 9 1050 (8) 950 (5) 1050 (3) 1000
(1)
800 +
40 10 1000 (5) 1000 (5) 950 (3) 1050
(1)
1200
-
50 11 950 (5) 1050
(1)
1000 (5) 950 (3) 800 +test
50 12 1050 (5) 950
(1)
1050 (5) 1000 (3) 1200 -test
30 13 1000 (8) 950 (5) 1000
(1)
1050 (3) 800 +
40 14 1050 (5) 1050
(1)
950 (3) 1000 (5) 1200
-
30 15 950 (5) 1000 (3) 1050 (5) 950
(1)
1200
-
50 16 1050 (8) 1000 (3) 950
(1)
1050 (5) 800 +
50 17 1000 (8) 1050 (5) 1000 (3) 950
(1)
800 +
30 18 950 (5) 950 (3) 1050 (5) 1000
(1)
1200
-
40 19 1050 (8) 1000
(1)
950 (3) 1050 (5) 1200 -test
40 20 1000 (8) 950
(1)
1000 (3) 1050 (5) 800 +test
50 21 950 (8) 1050 (3) 950
(1)
1000 (5) 1200
-
40 22 950 (5) 950 (3) 1050 (5) 1000
(1)
800 +
50 23 1050 (8) 1050 (5) 1000
(1)
950 (3) 1200
-
30 24 1050 (8) 1000 (3) 1050
(1)
950 (5) 800 +
117
Acquisition Trials— Continued
ITI Trial No. Ref. Stim. Choice Stimuli
1 ISI 2 ISI 3 ISI 4 UCS
30 25 1000 (5) 1050 (5) 950
(1)
1000 (3) 800 +test
30 26 1050 (5) 1000 (5) 1050
(1)
950 (3) 1200 -test
40 27 1000 (5) 950 (3) 1000 (5) 1050
(1)
800 +
40 28 950 (5) 1000 (5) 1050 (3) 950
(1)
1200
-
50 29 950 (8) 1050
(1)
950 (5) 1000 (3) 1200 -test
50 30 1000 (8) 950
(1)
1000 (5) 1050 (3) 800 +test
The time between adjacent tones is listed in parentheses. This list
is appropriate for Ss reinforced on the low (800 cps) tones. The only change
made for the Ss reinforced on the high (1200 cps) tones was that 800 and 1200
cps tones listed as the fourth choice stimulus were reversed.
APPENDIX C
POST-CONDITIONING INTERVIEW
1. Oral Portion
2. Written Portion
119
120
APPENDIX C
POST-CONDITIONING INTERVIEW
1. Oral Portion
1. The first thing I would like you to do is to review
what you observed, thought, and felt during the experi
ment. Just give me an overview of what happened.
2. How many shocks would you estimate that you got during
the whole experiment?
3. Now, I want you to think of a scale running from 0 to
10. The "0" end means not at all uncomfortable and the
"10" end means extremely uncomfortable. Where on that
scale would you put the shock and if it changed during
the experiment tell me how.
4. Was there any way you could predict ahead of time that
you were going to get the shock? If so, how?
5. (Asked only of the informed groups) Were my instruc
tions about when you were going to get the shock cor
rect?
121
2. Written Portion
Before the written portion of the post-conditioning
interview, Ss were instructed that each of the five tones
used during the experiment would be individually presented.
They were told to select one of the following seven state
ments for each tone and to write the number of their
selection in the appropriate slot on the next page. It was
explained that the GSR was being recorded and that they
should not move after hearing the tone until the speaker
background noise went off. The speaker background noise
went off approximately 5 sec. after the tone onset and then
the Ss were given as much time as necessary to make their
selection.
1. I am certain that this stimulus was never followed by
shock.
2. I don't think this stimulus was ever followed by shock.
3. I don't remember whether this stimulus was ever fol
lowed by shock.
4. I think this stimulus was sometimes followed by shock.
5. I am certain that this stimulus was sometimes followed
by shock.
6. I think that this stimulus was always or almost always
followed by shock.
7. I am certain that this stimulus was always or almost
always followed by shock.
122
Stimulus #1
If you marked either categories #4, #5, #6, #7 then check
one of the following categories (if you marked either cate
gories #1, #2, or #3 then signal to the experimenter to
present the next stimulus).
1. I don't have any idea of the time interval between this
stimulus and the shock.
2. I think the usual time interval between this stimulus
and the shock was (a) 1", (b) 2", (c) 5", (d) 8",
(e) 10", (f) 13".
3. I am certain that the time interval between this stimu
lus and the shock was (a) .5", (b) 2", (c) 5", (d) 8",
(e) 10", (f) 13".
Stimulus #2
If you marked either categories #4, #5, #6, or #7 then
check one of the above categories (if you marked either
categories #1, #2, or #3 then signal to the experimenter to
present the next stimulus).
Stimulus #3
If you marked either categories #4, #5, #6, or #7 then
check one of the above categories (if you marked either
categories #1, #2, or #3 then signal to the experimenter to
present the next stimulus).
Stimulus #4
If you marked either categories #4, #5, #6, or #7 then check
one of the above categories (if you marked either cate
gories #1, #2, or #3 then signal to the experimenter to
present the next stimulus).
Stimulus #5
If you marked either categories #4, #5, #6, or #7 then
check one of the above categories (if you marked either
categories #1, #2, or #3 then signal to the experimenter
to present the next stimulus).

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

The Effect Of Discriminability On The Partial Reinforcement Effect In Human Gsr Conditioning

PDF

The Effects Of Reinforcement Pattern, Intertrial Interval Regularity, And adaptation On The Orienting Response And Classical Gsr Discrimination Conditioning

PDF

Human Operant Eye Blink Conditioning, Awareness, And The Extraversion-Introversion Dimension Of Personality

PDF

Reward Expectancy Strength As Related To The Magnitude Of Frustration In Children

PDF

Inhibitory-excitatory interaction of conditioned autonomic responses through interoceptive and exteroceptive conditioning

PDF

The Effects Of Ether And Electroconvulsive Shock On One Trial Appetitive And Adversive Learning

PDF

Classical Discrimination Conditioning As A Function Of Probability Of Reinforcement

PDF

Autonomic and cognitive indices of semantic conditioning and generalization

PDF

The Modification Of Partial Reinforcement Effect As A Consequence Of Electrical Stimulation Of The Caudate Nucleus In Cats

PDF

Operant Discrimination Of An Interoceptive Stimulus In The Urinary Bladder Of Intact And Dorsal Root Transected Female Rhesus Monkeys

PDF

An Experimental Analysis Of The Relationship Between The Reliability Of Amultiple-Choice Examination And Various Test-Scoring Procedures

PDF

Use Of College Students In A Social Therapy Program With Chronic Schizophrenics To Produce Changes In Mood And Social Responsiveness

PDF

Dominance And Contiguity As Interactive Determinants Of Autonomic Conditioning

PDF

Human Performance As A Function Of The Joint Effects Of Drive And Incentive Motivation

PDF

The Effects Of Transverse Accelerations And Exponential Time-Lag Constants On Compensatory Tracking Performance

PDF

Effects Of Different Treatment Procedures On Reading Ability And Anxiety Level In Children With Learning Difficulties

PDF

An Effect Of Verbal Conditioning On Nonverbal Behavior

PDF

Noxious Auditory Environment And Psychomotor Performance

PDF

Habituation Of Aggression In The Three-Spined Stickleback (Gasterosteus Aculeatus L.)

PDF

The role of cues in the arousal of anxiety

Asset Metadata

Creator Dawson, Michael Edward (author)

Core Title Human Gsr Classical Conditioning And Awareness Of The Cs-Ucs Relation

Degree Doctor of Philosophy

Degree Program Psychology

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag OAI-PMH Harvest,psychology, experimental

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Grings, William W. (committee chair), Slucki, Henry (committee member), Wolf, Richard M. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-138006

Unique identifier UC11360018

Identifier 6713019.pdf (filename),usctheses-c18-138006 (legacy record id)

Legacy Identifier 6713019.pdf

Dmrecord 138006

Document Type Dissertation

Rights Dawson, Michael Edward

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA