You are on page 1of 6

Journal oj Experimental Psychology

1971, Vol. 87, No. 3, 367-372

LEARNED HELPLESSNESS IN HUMAN SUBJECTS:


JERRY W. THORNTON2 AND PAUL D. JACOBS
University of Oklahoma
The present investigation attempted to test the learned helplessness hypothesis
with human 5s. Four groups differing in shock contingency were given a fixed
level of shock, while a second four groups were given a variable level of shock.
Thirty choice reaction time training trials were given in which one group could
avoid shock, a second and third group received inescapable shock yoked to the
escapable group (one with a training task to perform and one without), and the
fourth performed the task but with no shock. Ten test trials, in a completely
different task, followed in which all 5s could avoid shock. Results revealed in
all phases of the experiment that variable shock was superior to a fixed level as
a stress inducer. Learned helplessness was offered as an explanation to the
yoked group's lack of responses in test trials. Implications for investigation of
learned helplessness in humans and for the use of variable shock were discussed.

In recent review of the literature on un- Further investigations have shown that
controllable aversive events, Seligman, prior training with escapable shock serves
Maier, and Solomon (1969) maintain that to immunize 5 against any interference of
5s, in addition to learning that a particular subsequent helplessness trials (Overmier
response controls reinforcement, are cap- & Seligman, 1967) and that the helpless-
able of learning that they cannot control ness syndrome is most effectively alleviated
reinforcement, a phenomenon Overmier by a therapeutic procedure of forced re-
and Seligman (1967) have labeled sponding (Seligman, Maier, & Geer, 1968).
"learned helplessness." Since all investigations conforming to the
The primary concern of the present study helplessness paradigm have used animals as
involves a typical learned helplessness 5s, the present investigation attempted to
paradigm where 5 is first given training quantify and validate helplessness in hu-
trials of inescapable/unavoidable shock mans. One problem inherent in such an
followed by test trials in which escape/ endeavor is that of S stressor levels. The
avoidance is possible. There has been a acquisition phase of all tests of helplessness,
variety of experiments using this procedure, thus far, has involved near traumatic shock.
e.g., where the nature of the inescapable To increase shock in humans to a traumatic
shocks differ, differing situations in which level is far from practical, not to mention
the shocks are given, variable time between ethical considerations. The typical stress-
the inescapable shock and escape/avoid- set instructions involve subjective setting
ance' responses, and differing species (An- of the stress level according to 5"s percep-
derson, Cole, & McVaugh, 1968; Braud, tion of an unpleasant but not painful level.
Wepman, & Russo, 1969; Overmier & Thornton and Jacobs (1970) call to mind
Seligman, 1967 ; Pickney, 1967; Seligman & contradictory reports using the subjective
Maier, 1967). setting of shock levels. The selection
The phenomenon has been replicated and/or adjustment of stressor levels has
with a variety of frequencies, densities, perhaps some of the most consequential
durations, and temporal distributions of effects on research of human performance
inescapable shock; and despite the multi- under stress (Thornton & Jacobs, 1970).
plicity of experimental procedures, a con- Previous research (Thornton & Jacobs,
sistent picture of "helplessness" emerges. 1970) suggests that the subjective setting
1
This article was based on a doctoral dissertation of a shock level may have inherent prob-
at the University of Oklahoma by the first author. lems; e.g., after 5 denotes a fixed level of
2
Requests for reprints should be sent to Jerry W.
Thornton, now at Angelo State University, Depart- shock to be unpleasant but not painful, he
ment of Psychology, San Angelo, Texas 76901. is aware that the shock will never be of
367
368 JERRY W. THORNTON AND PAUL D. JACOBS

greater intensity. By so defining the situa- The training apparatus were located in three dis-1
tion, S gains predictability, thus reducing tinctly separate rooms, each of which contained
identical choice reaction time units. Each unit con-
his stressed state. D'Amato and Gumenik sisted of a display board on which was located, at the
(1960) and Pervin (1963) have noted that top, a yellow warning light followed below by three
not only does S favor predictability in his horizontal green stimulus lights. The 5's manipu-
environment, but that increased predict- landum was a panel containing three buttons, each
ability appears to reduce to some extent a of which corresponded vertically to the stimulus
lights. A button in front of the stimulus buttons
stressed state. Also, this procedure for served the purpose of insuring 5's hand would be in
adjusting shock leaves E entirely dependent a consistent place preceding a trial. One-half-in.
on the willingness of 5 to "play the game," silver shock electrodes were located in two of the
so to speak, in adjusting a truly unpleasant three rooms containing the training apparatus, and
in a fourth. The fourth room contained only a chair
level of shock. and shock electrodes.
The present investigation attempted to Located in four distinctly separate rooms were the
eliminate these problems by using a vari- test apparatus, each of which consisted of an 8-in.-sq.
able range of shock in anticipation of reduc- box containing a light surrounded by seven white
buttons. Depression of both the first and fifth but-
ing 5 predictability over stressor amount, tons in any fashion broke all circuits (i.e., light,
thereby maintaining a facilitated stress shock, and clock). Silver shock electrodes were
state. located in all four test rooms.
It was hypothesized that groups adminis- Shock, generated by a Grason-Stadler 350-v. con-
stant-current shock apparatus, was delivered in
tered learned helplessness training would training trials for 1-sec. duration, with avoidance
perform significantly poorer (i.e., greater possible but not escape, and in test trials for 3 sec.
latency) on test trials, due to facilitated constant, with both escape and avoidance possible.
interference transfer, than groups adminis- Reaction time was recorded by four Standard timers
tered acquisition and test trials of escape/ (one-hundredth of a second).
Procedure.—Upon entering the laboratory, 5s
avoidance or no shock. It was further were assigned to rooms according to the PSI assign-
hypothesized that a stressed state would be ment variable. The assignment variable, involving
more resistant to extinction with a variable eight groups, consisted of a factorial combination of
level of shock than with a fixed level of 4 Shock Contingency Groups X 2 Stress-Set In-
structions. The 4 shock contingency groups in-
shock, with the former supporting helpless- cluded: (a) experimental reaction time (ERT) 5s,
ness in humans due to the increased and who could avoid shock during training trials; (b)
maintained stress level and the latter not yoked reaction time (YRT) 5s, who received un-
supporting the helplessness hypothesis due avoidable yoked shock to ERT 5s while performing
to a dissipation of the stress state early in the training task; (c) yoked (Y) 5s, who received
yoked unavoidable shock to ERT 5s, without a
acquisition. training task to perform; and (d) control reaction
time (CRT) 5s, who performed the training task
METHOD without shock.
The two stress-set instructions included: (a) a
Subjects.—The 5s were 80 introductory psy- fixed (f) level of shock according to a subjective level
chology volunteers who were randomly divided into setting as unpleasant but not painful administered
one of two instructional stress-set groups and then to the four shock contingency groups, denoted here-
blocked on four shock contingency groups on the after as ERTf, YRT,, Y f , and CRT t ; and (b) a
basis of Perceived Stress Index (PSI) base scores variable (v) level of shock, varying from low to
(administered in the class before the study was de- moderate intensity (e.g., from .5, 1.0, 1.3, 2.0, and
scribed or volunteers taken). The PSI (Jacobs & 2.5 ma.) administered to a second four shock con-
Munz, 1968), a reliable and validated (Jacobs & tingency groups, hereafter referred to as ERTV,
Thornton, 1970) check list instrument composed of YRTV, Yv, and CRTV. The five intensities of the
15 items derived by the Thurstone technique and re- variable shocks were assigned to each trial according
fined via Osgood's Semantic Differential, was used to to a random number table assignment.
detect perceived stress states in each instructional After electrodes were attached to the under side
stress set. of the arm 3 in. below the elbow, 5s were played a
Blocking consisted of assigning 5s to groups so tape recording of instructions.
that each group was homogeneous in respect to per- First, general instructions to the training task
ceived stress. informed all 5s their task was to depress the correct
Apparatus.—The apparatus consisted of two sepa- button corresponding to the onset of the stimulus
rate and distinctively different units, each unit used light. The second instruction informed ERTf and
for a different phase of the experiment. ERTV 5s about the contingency between shock and
LEARNED HELPLESSNESS IN HUMAN SUBJECTS 369
slow or incorrect responding (e.g., 5s were shocked
for responses with a latency > .5 sec.) and YRTf 5.00 •
and YRTv 5s that they would receive inescapable
shock, unrelated to their task. The third instruc-
tions informed 5s about the shock they would be
receiving (i.e., fixed or variable level). The 5s in ¥.50 • CRT f
groups Yf and YT received only the shock instruc- >-
o
tions, then were asked to remain seated while being
administered several inescapable shocks.
If fixed levels of shock were employed, shock levels •< ¥ . 0 0 -
were adjusted for each 5. The variable shock, how-
ever, involved no such adjustment. The range of
the fixed shock was .8 to 3.5 ma., with the average I 3.50•• ERT
level of 2.23 ma. The range of the variable shock UJ
cc
was .8 to 2.5 ma., with an average level of 1.81 ma.
After all instructions were completed, a PSI was
administered before Trial 1, after Trial 15, and after 3.00
Trial 30 of the training task.
Following the third PSI, 5s were individually
moved to the test rooms, and after having the elec-
trodes attached, were informed that they would be 2.50
receiving several shocks in accordance with the
shock instructions during the following minutes. ERT YRT Y CRT
They were given no instructions concerning how to SHOCK CONTINGENCY GROUPS
perform the task other than there was a task to be
solved. CRTf 5s were allowed first to adjust their FIG. 1. A (instructional stress set) X B (shock
shock level before test trials began. contingency) interaction of the learned helplessness
Ten test trials followed which consisted of 2 sec. test trials.
of CS (light) alone, overlapping 3 sec. of US (con-
stant shock). Any response greater than 5 sec. was
recorded as 5 sec. A response latency less than 2 Test trials.~A 2 X 4 ANOVA, performed
sec. avoided shock for any of the four groups, while on the test trials, revealed a nonsignificant
any response greater than 2 or less than 5 sec. A (instructional stress set) main effect
resulted in shock for that partial duration. To avoid
expectancy responding, random schedules of stimu- (p > .25), but a significant B (shock con-
lus-light occurrences, time intervals between onset tingency) main effect, F (3, 72) = 19.33,
of warning and stimulus lights, and time intervals p < .01, and an A X B interaction, F (3,
between trials were used. " 72) = 2.78,£ < .05. Due to the significant
Before 5s were dismissed they were asked (a) how A X B interaction (see Fig. 1), a simple
they performed the task and (6) if they did not re- main effects analysis was conducted to
spond, why not?
separately assess the variance at each fac-
tor. The results revealed significance of A
RESULTS at bi (instructional stress set at escapable
Training trials.—Dunn's multiple-com- shock, p < .05) and B at a2 (shock con-
parison procedure revealed no differences in tingency at variable shock, p < .01). To
reaction time scores between ERTf, further assess the significance of the shock
YRTf, and CRT f ; whereas ERTV < YRTV contingency factor, a Scheff6 5 test was
= CRTV (p < .01). It appears as though conducted on each level of shock contin-
a variable level of shock significantly gency at each level of stress instruction.
facilitated ERTV .S's performance (in order The fixed-shock instruction analysis re-
to avoid shock) and significantly decreased vealed no significant group differences.
YRTV S's performance (due to interference The variable-shock instruction analysis
from unpredictable shock). The t ratios revealed
for a priori orthogonal comparisons revealed -?ERTV < .XYBTV = X?v = ACBTV(£ < -01),
much the same effect of the two instruc-
tional stress sets in that ERTf > ERTV, which was considered to be supportive of
• t (54) = 3.61, p < .01; YRTf < YRTV, the learned helplessness hypothesis. The
t (54) = 2.42, £ < .01, and CRTf = CRTV. t tests for a priori orthogonal comparisons
370 JERRY W. THORNTON AND PAUL D. JACOBS

5.00 5.00
f.75 V.75
V.50 V.50
V.25 o V.25
0
LU V.QQ-
LU
p— 3.75 < 3.75
_J
3.50 1 11
3.50

,N RESPONSt
LU
3.25 3.25
O
Q.
3.00 3.00
00
LU 2.75 2.75

2.50 2.50
^2L
2.25 LU 2.25
3 2.00 2.00
1.75 1.75
1.50 1.50
-—-»YRT
1.25 1.25-

•-—»CRT,,

I 2 3 V 5 6 7 8 9 I O I 2 3 5 6 7 8 9 10
TRIALS TRIALS
FIG. 2. Graph of the 10 escape/avoidance test trials FIG. 3. Graph of the 10 escape/avoidance test trials
for the variable-shock contingency groups. for the fixed-shock contingency groups.

on latency measures yielded the following all of which reported significantly more
comparison : stress than

= 1.97, p < .05, ICET (P < .05).


>
The t tests for dependent measures con-
while all other comparisons were non- ducted on the stress-set-confounded C main
significant. effect revealed that PSIi indicated more
Figures 2 and 3 show the mean trial stress than PSIis, which equaled PSI3o
responses over the 10 test trials for the four (p < .05).
shock contingency groups. Both figures An analysis of the shock intensity (in
show the same group trends ; however, the milliamperes) administered per group in
variable-shock group means were signifi- training trials revealed that
cant, whereas the fixed-shock groups were
not. ERTf = YRTf = Yf > ERTV
PSI results.—A 2X4X3 mixed ANOVA = YRTV = Yv (p < .01).
conducted on the PSI scores (a repeated
measure) revealed a significant A main DISCUSSION
effect (p < .01) and perfectly confounded The test trial results of the variable-shock
with A, B, and C (PSI score) main effects 5s appear to be supportive of a phenomenon at
(p < .05) and (p < .01), respectively, least analogous to learned helplessness in
humans. Further, training trials, test trials,
while no interactions approached signifi- and PSI results suggest that variable shock was
cance. A Scheffe's S test, conducted on the superior in inducing and maintaining a stressed
B main effect to determine the relations state. Although the statistical evidence is of
between shock contingency means, revealed sufficient strength to support the learned help-
lessness hypothesis in humans, there are a few
that discrepancies between this investigation and
those using animals.
LEARNED HELPLESSNESS IN HUMAN SUBJECTS 371

Probably the most noticeable discrepancy yet untested effect of helplessness. That is,
concerns the fact that the CRT groups did not learned helplessness may vary in degree
differ from the YRT or Y groups with either depending upon the amount of shock used in
stress set during test trials. All previous tests training trials. Figures 2 and 3 reveal the simi-
of helplessness with animals have reported the larity of the two groups' performance. If one
ERT and CRT groups to be of near equality, accepts the idea that variable shock is more
both having smaller latencies than the YRT stressful than a fixed level, and results support
group. The present study with variable shock, this notion (i.e., ERTV < ERTf in training
however, found the CRT test trials to be sig- trials), then the difference in Fig. 2 and 3
nificantly different from the ERT groups but would suggest different degrees of learned
not the YRT or Y groups. helplessness.
It is suggested that a possible explanation for Some important relationships between Selig-
this discrepancy concerns the effects of instruc- man's work with dogs and the present investi-
tional set. The CRT groups in training trials gation are notable. Seligman et al. (1969) re-
were told only the nature of the task, while 5s ported that in test trials, dogs who initially
in groups ERT and YRT received additional received inescapable shock training made no
instructions concerning the relationship of the attempt later to escape or avoid. In the
task. In addition to instructional differences, present study, it was noted—by latency mea-
animal CRT 5s received no shock in training sures and postsession interviews—that with a
trials, but traumatic shock in test trials led to fixed level of shock, approximately 28% of all
inevitable escape. If one were to use a trau- YRTf, Y f , and CRTf 5s failed to make at least
matic level of shock with humans in the CRT one escape response during any of the 10 trials;
group, it is quite possible that their latencies whereas, with a variable-shock level, approxi-
would shortly conform to those of the ERT mately 65% of all YRTV, Yv, and CRTV 5s
group. Thus, the restriction of shock to mild failed to make at least one escape response.
levels in humans may repeatedly produce these This suggests that a variable level of shock
same findings, thus restricting a total replica- possibly acted to increase the interfering effect
tion of learned helplessness in humans. This of learned helplessness.
should not be taken, however, as failure to When asked why they did not respond,
validate helplessness in humans, since the more approximately 60% of the 5s in all YRT and Y
crucial evidence is derived from the three groups reported that they felt they had no con-
groups differing only in controllability, during trol over shock, so why try. These 5s reported
training, of the shock (i.e., ERT, YRT, and Y). that they spend the majority of their time in
The present investigation represents a form preparation for the upcoming shock. Approxi-
of transfer of instructions from one task to the mately 35% reported that they, after pushing
next. In reality, the whole basis of helpless- one or two buttons, abandoned the idea of
ness is transfer of a learned state from one task escape. The other 5% gave no reason for
to a second. Whereas animals transferred a response failure. All 5s in both ERTf and
self-learned helplessness state, humans trans- ERTV groups learned the task within four
ferred an instructionally set, internally veri- trials; however, after the sixth trial only 10%
fied, learned helplessness state. The point is of ERTf 5s were avoiding, while 84% of ERTV
that they did transfer "helplessness" to a 5s were avoiding. In postinterviews, when
second task which, in fact, offered control. ERTf and ERTV 5s were asked how they per-
It should be pointed out that the two stress- formed the task, more that 70% reported they
set groups performed under different shock felt that they had control over shock, and their
levels. The variable-shock group as a whole task was to find out how. Thus the perform-
avoided 82% of the training trial shocks, ance of ERTf and ERTV 5s reflects behavior
whereas, the fixed group as a whole avoided
similar to Seligman and Maier's (1967) dogs.
only 49% of the training trial shocks. This
gives reasoning behind why the variable-shock On several occasions, 5s would escape or
groups (i.e., ERTV and YRTV) received only avoid shock on one or more trials, but on sub-
about half as much shock as a group as did the sequent trials would again take the full 3-sec.
fixed-shock groups (i.e., ERTf and YRTf). shock. It appeared that these 5s did not asso-
The trends, but lack of support of the help- ciate their responding with the reinforcement
lessness hypothesis with fixed-shock levels and (i.e., the helplessness state was maintained).
the support for the hypothesis with variable- Seligman et al. (1969) described nearly identi-
shock levels may suggest an interesting and cal findings with dogs.
372 JERRY W. THORNTON AND PAUL D. JACOBS

One of the primary implications of the pres- JACOBS, P. D., & THORNTON, J. W. Scale sensitivity
ent study concerns Seligman's (1969) latest of the perceived stress index. Perceptual and
work on schizophrenic "therapy," using know- Motor Skills, 1970, 30, 944.
ledge of learned helplessness. Seligman hy- MAIER, S. F., SELIGMAN, M. E. P., & SOLOMON, R. L.
pothesized that schizophrenics are a product Pavlovian fear conditioning and learned helpless-
ness. In B. A. Campbell & R. M. Church (Eds.),
of the lack of control they have experienced Punishment and aversive behavior. New York:
either traumatically or developmentally. Sup- Appleton-Century-Crofts, 1969.
port for learned helplessness in humans would OVERMIER, J. B., & SELIGMAN, M. E. P. Effects of
seem to be a primary stepping stone for further inescapable shock upon subsequent escape and
research on this idea. avoidance responding. Journal of Comparative
The present study's heuristic value lies in and Physiological Psychology, 1967, 63, 28-33.
the fact that there are many parameters which PERVIN, L. A. The need to predict and control
under conditions of threat. Journal of Personality,
now must be investigated with human 5s, e.g., 1963, 31, 570-585.
stressor generalization, immunization, allevia- PICKNEY, G. Avoidance learning in fish as a func-
tion, etc. tion of prior fear conditioning. Psychological
Reports, 1967, 20, 71-74.
SELIGMAN, M. E. P. Can we immunize the weak?
REFERENCES Psychology Today, June 1969, 42-44.
ANDERSON, D. C., COLE, J., & McVAUGH, W., Varia- SELIGMAN, M. E. P., & MAIER, S. F. Failure to
tions in unsignaled inescapable preshock as deter- escape traumatic shock. Journal of Experimental
minants of responses to punishment. Journal of Psychology, 1967, 74, 1-9.
Comparative and Physiological Psychology, 1968, SELIGMAN, M. E. P., MAIER, S. F., & GEER, J. H.
65(3, Pt. 2). Alleviation of learned helplessness in the dog.
BRAUD, W., WEPMAN, B., & Russo, D. Task and Journal of Abnormal Psychology, 1968, 73, 256-
species generality of the "helplessness" phenome- 262.
non. Psychonomic Sciences, 1969, 16, 154-155. SELIGMAN, M. E. P., MAIER, S. F., & SOLOMON, R. L.
Unpredictable and uncontrollable aversive events.
D'AMATO, M., & GUMENIK, W. Some effects of In F. R. Brush (Ed.), Aversive conditioning and
immediate versus randomly delayed shock on an learning. Academic Press, 1969.
instrumental response and cognitive processes. THORNTON, J. W., & JACOBS, P. D. Analysis of task
Journal of Abnormal and Social Psychology, 1960, difficulty under varying conditions of induced
60, 64-67. stress. Perceptual and Motor Skills, 1970, 31,
JACOBS, P. D., & MUNZ, D. C. An index for measur- 343-348.
ing perceived stress in a college population. Jour-
nal of Psychology, 1968, 70, 9-15. (Received August 13, 1970)

You might also like