University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content HUMAN PERFORMANCE AS A FUNCTION OF THE
JOINT EFFECTS OF DRIVE AND INCENTIVE MOTIVATION
By
Donald N. Buckner
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 1959
UNIVERSITY OF SO U TH ERN CALIFORNIA
GRADUATE SC H O O L
U N IV ER SITY PARK
L O S A N G ELE S 7. C A LIFO R N IA
This dissertation, ‘written by
.............
under the direction of lAs.....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
D a te .....
DISSERTATION COMMITTEE
a Chairman
Acknowledgements
The writer wishes to thank Dr. J. P. Guilford,
chairman of hi s guidance committee, for his continued
help, understanding, and patience throughout the
execution of this research. The writer also appreciates
the help given him by the other committee members,
Drs - W. W. Grings, H. H. Grant, and L. E. Longstreth.
In addition, the writer wishes to thank hr. Albert
Harsfbedian for assisting him in collecting the experi
mental data.
The writer wishes to acknowledge the assistance
given him by Commander R. A. Rinker, Training Officer
of the Fleet Sonar School, San Diego, who provided the
subjects for this experiment and work space for its
execution. H e also wishes to acknowledge the support
given him by the Physiological Psychology Branch of the
Psychological Sciences Division of the Office of Naval
Research.
CHAPTER
I
II
III
Table of Contents
STATEMENT OF THE PROBLEM...............
The Hypothesis ........................
Description of Terms and Variables . .
Spence's Behavior Theory ...........
Incentive Motivation, k ...........
Drive, D ..........................
Reaction Potential, E .............
Assumptions ..........................
BACKGROUND IN THE LITERATURE ...........
Performance as a Function of Drive . .
The Use of the Taylor Scale of
Manifest Anxiety ...............
Miscellaneous Methods of Varying
Drive ............................
Fear as a Variation of Drive . . . .
Performance as a Function of
Incentive Motivation ...............
METHOD .................................
Experimental Design .................
Subjects ...............................
The Subjects' Task .................
Apparatus ............................
The Experimenter's Console .........
PAGE
1
1
5
n r
s
6
10
12
12
l5+
16
21
26
2?
33
*+3
5+3
5+6
5+6
5+6
ili
CHAPTER PAGE
The Response Panel.................... V/
The Stimulus Presentation Box .... 51
Auxiliary Equipment ................. 53
Electric Shocker .................... 53
Electric Timer ...................... 55
Relay Chassis...................... 55
Intercommunication System ......... 56
Ready Switch ...................... 56
Stimuli............................... 56
Installation............... 59
Procedure............................... 59
Training Phase ........................ 61
Conditioning and Test Phases......... 63
IV RESULTS............... 68
Description of the Response
Keasures............................... 68
Raw Scores............................. 68
Information Transmission Scores . . . 69
Reaction Time Scores................. 7*+
Difference Scores.................... 7*+
Adjusted Difference Scores ........... 76
Transmission-Time Scores ............. 79
Tests of Assumptions..................... 80
Tests of the Experimental Hypothesis 86
iv
CHAPTER PAGE
Analysis of Variance--!1 Scores . . . 67
Analysis of Variance--R!' Scores . . 91
Transmission-Time Score Analysis
of Variance ...................... 93
Incidental Analyses............... 96
Correlations Between the T' and
RT1 Scores........................ 9&
Changes in Errors as a Function
of Trials ........................ 97
V CONCLUSIONS AND DISCUSSION................ 99
Conclusions .......................... 99
Discussion............................ 101
Recommendations for Future Research . . 109
VI SUim-LARY ...................... 115
Introduction .......................... 115
me t h o d ................................. 117
Results................................. 119
Conclusions............................ 121
REFERENCES............................... 12^+
APPENDIX A. SUPPLEhXNTAEY DATA......... 132
V
List of fables
TABLE PAGE
I. The Experimental Conditions and the
Humber of Ss Assigned to Each............. *+5
II. Resistance Values in Ohms Used to
Equate the Brightness Levels of the
Three H u e s ............................ 58
III. Phases of the Experiment................... 60
IV. Total Errors by Stimulus and by
Experimental Group on the 63 Trials
Preceding the Test Trials............. 82
V. Humber of Ss in Each Experimental
Group who Made More Incorrect Than
Correct Responses to Each Stimulus
on the 63 Trials Preceding the Test
Trials................................. 83
VI. Errors and Proportions of Errors
Made on the Test Trials............... 85
VII. Mean T1 Scores for the Different
Experimental Conditions ............... 87
VIII. Analysis of Variance, T' Scores ..... 88
IX. Mean RT1 Scores for the Different
Experimental Conditions ............... 91
X. Analysis of Variance, RT1 Scores......... 92
TABLE PAGE
XI. Lean TRT' Scores for the Different
Experimental Conditions ............... 93
XII. Analysis of Variance, TRT* Scores .... 9*+
XIII. Correlations Between I1 and RT' Scores
for the Four Experimental Groups . . . 96
XIV. Absolute Numbers and Proportions (in
parentheses) of Errors by Experiment
al Condition in Blocks of nine Test
Trials................................. 97
vii
List of Figures
FIGURE PAGE
1. Experimenter's console ................. k6
2. Subjects' response panel and stimulus
b o x ................................. *+9
3. Schematic diagram of the apparatus .... 50
A. Diagram of the shocker circuit....... 5^
5. Kean number of errors in blocks of nine
trials............................. .. . 70
6. lie an s of median subject reaction times
in blocks of nine trials.......... 71
7. Distributions of T scores for the last
27 training and the test trials (H - 5-0) 73
8. Distributions of median hi scores on
training and test trials (N = ho) . . . 7^
9. Distribution of test T minus training T
scores plus 100 (K - *+0).......... 75
10. Distribution of training KT scores minus
test RT scores (N = ho) ........ 76
11. Distribution of adjusted T difference
scores (N = ho) ....................... 79
12. Distribution of adjusted median KT
difference scores (N = h o ) ........ 79
13. Distribution of the products of the T5
and KT' scores (N = ho) ............... 80
viii
FIGURE PAGE
lU, Hypothesized relationships between drive,
incentive and reaction potential,
hypothetical units ..................... 112
Chapter I
Statement of the Problem
The Hypothesis
The purpose of the study reported here -was to
investigate the joint effects of generalized drive and
incentive motivation on human performance. The study
represented an effort to investigate an extension of the
Hull-Spence incentive construct, K, to include an incentive
that is specific to human behavior.
Hull (33) made the assumption that generalized
drive, D, and incentive motivation, lv, combine in a multi
plicative manner in their action on H, the learning or
associative construct, and his resulting behavior equation
was E = H x D x K, where E is reaction potential or the
performance factor. At the time Hull postulated these
relationships, there was evidence indicating that D and H,
as well as X and H, combine in a multiplicative manner
(18, U-8, 68, 69). There was no evidence, however, concern
ing the relationship between D and K and their joint
effects on H. As Spence (56:196) has said, "Hull made the
multiplicative assumption he did purely as a working
hypothesis."
Spence (58:197)? on the other hand, made the
assumption that D and K are additive. He made it on the
2
basis of experiments conducted in his laboratory involv
ing, first, the joint variation of the number of trials,
N, and the time of delay of the reinforcer, Tg, and
second, the joint variation of the time of deprivation,
Ta, and Tg. It was also based on an experiment by Ramond
(50) that involved the joint variation of T^ and Tg, in
•which it was found that these two variables did not
interact♦
If Spence assumed that Tg was one of the experi
mental variables that determined K, then Ramond's results
indicated that D and K were additive, at least within the
range of variation of T^ and Tg he employed. If, on the
other hand, Spence assumed that variations in Tg produced
variations only in If the inhibitory construct, the re
sults of the Ramond study had no bearing on the relation
ship between the constructs D and K. To the writer's
knowledge, Spence has never clearly indicated whether or
not variations in Tg affect both K and 1^- or only If At
any rate, Spence (56:198) states "...that there still is a
considerable element of guessing or theorising" in his
formulation.
On the basis of incidental observations of human
behavior, the validity of the assumption of the additivity
of D and K seems questionable. Appropriate incentives
seem to have considerably more effect on behavior under
3
high drive levels than under low ones. For example, food
seems to have greater incentive value to a starving man
than to a moderately hungry one. Sex objects seem to have
less incentive value shortly after copulation than after a
long period of no sexual activity.
There is an additional question concerning both
the Hull and the Spence formulations that may have greater
research significance. In the Principles of Behavior.
Hull introduced the concept of generalized drive strength
and distinguished between two contributions to D, relevant
•
drive strength, D, and irrelevant drive strength, D. The
distinction concerns the relationship between the needs
of the organism and the reinforcing agent. Thirst, for
example, would be a relevant need if water were the rein
forcing agent, but it would be an irrelevant need if food
were the reinforcement.
Generalized drive, according to bo;th Hull and
Spence, multiplies all habit structures and both disting
uish between the actions of relevant and irrelevant drive
on H: the effects of D are less than those of D. How
ever, to the writer's knowledge, neither Hull nor Spence
distinguishes between the combined effects of irrelevant
drive and incentive, on the one hand, and relevant drive
and incentive, on the other.
It is difficult to believe that relevant and
If
irrelevant drives combine with incentives in the same way.
It could very well be that relevant drive and incentive
combine in some mathematically positive manner, that is,
additively, multiplicatively, or some complex monoto-
nically increasing function. Yet, at the same time, ir
relevant drive (and generalized drive, depending upon the
relative contributions of relevant and irrelevant drives)
could combine with an incentive in an entirely different
fashion.
Incidental observation of human behavior seems to
lend support to the need for a distinction between rele
vant and irrelevant drives in combination with incentives.
Many normally positive goal objects seem to diminish in
incentive value when the organism is under high irrelevant
drive conditions. The writer has observed that sex and
food objects seem to lose much of their incentive value
under highly stressful survival conditions.
The experimental operations performed in executing
this study were designed to produce variations in irrele-
vant drive, D, and, thus, in generalized drive, D, as
well. It was not the purpose of the study to test the
drive summation hypothesis or the possibility that a
distinction should be made between the effects on per
formance of irrelevant and relevant drives in combination
with incentives. Nevertheless, it was hoped that the
5
results might lead to a clearer understanding of these
problems.
The purpose of the study was to investigate the
hypothesis that generalized drive, D, and incentive
motivation, K, are additive. It was predicted that the
additive hypothesis is too simple to account for the re
lationship between D and K and that there is an inter
action- between D and K such that when D is large, adding
K will have less effect than when D is small.
Description of Terras and Variables
Snence1s Behavior Theory
Spence's (56) behavioral equation in terms of inter
vening variables is as follows:
R - f(I) = H x (D+R) - It
in which
R = the response
E = the effective excitory strength
H - the learning or associative factor
D = the general drive level
K = the incentive motivational factor
It = the inhibitory factor
6
and where
H = f(N), the number of trials or the training
variable
D » F(T^), the time of deprivation or the intensity
of noxious stimulation
K » f(N, Tg, Lg, Wg) and Tg is the time of delay
of the reinforcer, Lg is the length of the
response chain, and Wg is the amount of the
reinforcer
It = f(N, Tg).
Incentive Motivation. K
Spence (56:135) assumes that "...the basic mechan
ism underlying this incentive motivation factor, K, is the
classically conditioned rg*1 1 In other words, the manipu
lation of the experimental variables affects K by means of
the fractional anticipatory goal response, rg, and the
stimuli produced by that response, sg- K is regarded by
Spence (56:135) "...as representing, quantitatively, the
motivational property of the conditioned rg and sg
mechanism" and "...is defined in terms of the experimental
variables that determine the vigor of the latter." Spence
(56:137) also says that "...different experimental
variations of reinforcing agents will determine K, either
through the habit strength of rg or through the particular
rg, i.e., the particular vigor of rg being conditioned."
7
Thus, a large reward produces a larger K as a result of a
larger rg, and rg is larger either because of its greater
habit strength or the vigor of the rg response.
An effort was made to produce variations in K in
this study by giving one half of the Ss, Navy apprentice
seamen, an opportunity to obtain a three-day liberty de
pending upon their performance and not giving the other
half a similar opportunity. The two levels of K were thus
an opportunity for:
1. A three-day pass, or
2. a continuation of the usual routine,
in this case, performing manual labor, since the Ss were
'waits and holds,1 seamen waiting for their classes to
convene at the U.S. Fleet Sonar School, San Diego.
The instructions providing an opportunity for a
liberty which were designed to produce variations in K.
may not, strictly speaking, be considered in terms of the
Hull-Spence formulations as representing a variation in
incentive motivation, in spite of the fact that liberty
has been used for years by the military services as a
motivational factor. For this reason, this study was con
sidered to be a test of an extension of the Hull-Spence
b ehav i or the or y.
According to the theory, the fractional anticipa
tory goal response, rg, underlying the incentive
8
motivational factor K, develops while the organism con
sumes the incentive object. In the case of a food in
centive, for example, rg occurs during the consumption of
the food and becomes conditioned to stimuli that are
present while the food is being consumed. These fraction
al anticipatory goal responses are then elicited by other
stimuli along the instrumental path according to the
degree of their similarity with the stimuli that had been
conditioned to those responses in the goal box. For an
incentive to act according to the Hull-Spence paradigm,
these fractional anticipatory goal responses that are
developed during the consummatory act must be elicited
during the performance of the instrumental act required to
obtain the incentive object.
For an opportunity for a liberty to act in this
manner, a fractional anticipatory goal, in this case,
liberty response would have to occur during the per
formance of the task instrumental in obtaining the
liberty. For it to occur, the assumption would have to be
made that the Ss had had a liberty prior to the experiment
(this was actually the case) and that during their liberty
fractional anticipatory liberty responses had become
conditioned to stimuli that were also present in the
experimental situation, or that such responses had been
developed and generalized to stimuli that were present in
the experimental situation. It seemed reasonable to
assume, since human subjects were used in the study, that
the fractional anticipatory liberty responses could have
become conditioned as a result of previous liberty ex
periences to the word "liberty1 ' or the phrase used in the
instructions, "opportunity for a liberty." For these
reasons, and because freedom is such an obvious incentive
to military personnel, it was assumed that an opportunity
for a liberty would be a meaningful variation in incentive.
In spite of the fact that the experimental manipu
lation used in this study to produce variations in K has
been and is recognized by laymen and psychologists alike
as a variation of incentive motivation, there is also
reason to believe, from what Spence has said, that it
might have produced a variation in drive. Spence (55j137)
has stated, "The second point is that our theory of the
mechanism underlying D was developed in connection with
experimental situations involving some form of noxious
stimulation. Complex human learning tasks, on the other
hand, typically do not involve the use of a noxious
stimulus. Whatever stress is present in these situations
is usually produced by instructions that aim to create in
the subject the desire or need to make as good a showing
as possible." The instructions to the Ss in this study to
the effect that they would receive a three-day liberty if
10
they performed in the top half of their group could be
taken according to what Spence has said as instructions
that would produce variations in drive rather than in
incentive. If this be the case, then the study was really
concerned with human performance as a function of the
joint effects of two experimental Variables that are
subsumed under the Hull-Spence drive construct.
Drive. D
Spence (56:165) recognizes "...two types of needs
or drive states;
1. appetitional needs and
2. aversive or emotional drive states."
These primary motivational states are operationally de
fined in different ways. The appetitional needs are
specified in terms of deprivation of various kinds of
objects that are necessary to maintain life or the species,
and "...the primary emotional drive states are defined in
terms of the administration of some noxious or aversive
form of environmental stimulation, e.g., electric shock,
air puff, heat..." (56:166).
Spence (56:180) assumes "...that the basic mechan
ism determining the level of D in the case of aversive
forms of stimulation is an internal, emotional state or
response of the organism (re). This emotional response is
assumed to be aroused in different degrees with different
11
intensities of such aversive stimuli as shock."
An effort was made to vary D in this study by
conditioning one half of the Ss to a 1000 cps tone condi
tioned stimulus (OS) through the use of an electric shock
unconditioned stimulus (US) and then presenting the tone
on all test trials for all Ss. If Spence's assumptions
are correct and the conditioning procedure was adequate,
those Ss who received the tone paired with the shock prior
to the test trials responded emotionally (re) to the tone
during the test trials. Those Ss who did not receive
trials with the tone and shock paired gave either no such
emotional response to the tone during the test trials, or
if the tone elicited an unconditioned emotional response,
the responses of the Ss who received the conditioning
trials should have been quantitatively greater than of
those Ss who did not. In any case, if Spence's assump
tions are valid, and the conditioning procedure was
adequate, the drive level, D, of the Ss who received the
shock paired with the tone should have been higher on the
test trials because of a conditioned emotional response
than of those Ss who did not receive such conditioning
trials.
It was assumed, in line with the assumptions made
by both Hull and Spence (56:189), that the conditioned
emotional response contributed to a generalized drive
12
strength, which "...could energize or activate all hahit
structures present hy virtue of existing stimulus cues
regardless of the specific need state under which they
were acquired." The fact that the drive, D, produced by
the conditioning procedure was not relevant to the
incentive did not obviate testing the major hypothesis.
According to both Hull and Spence, both relevant drives,
13, and irrelevant drives, D, contribute to a total
effective drive strength, D, which combines with incentive,
K, in its action upon habit strength, H, to produce
performance, E.
Reaction Potential. E
Both the speed and the accuracy of each response
were recorded in this experiment. The records obtained
made it possible to use both the speed and the probability
of correct response as dependent measures, that is, as
indices of reaction potential, E. They also permitted the
computation of information transmission scores, bits-per-
signal-per second, for each S, which reflected in a single
measure both the latency and probability aspects of each
response.
Assumptions
The experiment involved some of the major experi
mental and intervening variables in the Hull-Spence theory
13
of behavior, and the assumptions made with regard to the
experiment were, in most instances, those of the theory
itself. It was assumed, for example, that noxious stimu
lation increases D which multiplies H and serves to in
crease E.
An assumption of the experiment was that the con
ditioning procedure actually resulted in a conditioned
emotional response (re) to the tone on the test trials.
It was necessary to assume that the shock (US) did not
produce incompatible responses that became conditioned to
the stimuli, se, produced by the conditioned emotional
response, re. It was also assumed that the habit
strength, H, for the correct response was equal and
dominant for all four groups at the beginning of the
experimental test trials. An effort was made in the
analyses to test these assumptions.
Chapter II
Background in the Literature
There is a great deal of* information available in
the psychological literature about the effects on per
formance of the Hull-Spence drive and incentive motivation
constructs. It is unfortunate that, in spite of its
abundance, little of the information is directly relevant
to the study described here. It Is difficult to under
stand, in light of the obvious importance of motivational
factors in human behavior, why more relevant information
is not available, particularly in the work of Spence and
his students in the Iowa laboratory.
It may be that more information directly relevant
to this study is not available because it is difficult to
manipulate drive and incentive motivation experimentally,
using human Ss, and still remain within the boundary con
ditions imposed by the Hull-Spence theoretical framework.
Many of the variations in motivation that are meaningful
to humans in their everyday lives apparently exceed those
boundary conditions. Furthermore, it is difficult to
manipulate motivational variables using human Ss in such a
manner that the conditions will be realistic not only inso
far as the laboratory conditions are concerned but also
insofar as the Ss’ lives outside the laboratory are
15
concerned. It remains to be seen whether the restrictions
imposed by the theory are indicative of limitations in it,
or whether the variations generally accepted by humans in
their everyday lives are inadequately defined.
Much of the research that has been done has involv
ed the use of rat 5s and it is difficult to generalize
from the behavior of rats to the human Ss used in this
study. In those Instances where human 3s have been used,
the experimental paradigm has often involved classical
conditioning which may or may not be directly relevant to
the discrimination performance involved In this study. A
great deal of research has been done using scores on the
Taylor Manifest Anxiety Scale to define different drive
levels, the assumption in these studies being that higher
scores on the Taylor scale are indicative of emotionality
under all situations or that they indicate a lower
threshold of emotionality in the face of stressful
stimulus conditions. In other instances attempts have
been made to vary drive by administering differential pac
ing or failure instructions or by varying the intensity of
noxious stimulation.
It is possible to outline the literature bearing on
the Hull-Spence formulations in a variety of ways. Studies
have been concerned with the effects of variations in drive
on learning, as well as on performance, and with the
16
effects of variations in incentive on learning and per
formance. The studies may he categorised according to
whether they involve classical or instrumental condition
ing or complex learning. They may he classified according
to the experimental variations that were used to produce
variations in the theoretical constructs, although if the
constructs were entirely adequate such a classification
might he unnecessary. Finally, the studies may be
dichotomized according to whether human or sub-human Ss
were used. Regardless of the classification used, the
writer was unable to find any study using human Ss that
had involved the joint variation of drive and incentive
motivation, although one study (Jo) involved the joint
variation of time of deprivation and delay of reward using
rat Ss.
Performance as a Function of Drive
Several studies have been concerned with the
effects on performance and learning of variations in
drive. The question in some instances has been whether or
not drive affects H, the associative factor, or only E,
the performance factor. In other studies, the object has
been to describe the relationship between selected drive
levels, using a variety of experimental variations, and
performance.
A comparison of the data obtained by Williams (66)
and Perin C*+B) , in which the extinction of a bar pressing
response in rats was plotted against the number of rein
forced trials, showed that the number of reinforced trials,
N, combined in some non-additive way with time of depriva
tion, T^, to produce the measure of response strength, in
this case, number of trials to produce extinction. In the
Perin study, the responses were extinguished under three
hours food deprivation and in the Williams study they were
extinguished under 22 hours food deprivation. Had drive
and habit strength been merely additive, the curves would
have remained parallel throughout; instead they diverged
indicating a non-additive relationship between drive, D,
and habit strength, H. Hamond (51), using a double bar
Skinner box, measured the bar pressing performance of two
groups of rats, one under 22 hours of food deprivation and
the other under four hours food deprivation. He found that
the high drive group responded significantly faster than
the low drive group and made superior choice behavior.
The results were interpreted as supporting the multiplica
tive relationship between drive and habit strength. Amsel
(1), on the other hand, using an instrumental escape situa
tion involving rat Ss, found that performance curves for
the high and low drive groups did not diverge, which would
indicate that drive and habit strength were additive rather
than multiplicative. Amsel was concerned, however, with
18
the drive summation hypothesis (i.e., that all of the
drives effective at a given moment summate to produce a
generalized drive strength), and his results with respect
to the relationship between D and H were incidental to the
primary purpose of the study.
In a classical eyelid conditioning study, Passey
(^-7) found that the mean number of conditioned responses
was related to the intensity of the unconditioned
stimulus, US. The curve for the lowest of four intensi
ties of air puff diverged from the other curves as the
number of trials was increased, indicating additional
support for the D x H hypothesis. Similar results have
been obtained using classical eyelid conditioning and
groups of anxious and non-anxious Ss as defined by scores
on the Taylor scale (58, 63).
The results given in an unpublished Master's thesis
(38), mentioned by Spence (58), showed that an instrument
al response to escape an electric shock was faster for the
more intense of two shock intensities. There was an in
creasing divergence of the performance curves with the
number of trials, again indicating a non-additive rela
tionship between D and H. Other investigators (12)
obtained results almost diametrically opposed to these in
that their performance curves, obtained using different
levels of shock as the variation of drive, converged as
19
the number of trials increased, until at the end of the
performance measuring period, all three experimental
groups were performing at approximately the same median
response speed.
Some investigators have been concerned with the
question of whether drive affects the development of
habit strength or only performance. Teel (66) trained
four groups of rats in a single-unit T-maze under four
levels of drive. He then divided each group into four
additional groups to make a total of 16 groups and ex
tinguished them under four levels of drive. The analysis
of variance of the data showed that variations in drive
strength during training were not related to the strength
of the habit developed, supporting Hull’s contention that
drive strength during training is not one of the determi
nants of habit strength. Kendler (37), also using rats
and food and water deprivation as experimental variables
determining drive level, concluded from two experiments
that drive was not a variable in the development of habit
strength.
Several investigators have been concerned with the
so-called drive summation hypothesis, that is, whether
relevant and irrelevant drives combine to produce a
generalized drive strength which is greater than the
magnitude of either. Amsel (1), using rats and the speed
20
of running as the dependent measure, combined an irrele
vant hunger with a primary need to escape an electric
shock and found that response strength was not any greater
than when the shock alone was used. He also combined an
irrelevant hunger drive with a secondary need to escape
anxiety, the anxiety being induced by stimuli that had be
come conditioned to the pain reaction to the shock. In
this instance, he found that the response strength was
greater than when only the drive to escape anxiety was
used. Another study (67) was designed to determine the
effects of an irrelevant thirst drive in activating per
formance that had been learned under a hunger drive. The
results indicated that the irrelevant, drive could be
assumed to contribute to a generalized drive strength and
that it was not as strong as a relevant drive induced by
comparable experimental procedures. Kendler (36), also
using rats and food and water deprivation as the experi
mental variations of drive, obtained results which showed
that the addition of an irrelevant thrist drive to a
relevant hunger drive served to increase resistance to
extinction up to about 12 hours of thirst deprivation.
Beyond that point, additional increases in irrelevant
drive served to decrease resistance to extinction. The
results of Ellis' (22) study were equivocal with respect
to the effects of introducing an irrelevant need into a
21
motivational complex. In his study, the relevant drive
was induced by depriving the animals of food for varying
numbers of hours, and the irrelevant drive by three levels
of electrical stimulation, the author stating that the
response induced by the shock was "...emotionality and
viewed as possessing motivational properties" (22:31 +1).
The Use of the Taylor Scale of Manifest Anxiety
Many studies have been performed in which an effort
has been made to produce variations in drive by selecting
Ss who obtained extreme scores on the Taylor Manifest
Anxiety Scale. In some instances, an effort has been made
to determine the strength of conditioned responses among
"high-anxiety" Ss as opposed to "low-anxiety" Ss as de
fined by scores at the extreme high and low ends of the
scale. Taylor (63) found that high-anxiety Ss showed a
more rapid growth in conditioned eyelid responses than
low-anxiety Ss. She interpreted her results "...to mean
that such sources of drive as those employed in the pre
sent experiment combine in some manner to produce a total
effective drive state, and that this value is a determiner
of the strength of the conditioned response" (63:91). By
the phrase "such sources of drive as those employed in
the present experiment," Taylor was referring specifically
to scores on the Manifest Anxiety Scale.
The assumption underlying the scale is that the
22
drive level of Ss varies directly with differing degrees
of anxiety and that the test measures anxiety. The test
was designed as a convenient means of manipulating drive
strength (6^-). It was not designed as a clinical tool and
no effort was made in its construction, according to the
author, to do anything more than obtain a convenient means
of selecting Ss representing varying drive levels to per
mit study of the effects of drive on behavior.
A number of the studies involving the use of scores
on the Taylor scale as a means of selecting Ss has been
concerned with the performance of so-called high- and
low-anxiety Ss on tasks in which the correct response may
not be the dominant one in the response hierarchy.
According to the Iiull-Spence formulation, when the correct
response is not the dominant response, higher drive should
lead to poorer performance. On the other hand, when the
correct response is the dominant one in the response
hierarchy, high drive should lead to better performance.
Farber and Spence (26) found that high-anxiety Ss did not
perform as well as low-anxiety Ss in a complex learning
situation, a ten choice stylus maze, but that these same
anxious Ss were superior in a conditioning situation. An
analysis of variance showed a significant interaction
between the high- and low-anxiety groups and the experi
mental conditions, that is, the complex versus the simple
conditioning task. They also found that the advantage of
the low-anxiety Ss over the high-anxiety ones in the com
plex learning situation was greater at the more difficult
choice points, there heing a significant correlation
between performance at a particular point and the groups.
An additional test was made to determine whether or not
the high- and low-anxiety groups could have differed in
terms of learning ability, and no differences were found.
Farber and Spence interpreted their results as indicating
that the groups differed in some way other than learning
ability and that the results were not incompatible with
the view that they differed in drive level.
Taylor and Spence (65) investigated the relation
ship between anxiety level, as determined by selecting
extreme groups on the basis of scores on the Taylor scale,
and performance on a serial learning task. Low-anxiety Ss
were superior to high-anxiety ones both in terms of the
number of errors and the number of trials to a criterion.
Others (59) have investigated the hypothesis that high-
anxiety Ss, again defined by responses on the Taylor scale,
would perform better than low-anxiety Ss in learning non
competitive material and that they would perform poorer in
learning material where there were strong competing res
ponse tendencies. Predictions from the Hull-Spence formu
lations were substantiated in that the high-anxiety Ss
2k
performed poorer than the low-anxiety Ss on the paired
associates that involved competition, and they were also
initially superior on the material in which there were
high associative connections.
Montague (*+2) compared the performance of high- and
low-anxiety Ss as defined hy extreme scores on the Taylor
scale on a nonsense syllable verbal learning task. There
were three lists of words: the first list contained items
of high similarity and low association value, the second
contained items of low similarity and low association
value, and the third contained items of low similarity and
high association value. The low drive (low-anxiety) Ss
were superior on the most difficult list in terms of the
number of correct anticipations. This superiority was
marked in the early trials and lessened as learning con
tinued. With the moderately difficult list, the low drive
Ss performed better but to a lesser degree than with the
more difficult list. The high drive (high-anxiety) Ss
performed better than the low drive Ss on the easiest list.
The results were interpreted as supporting the hypothesis
that anxiety has drive properties which combine with habit
strength to increase the differences between weaker and
stronger response tendencies.
Another group of investigators studied the perform
ance of children on a somewhat complex learning task as a
25
function of scores on the Taylor scale. They found, that
the low-anxiety Ss performed better than the high-anxiety
Ss, but they were unable to determine, as a result of the
design of their study, the strengths of the correct and
incorrect response tendencies, so the results did not lend
themselves to an unequivocal interpretation. The same
authors (15) performed another study in which it was
possible to specify on the basis of the responses of a
normative group the relative difficulty of the various
items in the task and found an interaction between anxiety
level and task difficulty. The high-anxiety Ss performed
better on the easy items, and the low-anxiety Ss performed
better on the more difficult ones.
In general, the results of the studies performed
using groups of Ss selected on the basis of scores on the
Taylor scale have supported the hypothesis that drive and
habit strength combine in some non-additive fashion,
although they cannot be said to support a straight multi
plicative assumption. The results tend to indicate a more
complex function is involved. With regard simply to the
direction of the effects as predicted from the Hull-Spence
formulations, the results have been favorable. Many of
the studies have suffered from a failure to control the
possibility of differential learning ability in the high-
and low-anxiety groups. In the one case (26) where an
26
effort was made statistically to control learning ability,
an internal consistency type control was used, and it
cannot be said to have been entirely adequate.
Buchwald and Yamaguchi (11) also performed a study
to test the hypothesis that increasing drive level will
lead to poor performance when the correct response is not
initially the strongest one in the response hierarchy.
They used rats as Ss and defined drive experimentally
through variations in the amount of water deprivation.
The authors described the study as an analogue of the
Taylor-Spence verbal maze learning experiment and the
Farber-Spence stylus maze learning experiment, which were
previously discussed. These investigators found that high
drive level Ss performed better than low drive level Ss
even when the correct response should not have been the
dominant one. They interpreted their results as indicat
ing possibly “...the necessity of re-interpretation of the
theoretical significance of Taylor scale scores" (11:268).
Miscellaneous Methods of Varying Drive
Drive has been varied in other studies by various
types of instructions. In some instances, variations in
drive go under the name of variations in stress. In one
case (1^), stress was varied by emphasizing the need for
speed on the task; in another, it was varied by imposing
pacing and non-pacing conditions (*+5) ; and in another (13),
27
it was varied by instructing the Ss that failure to res
pond within a specified period of time would constitute
failure (13)•
Fear as a Variation of Drive
Farber, in an article in the Nebraska Symposium on
Motivation (2*+), discusses two types of anxiety in rela
tion to two characteristics that, according to him,
determine the drive status of a variable;
1. Its reinforcing properties through its
elimination or reduction and
2. its dynamogenic effect upon response tendencies.
Stimuli that become associated with noxious stimulation
satisfy both of these requirements, according to Farber,
and anxiety, as defined by responses on the Taylor scale,
satisfies at least one of them. In line with this is
Spence's statement that "...the second point is that our
theory of the mechanism underlying D was developed in
connection with experimental situations involving some
form of noxious stimulation" (55*137)*
A great many studies have been performed using fear
or conditioned fear as a variation of drive. The under
lying mechanism of a conditioned fear response has been
described, generally, as follows; Aversive or noxious
stimuli create a hypothetical pain or emotional response
which can become conditioned to previously neutral stimuli.
28
Once conditioned to the fear or emotional response, these
stimuli become capable of eliciting fear responses which
act as a drive. As such, according to the Hull-Spence
formulations, they multiply all habit structures existing
in the organism.
Miller (*+0) concluded from a study of rat behavior
that the acquired drive of fear was sufficiently strong to
initiate new learning and that the reduction of the fear
by escape was reinforcing. He stated that his results
confirmed Mowrer's (^3) hypothesis that fear or anxiety
can play a role in learning comparable to that of primary
motives, such as those resulting from food or water de
privation. Kalish (35) performed a study to determine
whether or not conditioned fear responses showed acquisi
tion and extinction properties similar to other responses.
He used rat Ss in a b x b factorial design with four
levels of numbers of acquisition trials and four levels of
numbers of extinction trials. The dependent measure was
based on hurdle jumping. The results provided support for
the assumption that conditioned fear increases as a
monotonic function of the number of conditioning trials
and weakens with extinction trials. Amsel and Cole (3)
demonstrated that the conditioned fear response shows
generalization similar to other responses depending upon
the degree of similarity between the test and the shock or
29
acquisition situation. Brown and Jacobs (9) concluded
from a study in which rats were conditioned to a buzzer
using an electric shock US that fear reduction serves to
reinforce new responses like other drive reductions and
that fear acts to intensify whatever responses are
dominant at the moment. Amsel (2) trained rats to drink
following a deprivation period and then for a four day
period shocked them while they were drinking. On subse
quent test trials the rats who had been shocked while
drinking showed considerably less consumption than a
comparable group which had not been shocked. Amsel in
terpreted his results as indicating that the experimental
group consumed less water because of competing anxiety
reducing responses that were present during the drinking
situation. Amsel and Maltzman (5) found that if rats were
shocked in a different situation prior to making a con-
summatory drinking response, water consumption increased
significantly. The results in this instance were in
terpreted in terms of a generalized drive strength result
ing from the emotionality induced by the noxious shock
stimulation. Brown, et al, (10), feeling that most of the
experiments done prior to theirs had investigated only the
reinforcing properties of fear reduction, investigated the
drive characteristics of conditioned fear. They concluded
from their study that:
30
1. Pairing a neutral and a noxious stimulus in a
typical conditioning paradigm resulted in the neutral
stimulus becoming capable of eliciting an emotional res
ponse of fear or anxiety.
2. The conditioned fear response possessed
motivating or energizing properties akin to those of
primary drive.
3. The conditioned fear response exhibited an
increase as a function of the number of trials as well as
extinction and spontaneous recovery, similar to other
classical conditioned responses. The dependent measure in
their study was a startle response, and they found that the
conditioned fear response increased the startle reaction.
Using a differential conditioning paradigm in which
the positive stimulus is reinforced and the negative
stimulus is not, Runquist, et al, (52) found that the num
ber of conditioned responses to both the positive and
negative stimuli was greater for Ss receiving a stronger
noxious US and that discrimination between positive and
negative stimuli was greater, although not significantly
so, for the stronger US group. The results were similar to
those of the many studies in which Taylor scale scores
have been used as a variation in drive. The assumption
made in this study was that emotional response evoked by
the US persisted from trial to trial.
31
Spence and Runquist (57) > investigating the temporal
effects of conditioned fear on the conditioned eyelid res
ponse, concluded that "...the assumption that the condi
tioned emotional response which develops to the CS in
classical eyelid conditioning experiments has too long a
latency to influence the drive level at the time of
occurrence of the eyelid CR' 1 (57*616) is supported. They
further conclude that their data support the assumption
that a conditioned emotional response "...can he establish
ed to a neutral stimulus by pairing a noxious US with that
stimulus" (57*616).
It is evident from the foregoing that a variety of
experimental manipulations have been used to produce
variations in drive strength. The most obvious variation
and the one that has been used frequently in animal
studies is the deprivation of food and water. The need to
escape noxious stimulation has also been used frequently
in studies where animal Ss have been used, and noxious
stimulation has been used in some studies involving human
Ss. Instructions to human Ss indicating the threat of
shock have been used, and in other instances, an effort
has been made to develop a conditioned emotional response
to some neutral stimulus through the use of an electric
shock or other noxious stimulation as a US. Finally, the
Taylor Manifest Anxiety Scale was developed to provide a
32
convenient means of obtaining Ss who differed in general
ized drive level.
The literature reported here on performance as a
function of different variations in drive was only par
tially relevant to this study. However, it illustrates
that developing a conditioned emotional response through
the pairing of a neutral stimulus with a noxious electric
shock US has been used by others to produce variations in
drive and is generally accepted by proponents of the Hull-
Spence position as an acceptable means of manipulating
drive. In fact, Spence (57) used a procedure that was
almost identical to the one employed in this study to
produce different levels of drive in Ss participating in
an eyelid conditioning study. With respect to the con
ditioned response that was developed, it was stated in the
description of this study that "...an emotional response
produced by a shock US to the finger was conditioned to a
light CS with a 500 millisecond CS-US interval" (57;6l6).
The literature reported on drive also seemed to be
relevant to the question of whether the variation in in
centive produced in this study was really a variation in K
or in D. Several investigators who are closely related to
the Iowa laboratory have used comparable instructions con
cerning the possibility of failure in the performance
situation as a variation in drive (13, 1*+, 1 +5) • Spence
33
(55:137) also indicates that variations in instructions
can produce variations in drive when he says "...whatever
stress is present in these situations (complex human
learning tasks) is usually produced hy instructions that
aim to create in the subject the desire or need to make
as good a showing as possible."
Performance as a Function of Incentive Motivation
Many of the studies in which experimental varia
tions in the incentive motivational construct, h, have
been used have been concerned with whether or not varia
tions in incentive affect the learning factor H or only
the performance factor E. Cowles and Nissen (17) and
Nissen and Elder (M+) have performed studies using
chimpanzees in which K was varied by varying the magnitude
of reward. Both studies showed that after obtaining a
given level of performance with a particular reward, a
reduction or an increase in the size of the reward pro
duced shifts in performance, either up or down, depending
upon whether the reward represented an increase or a de
crease over the magnitude under which the performance
asymptote was reached. Perin (^9) found that the
asymptotes of performance were different under different
delays of reinforcement after 50 trials. Crespi (18, 19),
in two different studies using rats, found that the speed-
of-running curves diverged with the number of reinforced
31 * -
trials as a function of the magnitude of the reward, in
dicating that the relationship between K and H was a
multiplicative one; he described the relationship between
the level of performance and the amount of incentive as a
negatively accelerated, positive function. Zeaman (69)
found support for the Crespi data on the relationship
between reward and performance level. The general con
clusion from all of these studies was that variations in
incentive produce variations in the performance factor and
are not a determining factor with respect to the develop
ment of habit strength.
Ramond (50) performed a study using rats in which
both K and D were varied experimentally. He was interest
ed in instrumental performance as a joint function of de
lay of reinforcement and the time of food deprivation.
The results showed that speed of running was directly re
lated to both variables. The rats that had been deprived
of food for 22 hours and had a one-second delay of reward
performed the best; those that had been deprived of food
for four hours and had a one-second delay were next best;
those that had been deprived of food for 22 hours and had
a five-second delay were next; and those rats that had
been deprived of food for four hours and had a five-second
delay of reward showed the slowest speed of running. It
is, perhaps, primarily on the basis of this study that
35
Spence hypothesized that the relationship between D and K
is additive, for the results indicated that the effects on
running performance of time of deprivation and time of de
lay of reward were additive.
Spence (56) assumes that the mechanism underlying
the incentive motivational factor K is the classically
conditioned fractional anticipatory goal response, rg . He
regards these responses as motivators, and since they are
classically conditioned, their strength should vary as a
function of the number of conditioning trials in the goal
box, the similarity of the environmental cues along the
instrumental path to those in the goal box, and the vigor
of the unconditioned consummatory responses, since the
vigor of the response may be conditioned per se. Thus the
number of trials and the amount of the reward and other
factors that determine the vigor of the consummatory res
ponse will produce changes in the habit strength of rg and
thus changes in K.
Spence (56) cites evidence that showed that the
speed of running in the runway varies with the consumma
tory time in the goal box and was not a function of the
size of the food reward per se. He attempted to explain
this result by saying that there is more conditioning of
the fractional anticipatory goal response, rg, and thus
greater habit strength of rg the longer the time in the
goal box. Stein (60) has attempted to study variations in
rg, but his results vie re, at best, equivocal.
The vast majority, if not all, of the studies that
Spence cites to support his assumptions regarding the
actions of and the mechanism underlying the incentive
motivational factor were not directly relevant to the
study proposed here. As was previously mentioned, there is
reason to believe that the incentive employed in the pre
sent study could not be assumed to be a proper manipula
tion of the Hull-Spence motivational factor K. What are
normally assumed to be incentives in our society may not
be considered incentives within the Hull-Spence formula
tions. The reason for this could be, first, that what are
normally described as incentives in everyday life are not
sufficiently well defined to be used experimentally within
the Hull-Spence boundary conditions or, second, that the
theoretical formulations themselves may be severly limited
by their inability to deal with such incentives.
At any rate, two studies (6l, 62) not directly con
cerned with testing the Hull-Spence behavior theory used
what are normally accepted as meaningful incentives in our
society as motivational factors. Surwillo (62) investigat
ed the relationship between electromyographic gradients
and the level of motivation, the difficulty of the task,
and the goal structuring of the task. Level of motivation
37
was varied by instructing the Ss, under one set of condi
tions, that the pursuit tracking task they were performing
represented the test of their performance and instructing
them, under the other set of conditions, to perform the
task so that the experimenter could calibrate the equip
ment. The difficulty of the tracking task was varied by
having both a compensatory tracking task (the more diffi
cult one) and a pursuit tracking task. Variations in what
the author called "goal structuring of the task1 ' were
obtained by offering the Ss different monetary rewards
depending upon their performance on the task. The results
showed that the money incentive was the primary factor in
raising the electromyographic gradient.
Stennett (61) was interested in the relationship
between the level of arousal and performance level; in
particular, he tested the hypothesis that the relationship
between the two may be represented by an inverted-U. To
produce different levels of arousal, Stennett employed
three different motivation-incentive conditions. To
achieve the optimum level of arousal, he encouraged the Ss
while they were performing and gave them small financial
rewards for improving their score. To achieve a low level
of arousal (calibration level), the Ss performed the
auditory tracking task under the impression that the ex
perimenter was merely calibrating the equipment while they
were performing. The high level of arousal (incentive
condition) was achieved in two different ways. In one
instance, the Ss were instructed that they would he given
$5*00 for each time they equaled or surpassed their best
score; in another, the Ss were instructed that they would
be given $2.00 each time they equaled or bettered their
score and also avoid a 150 volt shock. When Palmar
conductance level or the electromyographic responses of
any one of four different muscle groups were used as the
criteria of arousal, the data strongly supported the
inverted-U hypothesis. They further supported Surwillo's
finding that the monetary incentive was the most important
experimental variable with regard to the steepness of the
electromyographic gradient. When considered in terms of
error scores, and assuming that the experimental variables
did produce variations in level of arousal, the hypothesis
was also strongly supported. The difference between the
optimal and calibration conditions in terms of error
scores was significant beyond the .001 level with far
fewer errors being made under the optimal conditions.
Significantly more errors were also made under the in
centive conditions than under the optimal conditions; this
result was significant beyond the .02 level.
A series of studies has been conducted in which the
experimental operations, although designed to produce
39
variations in frustration, were very similar to experi
mental operations that have "been used to produce varia
tions in incentive. Brown and Farter (8) have suggested a
theory of frustration in which they attempt to relate
frustration to antecedent experimental conditions and
treat it in much the same way that the drive concept is
treated in the Hull-Spence formulations. According to
these authors, reactions produced hy frustration will act
like drive by intensifying responses, by reinforcing S-R
connections when frustration is reduced, and by generaliz
ing to other situations.
Generally speaking, variations in frustration have
been produced experimentally by delaying a rat, for
example, as it moves down the runway toward the goal box
or by not rewarding the rat when he reaches the goal box
after he had previously been rewarded there. It is easy
to see the similarity between these experimental manipula
tions and varying the magnitude of reward, increasing the
length of the instrumental response chain, delaying the
administration of the reward, and varying the time in the
goal box. Amsel and Roussel (6) tested hypotheses that
were related to the assumptions that frustration is a
motivational condition and that the strength of frustra
tion will vary with the time in the frustrating situation.
They found in using rat Ss that the strength of response
ho
on a frustration trial was greater than on rewarded trials,
"but they found no variation in its strength as a function
of the amount of time spent in the frustrating situation.
Amsel and Hancock (*+) , assuming that the important factor
in the development of a frustration response is the
fractional anticipatory goal reaction, conducted two
studies using rats in a runway situation. In one,
several rewarded trials were given, presumably to build up
the strength of the fractional anticipatory goal response;
in the other study, the rg response developed was pre
sumably less strong. They found that when there was a
strong rg, the response to frustration was more immediate;
where the Tg mechanism had not been developed strongly, it
was assumed that the non-rewarded trials were not frustrat
ing. Holder, et al (31) trained rats to run down a runway
to a goal box. During the training trials, they delayed
the rats in the middle of the runway for one second.
Following the training, the rats were divided into three
groups; one group had a one-second delay, a second group
had a 15-second delay, and a third group had a ^-5-second
delay in the middle of the runway; a buzzer was sounded
during the delay period for both the 15 and ^-second
groups. All three groups were then run on a second runway
and the buzzer was sounded throughout the entire run. The
results showed that both the 15-second and 1 +5-second rats
bl
ran significantly faster than the one-second control
group, but there was no difference in the speed of running
between the two experimental groups. They also found that
there was a decrease in running speed in the experimental
groups preceding the portion of the runway in which the
delay was administered and an increase in running speed
following the delay period. These results were interpret
ed in terms of a fractional anticipatory goal response
with inhibitory properties to explain the decrease in res
ponse strength prior to the delay period, and in terms of
the irrelevant drive properties of frustration to explain
the increase in response strength following the delay
period.
The literature available on performance as a func
tion of variations in incentive seemed, in general, to be
irrelevant to the study described here with the exception
of the studies done by Stennett (61) and Surwillo (62).
Although neither of these studies was designed to test the
Hull-Spence formulations, in both instances, what is
generally considered to be a meaningful incentive in our
society to human Ss was used. In both studies, it was
found that a monetary incentive was most highly related to
the steepness of electromyographic gradients. In one (61),
it was found that when an opportunity was given to obtain
monetary rewards for improvements in performance, the
number of errors in performance increased over what were
described as optimal arousal conditions. Similar to the
study described here, this monetary variation of in
centive could be interpreted by Spence as a variation in
drive rather than a variation in incentive. To consider
a monetary reward as a variation in incentive, it would
be necessary, as in the case of the opportunity for a
liberty used in this study, to assume that rg, the
fractional anticipatory response mechanism, had been de
veloped on previous occasions, and that the stimuli that
had been conditioned to these responses were present in
the experimental situation or were similar to stimuli
present in the experimental situation.
Chapter III
Method
Experimental Design
A two-by-two factorial design was used. Experi
mental manipulations designed to produce variations in the
intervening variables D and K were the two independent
variables. One half of the Ss, hereafter to be called the
high incentive groups, were instructed that they would be
given a three-day liberty (a pass, in Air Force or Army
terminology) depending upon how well they performed on the
test trials. The remaining half of the Ss, the low in
centive group, was not given a similar opportunity. Thus,
an attempt was made to produce two levels of the inter
vening variable K, incentive motivation, by giving one
half of the Ss an opportunity to obtain a liberty and not
giving the other half a similar opportunity.
An attempt was made to produce two levels of drive,
D, by giving one half of the Ss conditioning trials in
which a 1000 cycle per second (cps) tone, CS, was paired
with an electric shock, US. The other half of the Ss were
not given trials in which the CS was paired with the US.
Thus, as shown In Table I, there were four experi
mental groups: low drive— low incentive, DjKp; low drive—
high incentive, DpK.2; high drive— low incentive, D2Kj_; and
high drive— high incentive, D2K2*^
There were two independent measures, the speed and
the accuracy of response.
Subjects
Forty Wavy apprentice seamen between the ages of 17
and 19 served as Ss. They were asked to participate in
the experiment by their commanding officer; none seemed
1. It is recognized that it is necessary to make
assumptions concerning whether or not the experimental
manipulations produced variations in the intervening
variables and that it would be more accurate to speak in
terms of the experimental rather than the intervening
variables. However, the assumptions regarding the rela
tionship between the experimental and intervening varia
bles have already been discussed, and for convenience the
designations “low" and "high incentive" and "low" and
"high drive" rather than "no opportunity for a pass" and
"opportunity for a pass" and "no conditioning trials" and
"conditioning trials" will be used.
b5
reluctant. At the time the study was conducted, the Ss
were assigned to the U.S. Fleet Sonar School, San Diego,
waiting for their classes to convene. During this waiting
period, the Ss normally performed various routine manual
tasks around the school, mowing lawns, painting, cleaning
and waxing floors, and so on.
The Ss, all of whom had normal color vision and
had previously passed the Sonar School visual and auditory
screening tests, were randomly assigned to one of the
four experimental groups. At the same time, they were
randomly assigned a particular time during a five-day
period to participate in the study. The experimental
design and the number of Ss assigned to each experimental
condition are illustrated in Table I.
Table I
The Experimental Conditions and the
Number of Ss Assigned to Each
h
Low Incentive
K2
High Incentive
Dl
D1K1
DiK2
20 Ss
Low Drive 10 Ss 10 Ss
d2 D2K1
D2K2
20 Ss
High Drive 10 Ss 10 Ss
20 Ss 20 Ss bo Ss
* + 6
The Subjects' Task
The 3s1 task was to discriminate among nine stimuli,
three hues (red, green, and yellow) appearing at any one
of three brightnesses, ana then to throw the appropriate
toggle switch among nine such switches arranged in three
rows and three columns of three switches each. Each hue
corresponded to one column and each brightness level to
one row of switches. They were instructed to pull the
correct switch as rapidly and accurately as they could.
Apparatus
Not all of the functional characteristics of the
apparatus were employed in the present study, and only
those that were used will be described. The major com
ponents of the apparatus are illustrated in Figures 1 and 2.
Figure 3 is a schematic representation of the re
lationships among the major chassis and the auxiliary
units. The arrows indicate the direction of flow of the
electrical signals .
The Experimenter ' s Console
The experimenter's (E's) console contained switches
for activating the lights (feedback lights) located below
each of the response switches on the 3s' response panel to
indicate the correct response during the training trials;
multi-position controls for selecting the desired stimulus;
lights in series with the Ss1 response switches to enable
E to determine which switch S had thrown; a "go” switch
that simultaneously activated a relay controlling the on
set of the stimulus and started the electric clock; and a
light in series with the Ss' ready switch.
The interior of E's console chassis contained a
power supply, an audio amplifier, and an audio oscillator.
The power supply provided power for the oscillator,
amplifier, relays, response indicator lights on E's con
sole, feedback lights on the Ss1 response panel, and the
ready circuit.
The audio oscillator and amplifier produced the
1000 cps tone CS, which was presented over the speaker on
the stimulus producing box.
The lights in the stimulus producing box were
powered by the ordinary, 110 volt A.C. "house" outlets.
To avoid the danger associated with having such power in
E's console, the visual stimuli were controlled using 6
volt D.C. power in E's console that controlled 110 volt
relays on the relay chassis, which, in turn, powered the
lights in the stimulus box.
The Response Panel
The Ss' response panel contained nine response
toggle switches and below each was an enclosed dim red
feedback light. The response switches were normally in
1+8
response indicator lights
o o o
o o o
o o o
§ O
/ ° c
f t
3 O
1 ^ °
ready light
stimulus
control
switches
feedback go
switches button
Figure 1. Experimenter's console
* + 9
stimulus
aperture
feedback
lights
hand
restra ining
strap
ready switch
Figure 2. Subjects' response panel and stimulus
box
visual
stimuli
tone CS
go relay
go
relay
stimuli
go relay
sneaKerl-*
ready
switch
intercon
intercom
electrode strap
stop
clock
shocker
condenser
chassis
shocker
trans.
chassis
relay
chassis
stimulus
aperture
s' response panel
experimenter 1s
console
Figure 3. Schematic diagram of the apparatus
51
the "up" position. Pulling a switch down activated one of
the indicator lights on E's console, the light isomorphic
to the switch thrown, and, at the same time, opened the
"go" relay and stopped the electric clock. The lights be
low each of the toggle switches were activated from E's
console.
The Stimulus Presentation Box
The stimulus presentation box contained an 8 inch,
round Stentorian speaker, Model HF812, with a 12 ounce
magnet. Beneath the speaker was a 1/2 inch aperture
through which the visual stimuli were presented.
The interior of the box contained a mounting for
three 120 volt, 100 watt, movie projector lamps, GE Model
PH/100 TbSC. The filaments were in the same horizontal
plane as the center of the aperture in the face of the box.
One lamp was on a line running directly to the rear of the
box and was mounted six inches behind the aperture. The
other two lamps were 5 I/*4 - inches behind the aperture at an
angle of 27 degrees from the center line or middle light.
The mounting for the lamps and the shielding between them,
which ran from 1 1/2 inches to the rear of the aperture to
a point just to the rear of the lights, was painted with a
flat, black paint.
The aperture was covered with frosted glass. A
Kodak, number 8, K2 yellow Wratten gelatin filter glued
52
between clear glass was mounted immediately behind the
frosted glass. Additional filters were placed perpendicu
lar to the line between the aperture and each projector
lamp at a distance of two inches from the aperture. For
the red hue, a Kodak number 25A filter was used; for the
yellow, a Kodak 15G filter; and for the green, a Kodak *+713
filter.
It was found in constructing the equipment that the
voltages necessary for producing the lowest brightness
levels were not sufficient to make the onset of the low
brightnesses equal to the onset of the high brightnesses.
1’ his was corrected by supplying a low voltage to the fila
ments of the projector lamps between trials.
During the construction of the equipment, it became
apparent that the hue of the light through the U-7B filter
changed as a function of the voltages applied to the
filament of the lamp behind that filter. Subjectively,
the hue changed from green to blue as the brightness level
increased. The yellow, number 8, K2 filter was placed
immediately to the rear of the ground glass covering the
aperture to offset this change in hue. It did not
perceptibly change the hue produced by the red and yellow
filters, and it made the hue produced by the *+7B filter
appear to be an equally saturated green at all brightness
levels.
53
Auxiliary Equipment
There were five auxiliary equipments in addition to
the three major chassis just described: an electric
shocker, an electric timer, a relay chassis, an inter
communication system, and a "ready" switch.
Electric Shocker. The electric shocker consisted of
two small (*+ by 6 inch) chassis, interconnecting wires, a
small leather strap on which two dimes were mounted for
delivering the shock to S's forearm, and one 90 volt
battery. One of the chassis contained five different size
condensers, a toggle switch for selecting the condenser to
be charged from the battery, and a microswitch for dis
charging the condenser through the skin of S's forearm.
This chassis was placed adjacent to E's console during the
experimental sessions. The other chassis held a trans
former and connections for the leads to the dimes on the
leather strap. This transformer chassis was placed on the
floor beneath the table on which the Ss1 response panel and
the stimulus producing box rested.
The intensity of the shock was established by having
five Ss, who did not participate in the main experiment,
set it at a level that was "more than annoying but not
painful." The most intense level established by these five
Ss was used for all Ss.
Figure ^ is a diagram of the shocker circuit.
charge
switch
-• > 1 m-
shock switch,
momentary contact
1.25mfd,
200v
6 ,3v
sec
►HOv
pr i
6.3v -- HOv
2A filament
transformer
Figure *+. Diagram of the shocker circuit
55
Electric Timer. A Standard Electric Time Company
synchronous-motor stop clock was connected to the "go"
relay on the relay chassis. The "go" button on E's con
sole activated the "go" relay which, in turn, illuminated
the stimulus light and started the electric stop clock.
Pulling down any of the nine toggle switches on the Ss1
response panel again activated the "go" relay, turning off
the stimulus and stopping the timer. Power for starting
and stopping the timer was obtained from a 6 volt D.C.
battery.
Belay Chassis. The relay chassis contained six re
lays for controlling the visual stimuli and one, the "go"
relay, for starting and stopping the visual stimuli, the
electric timer, and the 1000 cps tone, CS. Positioning
the controls for brightness and hue on E's console served
to select corresponding relays on the relay chassis.
Pushing the "go" button produced a 6 volt D.C. signal that
activated the "go" relay. It, in turn, activated the
selected stimulus relays, allowing a 110 volt alternating
current to go through appropriate resistors to the pro
jector lamps. The auditory signal, the CS for the con
ditioning trials, went directly through the "go" relay to
the loudspeaker on the stimulus producing box.
56
Intercommunication System. The intercommunication
system consisted of two speaker-microphones, one in the
Ss1 tooth and one at S's station. It was used only to
warn the Ss when the training or test trials were about to
begin or when they were completed.
Ready Switch. A microswitch on a small chassis was
mounted below and in front of the Ss1 response panel.
This 'ready' switch was connected in series with a 6 volt
D.C. lamp on S's console. When S placed his finger on the
microswitch, the light on E's console illuminated. The
purpose of the switch was to insure that all Ss kept their
preferred hand in the same place between trials.
Stimuli
The filters used to produce the three hues were
selected for two reasons. First, the saturation of the
hue they produced at the stimulus aperture did not change,
according to the judgments of 10 Ss, as their brightness
levels were varied over a wide range. Second, disjunctive
reaction time tests made using these same 10 Ss indicated
that the hues were not significantly different from each
other in terms of either the number of incorrect responses
or the mean reaction times.
An effort was made prior to conducting the experi
ment to make the three brightness levels of the hues equal.
57
The red, Kodak 2^A, gelatin filter permitted the least
passage of light, so it, with no additional resistance in
the line, was taken as the standard for the brightest
hues. Seven Ss decreased the brightness of the lamp for
the red filter to the lowest possible level without chang
ing, according to their judgment, the saturation of the red
stimulus. This was done by adjusting a variable resistance
that was in series with the lamp, and the mean resistance
value obtained was taken as the standard for the least
bright hues. The Ss next bisected the least bright and
brightest red hues, again by adjusting a variable re
sistance. The resulting middle brightness for red was
arbitrarily taken as the standard for the middle bright
ness of the remaining two hues. The three brightness
levels of the yellow and green hues were then matched to
the three levels of the red and to each other using the
method of adjustment. In addition, the brightest and
least bright yellows and greens were bisected to obtain
values using a different psychophysical procedure.
Variable resistors set at the means of the re
sistance values in ohms obtained using both the adjustment
and bisection methods were placed in series with the
appropriate lamps to produce three brightness levels for
each hue. The values in ohms of the resistances used are
given in Table II.
58
Table II
Resistance Values in Ohms Used to Equate
the Brightness Levels of the Three Hues
Red Green Yellow
Brightest 0
^3
Middle Brightness 6o 108 100
Least Bright 190 190 200
With the resistances given in Table II placed in
the appropriate circuits, six Ss were given a three choice
disjunctive reaction time test using all three hues at the
three brightness levels of each. They were instructed to
pull down the top switch of a column of three in response
to the brightest lights, the middle switch in response to
the middle brightnesses, and the bottom switch in response
to the least bright lights. They were instructed to
respond only to the brightness levels of the lights and
not to the hues. Each S was given a total of 27 trials,
three trials on each brightness-hue combination. No
significant differences were found among the responses to
the three hues, the three brightness levels, or the
specific hue-brightness combinations in terms of either
the number of errors or the mean reaction times.
None of the Ss who participated in the selection of
filters or matching of the stimuli participated in the
main study.
59
Installation
The experimental equipment was installed in an un
used, sound insulated building at the U.S. Fleet Sonar
School, San Diego. The stimulus producing box, the Ss1
response panel, the transformer chassis of the shocker,
and the ready switch were placed in a sound insulated,
darkened, training booth, seven by five feet at the floor
and nine feet in height. E's console and the remaining
auxiliary equipment were placed in an adjoining room U-5
feet away from the Ss* booth.
The ambient noise level in the experimental booth
was 57*0 db as measured by a General Radio Company sound
pressure level meter, type 1551-A, calibrated at 8U-.5 db.
Procedure
The data were gathered in a five-day period. It
took 35 minutes to run each S, and the Ss arrived one at a
time at the experimental room at b'y minute intervals.
The experimental procedure was divided into three
phases: training, conditioning, and test. Instructions
were given prior to the training trials and following the
training, prior to the conditioning trials. There was no
pause and no additional instructions were given between
the conditioning and the test trials. Table III shows the
phases of the experiment.
60
Table III
Phases of the Experiment
Grouo
Training
Phase
Conditioning
Phase
Test
Phase
D1K1 Trials
xn
s
o
n
E - i
O
P
CO
P
Id trials, tone
on 9 trials,
no shock
3b trials, tone
appeared on all
trials, no shock
DiK2 •+5 Trials
1 1 t t
D2Ki b$ Trials
18 trials, tone
followed by shock
on 9 trials
36 trials, tone
appeared on all
trials, shock on
9 trials
d2k2 b$ Trials
i i 1 1
Two experimenters were required in all phases of
the study. One read the instructions to the Ss, set up
the appropriate stimuli on E's console, operated the "go"
button, recorded the response made to each stimulus, and
flashed the light indicating the correct response after
training each trial. The other recorded the speed of each
response, reset the electric timer, and, during the con
ditioning trials, charged the condenser and administered
the electric shock. This second experimenter timed the
onset of the electric shock by observing the electric
timer after it had been started by the "go" button. The
shock was to follow the onset of the visual stimulus and
the tone, CS, by one-half second. It was found by con
ducting pre-experimental trials that E could administer
61
the electric shock within plus or minus three hundredths
of a second of the desired time, using this method.
Training Phase
Ten minutes before entering the experimental booth,
S was seated in an almost entirely dark room to permit
some dark adaptation to take place before he entered the
partially darkened experimental booth. When it was time
for him to act as a subject, he was conducted to the ex
perimental booth and told to sit comfortably in front of
the response panel and look directly into the stimulus
aperture. He was asked whether he was left or right hand
ed, and the wrist of his non-preferred hand was fastened
with a leather strap to the table on which the response
panel rested. E then read the following instructions:
"Part of the men who participate in this
experiment will have an opportunity to get a
72 hour liberty. I have the liberty cards
here with me now; they have already been sign
ed by Commander Rinker.
"The experiment is divided into two
phases, a training phase and a test phase
with the training phase coming first. I will
let you know after you have completed the
training phase, which takes about 15 or 20
minutes, whether or not you will have an
opportunity to get the 72 hour liberty.
"This is a task to see how quickly and
accurately you can tell the difference
between various colors and brightnesses.
You will be presented with 3 colors. Each
color will occur at 3 brightness levels.
Thus, there will be 9 combinations of color
62
and "brightness.
"Now look at the switches on the panel
in front of you. Notice there are 9 switches.
One switch is correct for a particular com
bination of color and brightness. The
columns (show S) represent the different
colors, and the rows (show S) represent the
different brightnesses. The correct switches
for the red light are in the LEFT HMD COLUMN
(show S); the correct switches for the green
are located in the MIDDLE COLUMN (show S):
and the correct switches for the yellow light
are in the RIGHT HAND COLUMN (show S) . The
correct switches for the brightest lights are
in the TOP ROW (show S); the correct switches
for the middle brightnesses are in the MIDDLE
ROW (show S); and the correct switches for
the least bright lights are in the BOTTOM ROW
(show S).
"For example, if you see the red light and
it is the brightest, pull down the switch in
the upper left corner (show S). If you see
the green light and it is the least bright,
pull down the bottom switch in the middle
column (show S). If you see the yellow light
and it is the middle brightness, pull down the
middle switch in the right hand column (show
S).
"When you see the light, react as quickly
and as accurately as you can. After you have
pulled the switch down, push it back up (show
S). Immediately after you push the switch
back up a red light will flash under the
correct switch.
"Put your finger on the button on the table
in front of you (show S) before each trial.
Your finger must be on the button at all times
between trials and before the next light will
be given to you.
"Remember, you will have the training trials
first and then the test trials.
"Do you have any questions?"
63
After these instructions had "been read, E shoved S
the nine stimuli.
All Ss received h-5 training trials, five presenta
tions of each stimulus. The stimuli were presented in a
random order in Hocks of nine trials each. The interval
between trials in all phases was randomly varied between
six and ten seconds.
Conditioning and Test Phases
The test trials followed immediately after the con
ditioning trials with no pause, so both of these phases of
the experiment will be described in this section.
There was a five minute (plus or minus 15 seconds)
rest interval for all Ss between the training and the con
ditioning trials. Immediately before starting the condi
tioning trials, E administered one of four sets of in
structions depending upon the group to which the particular
S had been assigned. The instructions given to the
different groups were as follows;
Group I; Low D - Low K
"Now we are ready to begin the test trials
and you will perform the task in the same
manner as you did before. The only difference
is that now the red light will not flash under
the correct switch after-you make your response.
"You may have heard from some of the men
who have already participated in the experiment
that they received a shock. You, however, will
not get one.
"You have fallen into the group that will
not have the opportunity to get the 72 hour
pass, but you will continue on with the test
task. Your score will be based on both how
accurately and how quickly you respond.
"Do you have any questions?"
Group II: Low D - High K
"Now we are ready to begin the test trials
and you will perform the task in the same
manner as you did before. The only difference
is that now the red light will not flash under
the correct switch after you make your response.
"You may have heard from some of the men
who have already participated in the experiment
that they received a shock. You, however, will
not get one.
"You have fallen in the group that has an
opportunity to receive a 72 hour liberty de
pending on how well you perform on these test
trials. I have the liberty cards here and they
have already been signed by Commander Rinker
(show S). 20 men are taking this test under
the same conditions as you. The 10 men who make
the highest scores will receive 72 hour liberty
cards. They will be good for any 3 days of the
week you select up until the time your Sonar
School class convenes. Remember, to get the
pass you must score in the upper half of the
group. Your score will be based on both how
accurately and quickly you respond.
"Do you have any questions?"
Group III: High D - Low K
"Now we are ready to begin the test trials
and you will perform the task in the same
manner as you did before. The only difference
is that now the red light will not flash under
the correct switch after you make your response.
In addition, this time when the light goes on,
you will receive a mild electric shock on your
arm. This shock will be annoying but will not
65
harm you in any way.
"You have fallen into the group that will
not have the opportunity to get the 72 hour
pass, but you will continue on with the test
task. Your score will be based on both how
accurately and how quickly you respond.
"Do you have any questions?"
Group IV; High D - High K
"Now we are ready to begin the test trials
and you will perform the task in the same
manner as you did before. The only difference
is that now the red light will not flash under
the correct switch after you make your response.
In addition, this time when the light goes on,
you will receive a mild electric shock on your
arm. This shock will be annoying but will not
harm you in any way.
"You have fallen in the group that has an
opportunity to receive a 72 hour liberty de
pending on how well you perform on these test
trials. I have the liberty cards here and
they have already been signed by Commander
Rinker (show S). 20 men are taking this test
under the same conditions as you. The 10 men
who make the highest scores will receive 72
hour liberty cards. They will be good for any
3 days of the week you select up until the
time your Sonar School class convenes» Remember,
to get the pass you must score in the upper
half of the group. Your score will be based on
both how accurately and quickly you respond.
"Do you have any questions?"
After giving the instructions, E attached the
leather strap to the men in the high drive groups, so that
the dimes pressed against the inside of their forearms.
The electrode strap was not attached to the arms of the Ss
in the low drive groups.
66
All groups received 18 conditioning and 36 test
trials, two and four presentations of each stimulus,
respectively. For the low drive Ss, the conditioning
trials were identical to the training trials except that
no indication of the correct response was given. On one
half of the conditioning trials at random intervals, and
only once per hue-brightness combination, a 1000 cps tone
was presented simultaneously with the stimulus. This same
tone, whose loudness as measured by a General Radio
Company sound pressure level meter, type number 1551-A,
was 76.O db, came on simultaneously with the stimulus on
all 36 test trials.
The conditions for the high drive groups were
identical with those for the low drive groups except that
one-half second after the presentation of the tone, which
for all groups appeared on 9 of the 18 conditioning trials,
an electric shock was delivered to S's forearm. Like the
conditions for the low drive groups, the tone appeared on
all 36 test trials given the Ss in the high drive groups.
On nine of these trials at random intervals and only once
per hue-brightness combination, the shock was administered
one-half second after the presentation of the tone and
visual stimulus. The purpose of this was to maintain the
emotional response to the tone assumed to have been de
veloped during the conditioning trials.
67
The training, conditioning, and test trials were
thus identical for all groups with the exception of the
differences in the instructions given prior to the con
ditioning trials and the shock that was given the h’ igh
drive Ss following the presentation of the tone on nine of
the conditioning trials and nine of the test trials.
It should be noted that the distinction made In
this report between conditioning and test trials was not
made in the instructions given to the Ss. The Instructions
indicated that the test trials began immediately following
the instructions. Of course, the 18 trials designated
here as conditioning trials were not used in determining
the effects of the experimental variables.
Chapter IV
Results
Description of the Response Measures
A record was made of the speed of each response, in
hundredths of a second, and the particular switch thrown in
response to each stimulus during the training, condition
ing, and test phases of the study. The response records
obtained during the conditioning phase were not used in the
analyses, though the group mean scores for the nine condi
tioning trials on which the high drive Ss were not shocked
are represented in Figures 5 6. Of the 36 trials ad
ministered during the test phase, the nine on which the
high drive Ss received the electric shock were omitted from
all analyses for all Ss, leaving a total of 27 test trials
for all four experimental groups.
Raw Scores
Figure *+ shows the mean number of errors in blocks
of nine trials made by each of the four experimental groups
on the last 27 training trials, the nine conditioning
trials on which the shock was not administered, and the 27
test trials.
Median reaction time (RT) scores were computed for
each S for each block of nine training and nine test
trials. In addition, median RT scores were computed for
69
each S for the nine conditioning trials on which the shock
was not administered to the high drive Ss. The means of
the median RT scores made by the Ss in each group on
blocks of nine trials are given in Figure 6.
As can be seen from Figure 5> only the low drive—
high incentive group performed as accurately on the test
trials as on the training trials. The remaining three
groups responded less accurately during the test trials
than during the training trials.
Figure 6 shows that all groups responded more
rapidly on the test trials than on the training trials.
Information Transmission Scores
Information transmission scores (T scores) were
computed for each S for the last 27 training and for the
27 test trials. These scores, which were based solely on
the accuracy and not the speed of response, indicated the
amount of information transmitted between the stimuli, on
the one hand, and Ss’ responses, on the other. They may be
taken as indicating the correlation or covariation between
the stimuli and the responses. The scores were computed
using the following formulation (28) :
Mean number of errors
Experimental Groups
High drive— high incentive O-----
High drive--low incentive O-------O
Low drive— high incentive
Low drive--low incentive
2.0
n 1 --------- 1 ---------L
u 19-27 28-36 37-I +t No shock 1-9 10-18 19-27
Training trials ^^rials” *^ Test trials
Figure 9* Mean number of errors in blocks of
nine trials
Reaction time in hundredths of a second
71
Experimental Groups
•--------O high drive--high incentive
O--------O High drive--low incentive
O--------O Low drive--high incentive
O--------O ^ow drive--low incentive
170
160
150
130
120
* 1 -| P j 1 ! ■ I .'■■■ ■ ■ !■>— I . ■ ■ ■ »■ —
19-27 28-36 37-^5'N? shock 1-9
Training trials
10-18 19-27
Test trials
conditioning
trials
onal
al Tra
Figure 6. Means of median subject reaction times
in blocks of nine trials
72
where
T(u,V5y) " H(u,v) ^ H(y) “ H(u,v,y)
T/ ^ = the amount of information transmitted
(u,v;y)
between the two stimulus dimensions,
hue and brightness, u and v,
respectively, and the responses, y.
and where
H
(u,v)
= the amount of information in u,v
(event j,k)
= 2 p(3,k) log2 p(j,k)
H(y) = the amount of information in y
(event m)
= 2 p(m) log2 p(m)
m
H^u v ■ the amount of information in u,v,y
(event j,k,m)
= 2 p(j,k,m) log2 pU,k,m)
3,k,m
The unit of measurement resulting from such compu
tations in the terminology of communication theory is bits
per signal. Since there were nine equally probable
stimuli used and nine possible responses, the maximum
score was log2 9 or 3.17 bits per signal. The distribu
tions of T scores obtained for the test and training
trials are given in Figure 7. The mean T score for the
training trials was 2.651 bits per signal, and for the
73
test trials it was 2,623 tits per signal.
Training Trials Test Trials
2.95
—
2 .9 9
2 **
b
2.90
-
2.99- 1 *
2 .8 5
—
2.89
^ * * *
6
2.80
- 2.ob 1 * 1
*
2 .7 5
—
2.79
0 ' ■ / * 4 s - t - ^ - K - 4 s
3
***
2.7 0
-
2.79- 2
**
2.65
-
2 .69
^ ^
5
* * * *
2.60 2 • 6*+ 1
*
2 .5 5
-
2.59
9 # * * * *
3
2.50
—
2.5b
+ s |c * % : s ic
1
*
2.9-5
-
2.9-9
2 **
9-
* * *
2 .bo
-
2.9-9-
2 **
1
*
2.35'
-
2.39 3
***
2.30
- 2 .3^ 1 * 2
**
2 .2 5
-
2.29
9-
= M - * *
2 .2 0
- 2.29-
1 *
Figure 7. Distributions of T scores for the last
27 training and the test trials (N = >+0)
The T scores indicated the accuracy as well as the
probability of response, and thus represented indices of
reaction potential, E. They were used because the
stimulus-response conditions of the study ideally suited
the information theory model (59-) » and because they not
only reflected the number of errors but also the magnitude
of the errors. An additional minor advantage was the fact
that larger T scores indicate better performance, whereas
the opposite is true with regard to errors. The Pearson
product-moment correlation between the T scores computed
for each S on the 27 test trials and the number of errors
each S made on those trials was -.92.
7^
Reaction Time Scores
A median reaction time score (.RT score) was computed
for each S for the last 27 training and the 27 test trials.
The distributions of these scores are shown in Figure 8.
The median RT for the training trials was 139*0 hundredths
of a second, and for the test trials, the median RT was
111.0 hundredths of a second.
Last 27
Training Trials
306 1
*
190 -
199 3
***
180 -
189 1
*
170 - 179 3
***
160 - 169 3
***
i?o - 159
V
****
1*4-0 - 1*4-9 *4-
****
130 - 139
10
********;
120 - 129 7
*******
110 -
119 3
***
100 - 109
90 - 1
*
80 - 39
Figure 8* Distributions of
Test Trials
1 *
1 *
If ****
+ * * * *
* * * * *
******
12 ************
^ *****
1 *
training and test trials (N = *+0)
Difference Scores
Difference or improvement scores were computed to
compensate for differences observed during the training
trials among individual as well as experimental group
performance levels. The difference scores reflected each
S's improvement from the training to the test trials. It
was assumed that improvement in performance would be a more
75
sensitive indicator of the effects of variations in the
experimental variables than absolute performance levels on
the test trials, which, it was assumed, reflected more in
the way of basic aptitudes and skills as well as positive
transfer from previous experience.
In the case of the T scores, each S's score on the
training trials was subtracted from his score on the test
trials. A constant of 100 was added to each S's T
difference score to avoid using negative numbers in those
instances where an S ' s test score was less than his train
ing score. The distribution of the T difference scores is
given in Figure 9.
150 -
159
2
**
1*4-0 - 1*4-9 2
**
130 -
139
2
**
120 - 129 2
**
110 -
119 5
*****
100 -
109
6
* * * * *
90 -
99 3
***
80 - 89 6
******
70 -
79 5
*****
60 - 69 5
*****
50 -
59
1
*
*+0 - * 4 - 9 1
*
Figure 9. Distribution of test T scores minus
training T scores plus 100 (K - *+0)
In the case of the RT scores, each S's test score
was subtracted from his training score to obtain the RT
difference score. It was not necessary in this instance
to add a constant since all Ss increased the speed of
76
their responses in going from the training to the test
trials. The distribution of the RT difference scores is
given in Figure 10.
10b
—
117
" 1
- j -
Cj
-
69 1
6C
—
6h
p ; c
s y
—
C-O
S /
2
❖ ' i ~
50
-
i : -k 1
ho
— bo 2
* *
hh
.....
— 6
* T ‘ v • ? / *
-
k
^
30
- 34- 1
*
' 3 t h
—
29
h
' i ' ^
20
—
2k 6
4 * * *
15
- 6
* n - # ' f ' > f ^
10
—
Ik 6
% - 1 - * * * *
5
-
9
Figure 10. Distribution of training KT scores
minus test RT scores (N = ho)
Adjusted Difference Scores
Computing difference scores compensated for
differences in individual and group performance levels
prior to the test trials. At the same time, however, it
handicapped the Ss who performed well on the training
trials. For example, an S who made a good score on the
training trials could be expected to have less room to
improve than an S who made a poor score. This was particu
larly true of the RT scores where a physiological limit may
have been involved. An adjustment was made, therefore, in
the difference scores to compensate for the amount of im
provement that could be expected on the basis of the
77
training scores. Stated in another way, the difference
scores were adjusted so as to remove that portion of the
variance in them that was common with or could he pre
dicted from the training scores. The procedure used was
directly analogous to the analysis of covariance.
Pearson product-moment correlations between the
training and difference scores were computed. The corre
lation obtained using the T scores was -0.58» and using
the RT scores it was O.B^. Both of the coefficients were
significant beyond the .01 level and they indicated the
extent to which the T and RT difference scores could be
predicted from the training scores. The T score correla
tion indicated that an S who scores high on the training
trials was less likely to obtain a large T difference
score than an S who scored low on the training trials.
The RT score correlation, on the other hand, indicated
that an S who had obtained a high score (slower reaction
time) on the training trials was more likely to obtain a
large RT difference score than an S who had obtained a low
score (faster reactions) on the training trials.
The difference scores were adjusted to take out the
variance that was common with the training scores by first
computing the coefficient of regression of the difference
scores on the training scores. Then, each S's difference
score was adjusted by an amount equal to the expected
78
deviation in that score as the result of the training
score he had obtained. An amount equal to the amount of
an expected deviation was added or subtracted from the
difference score as appropriate. The adjustment procedure
produced a zero correlation between the training and
difference scores.
The adjustment procedure had little effect on the
T scores; the Pearson product-moment correlation between
the T difference scores and the adjusted T difference
scores was 0.98. It had a rather appreciable effect on
the RT scores; the Pearson product-moment correlation
between the RT difference scores and the adjusted RT
difference scores was 0.57*
To illustrate the effects of the adjustment pro
cedure, one S had a training RT score of 131, a test RT
score of 120, to make an RT difference score of 11; his
adjusted RT difference score was 19. Another S had a
training RT score of 93 j a test RT score of 82, to make
the same RT difference score of 11; his adjusted RT
difference score was 38. It can be seen from this illus
tration that the adjustment was greater for the S who had
less room, so to speak, to improve his score from the
training to the test trials.
The adjusted T difference scores and the adjusted
RT difference scores were used in the analyses designed to
79
test the experimental hypothesis. The distributions of
these scores, which hereafter will be designated as T1
and RT1 scores are shown in Figures 11 and 12.
i b o - l*+9
2
9 - ¥
130 -
139 3
■ jf. * *
120 - 129
3
* **
110 -
119
b
* > ) ; * *
100 - 109
0
*** * * *
90 -
99 5
*****
bo -
89
6
******
70 -
79
6
* * ^ " i " *
60 - 69 3
if. * > f
50 -
59
2
**
Figure 11. Distribution of adjusted 1 difference
scores (N = ho)
50 - 52
1 *
h7 - 9-9 2
**
bb - ! +6 1
*
Vi -
9-3 5
* * * * *
3& -
^0
5
*****
35 -
37
***
32 _
3^ 3
***
29 -
31 5
*****
26 - 28 2
* *
23 - 25
6
* * * * * *
20 - 22
17 - 19 3
* * *
lb - 16 1
*
11 -
13 3
***
Figure 12. Distribution of adjusted median RT
difference scores (W = M-0)
Transmission-Time Scores
Analyses were performed using a third measure which
simultaneously reflected the difference between training
80
and test performance in both the speed and the accuracy of
response. Both the T1 and RT1 scores were converted to a
standard scale using the standard deviation of each dis
tribution as the unit of measurement. The deviation of
each S's T' score from the mean of the T' scores was
divided by the standard deviation of the T1 distribution
and then a constant of 10 was added to it. The RT1 scores
were treated in the same way. The resulting deviation
scores were multiplied together to produce an increase-in-
transmission multiplied by increase-in-speed scores. The
distribution of these TRT* scores is shown in Figure 13.
130 - 13^ 2 **
125 - 129
120 - 12V
115 - 119 2 **
110 _ 114- 9 * * * * * * * * *
105 - 109 2 **
100 - ion- 3 ***
95 — 99 7 *******
90 — yb 8 ********
85 - 89 3 ***
80 - 8V
75 - 79 1 *
70 - 7b 2 **
65 - 69 1 *
Figure 13. Distribution of the products of the
T1 and RT1 scores (N = Vo)
Tests of Assumptions
One of the assumptions of the study was that the
strength of the correct responses would be equal and
dominant for all four experimental groups. Stated in terms
of the intervening variables, it was assumed that habit
strength for the various stimuli would be equal and
dominant for all groups at the time the test trials began.
According to Spence, if an incorrect response were
dominant, increasing either D or K or both would have
served to increase the reaction potential for an incorrect
response. Thus, in this experiment, if an incorrect res
ponse had been dominant at the beginning of the test
trials, the high drive and the high incentive groups
should not have performed as well as the low drive and low
incentive groups.
To test the assumption, a tabulation was made of
the number of incorrect responses each group made on the
^5 training and 18 conditioning trials that preceded the
test trials. Table IV is a summary of those tabulations.
82
Table IV
Total Errors by Stimulus and by Experimental Group
on the 63 Trials Preceding the Test Trials
Stimulus
GROUP 1 2
. 8
b
5
6
7
8
9
TOTALS
D1K1 19 25
16
13
32
39
11 26
9
190
D1K2
18 2b 11
7 36 16 30 186
D2Ki
lb 23 21 11 38
^7 13 31
6 20b
16 20 12 8
31 31 11 27 5
161
TOTALS 67
92 60
39 137 157 51
11^ 2b 79-1
Analyses showed that there were no significant
differences among the four experimental groups with
respect to the number of errors made in response to any
particular stimulus. It was felt, therefore, that the
assumption could safely be made that there were no
differences among the four groups in terms of the habit
strength developed for the nine stimuli. And, when all
stimuli were combined, the assumption that H was equal for
all groups seemed justified.
It is evident that an analysis of this sort, in
which errors were summed across Ss within each group for
each stimulus, did not account for individual differences
and that it was possible that some Ss in each group made
more incorrect than correct responses to a particular
stimulus, in spite of the fact that the group as a whole
33
made considerably more correct than incorrect responses to
it. Table V shows the number of men in each group who
made more incorrect than correct responses to each
stimulus.
Table V
Number of Ss in Each Experimental Group Who Made
More Incorrect Than Correct Responses to Each Stimulus
on the 63 Trials Preceding the Test Trials
Stimulus
GROUP 1 2
3
b
5
6
7
8
9
D1K1 1 0
-----— — —
1 0
b 6 0 1 1
d 1k2
2
1
0 0 7 5
0
3
0
D2K1
1 0
3
0 6 6 1 b
0
D2k2
0 1 0 0 b
0
0
1
5
0
TOTALS If 2 b
0 21 20 2
13
1
Other than stimuli 5, 6, and 8, the majority of Ss
in each group made more correct than incorrect responses
to each stimulus. Half of the Ss in the D2N2 6rouP m&de
more incorrect than correct responses to stimulus 8.
Stimuli 5 and 6 were apparently the most difficult of the
nine. The majority of the men in two of the experimental
groups, DpK-) and 02^1? made more incorrect responses to
stimuli 5 and 6 than correct ones. The majority of the
men in the group made more incorrect than correct
responses to stimulus 6. Only in the £> 2^2 SrouP the
8*+
majority of the men make more correct responses to "both
stimuli 5 and 6.
It should be noted, that these tabulations did not
show that the correct response was not the dominant one
for even the most difficult stimuli. Tabulations were
not made of the particular incorrect responses made to
each stimulus. If the incorrect responses were distri
buted even partially over the remaining eight response
switches, the correct response to even the most difficult
stimulus would probably have been the dominant response.
In any case, the analysis suggested by these last comments
was not performed*, another analysis seemed more
appropriate.
The tabulation of errors by group and stimuli
indicated that it was more likely that the correct res
ponses to stimuli 5 and 6 were less dominant than the
correct responses to any of the remaining stimuli. A
comparison was made, therefore, to determine whether or
not there were differences in the number of errors made
during the test trials to stimuli 5 and 6, on the one
hand, and the remaining stimuli, on the other. The re
sults are shown in Table VI. Both numbers of errors and
proportions are presented in the table.
Table VI
Errors and Proportions of Errors Wade on the Test Trials
Stimuli 5 and 6
Kq_ , Low
Incentive
K2, High
Incentive
Dl
Mt 26 70
Low
Drive .26
.15
.hi
High
52 100
Drive
•31
.28
.59
96
7*+ 170
.57 .^3
1.00
Stimuli 1,2,3,^,7,8, and 9
K;l , Low
Incentive
K2 j High
Incentive
Di
26 18 ¥+
Low
Drive .26 .18 .bb
D2
31
25 56
High
Drive
•31
ir\
CM
•
.56
57 *+3
100
.57 •^3
1.00
86
Tests of the differences between proportions in
dicated there were no significant differences among the
four experimental groups between the accuracy of the res
ponses to stimuli 5 and 6 and to the remaining seven
stimuli combined.
It was concluded from the results of these analyses
that H was not different for the four experimental groups
at the time the test trials began. It was also concluded
from the analysis shown in Table VI that the possibly
differing habit strengths for the nine stimuli apparently
had no significant effects on the results of the study,
since the results in terms of error scores were not
significantly different for the two most difficult stimuli
as opposed to the remaining seven stimuli.
The assumptions regarding the effects of the con
ditioning procedure and the incentive instructions are
discussed with the descriptions of the analyses of
variance that were performed to test the experimental
hypothesis, since the tests of the main effects in the
analyses of variance represented the tests of these
assumptions.
Tests of the Experimental Hypothesis
The study was designed to permit a test of the
hypothesis that the effects of the two intervening
variables D and K, as defined by operations previously
87
described, are additive in their actions upon reaction
potential, E. The hypothesis was tested by determining
the significance of the interaction term in a two by two
factorial design involving two levels of iv and two levels
of D. Analyses were performed using the three response
measures previously described.
Analysis of Variance — T_[_ Scores
The mean adjusted T difference scores, that is, T'
scores, for the four experimental groups are given in
Table VII.
Table VII
Mean T' Scores for the Different
Experimental Conditions
K1
low Incentive
k2
High incentive
D1
Low Drive
93-2
117.7
105.1 +
d2
High Drive
91.8 86.8
89.3
92.5
102.2
97->+
An analysis of variance was performed to test the
significance of the differences among the mean scores
88
2
shown in Table VII . The results are shown in Table VIII.
Table VIII
Analysis of Variance, T* Scores
Source .
_.ir:
Mean Sauare ....E----
D i 2.608 5.21*
K i 9*0 1.90
D x K i 2,176
Error
,36 . . *01
* Significant beyond the .05 point
2. A similar analysis using the differences between
the number of errors made on the last 27 training and the
27 test trials produced the same results. The main effect
due to D and the interaction effect were significant. The
low drive--high incentive group improved its error score
significantly more than any of the other three groups,
whose improvements-~actually decrements--were not signifi
cantly different from each other.
The analysis of variance using the T difference
scores that had not been adjusted to take out the variance
that was common with the variance in the T scores on the
last 27 training trials also produced essentially the same
results. The main effect due to D was significant beyond
the .025 level and the interaction effect was significant
beyond the .05 level. The low drive--high incentive group
improved to a significantly greater extent than any of the
other three groups.
As shown in Table VIII, the effects due to the ex
perimental manipulations designed to vary D were signifi
cant beyond the .05 level. It should be noted from Table
VIII, however, that the effects were in a direction
opposite to that predicted by the Hull-Spence formula
tions; the low drive groups performed significantly better
3
than the high drive groups .
3. A similar analysis of variance of the original T
scores computed for just the 27 test trials (in contrast
to the adjusted T difference scores given in Table VII)
produced no significant effects. The significance levels
of the two main effects and the interaction were all
between .20 and .10. The difference between the scores for
the two drive levels was identical to the difference
between the scores for the two levels of incentive.
The low drive groups were superior, however, to the
high drive groups, and the high incentive groups were
superior to the low Incentive groups. An overall F test of
the differences among the four treatment means was signifi
cant beyond the .001 level. The performance of the low
drive— high incentive group was significantly better than
the performances of any of the other groups beyond the
.05 level.
90 ;
The main effect due to the experimental manipula
tions designed to produce variations in Iv "was not signifi
cant. It should he noted, however, that the interaction
effect was significant beyond the .05 level, and that the
effects of providing an opportunity for a 72 hour liberty
were in opposite directions at the two drive levels.
Increased incentive at the low drive level served to im
prove performance and at the high drive level possibly to
make performance poorer.
Tests of the significance of the differences
between individual treatment means, using the mean square
for error obtained in the analysis of variance, showed
that the low drive— high incentive group performed
significantly better than the high arive--high incentive
group at the .01 level and better than the other two
groups at the .05 level. None of the differences between
the other treatment means was significantly different from
zero, using the .05 level as the criterion.
An assumption of the study was that the condition
ing procedure would produce significant behavioral changes.
The results indicated that significant changes were pro
duced, but they were not in the expected direction insofar
as the Spence position is concerned. Undoubtedly, there
are innumerable alternative explanations of the results
due to the conditioning procedure, one of which is, of
91
course, that a conditioned emotional response does not
increase the accuracy or -probability of response.
Analysis of Variance— RT1 Scores
The mean adjusted RT difference scores, that is,
RT1 scores, for the different experimental conditions are
shown in Table IX.
Table EC
Mean RT1 Scores for the Different
Experimental Conditions
kl
Low Incentive
L2
High Incentive
D1
Low Drive
28.7
30.9
29.8
d2
High Drive
32.5
3^.2
33 A
30.6 32.6 31.6
The results obtained using the RT' scores were
consistent with what would be predicted from the Hull-
Spence behavior theory with regard to the actions of D and
1C on performance. The Ss who received the conditioning
trials showed greater improvement in performance at both
levels of incentive. And, the Ss who had the opportunity
to receive the pass showed greater improvement in per
formance regardless of whether or not they had received
the conditioning trials.
92
The results of the analysis of the variance using
the RT1 scores, which are given in Table X showed, however,
that none of the differences between the mean scores shown
in Table IX was significantly different from zero.
Table X
Analysis of Variance, RT’Scores*4"
Source df Mean Square F
D 1 126 1.08
K 1 38 0.32
D x K 1 1 0.01
Error 36 117
The analyses of reaction time scores thus indicated
that the experimental manipulations failed to produce any
significant variations in behavior with regard to in
creases in speed of response. Neither of the two main
effects, the conditioning procedure and providing an
opportunity for a liberty, nor the effect due to the
joint action of these two experimental manipulations was
b. A similar analysis was performed using adjusted
mean RT difference scores; it also produced no significant
results.
significantly different from zero using reaction time as a
criterion.
Transmission-Time Score Analysis of Variance
The mean transmission-time TRT1, scores for the
experimental groups are given in Tahle XI.
Table XI
Mean TRT’ Scores for the Different
ExperI mental Conditions
L1
Low Incentive
k2
High Incentive
D1
Low Drive 95., 2 1Q7..3
101.2
D2
High Drive 98.6 98.3
98.H
96.9 102.8 99-8
These mean scores, which represented increases in
information transmission multiplied by increases in speed
from the last 27 training to the test trials, indicated
that the low drive--high incentive group improved its
performance over all other groups.
The results of the analysis of variance performed
using the TRT1 scores are shown in Table XII.
9^
Table XII
Analysis of Variance, TRT1 Scores
Source df Mean Square F
D 1 78
0.39
K 1 31+8
1.7^
D x K 1
385 1.92
Error 36 200
Neither of the main effects nor the interaction
effect was significantly different from zero at the .05
point, however, both the effect due to the experimental
variation of K and the interaction effect approached
significance, P = .20 and .18, respectively. The overall
test of the significance of the differences among the four
treatment means produced an F ratio of *+.06, which was
significant between the .025 and the .01 levels. Tests of
individual mean differences, the simple effects, showed
that the difference between the performance of the low
drive--high incentive group and the performance of the low
drive--low incentive group could have occurred by chance
7 times in 100.
To summarize, the analysis of the T1 scores in
dicated that there was a significant interaction due to
the joint variation of the experimental variables determin
ing D and K. At the low drive level, the increase in K
served to increase transmission performance. At the high
drive level, the increase in h served either to decrease
or not to add to transmission performance. The main
effect due to the variation in D, the conditioning pro
cedure, was significant. However, the direction of the
effect was opposite to what would he predicted from the
Hull-Spence formulations. The main effect due to the
variation in K, the opportunity for a liberty, was not
significant. However, such significance could not he
expected in light of the significant interaction and the
fact that the effects of K were in opposite directions at
the two drive levels.
The analyses of the RT' scores produced no signifi
cant results although the differences due to the experi
mental manipulations were in the directions that would be
predicted from the Hull-Spence formulations.
The analysis of variance using the TRT' scores
produced no significant results, hut the interaction effect
came closest to being significant. An overall test of the
differences among the four treatment means was significant
beyond the .025 level, and the performance of the low
drive— high incentive group was superior to all other
groups.
96
Incidental Analyses
Correlations Between the 11 and RT1 Scores
Pearson product-moment correlations were computed
Between the T* and the RT1 scores for the individual ex
perimental groups and for all groups combined to determine
whether or not the Ss had similar sets toward speed and
accuracy as measures of performance. The correlation for
all groups combined was .00. The correlations for each
of the four experimental groups are shown in Table XIII.
Table XIII
Correlations Between T' and RT1 Scores
for the Four Experimental Groups
k2
Low Incentive High Incentive
D1
Low Drive
-.27
d2
High Drive
+ .19
4 .*+6
None of the coefficients given in Table XIII was
significantly different from zero. However, taken simply
as descriptive statistics, they indicated a tendency for
the low drive Ss to improve one score at the sacrifice of
the other and for the high drive Ss to improve or fail to
improve both. This tendency was more pronounced under the
high incentive conditions.
97
Changes in Errors as a Function of Trials
Tahle XIV shows the absolute numbers and proportions
of errors made by each group in blocks of nine test trials.
Table XIV
Absolute Numbers and Proportions (in parentheses)
of Errors by Experimental Condition
in Blocks of Nine Test Trials
K1
Low Incentive
%
High Incentive
Dl
27
Ik 4-1
Low Drive (.26)
(.13) (.39)
d2
34- 31 65
High Drive (.32) (.29)
(.61)
61
^5
106
(.58) (.4-2) (1.00)
K1
Low Incentive
*2
High Incentive
Dl
19
12
Low Drive (.26) (.16) (.4-2).._
d2
25
18
4-3
High Drive (.34-)
....L.2V) . . .L-W
kk 30
7k
(.60) (.4-0) (1.00)
Test
Trials
1 - 9
Test
Trials
10 - 18
K1
Low Incentive
K2
High Incentive
Dl 2k 18 4-2
Low Drive (.27) .(.20) .1.4-7.)
d 2
2k 24- 4-8
Hieh Drive ( .27) . (.27) X.53L.
4-8 4-2
90
(.53) (.4-7) (1.00)
Test
Trials
19 - 27
As can be seen from Table XIV, there appeared to be
a tendency for the effects due to the conditioning trials,
as well as the Incentive instructions, to diminish as the
test trials continued, in spite of the fact that the total
number of errors in the first and last nine test trials was
about the same. The change in the effects due to the
variations in D was greater than the change in the effects
due to lv from the first to the last nine test trials. All
groups reduced the absolute number of errors made from the
first to the last nine test trials with the exception of
the low drive--high incentive group.
These results seerned to indicate that when the
effects due to the conditioning trials--the assumed con
ditioned emotional response--shouid have been greatest,
that is, immediately after the conditioning trials, the
high drive groups had a greater tendency to make more
errors than the low drive groups. At the same time, the
high incentive groups had a greater tendency to make fewer
errors than the low incentive groups immediately following
the incentive instructions. Also, the interaction effect
was more pronounced on the first nine test trials than on
subsequent trials.
Chapter V
Conclusions and Discussion
Conclusions
The following conclusions seem warranted on the
basis of the results of the data analyses.
1. The Ss who received the conditioning trials
prior to the test trials improved their per
formance significantly less than the Ss who
received no such trials, when transmission
scores which reflected accuracy or probability
of response were used as the dependent
measures.
2. Considered over both levels of drive, condi
tioning versus no conditioning trials, the Ss
who received an opportunity to obtain a 72 hour
liberty did not improve their performance
significantly more than the Ss who received no
such opportunity, again when transmission
scores indicating the accuracy or probability
of response were used as the dependent measures.
Significant main effects due to variations in
incentive could not have been expected, how
ever, in light of conclusion number three which
follows.
100
3. When transmission scores were used as the
dependent measures, there was a significant
effect due to the interaction between the two
experimental variables. At the low drive level,
the improvement in performance from the train
ing to the test trials of the high incentive
group was significantly greater than that of
any of the remaining three groups. At the high
drive level, the improvement in performance of
the high incentive group was less than the im
provement in performance of the low incentive
group but not significantly so. At the low
incentive level, drive had no measurable
effect. At the high incentive level, the low
drive group improved its performance signifi
cantly more than the high drive group.
b. When improvement in reaction times was used as
the dependent measure, there were no signifi
cant differences due to the experimental
variations.
5. When scores reflecting improvement in both
accuracy and speed were used as the dependent
measures, the low drive— high incentive group
appeared to improve its performance more than
any other group. However, the results were
101
somewhat equivocal in that an analysis of
variance showed no significant effects and an
overall F test did.
It appeared that the validity, or lack of validity,
of the Spence hypothesis concerning the relationship
between D and li depended on the response measure used.
The results indicated a possible need for a distinction to
be made between speed and accuracy of response as indices
of reaction potential, E.
There was a tendency for low drive Ss to sacrifice
speed of response to some extent for the sake of accuracy
of response. The high drive Ss appeared to attempt not to
sacrifice either at the expense of the other. The results
suggested, therefore, that the experimental conditions may
have produced different sets toward the importance of
speed and accuracy as measures of performance.
Discussion
According to Spence's formulations, increases in
drive should improve performance, make responses both more
probable and rapid. The effect in this study of increas
ing drive was to increase the number of errors made or, in
other terms, to decrease information transmission or the
accuracy and probability of response. -This statement may
be made on the basis of either one of two assumptions:
first, that the conditioning procedure resulted in a
102
higher drive level or, second, that simply saying to the
Ss "we will now begin the test trials" resulted in a
higher drive level. The only group among the four experi
mental groups that performed as well (in terms of accuracy
or probability of response or information transmission)
on the test trials as on the training trials was the low
drive— high incentive group. The performance of the re
maining groups was poorer on the test trials than on the
training trials.
Though not significant, the results obtained using
improvement in reaction time as the dependent measure were
in the opposite direction. Increasing either drive or in
centive served to increase the speed of response, and the
effects of the two appeared additive.
These results appeared to raise the general
question as to whether or not speed and accuracy scores
can be subsumed under the same construct in dealing with
human behavior. The fact that speed and accuracy scores
were negatively correlated in the low drive groups and
positively correlated in the high drive groups seemed to
emphasize possible differences in attitudes or sets toward
the two aspects of performance as a result of variations
in drive. Under the high incentive conditions these
differences were even further emphasized.
The conditioning procedure did not produce
significantly greater improvements in the speed, of res
ponse, but it did produce significantly greater decrements
in the accuracy of response. One possible explanation
might be that the conditioned emotional response, assuming
that was what was developed in the high drive Ss, produced
a set toward escape or avoidance. The Ss could have felt
that if they moved fast enough they would not get shocked,
in spite of the fact they obtained no indication to that
effect. They had already approached a limit in the speed
of their reaction, and the possibility of significant
differences among the mean improvement scores was very
small. At the same time, by attempting to avoid or
escape, they made errors, either by exceeding their speed-
accuracy capacity - if there is such a thing - or because
of incompatible responses that had become conditioned to
the stimuli, se, from the conditioned emotional response,
re •
With regard to this last point, Child (16:15^) has
said, "...in complex situations, where the S is already
in conflict between various response tendencies relevant
to the task, the presence of irrelevant responses made to
anxiety heightens the conflict and interferes-with per
formance to a greater extent than increased drive improves
it." It seems unlikely in this study, however, that an
incompatible response could have become conditioned to the
10*+
stimuli resulting from the conditioned emotional response
in view of the fact that while the number of errors for
the high drive groups increased from the training to the
test trials, the speed of their responses also increased.
It should also be noted that not only did the high drive
groups that had received the conditioning trials make more
errors on the test trials than on the training trials but
so did the low drive--low incentive group. Furthermore,
the low drive— low incentive group, like the high drive
groups, also showed greater speed of response in the test
trials as compared to the training trials.
The poor performance in terms of accuracy of res
ponse of the high drive--high incentive group might also
have been due to incompatible responses conditioned to
stimuli from conflict or frustration responses. The high
drive groups, and particularly the high drive— high in
centive group, could have made a frustration response to
the test conditions. The Ss in these groups could have
reverbalized the test instructions. The high drive— low
incentive group Ss might have reverbalized them as
follows: "Now that I've learned what to do and could do
well on the test trials, they give me a shock to prevent
my doing well." The high drive--high incentive group Ss
might have reverbalized them in this way: “Now that I
know what to do and have an opportunity to get liberty,
they give me a shock to prevent my doing well." In either
case, the reverbalization might have tended to produce a
conflict or frustration type of response. Such a res
ponse, and the stimuli from it, probably should, according
to Spence, increase the drive levels of these Ss. If it
did, the results of the study were even more at variance
with Spence's formulation concerning the relationship
between E, H, D, and K. However, it is possible that
there could have been responses that had previously been
conditioned to the stimuli resulting from the frustration
response, and those responses were incompatible with the
test response. If there were such responses and they were
comparable to the aggressive responses that have often
been observed in humans faced with a frustrating situa
tion, it is not difficult to imagine how the high drive
groups and particularly the high drive— high incentive
group improved their performance less than the other
groups in terms of accuracy of response.
Spence and Hunquist (57) found in a study of the
conditioned eyelid response that the effects of a condi
tioned emotional response induced by pairing a tone with
an electric shock to the finger was sufficiently slow so
as not to be reflected in a conditioned eyelid response
occurring approximately 500 milliseconds after the onset
of the conditioned stimulus. The conditioned emotional
response did show drive effects after a h^OO millisecond
interval following the onset of the emotional conditioned
stimulus. In light of these results, it might have been
more appropriate to have had the conditioned stimulus in
the study reported here presented at some time before the
onset of the stimulus that was to be discriminated. If
the Spence and Runquist results held, then it is reason
able to assume that the effects due to the conditioned
emotional response would have been greater than they were
in this study. However, the fact that significant effects
due to the experimental variations designed to produce
different levels of drive were observed seems to lend
justification for the procedure that was used. It could
very well be that the effects due to the conditioning pro
cedure perseverated from trial to trial, particularly be
cause of the partial reinforcement that was given during
the test trials. As a matter of fact, Runquist, Spence
and Stubbs (52) have made the assumption that the effects
of conditioned emotionality do perseverate from trial to
trial.
It also appears that there may be a disadvantage to
having the stimulus that has been conditioned to the
emotional response to shock precede the stimulus that is
to be discriminated by a period that would conform to the
greatest intensity of emotionality as indicated by the
Spence-Runquist study. If a response that is competitive
•with the response required by the instrumental task em
ployed in the study becomes conditioned to the stimuli
resulting from the conditioned emotional response, then
requiring the test response at the time of the point of
maximum emotionality would tend to maximize the possi
bility of competing responses occurring at the time the
test response was required. It might be possible to
demonstrate the existence of such competing responses by
using an interval as indicated by Spence and Runquist
prior to the presentation of the test stimuli and observ
ing whether or not there were decrements in both error and
reaction time scores or whether, as was the case in this
experiment, reaction time scores improved while error
scores became poorer.
It is reasonable to assume in light of a statement
made by Spence, which was quoted in the introduction to
this report, that the instructions used in this study to
produce variations in incentive actually produced varia
tions in drive. If this be true, the results of the study
suggested that different experimental variations normally
subsumed under the drive construct may produce different
effects on behavior. The results may have indicated a
need for a distinction between the effects of drive as
produced by aversive stimulation or by stimulation
108
produced by responses that had been conditioned as a re
sult of aversive stimulation, on the one hand, and drive
as produced by prestige or achievement type motives, on
the other.
The results obtained when either error or trans
mission scores were used as the dependent measures support
a hypothesis advanced by Freeman (27), Hebb (30), and
Harris, Mackie, and Wilson (29). Duffy (20, 21) has pro
posed that the concepts of emotion and motivation should
be subsumed under a single concept of arousal or activa
tion j and Freeman and Hebb have proposed that the rela
tionship between performance and this level of arousal or
activation takes the form of an inverted-U. At low levels
of arousal, response strength increases or performance
improves as arousal is increased; performance reaches an
asymptote at some middle value of arousal; and as the
level of arousal is increased beyond the optimum level,
performance begins to get poorer. Stennett (61) obtained
results supporting this hypothesis. He found that under
low levels of arousal (when the subjects were performing
under the impression that the experimenter was simply
calibrating the equipment) performance was inferior, in
terms of the number of errors made, to performance when
the subjects were given encouragement by the experimenter
and small monetary rewards for improving their
109
performance. He also found that, as the level of arousal
was presumably increased by providing large monetary re
wards for improved performance, the number of errors again
increased over the optimum performance level.
The results of the study performed here can be con
sidered to represent additional support for the inverted-U
hypothesis if it is assumed that the low drive— high in
centive subjects were at the optimum level of arousal and
the high drive— high incentive subjects were at a greater
level of arousal. Since the variables employed in this
study, as well as those generally employed by Spence and
many others, have not been scaled, it is impossible to
make anything more than suggestive statements with respect
to the motivational continua involved in the various
studies. As a matter of fact, in the opinion of this
writer, one of the most fruitful areas for additional re
search would be the task of scaling various experimental
variables normally employed in studies designed to test
the Hull-Spence formulations or to test other motivational
hypotheses .
Recommendations for Future Research
The variations in drive used in this study were not
relevant to the incentive. The writer is not aware of any
evidence concerning the comparison of relevant drive and
incentive with irrelevant drive and incentive. The
110
results reported here do not, by themselves, indicate
whether or not such a distinction should be made. But, if
the results are accepted as a negative instance for the
Spence formulation, the use of a relevant drive might
yield results that could be accepted as a positive one.
The writer assumes, partly, at least, on the basis
of the results of the study reported here, that a dis
tinction should be made between the effects of relevant
and irrelevant drives on incentives. The following
hypotheses are proposed as a first approximation to the
distinction, and they are considered to be descriptive of
performance under very high drive levels. The writer
believes, along with Freeman (27), Hebb (30), and Harris,
Mackie, and Wilson (29), that high levels of arousal may
be associated with decrements in performance:
1. At low levels of relevant drive, the effects
on E of increases in K are additive.
2. At high levels of relevant drive, the effects
on E of increases in A are additive and
positively accelerated.
3. At low levels of irrelevant drive, the effects
on E of increases in K are additive, but the
slope of the function is not as great as when
the drive is relevant.
1 +, At high levels of irrelevant drive, the effects
Ill
on IS of increases in K are subtractive and
positively accelerated.
5. At any level of irrelevant drive, the effects
on E of a given level of IC are less than the
effects K would have if the drive were rele
vant .
6. The higher the level of the irrelevant drive,
the greater is the discrepancy between the
effects of l i acting jointly with it and with
a relevant drive of comparable magnitude.
These hypotheses, which are little more than
guesses based on incidental observations of human be
havior, are illustrated in Figure 1*+.
It is difficult when working with human Ss to
produce meaningful variations in drive, either relevant
or irrelevant, and incentive as they are experimentally
defined by Spence. Any variations produced should be
meaningful in the "real1 1 as well as the laboratory en
vironment. Of course, animals can be used, but observa
tions of animal behavior can lead, at best, only to
hypotheses concerning human behavior.
112
Reaction
uotential,
E
120
110
100
90
80
70
60
50
^0
30
20
10
0
relevant
relevant
irrelevant
Do
, irrelevant
Do
relevant
D1
irrelevant D]_
10 20 30 *K) 5o
Incentive, K
Figure 1>+. Hypothesized relationships between
drive, incentive and reaction potential, hypothetical
units
113
It is suggested that it would be worthwhile in any
future research on the effects of drive and incentive
motivation on behavior to determine whether or not rele
vant and irrelevant drive produce the same effects in
combination with different levels of incentive. It would
also be desirable to use more than two levels of each
variable, which is exceedingly difficult to do when human
Ss are concerned. Furthermore, it would be desirable to
scale the experimental variables to be used before attempt
ing to describe the relationships among them. Much of the
controversy concerning the Hull-Spence formulations re
sults, in this writer's opinion, from a failure to scale
the experimental variables in terms of their effects,
acting singly, on performance.
It is felt that the methods employed in this study
could be extended to make possible the testing of some of
the hypotheses presented above. More than two levels of
incentive might be achieved in this type of study by:
1. Taking away regularly scheduled liberty if
performance is poor.
2. Neither taking away scheduled liberty nor
providing the opportunity for extra liberty.
3* Providing an opportunity for different amounts
of liberty— one, two, or three 72 hour
liberties or two, three, four, and so on, days
lib
of liberty.
Relevant drive might possibly be varied by;
1. Having paced and unpaced test conditions.
2. Varying the discriminability of the test
stimuli.
3. Interjecting erroneous comments during the
test trials concerning the quality of the S's
performance.
Irrelevant drive might be varied by;
1. Using the conditioning procedure that was
employed in the study reported here.
2. Varying the intensity of the unconditioned
noxious stimulus.
3. Selecting different groups on the basis of
scores on the Taylor Manifest Anxiety Scale.
Obtaining Ss who would volunteer to undergo
differing periods of deprivation of food, water,
and so on.
In the writer's view, a great deal of profitable
research could be conducted in the area of motivation
using the Hull-Spence formulations as a model. While they
provide a convenient framework within which to work, it is
felt they will not stand up if human Ss are used and
meaningful variations in D and K are produced.
Chapter VI
Summary
Introduction
The purpose of this study was to investigate the
joint effects of the Hull-Spence motivational construct,
D or generalized drive, and incentive motivation, K, on
performance. Hull (33) made the assumption on the basis
of very little, if any, experimental evidence that D and
K combine multiplicatively in their action on H, the
associative or learning factor. This resulted in the
following formulation, where E, reaction potential, is the
construct corresponding to measures of performance:
E = H x D x K
Spence (56), on the other hand, made the assumption,
again on the basis of very little experimental evidence,
that D and K combine additively. His formulation is ex
pressed as follows:
E = H x (D+K)
In both of these formulations, D symbolizes
'generalized drive,1 which multiplies all habit structures.
When he introduced the concept of generalized drive, Hull
hypothesized that the total effective drive strength at
any moment consists of the relevant drive strength, D, and
t
the irrelevant drive strength, D, existing in the
116
organism. The distinction between these two types of
drives, relevant and irrelevant, concerns the relation
ship between the needs of the organism and the reinforcing
agent. Thirst, for example, would be a relevant need if
water were the reinforcing agent, but it would be irrele
vant if food were the reinforcing agent.
The experimental operations performed in this
study produced variations in irrelevant drive. It was
assumed, as Hull and Spence assumed, that the variations
in irrelevant drive resulted in variations in generalized
drive, D. The hypothesis tested was that D and K are
additive. It was predicted that the additive hypothesis
is too simple to account for the relationship between D
and K and that there is an interaction between D and K
such that when D is large, adding a will have less effect
than when D is small.
The experimental manipulations designed to produce
variations in K were probably not within the boundary con
ditions of the Hull-Spence formulations, in spite of the
fact that they are generally recognized in our society as
meaningful variations of human incentive. If they were
not within those boundary conditions, the study must be
considered as a test of an extension of the Hull-Spence
formulations.
117
Method
An effort was made to produce variations in K by
giving one half of the Ss, Wavy apprentice seamen, an
opportunity to obtain a three-day liberty and not giving
the other half a similar opportunity. An effort was made
to produce two levels of D by giving one half of the Ss
conditioning trials in which a 1000 cps tone, CS, was
paired with an electric shock, US, and then presenting the
tone to all Ss on all test trials. Thus, there were four
experimental groups; low drive— low incentive, DpK^, low
drive— high incentive, D]_K2; high drive--low incentive,
D2ivl$ and high drive--high incentive, D2&2*
The Ss were assigned randomly to the experimental
conditions to produce four groups of ten Ss each. All Ss
first received training instructions followed by train
ing trials. Five minutes after the training trials, they
were given additional instructions appropriate to their
drive-incentive conditions.
The Ss in the two high incentive groups were told
they would receive a three-day liberty if they performed
in the upper half of their group. The Ss in the low in
centive groups were told they would not have an oppor
tunity to obtain a liberty.
The Ss in the high drive groups were given 18 con
ditioning trials. On nine of these trials a 1000 cps
118
tone, CS, came on simultaneously with the visual task
stimulus. One-half second after the tone and visual
stimulus came on, the high drive Ss received a mild
electric shock on the forearm of their non-preferred hand.
The tone appeared on the same nine of the 16 con
ditioning trials for all groups. The tone also appeared
simultaneously with the visual stimulus on all 36 test
trials for all *+0 Ss. The low drive Ss were told prior to
the start of what were the conditioning trials for the
high drive Ss that they would not receive a shock, and they
did not at any time receive a shock.
The test trials followed immediately after the
conditioning trials with no pause and no additional
instructions. On nine of the 38 test trials, the high
drive Ss received the shock US. This partial reinforce
ment was designed to maintain the emotional response it
was assumed was developed during the conditioning trials.
The distinction hetween conditioning and test
trials is only made in the report of the experiment to aid
in explanation. The Ss were not aware of any such
distinction. The/ were instructed that what are described
here as the conditioning trials were test trials.
The Ss task was to discriminate among nine combina
tions of hue and brightness, three hues (red, green, and
yellow) each at three brightness levels, and select the
119
appropriate switch among nine toggle switches arranged in
three rows and three columns. Each column corresponded to
one of the three hues and each row to one of the three
brightness levels. All Ss were instructed to respond as
rapidly and accurately as they could.
Both the speed, in hundredths of a second, and
accuracy of each response were recorded. Performance data
for the nine test trials on which the high drive Ss re
ceived the shock were omitted from the analyses for all
Ss. The response records permitted the computation of
three dependent measures: the differences in information
transmission (T scores) from the last 27 training to the
27 test trials-, the differences between each S's median
reaction times on the last 27 training and the 2? test
trials (RT scores); and a combination of these two scores
that reflected improvement in information transmission
and improvement in reaction time from the training to the
test trials (TRT scores). Prior to the analyses, the T
and RT scores were adjusted so as to remove the variance
in them that could be predicted from the training scores.
The resulting T1 and RT' scores were converted to standard
scores before being multiplied together to obtain the
TRT* scores.
Results
Only one group, the low drive— high incentive
120
group, improved its mean T score from the last 27 training
to the 27 test trials. The mean T scores for the remain
ing three groups were less on the test trials than on the
training trials.
All groups increased their mean RT scores from the
training to the test trials.
The analysis of variance using the T' scores
indicated a significant interaction due to the joint
variation of the experimental variables designed to pro
duce different levels of D and K. At the low drive level,
the increase in h served to increase transmission per
formance. At the high drive level, the increase in Iv
served either to decrease or not to enhance transmission
performance. The main effect due to the variation in D,
the conditioning procedure, was significant; but the
direction of the effect was opposite to what would be pre
dicted from the Spence formulation. That is, the high
drive group improved its performance significantly less
than the low drive group. The main effect due to the
variation in K, providing an opportunity to obtain a
liberty, was not significant. However, significance
could not be expected since the interaction effect was
significant and the effects of K were in opposite
directions at the two drive levels.
The analyses of the RT* scores produced no
121
significant results. However, the differences in per
formance due to the variations in K and D were in the
directions that would be predicted from the Hull-Spence
behavioral equation.
The analyses of variance of the TRT1 scores showed
no significant results. An overall F test of the signifi
cance of the differences among the four treatment means,
however, was significant, and the performance of the low
drive— high incentive group was superior to all other
groups.
Conclusions
When information transmission scores, which reflect
the probability or accuracy of response, were used, the
effects due to variations in experimental variables
determining D and K were not additive. The effects of
variations in K were different at different levels of D.
When reaction time measures were used, differences
in performance were in the direction predicted from the
Hull-Spence behavior theory. However, none of the
differences was significantly different from zero.
When combined transmission-time scores were used,
the differences among the four group means was signifi
cant. The low drive— high incentive group was superior to
all others, contrary to what would be predicted from the
Hull-Spence formulation.
122
The validity, ox lack of validity, of the Spence ;
hypothesis concerning D ana K depended on the response
measure used. The results indicated a possible need for a
distinction to be made between speed and accuracy of res
ponse as indices of reaction potential, E.
There was a tendency for low drive Ss to sacrifice
speed of response to some extent for the sake of accuracy
of response. The high drive Ss appeared to attempt not to
sacrifice either at the expense of the other. It might be
concluded, therefore, that the experimental conditions
produced different sets toward the importance of speed and
accuracy as measures of performance.
The experimental manipulations were designed to
produce variations in irrelevant drive and, thus, also in
generalized drive, assuming both relevant and irrelevant
drives contribute not equally but in the same way to
generalized drive. It could well be, however, that if
variations in relevant drive instead of irrelevant drive
had been produced the results would have been different.
Nevertheless, the results obtained in this study indicate
either that generalized drive, D, and incentive, K, are
not additive (nor multiplicative for that matter), or that
irrelevant and relevant drives do not combine with in
centive in the same manner, or that speed (or latency) and
accuracy (or probability) of response are not related to
123
reaction potential in the same manner.
The methods employed in this study seemed to he
well suited to the task of determining the applicability
of many of the Hull-Spence formulations to human behavior.
It is recommended that additional studies be con
ducted using human Ss to determine whether or not relevant
and irrelevant drives combine with incentives in the same
manner. Studies could be designed and conducted in which
relevant and irrelevant drives as well as incentives were
varied simultaneously. It would be desirable to use more
than two levels of each variable, in spite of the
difficulties of such a procedure when human Ss are used.
It would also be desirable to scale all of the experi
mental variables to be used in terms of their effects on
performance before conducting any studies designed to
determine the interrelationships among different variables.
References
References
Amsel, A. The combination of a primary appetition-
al need with primary and secondary emotionally
derived needs. J. exp . Psychol.. 1950, L+0,
1 - 15-.
Amsel, A. The effect upon level of consummatory
response of the addition of anxiety to a
motivational complex. J. exp. Psychol., 1950,
1+0, 709 - 715.
Amsel, A. and Cole, Iv. F. Generalization of fear-
motivated interference with water intake.
J* » Psychol.. 1953? 1+6, 2*+3 - 2V7.
Amsel, A. and Hancock, W. Motivational properties
of frustration: III. Relation of frustration
effect antedating goal factors. J. exp.
Psychol., 1957} .13, 126 - 131*
Amsel, A. and Maltzman, I. The effect upon gener
alized drive strength of emotionality as in
ferred from the level of consummatory res
ponse. J. exp. Psychol. „ 1950, 1+0, 563 - 569-
Amsel, A. and Roussel, J. Motivational properties
of frustration: I. Effect on a running res
ponse of the addition of frustration to the
motivational complex. J. exp. Psychol.,
1952, i+3, 363 - 368.
Brown, J. S. Problems presented by the concept of
acquired drives. In Current Theory and
Research in Motivation: A Symposium. Univer.
Nebraska Press, 1953, PP• 1 - 10.
Brown, J. S. and Farber, I. E. Emotions concept
ualized as intervening variables--with
suggestions toward a theory of frustration.
Psychol. Bull., 1951, i±6, 5-65 - *+95*
Brown, J. S. and Jacobs, A. The role of fear in
the motivation and acquisition of responses.
J. exp. Psychol.. 195-9, 3i3, 75-7 - 759*
126
10.
11.
12.
13-
1>+.
15-
16.
17-
lo.
19-
20.
Brown, J. S., kalish, H. I. and Farter, I. E. The
condition fear as revealed by magnitude of
startled response to an auditory stimulus.
J. exp. Psychol.. 1951, 1+1, 317 - 326.
Buchwald, A. M. and Yamaguchi, H. G. The effect
of change in drive level on habit reversal.
1* exp. Psychol.. 1955, 50., 265 - 269.
Campbell, B. A. and kraeling, Doris. Response
strength as a function of drive level and
amount of drive reduction. J. exp. Psychol.,
1953, 25, 97 - 101.
Castaneda, A. Effects of stress on complex learn
ing and performance. J. exp. Psychol.. 1956,
52, 9-12.
Castaneda, A. and Palermo, D. 3. Psychomotor
performance as a function of amount of train
ing and stress. J. exp. Psychol., 1955, 50.
175 - 179-
Castaneda, A., Palermo, D. and PlcCandless, B.
Complex learning and performance as a function
of anxiety and task difficulty in children.
Child Development. 1956, 22, 327 - 332.
Child, I. L. personality. In C. P. 3tone (Ed.)
Ann. Rev. Psychol♦. Vol. 5, Stanford, Calif.:
Annual Reviews, 195^.
Cowles, J. T. and Nissen, N. H. Reward-expectancy
in delayed responses of chimpanzees. J. comp.
Psychol., 1937, 22, 3^5 - 356.
Crespi, L. P. quantitative variation of incentive
and performance in the white rat. Amer. J.
Psychol.. 19^2, 55, U-67 - 517.
Crespi, L. P. Amount of reinforcement and level of
performance. Psychol. Rev. . 19*+*+, 51, 3^1 -
3 57-
Duffy, E. The concept of energy mobilization.
Psychol. Rev., 1951, 55, 30 - *+0.
127
21.
22.
23.
2b.
25.
26.
27.
2b.
29 •
30.
31-
Duffy, Elizabeth. The psychological significance
of the concept of "arousal1 1 or "activation.1 1
Psychol. Hev., 1957, 6]+, 265 - 275*
Ellis, N. R. The immediate effects of emotionality
uoon behavior strength. J. exp. Psychol.,
1957, £±, 339 - 3>bb.
Farber, I. E. Response fixation under anxiety and
non-anxiety conditions. J. exp. Psychol..
i>5c, 3b, ill - 131.
Farber, I. E. Anxiety as a drive state. In
Nebraska Symposium on motivation. Lincoln,
Neb.: Nebraska Dniver. Press, 195^, 1 - ^5-
Farber, I. E. The role of motivation in verbal
learning and performance. Psychol. Bull..
1955, £2, 311 - 327.
Farber, I. E. and Spence, K. w. Complex learning
and conditioning as a function of anxiety.
J. exp. Psychol.. 1953, ^5, 120 - 125.
Freeman, G. L. The relationship between perfor
mance level and bodily activity level. J. exp.
Psychol., 19*+0, 26, 602 - 608.
Garner, W. R. and McGill, W. J. The relation
between information and variance analysis.
Psychometrica. 1956, 21, 219 - 228.
Harris, W. , Mackie, H. R. and Wilson, C. L.
Performance under stress: a review and
critique of recent studies. Technical Report
VI: Research on the development of performance
criteria. Los Angeles: Human Factors Research,
Incorporated, 1956.
Hebb, D. 0. Drives and the C.N.S. (Conceptual
Nervous System). Psychol. Rev., 1955, 62,
2^3 - 253.
Holder, W. B., Marx, M. H. , Holder, E. E. and
Collier, G. Response strength as a function
of delay of reward in a runway. J. exp.
Psychol., 1957, £3, 316 - 323.
128
32.
33.
3H.
35-
36 •
37.
36.
39-
HO .
Hi.
H2 .
H3.
Hull, C. L. Principles of Behavior. hew York;
Apple ton-Century-Cr of ts, I9P3.'
Hull, C. L. Essentials of Behavior. hew Haven;
Yale University rress, 19^1. ’
Hull. C. L. A Behavior System. New Haven: Yale
'University Press, 1950.
ivalish, H. I. The strength of fear as a function
of the number of acquisition and extinction
trials. J. exp. Psychol. , 195*+, 1-9*
ivendler, H. H. Drive interaction: I. Learning is
a function of the simultaneous presence of the
hunger and thirst drives. J. exp. .^sychol. .
19H5, 3l5? 96 - 109.
Kendler, H. H. Drive interaction: II. Experi
mental analysis of the role of drive in learn
ing theory. J. exp. Psychol. . 19^5, 3^j 3-66 -
198.
Ketchel, Rhoda. Performance in instrumental learn
ing as a function of shock intensity.
(Unpublished master's thesis. State University
of Iowa) Reported in Spence.
killer, N. E. Learnable drives and rewards. In
Handbook of Experimental Psychology, Ed. by
S. S. StevensT Pp. H35 -
killer, N. E. Studies of fear as an acquirable
drive; 1. Fear as motivation and fear re
duction as reinforcement in the learning of
new responses. J. exp. Psychol., 19^6, 38.
89 - 101.
Moltz, H. Resistance to extinction as a function
of variations in stimuli associated with
shock. J. exp. Psychol., 195H, H8, 123 - 126.
Montague, E. k. The role of anxiety in serial rote
learning. J. exp. Psychol. , 1953, i±5, 91 - 96.
Mowrer, 0. H. A stimulus response analysis of
anxiety and its role as a reinforcing agent.
Psychol. Rev., 1939, H6, 553 - 565.
129
bb.
i f 5.
1+6.
*+7 •
1+8 .
if 9-
50.
51.
52.
53-
5b.
Nissen, h. W. and Elder, J. II. The Influence of
amount of incentive on delayed response
performances of chimpanzees. J. genet.
Psychol., 1935, hi, *+9 - 72.
Palermo, D. 8. Proactive interference and
facilitation as a function of amount of train
ing and stress. J. exp . P s ychol. , 1957, 5.3 .
293 - 296.
Palermo, D., Castaneda, A. and McCandless, B. The
relationship of anxiety in children to perfor
mance in a complex learning task. Child
Development, 1956, 2£, 333 - 337*
Fassey, G. E. The influence of intensity of un
conditional stimulus upon acquisition of a
conditioned response. J. exp. Psychol.,
19*+8, *+20 - *+28.
Perin, C. T. Behavior potentiality as a joint
function of the amount of training and the
degree of hunger at the time of extinction.
J. exp. Psychol., 19^2, 30, 93 - 113*
Perin, C. T. A quantitative investigation of the
delay-of-reinforcement gradient. J. exp.
Psychol., 19^3, 12, 37 - 51-
Ramond, C. k. Performance in instrumental learning
as a joint function of delay of reinforcement
and time of deprivation. J. exp. Psychol.,
195*+, hi, 2lH6 2 250.
Ramond, C. K. Performance of selective learning as
a function of hunger. J. exp. Psychol., 195*+,
hi, 265 - 270.
Runquist, W. H., Spence, K, W. and Stubbs, D. W.
Differential conditioning and the intensity of
U.C.S. J. exp. Psychol., 1958, 21, 51 - 56.
Seward, J. P. Drive, incentive and reinforcement.
Psychol. Rev. , 1956, 63, 195 - 203*
Shannon, C. E. and weaver, W. The Mathematical
Theory of Communication. Urbana; University
of Illinois Press, 19^9 •
130
55-
56.
57.
5&.
59-
60.
61.
62.
63-
6*+.
65 .
Spence, A. W. A theory of emotionally based drive
(D) and its relation to performance in simple
learning situations. Am. Psychol., 1956, 13.
131 - 1^1.
Spence, A. W. Behavior theory and conditioning.
Yale university Press, 1956, New Haven, pp.
12*+, l*+8.
Spence, A. W. ana Runquist, W. N. Temporal effects
of conditioned fear on the eyelid reflects.
ex'P» Psychol., 1956, 553? 613 - 616.
Spence, A, A. and Taylor, Janet. Anxiety and
strength of the UCS as determiners of the
amount of eyelid conditioning. J. exp.
Psychol. , 1951, 1+2, 163 - 188.
Spence, A. W. , Farber, I. E. and HcFann, H. H.
The relation of anxiety (drive) level to
performance in competitional and non-competi-
tional paired associates learning. J. exp.
Psychol., 1956, 52, 296 - 305.
Stein, L. The classical conditioning of the
consummatory response as a determinant of
instrumental performance. J. comp. physiol.
Psychol., 1957, 50, 269 - 278.
Stennett, R. G, The relationship of performance
level to level of arousal. J. exp. Psychol.,
1957, 53+? 5*+ - 61.
Surwillo, W. W. Psychological factors in muscle-
action potential; BAG gradients. J. exp.
Psychol. , 1956, .52, 263 - 272.
Taylor, Janet A. The relationship of anxiety to
the conditioned eyelid response. J. exp.
Psychol., 1951? 3+1, 81 - 91*
Taylor, Janet A. Drive theory and manifest
anxiety. Psychol. Bull., 1956, 5-3? 303 - 320.
Taylor, Janet A. and Spence, A. W. The relation
ship of anxiety level to performance in serial
learning. J. exp. Psychol., 1952, 3 +3 +? 61 -
6*+.
Teel, K. S. Habit strength as a function of
motivation during learning. J. comp. Psychol.,
1952, 188 - 191.
Webb, W. B. Motivational aspects of an irrelevant
drive in the white rat. J. exp. Psychol.,
19^9, 32> 1 " ll+*
Williams, S. B. Resistance to extinction as a
function of the number of reinforcements.
J* exp. Psychol.. 1938, 23., 506 - 522,
Zeaman, D. Response latency as a function of the
amount of reinforcement. J. exp. Psychol.,
19^9, 32 j 5+66 - 1+86 •
Appendix A
Supplementary Data
133
Table A
Median Reaction Times in Hundreds of a Second Made by
the Ss in Each Experimental Group for the last 27
Training and the 27 Test Trials
Group I, Low Drive--Low Incentive
Subject Median Reaction Time Median Reaction Time
Lumber Last 27 Training Trials 27 Test Trials
1 ib3
116
2 16b Ibb
3
138 126
b
131
loV
5 125 111
6
179 139
7 175 125
8
u? 103
9
lib 91
10 129 10^
Group II, Low Drive--Eigh Incentive
Subject Median Reaction Time Median Reaction Time
Number Last 27 Training Trials 27 Test Trials
1
13^
116
2 I80
125
3
121 111
lb2
105
5 131
116
6 120
10*f
7 131
96
8 168
lb6
9 139 101
10 126 106
13*+
Table A (Continued)
Group HI, High Drive— Low Incentive
Subject Median Reaction Time Median Reaction Time
Number Last 27 Training Trials 27 Test Trials
1 H 9 131
2 152 103
3
156 1 H
b 190
13^
5
156 120
6
155
H o
7 19*+
i5 o
8 132 107
9
178 138
10 139 91
Group IV, High Drive--High Incentive
Subject Median Reaction Time median Reaction Time
Number Last 27 Training Trials 27 Test T]
1 123
92
2 198 130
3
131 120
U -
93
82
5
160 H o
6 H 7 107
7
136 93
8 306 198
9
127
106
10
118
107
135
Table B
Number of Errors Made by the Ss in
Each Experimental Group on the Last 27 Training
and the 27 Test Trials
Group I, Low Drive— Low Incentive
Subject Number of Errors Number of Errors
Number Last 27 Training Trials 27 Test Trials
1 7
8
2 5
6
3 5
2
h
8 10
5
h
7
6 7
6
7
8 11
8 5 3
9 9 5
10 6 12
Group II, Low Drive— High Incentive
Subject Number of Errors Number of
Number Last 27 Training Trials 27 Test 1
1
2
3
2 8 8
3
1 +
3
f+ 8 6
5 5 3
6
7
2
7
11 >+
8
7
2
9
7
8
10
13 5
136
Table B (Continued)
Group III, High Drive— Low Incentive
Subject Number of Errors Number of Errors
Number Last 27 Training Trials 27 Test Trials
1
8 11
2
9
lb
3
2
5
£
b 8
5 9 3
6 5
10
7 9 5
8 8 lb
9
6
5
10
b 8
Group IV, High Drive--High Incentive
Subject Number of Errors Number of Errors
Number Last 27 Training Trials 27 Test Trials
1 5 13
2 6
3
3
3
8
5
10
5
i+ 10
6
3
k
7 7
11
8 6
3
9
2 6
10 7 5

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

A Preliminary Study Of The Dimensions Of Future Time Perspective

PDF

The Effects Of Prior Part-Experiences On Visual Form Perception In The Albino Rat

PDF

Effects Of Task Performance Upon The Acoustic Reflex

PDF

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

PDF

Dominance And Contiguity As Interactive Determinants Of Autonomic Conditioning

PDF

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

PDF

The Effect Of Subject Sophistication On Ratio And Discrimination Scales

PDF

Recognition Accuracy In The Method Of Single Stimuli: A Test Of An Operational Definition Of The Distinctiveness Of Stimuli

PDF

Classical Discrimination Conditioning As A Function Of Probability Of Reinforcement

PDF

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

PDF

A Factor-Analytic Study Of Problem-Solving Abilities

PDF

The Influence Of Order Information And Stimulus Display Time On Short-Term Retention

PDF

A Factor-Analytic Study Of Military Leadership

PDF

On The Relationship Between Anxiety And Aggression In Nine-Year-Old Boys

PDF

Intellect After Lobotomy In Schizophrenia: A Factor-Analytic Study

PDF

A Factor Analysis Of The Semantic-Evaluation Abilities

PDF

Ucr Diminution As A Function Of Isi

PDF

Measures Of Pilot Performance: Comparative Analysis Of Day And Night Carrier Recoveries

PDF

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

Asset Metadata

Creator Buckner, Donald N. (author)

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

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

Format dissertations (aat)

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Guilford, Joy P. (committee chair), Grant, Homer H. (committee member), Grings, William W. (committee member), Longstreth, Langdon E. (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c18-33622

Unique identifier UC11357620

Identifier 5903060.pdf (filename),usctheses-c18-33622 (legacy record id)

Legacy Identifier 5903060.pdf

Dmrecord 33622

Document Type Dissertation

Format dissertations (aat)

Rights Buckner, Donald N.

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