A STUDY OF ANXIETY AND LEARNING

GEORGE MANDLER AND SEYMOUR B. SARASON

Vale University

HE present study is an attempt to investi- and strength of internal or covert drive states,
T gate the role of drive states in a testing
situation; more specifically, the extent to
which anxiety responses are evoked in the
In the present investigation certain aspects o£
the testing situation are emphasized: the
drive states and the resultant responses. The
situation and the relation of such responses following assumptions refer to the drives
to performance and learning. In addition evoked and types of responses elicited in the
to studying the role of anxiety, consideration test situation. These drives will be considered
is given to the effects of subjective success as falling into two major categories:
and failure on the anxiety state and test Assumption i. Learned drives which are a
performance. function of the nature of the task, test ma-
Comparatively little attention has been paid terials, and instructions. These include the
in the literature to the influence of various need to achieve and to finish the task; in
drive states on the performance of typical short, drives which evoke responses relative
intelligence test items. Thus, Mursell (8) to satisfying the requirement set by the task
mentions motivation only in comparing edu- or the experimenter. We shall refer to these
cational with mental tests, and Terman and as task drives (ST). It is assumed that these
Merrill (10) refer briefly to "hunger restless- drives are reduced by task responses (Ri),,
ness, desire to please . . ., timidity and a which are responses or response sequences
hundred other motives...." which lead to completion of the task.
Diethelm and Jones (3) found that the Assumption 2. A learned anxiety drive
presence of clinical anxiety significantly de- which is a function of anxiety reactions pre-
creased scores on the Kohs Block Design viously learned as responses to stimuli present
Test for most of their adult subjects, and in the testing situation. Anxiety is here con-
that maze learning was reliably slower under sidered as a response-produced strong stimulus
anxiety conditions. They conclude that the with the functional characteristics of drives
"possible influence of anxiety . . . in psy- as discussed by Miller and Dollard (7). Anx-
chological examinations must be considered iety reactions are generalized from previous
more seriously." Malmo and Amsel (6) in- experiences to testing situations. The anxiety
vestigated the effect of neurotic anxiety upon drive (SA) primarily elicits responses that
rote learning and demonstrated the role of tend to reduce the drive. These responses
anxiety-produced interference upon learning. are considered to be of two general types:
They concluded that the "forgetfulness of the (a) Anxiety responses which are not spe-
anxious patient is due to anxiety-produced cifically connected with the nature of the
interference between the relevant responses task or materials. These responses (desig-
and the irrelevant responses generated out nated as R A ) may be manifested as feelings
of the patient's anxiety state." of inadequacy, helplessness, heightened so-
In the present study, in contrast to previous matic reaction, anticipations of punishment
ones (i, 6), the measure of anxiety was based or loss of status and esteem, and implicit
on a questionnaire which was specifically attempts at leaving the test situation. It
concerned with the subjects' attitudes and ex- might be said that these responses are self
periences in a testing situation. rather than task centered.
(b) Anxiety responses which are directly
THEORETICAL CONSIDERATIONS AND HYPOTHESES related to the completion of the task and
In a previous paper Sarason (9) has pointed which reduce anxiety by leading to comple-
out that in both theoretical and practical ap- tion of the task. These responses (desig-
proaches to the testing situation insufficient nated RAT) are functionally equivalent to
attention has been given to the nature, role, RT responses. More specifically, intervening
166
 A STUDY OF ANXIETY AND LEARNING 167
responses rat and rt lead to the final response each trial and when improvement in succeed-
jK-X' ing trials is an easily observable cue, more
Assumption 3. Responses of the type RT responses of the order RAT will be reinforced
and RAT lead, in point of time, to task com- than responses of the order RA. Such final
pletion, while RA type of responses interferes task completion responses (RAT) will be anx-
with task completion. It is assumed that in- iety reducing and will become progressively
tervening ra responses are incompatible with stronger and dominant.
rt and rat responses. Hypothesis 3. When improvement in a
Assumption 4. RAT responses, which are task from trial to trial is not readily apparent
considered to be specific to the task, are not to the subject, responses of the order RAT
available in the response repertory of an indi- will not be substantially reinforced and more
vidual, but are evoked and learned in the random responses of the order RA will be
course of task performance. RA responses, evoked, especially if the anxiety drive is strong
which are not considered to be specific to to begin with.
the task, are available in the response reper- Hypothesis 4. When the dominant response
tory and are, by the process of generalization, tendency has been of the order RA (responses
readily evoked. Thus, individuals with a not relevant to the task), then any further
high anxiety drive and a large number of RA reference to the test situation (such as suc-
anxiety responses in their response repertory cess or failure) will evoke predominantly RA
will tend to make more RA responses ini- responses.
tially than individuals with a low anxiety Hypothesis 5. When there is no initial
drive and, conversely, in relation to the total tendency to make RA responses and when
number of responses available, individuals RAT responses have been predominantly
with a low anxiety drive will tend to make evoked in a task, then presentation of a new
more task relevant anxiety responses (RAT) but similar task preceded by a reference to the
than those with a high anxiety drive. test situation will elicit further RAT responses.
Figure i is a diagrammatic presentation of
the drives and responses evoked in the testing METHOD
situation. Section I
STIMULUS INTERVENING FINAL One hundred and fifty-four students in an intro-
ductory psychology course in Yale College were-
SITUATION RESPONSES RESPONSES given an anxiety questionnaire.1 The questionnaire
was presented to the students as a "questionnaire-
on attitudes toward test situations." It consisted
of 67 questions and was divided into four sec-
tions: (a) individual intelligence tests, (b) group
TASK DRIVES' intelligence tests, (c) course examinations, and (d)
-••it - ' general questions.
C RELEVANT Each section contained questions ("anxiety ques-
A
TO TASK > tions") dealing with the student's subjective ex-
ANXIETY periences in the testing situation such as un-
DRIVE easiness, accelerated heartbeat, perspiration, emo-
C NOT RELEVANT tional interference, and "worry" before and during-
TO TASK ) a testing session. The questionnaire also contained
FIG. I. DIAGRAMMATIC PRESENTATION OF DRIVES AND questions relating to attitudes (likes and dislikes)
RESPONSES EVOKED IN THE TESTING SITUATION toward tests. The 5s were requested to mark their
answers for each question anywhere along a
15-centimeter, graphic scale with endpoints and
These assumptions lead to the following midpoint indicated. Only nonveteran sophomores
hypotheses: and juniors who completed the questionnaire were
Hypothesis i. When an anxiety drive (SA) included in this study. The resulting 101 question-
evokes intervening responses predominantly naires were then scored by assigning a score, ex-
pressed in millimeters, to each question. The
of the nature rat, learning will, because of resulting 101 scores for each question were then
summation of drives (ST and SA), proceed tabulated and a subject was given a score of o if
at a faster rate than under the condition when 1
The authors wish to thank Dr. Claude E. Buxtoti
incompatible ra and rat responses are evoked. and the students of his introductory psychology course
Hypothesis 2. When a task is completed in for their wholehearted cooperation.
i68 GEORGE MANDLER AND SEYMOUR B. SARASON
he fell below and a score of i if he fell above given the Wisconsin Card Sorting Test2 (2, 4)
the group median. A total score for each subject and informed at its conclusion that this was not
based on the o and i scores was then determined part of the intelligence testing but an additional,
for 42 anxiety questions and 25 attitude questions. purely exploratory, study.
For the purpose of this study only the anxiety
scores were used. The distribution for each of the Section III
42 questions was then compared with the At this point the experimenter secretly opened
distribution based on all 42 questions. For an envelope, previously prepared for each subject,
this purpose the subjects were divided into informing die experimenter as to which of the
two groups at the total score median, and the three experimental subgroups (success, failure, or
consistency with which each of the 42 questions neutral) the subject belonged. The procedure
divided the subjects into these two groups was was then as follows:
determined by chi-square. All questions which Success group. The subject was informally told
divided subjects on this basis at better than chance that he had done extremely well on the Kohs
expectancy (at the .03 level of confidence) were Block Design and Digit Symbol tests. The ex-
retained. Five questions did not fulfill this condi- perimenter looked through several sheets of paper
tion and were discarded. From the remaining 37 (apparently score and norm data), leafed through
questions each subject obtained a revised score Wechsler's The Measurement of Adult Intelligence,
theoretically ranging from o to 37. The split- and finally informed the subject that he had scored
half reliability (odd vs. even questions) of the at approximately the 93rd to g6th percentilc for
anxiety questionnaire was .91 (Spearman-Brown a comparable college population. This was con-
prophecy estimate). cluded by the experimenter's saying: "Let's see
The final anxiety questionnaire scores yielded a dis- whether you can do as well in the second part."
tribution of the 101 subjects along a presumed anxiety Failure group. The procedure was the same as
continuum. Forty-two subjects with the 21 lowest for the success group except that the subject was
.and 21 highest scores were then selected for the told that he had done much worse than would be
actual study, i.e., one high-anxiety group (desig- expected from his aptitude scores, and that his
nated HA) and one low-anxiety group (LA). performance fell below the 10th percentile. The
Each of these two groups was divided by random experimenter then said: "Let's see whether you
procedure into 3 subgroups (Success, Failure, and can do better in the second part."
Neutral). Approximately ^A months after the Neutral group. The experimenter only said:
administration of the questionnaire these students "Let's go on to the second part."
were asked to participate in a psychological experi-
ment concerning intelligence tests. Apparently none Section IV
•of the subjects was aware of any connection be- Each subject was then presented with six trials
tween the questionnaire and the testing session. of the Kohs Block Design #16 and a specially
prepared Digit Symbol test of the same form as
Section II the Wechsler-Bellevue subtest but widi different
symbols. Procedure and scoring were the same
At the beginning of the testing session the sub- as described in Section II. At the conclusion of
ject was told that he was going to be given a the session each 'subject was fully informed of the
number of intelligence tests as part of a study purpose and design of the experiment and asked
•relating to the aptitude tests given to every Yale to fill out a short self-rating schedule on anxiety
freshman. He was also told that his scores in and quality of performance. He was also requested
the session would be compared with his aptitude not to discuss the experiment with other students
test scores to make possible a better interpretation until the end of the testing sessions.
•of the latter. He was presented with the Kohs Of the initial 42 subjects, 9 (5 HA and 4 LA
Block Design Test and asked to do sample design subjects) failed to participate in the testing sessions.
#2 of the Wechsler-Bellevue modification. He The final design and distribution of subjects are
was then presented with design #13 from the Kohs presented in Table i. The procedure was de-
Block Design Test (5) and the total time (in signed in such a way that the experimenter could
•seconds) needed for completion was recorded. The not know to which of the two anxiety groups any
subject was then presented with the Digit Symbol subject belonged.
subtest from the Wechsler-Bellevue Intelligence
'Test (n), doing the practice part first. A time RESULTS AND DISCUSSION
limit of one minute was used for the main part,
scoring number right minus number wrong. The In an attempt to establish face validity o£
subject was then again presented with Kohs Design the questionnaire, the experimenter (without
#13 and, after that, again with the' Digit Symbol knowledge of the subject's questionnaire
test. This was continued until each S had com- standing) rated each subject's behavior on a
pleted six trials of each of the tests. Six trials five-point scale according to five criteria of
were chosen in order to obtain both a representative
performance curve and to minimize boredom that overt anxiety manifestation (perspiration, ex-
•could appear with a large number of trials. After 2
The analysis of the subjects' performance on this
the completion of these trials each subject was test will not be presented in this study.
 A STUDY OF ANXIETY AND LEARNING
TABLE 1
EXPERIMENTAL DESIGN AND DISTRIBUTION OF SUBJECTS (N)

HIGH-ANXIETY GROUP LOW-ANXIETY GROUP

(N=i8) (N=I 5 )

PART I 6 trials Kohs Block Design Test (Design No. 13)

6 trials Digit Symbol Test (Wechsler-Bellevue subtest)

PART II Wisconsin Card Sorting Test

HA» HA HA LA LA LA
SUCCESS FAILURE NEUTRAL SUCCESS FAILURE NEUTRAL
(N=5) (N=6) (N=7) (N=5) (N=5) (N=5)
PART III **
6 trials Kohs Block Design Test (Design No. 16)
6 trials Digit Symbol Test (new form)

* HA=High anxiety group, LA=Low anxiety group.

** The 5s in each of the two anxiety groups were assigned by random procedure to the 3 subgroups (success,,
failure, neutral). For the purpose of statistical analysis only five 5s in each subgroup were used.

cessive movement, inappropriate laughter and The following predictions arise out of the
exclamations, questioning of instructions, assumptions and hypotheses stated previously:
hand movement). These ratings were di- Prediction i (Hypotheses i and 2). The
chotomized into a high anxiety and a low high anxiety group (HA) is expected to have
anxiety group and compared with question- higher mean time scores and greater varia-
naire ratings. The agreement was significant bility than the low anxiety group (LA) on
at better than the .001 level of confidence with the Kohs Block Design Test because of the
a point correlation (phi) of .59. evocation of more responses of the order RA.
In line with the analysis presented in the Table zA and Fig. 2 and 3 show the mean
introduction, the high anxiety group can scores and variability for the two groups in.
be characterized as individuals with a high the six trials of Part I. The mean score dif-
anxiety drive in testing situations and these ferences all lie in the predicted direction,,
subjects would be expected to react to the ^-values for the first five trials ranging from
testing situation with a large variety of in- .04 to .14. The variability of the HA group-
tervening responses of the order ra (responses is significantly larger at better than the .001
not relevant to the task). For the low anxiety level of confidence for trials 2 to 5. Trials-
group the anxiety drive is much weaker. i and 6 will be discussed separately below.

TABLE 2A
MEAN TIME SCORES AND STANDARD DEVIATIONS FOR BOTH ANXIETY GROUPS FOR Six TRIALS
OF FIRST KOHS DESIGN

HIGH ANXIETY GROUP Low ANXIETY GROUP

TRIAL / P* F P*
MEAN SD MEAN SD

i 199.4 114 158.2 107 1.07 .14 I.I5 >.o5

2 in. i 78 82.6 25 I. 3 6 .08 9.84
3 76.1 36 65.1 12 1. 13 •13 8.83 <.OOI
4 76.2 35 58.9 12 1.86 .04 8.06 <.OOI
5 74.4(60.5)** 63(30)** 55-4 II 1.17 .12 32.88 <.OOI
6 58.9 20 54-3 12 0.78 .22 2.44 .05

•One tail of the distribution of / used.

** Figures in parentheses exclude one extreme case, see also Fig. 2 and 3. The respective / and F values aret-
2=0.94 P=.i7
F=7-49 p=.ooi
I70 GEORGE MANDLER AND SEYMOUR B. SARASON

200
150-

• • LOW ANXIETY GROUP » oLOW ANXIETY GROUP

( N • 6) < N- 15 )
W 150- \ - - •» HIGH ANXIETY GROUP •- - «HIGH ANXIETY GROUP
( N.I8) t N-18 ]
•i loo-

'100

50
25

I 2 J 4 5 6 1 2 3 4 5 6
TRIALS TRIALS
Fie. 2. PERFORMANCE CURVE FOR HIGH ANXIETY AND FIG. 3. STANDARD DEVIATIONS FOR HIGH ANXIETY
Low ANXIETY GROUPS FOR Six TRIALS OF FIRST AND Low ANXIETY GROUPS FOR Six TRIALS OF
KOHS BLOCK DESIGN. (X=MEAN SCORE FIRST KOHS BLOCK DESIGN. (X=STANDARD
FOR HIGH ANXIETY GROUP INCLUDING DEVIATION FOR HIGH ANXIETY GROUP
ONE EXTREME CASE.) INCLUDING ONE EXTREME CASE.)

Prediction 2 (Hypotheses i and 3). The the .05 to .17 level of confidence. Figure 5
mean scores on the Digit Symbol test (where shows the predicted rise in variability for the
high scores indicate better performance) are HA group accompanied by an apparent drop
•expected to be lower for the HA group in in variability for the LA group. None of the
line with Prediction i. However, variability differences in variability between the two
is expected to rise for the HA group in line groups was significant. In order to test the
with Hypothesis 3 since the Digit Symbol difference between the two anxiety groups
test was never completed by any subject and in variability change from trial i to trial 6,
improvement in this test was slight for suc- the following statistical procedure was fol-
cessive trials. The variability of the HA lowed. In the HA group the 18 5s were as-
groups should increase because of the absence signed randomly to one of six groups. For
of anxiety reduction resulting from comple- each of these groups of three subjects an esti-
tion of the task at hand. Table 3 and Fig. mate of variability was obtained for trials
4 and 5 show the scores for the Digit Symbol i and 6. The variability on trial 6 was then
test. The mean scores for the first five trials subtracted from that on trial i, resulting in
result in differences which are significant at six values. The same procedure was employed

TABLE 2B
MEAN TIME SCORES FOR THE EXPERIMENTAL SUBGROUPS ON THE FIRST TRIAL OF SECOND KOHS DESIGN **

HIGH-ANXIETY GROUP LOW-ANXIETY GROUP

t P*
MEAN ADJUSTED MEAN MEAN ADJUSTED MEAN

Neutral 104.8 126.0 143.4 165.6 1.46 .08

Success 197.6 163.6 106.0 138-0 0.94 •i?
Failure 204.8 162.2 146.2 108.6 1.98 .03

* See footnote to Table 2A.

** Tables 26 and 2C are based on an analysis of covariance of the time scores on trial i of the second
Kohs design for the six experimental subgroups. The Ss' scores on trial I of the first Kohs design were used
:as a statistical control for initial differences in Kohs Block Design performance. The resulting ^-values have 23
•degrees of freedom.
 A STUDY OF ANXIETY AND LEARNING 171
TABLE 2C
SIGNIFICANCE OF DIFFERENCE BETWEEN MEAN TIME SCORES OF EXPERIMENTAL SUBGROUPS' WITHIN THE
Two ANXIETY GROUPS **

HIGH-ANXIETY GROUP LOW-ANXIETY GROUP

t P* t t*
Neutral vs. success 1.38 .09 1.02 .16
Neutral vs. failure 1.34 .09 2.10 .02
Success vs. failure 0.05 I .01 .16

* See footnote to Table zA.

** See footnote to Table 28.

for the LA group, except that here there were is least apparent. However, when the Neu-
five groups with three subjects in each. A tral groups (no intervening stimulation be-
/-test of the difference in variability change tween parts I and III) are compared, this
between the HA and LA groups was signifi- predicted difference is striking. The HA
cant at the .04 level of confidence. Thus, group, which has been performing at a lower
when compared with the low anxiety group, level than the LA group for the six previous
the high anxiety group showed a significant trials, has a mean time score which is better
increase in variability. (at the .08 level of confidence) than the LA
Prediction 3 (Hypotheses i and 2). In the group on trial i of the second design.
Kohs Block Design Test, where each trial Prediction 4 (from Hypothesis 4). For the
was necessarily concluded by successful com- HA group any further reference to the test-
pletion and where improvement is easily ap- ing situation in terms of success or failure
parent, RAT responses are strongly reinforced will elicit internal responses which will in-
and become increasingly stronger as com- crease the strength of the anxiety drive and
pared to RA responses elicited by the anxiety evoke random responses not relevant to the
drive. It can be predicted that eventually the task (RA). Thus, the neutral HA group is
high anxiety group will be equal to or better expected to perform better than either the
than the low anxiety group because of the failure or success group. Their mean time
summation of drives SA and ST. This is score on the Kohs Block Design Test is
partly confirmed by trial 6 of the first part, lower, but this difference is not highly sig-
where the difference between the two groups nificant (at the .09 level of confidence). There

TABLE 3
MEAN SCORES AND SD's FOR BOTH ANXIETY GROUPS FOR Six TRIALS OF DIGIT SYMBOL TEST
(WECHSLER-BELLEVUE SUBTEST) WITH TIME LIMIT OF ONE MINUTE

HIGH ANXIETY GROUP Low ANXIETY GROUP

t P*
TRIAL
MEAN SD MEAN SD

i 38.5 4-4 40-3 6.6 ' 0.95 •17

2 41.8 5.2 44-3 7.0 1.19 .12
3 44-4 5-3 47-7 6.3 1.64 .05
4 47-9 5-5 49.9 5-9 I. 00 .16
5 49-3 5-7 51 .5 5.8 i. 09 .14
6 51-3 7.2 52.3 5-3 0.45 •33

* One tail of the distribution of t used.

(None of the F-values testing the significance of difference of the variances of the two groups on the first
6 trials was significant (p values >.05 throughout). None of the t- or F-values comparing the first trials of
the second Digit Symbol test of the 6 subgroups was significant.)
172 GEORGE MANDLER AND SEYM«UR B. SARASON

50

45..

LOW ANXIETY
GROUP • LOW ANXIETY GTOUP
< N- 15) < N» 15 )

-• HIGH ANXIETY • HIGH ANXIETY CWOUP

GROUP I N - 18 )
(N-B)

35
3 4
1 2 3 4 .
TRIALS
TRIALS
FIG. 5. STANDARD DEVIATIONS FOR HIGH ANXIETY
FIG. 4, PERFORMANCE CURVE POR HIGH ANXIETY AND AND Low ANXIETY GROUPS FOR Six TRIALS OF
Low ANXIETY GROUPS FOR Six TRIALS OF DIGIT DIGIT SYMBOL TEST (WECHSLER-BELLEVUE
SYMBOL TEST (WECHSLER-BELLEVUE SUBTEST, SUBTEST, TIME LIMIT OF ONE MINUTE)
TIME LIMIT OF ONE MINUTE)

is no difference, or trend toward a difference, groups are not differentiated by the over-all
between the success and failure groups. median when the two groups are thrown to-
Prediction 5 (from Hypothesis 5). Any gether. A chi-square test allowed the rejec-
further reference to the testing situation in tion of the null hypothesis at the .03 level of
terms of success and failure is expected to confidence.
improve the performance of the LA group No data have been given as to the effect
due to the summation of drives SA and ST. of the different types of report on performance
Table zC shows that the difference between on die Digit Symbol test. All the relevant
the neutral and failure groups is significant comparisons were not significant, which may
(at the .02 level of confidence), whereas the be due to a lack of sensitivity and small range
mean time score of the success group lies of scores on this test.
between the other two groups.
CONCLUSIONS
Further Observations The results of the present study suggest that
A comparison of the two failure subgroups anxiety present in the testing situation is an
shows that the low anxiety group is signifi- important variable in test performance. It
cantly better than the high anxiety group is questionable whether intelligence test scores
(at the .03 level of confidence). This also adequately describe the underlying abilities of
lends support to the hypothesis that pre- individuals who have high anxiety drive in the
sumably anxiety-evoking stimulation can have testing situation, particularly since the relation
a differential effect (interfering or supporting) of the type of the test to the test performance
on different individuals. seems to play an important determining role.
It appears that, in general, the high anxiety Furthermore, anxiety does not necessarily de-
group performs better when no further ref- press scores, but can serve to elicit improve-
erence is made to their performance, while ment. Further study of the nature of the
the low anxiety group profits by such ref-
various anxiety responses is necessary, par-
erences, apparently in the direction of in-
ticularly the two types of responses (compati-
creased motivation.
The lack of significant differences between ble and incompatible with test responses)
the two groups on trial i of the first Kohs suggested in this study. In demonstrating the
Design could be attributed to the factor of role of anxiety-produced interference, no at-
unfamiliarity which results in large variability tempt has been made to study the origin and
for the two groups. However, the null hy- establishment of the interfering anxiety re-
pothesis was tested that the HA and LA sponses.
 A STUDY OF ANXIETY AND LEARNING 173
Further research into the factors discussed 1. The mean time scores on die Kohs Block
above is necessary and at present is being Design of the low anxiety group were better
undertaken by the authors. Among the more dian those of the high anxiety group for
important variables being subjected to further the first five trials.
scrutiny is the specific effect of instructions, 2. The variability of the high anxiety group
i.e., does die test situation per se produce the was significantly larger than diat of the low
differences between the two groups or is anxiety group.
this difference a function of specific instruc- 3. As the learning process proceeded, the
tions given by the examiner? Another vari- anxiety drive of the high anxiety group
able subject to further experiment is die factor tended to improve performance scores.
of finishing a task and improving on suc- 4. An intervening report (success or fail-
cessive trials. Furthermore, the influence of ure) elicited improved performance for the
familiarity with die test material and its low anxiety group but depressed scores for
•effect upon differences between the two anx- the high anxiety group.
iety groups appears to be an important It appears that the optimal conditions
variable in interpreting test results in the for a high anxiety group are those in which
light of the data presented in this study. no further reference is made to the testing
Another problem presented by our data is situation, and that the optimal conditions
the differential effect of success and failure. for a low anxiety group are those in which
It is presumed that this effect is to some the subjects are given a failure report.
degree dependent upon S's own evaluation The above processes were discussed in
of his performance. Finally, it is suggested terms of responses relevant to the task and
that an examination of S's previous experi- responses not relevant to the task, both types
ences and performances in testing situations, being evoked by an anxiety drive.
as well as specific personality factors and life Further research problems arising out of
history data, may be valuable in shedding the present study are discussed.
further light on this problem.
REFERENCES
SUMMARY 1. BEIER, E. G. The effect of induced anxiety on some
The present study was designed to investi- aspects of intellectual functioning; a study of the
relationship between anxiety and rigidity. Amer.
gate the influence of anxiety, as evoked by a Psychologist, 1949, 4, 273-274. (Abstract.)
testing situation, on the performance of typical 2. BERG, E. A. A simple objective technique for meas-
uring flexibility in thinking. /. gen. Psychol.,
intelligence test items. Relevant hypotheses 1948, 39, 15-22-
were postulated on the basis of S-R behavior 3. DIETHELM, O., & JONES, M. R. Influence of anxiety
theory. on attention, learning, retention and thinking.
Arch. Neural. Psychiat., Chicago, 1947, 58, 325-
Two groups of 5s (a high anxiety group 336.
and a low anxiety group), all sophomore and 4. GRANT, D. A., & BERG, E. A. A behavioral analysis
junior college students, were selected on the of degree of reinforcement and ease of shifting
to new responses in a Weigl-type card sorting
basis of an anxiety questionnaire and each problem. /. exp. Psychol., 1948, 38, 404-411.
group was randomly divided into three sub- 5. KOHS, S. C. Intelligence measurement. New York:
groups (success, failure, and neutral). All Macmillan, 1923.
6. MALMO, R. B., & AMSEL, A. Anxiety-produced
subjects were given six trials each on the interference in serial rote learning with observa-
Kohs Block Design #13 and die Digit Sym- tions on rote learning after partial frontal lobot-
bol subtest of the Wechsler-Bellevue Intelli- omy. /. exp. Psychol., 1948, 38, 440-454.
7. MILLER, N. E., & DOLLARD, J. Social learning and
gence Test. The subjects were then told either imitation. New Haven: Yale Univ. Press, 1941.
that they had done very well, or very badly, 8. MURSELL, J. L. Psychological testing. (2nd Ed.)
New York: Longmans, Green, 1949.
or were told to go on to the second part. 9. SARASON, S. B. The test situation and the problem
In the second part all 5s were give six trials of prediction. /. din. Psychol., 1950, 6, 387-392.
each of the Kohs Block Design #16 and a 10. TERMAN, L. M., & MERRILL, M. A. Measuring
intelligence. Boston: Houghton Mifflin, 1937.
comparable variation of the Digit Symbol 11. WECHSLER, D. The measurement of adult intelli-
Test. gence. (3rd Ed.) Baltimore: Williams and
The following predictions based on the Wilkins, 1944.
theoretical hypotheses were borne out: Received February 9, 1951.