- Email: [email protected]
Learned helplessness in groups
Behar. Re.r. Thu. Vol 21, No. 6, pp. 613-622. 1983 Printed in Great Britain
LEARNED
0005-7967:x3 53.00 + 0.00
Pergamon Press Ltd
HELPLESSNESS
DAVID K. SIMKIN, JAN P. LEDERER Department of Psychology,
University
and MARTIN E. P. SELIGMAN*
of Pennsylvania. U.S.A. (Rewired
IN GROUPS
3815 Walnut
Street.
Philadelphia,
PA 19104,
12 Fehruar~ 1983)
Summary-Group learned helplessness 1s demonstrated in Experiment I. Groups of 2 tried to turn off noise by their joint action. In the solvable group (S), noise ofTset was contingent on their sequence of button pushing. In the yoked, unsolvable group (U), noise ofTset was independent of all sequences of button pushes they produced. In a practice group (0). subjects practiced coordinated sequences of button pushes with their partners, but heard no noise. Later, all 3 groups were tested in pairs in a shuttlebox which required coordinated joint responding to turn off noise. The unsolvable group escaped more poorly than the other 2 groups, paralleling helplessness effects with individuals. Experiment 2 and 3 found no transfer from individual helplessness training to group testing and no transfer from group helplessness training to individual testing. We suggest that the same mechanism, the expectation of response ineffectiveness. may mediate both individual and group lcarncd helplessness.
INTRODUCTION
In principle not only can all the voluntary actions of 1 individual be futile, but so also can the concerted actions of 2 or more individuals. Will such a group learn to expect that its future concerted actions will be in vain? Can a group learn to become helpless? To date, the entire helplessness experimental literature involves single subjects, be they human or animal. Yet speculation about what learned helplessness might explain has occasionally extended to interpersonal relations, families, races and even nations (e.g. Sacco and Hokanson, 1978, 1982; Garber and Seligman, 1981. We set out to determine if the standard helplessness procedure used with individuals (i.e. Hiroto and Seligman, 1974) could be applied with parallel results to 2 people acting in concert? This study reports the results of 3 experiments. In the first experiment pairs or subjects received pretreatment as a group with escapable noise, inescapable noise or no noise followed by a test for helplessness as a group (Gr-Gr). In this test, only concerted action by both individuals produced termination of loud noise. In the second experiment subjects received pretreatment as individuals followed by a test for helplessness as a group (In-Gr). The third experiment used a group pretreatment and individual testing (Gr-In).
EXPERIMENT
1 (Gr-Gr)
Method
Ss were 36 male and 36 female University of Pennsylvania undergraduates. They were recruited from class solicitation and advertisements asking for volunteers for a group problem-solving experiment which would pay $2.50? for approx 1 hr of participation. This pool consisted predominantly of freshman and sophomore students enrolled in non-psychology courses. Ss were recruited until 12 males and 12 females were obtained for each of 3 experimental conditions. Assignment to these groups was random. Ss were run in same-sex pairs resulting 12 groups per condition (6 male groups, 6 female groups per condition).
*To whom all correspondence should be addressed tRun before 1978. You should be so lucky! 613
The 3 experimental conditions were: (a) solvable problems (S); (b) unsolvable problems (U); and (c) a practice pretreatment control group
(0).
After assignment. Ss were administered the Beck Depression Inventory (BDI), a test of affective, cognitive and somatic symptoms of depression (Beck, 1967). There were no significant differences in mean depression ratings among the group upon their arrival at the laboratory (P > 0. I ).
Prrtvcwtmmt tmk. The apparatus consisted of a stimulussresponse panel mounted in front of each S‘s chair and a Pioneer headset (model SE-205) for each S (See Fig. 1). The horizontal panel consisted of 3 black spring-loaded buttons mounted on a wooden block 12 x 3 x I!” and arranged in a horizontal line spaced 3” apart. Above each button was a white light mounted on a wooden backboard 12” in height which corresponded to each of the 3 buttons of the partners’ push-button apparatus. It was by this arrangement that Ss were able to communicate. that is when Sl pushed the left-most button, the left-most white light on the panel of S2’s was lit; when S2 pushed the middle button, the middle white light on S l’s panel was lit etc. This allowed each S to know both his own movements as well as those of his partner. Above these white lights were located a series of colored lights, I red, 1 yellow and 1 green. These lights served as special signals from the experimenters to the Ss and will be further explained in the Procedure section. The aversive stimulus used was white noise played through Pioneer headsets. emanating from a Testan Deluxe Audio Generator set at 90 dB. Noise onset was controlled by a tape timer which allowed a 5-set presentation of the aversive stimulus according to a preset random interval schedule [average inter-trial interval (ITI) of 7 XC]. Experimenters were informed of all activities by lights located in an adjacent equipment room. Test /u.sli. The apparatus in the testing phase was a modified Manipulandutn Type S test originally designed by Turner and Solomon (I 962) and used by Hiroto (1974) as a human analog to the two-way shuttle box used in animal studies. The manipulandum was 23 x 5 x 6” with a 4” Plexiglas knob protruding from the top. The knob slid along a 12 x I” straight channel on the cover of the box. Attached to the knob on the underside of the channel was a Plexiglas disc which made contact with the 6 hidden switches (see Fig. 1).
Colored
slgnal
lights
White communication
\
\
(A) Pretreatment button
(B) task
-
Test
push
Fig. I
task
-
\ shuttle
box
Learned
helplessness
615
in groups
Once again communication between the Ss was by way of white lights (6 lights), located on a wooden backboard 18y in height, which were connected with the 6 hidden switches of the partner’s shuttle. The 6 lights traced the movements of the partner’s shuttle handle; e.g. when S 1 moved his shuttle handle from the extreme left to the extreme right. the white lights of Ss panel lit one at a time showing the position of the partner’s handle. This allowed each S to know the movements of his partner. Above these communication lights were the series of colored signal lights: 1 red, 1 yellow and 1 green. The aversive stimulus was the white noise played through the headsets at 90dB.
Ss were introduced to the experimenter and seated. Each S’s apparatus was located on a desk in front of him. Between the two desks was a barrier that prevented the Ss from seeing each other. The Ss were instructed that they were to use no form of communication other than the apparatus that was given them (no talking, hand signals etc.). The Ss then filled out the BDI, a Multiple Affect Adjective Check List (MAACL) and a consent form. The MAACL (Zuckerman and Lubin, 1965), which indexes depression, anxiety and hostility, was also administered after the pretreatment and after the test task. Pretwmtmrnt phase. All Ss in these 2 conditions received the same instructions. They were told:
“During this part of the experiment you will both be wearing these headsets that are to the left of you, and from time to time a loud noise will come on for a while. When that noise comes on, there is something that you two as a team can do to stop it. There is some predetermined, alternating sequence of button pushes that you and your partner can perform that will stop the noise. By alternating, we mean that each button push of one subject must be followed by a button push of the other subject. It is your collective task to try to figure out the solution to how the noise can be turned off. By finding the solution to the problem quickly and consistently you can minimize the amount of unpleasant noise that you will receive. There will be a randomly predetermined number of trials in which you are to solve one sequence, and this sequence will remain constant throughout the entire set of trials. The noise is scheduled to come on at predetermined random intervals. Now there are several lights located on the base in front of you. The three white lights are synchronized with the set of buttons that your partner operates. When he pushes his right button, your right light goes on etc. (demonstration). This will enable each of you to know what button you and your partner are pushing. On the panel in front of you. there is also a set of colored lights. These lights will serve as signals for you. If the red light goes on yfter the noise stops, then your response did not turn off the noise, but rather the noise has stopped automatically according to a preprogrammed schedule. If, on the other hand, the green light comes on after the noise stops, then your response has turned off the noise and you have found the solution to the problem. You should attempt to have as little aversive noise as possible, so that by consistently turning off the noise you will suffer from the noise as little as possible. Taking the headphones off or dismantling the apparatus in any way are not acceptable ways of turning off the noise. Only responding while the noise is on counts. O.K. are are there any questions before we begin? Before we get started, I would like to give you a sample of what the noise will be like, so that you can decide whether or not you wish to continue (one presentation of white noise at 90 dB). At the sound of the next noise the first trial will begin.” A sequence of three button pushes (FR-3) was set up for the S condition. required to push his/her left-most button, followed by the other S pushing finally the first S pushing his middle button. This alternating sequence of to be performed without any other responses intervening, but responses sequence did not detract. This alternating sequence was required to occur
The S on his left-most three button made before within 5 set
the left was button and pushes had the correct to terminate
the noise and obtain a green signal light which indicated that they had found the correct sequence. If the correct response was not performed. the noise stopped automatically at 5 set and the S’s’ received a red signal light. indicating that their response did not terminate the noise. Response latencies were measured by experimenters with a hand-held stopwatch. Ss were allowed a minimum of 30 trials and a maximum of 50 trials in which to achieve IO consecutive correct responses. This was to ensure that all groups learned the task and ended with IO consecutive correct responses. All Ss met this criterion. Groups in the U condition were yoked to groups in the S condition. Here the amount of aversive noise the Ss received was determined by the recorded response latencies of the matched S group. There was no correct response for the U groups so Ss always received the red signal light at the termination of the noise. The 0- conditon is not the conventional control condition used in helplessness research. Usually, the control condition receives no pretreatment. but in the S and U group prior experience with our means of communication was a considerable advantage. So, some form of practice was in order. The Ss were instructed as follows: “You have been randomly chosen to be in a group that will practice in sort of a game. As you can see, there is a set of buttons and lights in front of you. The yellow light will come on at predetermined random intervals. Each time the yellow light does come on. we would like both you and your partner to begin pushing buttons. The only requirement is that you press buttons in what we call an alternating sequence. By alternating. we mean that each button push of one subject must be followed by a button push of the other subject. Now there are no correct or incorrect responses. that is no success or failure. We merely want you to practice different sequences of button pushes with your partner. We are interested in seeing what different types of sequences you both can come up with. The yellow light will remain on for five seconds at a time. Please respond only while the yellow light is lit. There will be a randomly predetermined number of trials. Now you can see there are three white lights located on the base in front of you and these three white lights are synchronized with the set of buttons your partner operates. When he pushes his right button your right light goes on etc. (demonstration). This will enable each of you to know what buttons both you and your partner are pushing.” Twenty trials in which the yellow light was lit took place. After the pretreatment task. Ss in all conditions were again administered the MAACL Form and were told that they were to “check off each and every adjective that applies to how you are feeling at this moment”. Test phrc~. The test phase was the same for all conditions. The shuttle-box task was used and the operation of the tracking lights was demonstrated. Instructions were similar to the S and U pretreatment instructions and the operation of the signal lights was explained. Ss were told that there would be 20 trials Again, Ss were given 5 set on each trial in which to perform the correct response. The correct response was for 1 S to move the shuttle handle to one end of his shuttle box. and his partner to move the handle to the opposite end. Thus, there were actually four correct patterns. Six dependent measures were obtained for each group: (a) number of successes (escapes); (b) mean response latencies (with a maximum of 5 set per trial); (c) trials to criterion. defined as the trial number of the fifth successive trial with a response latency of less than 5 set: (d) Ss subjective recall of correct response; (e) conditional probability of performing a correct rcsponsc given a correct response on the previous trial; (f) MAACL scores. After 20 trials were completed all Ss were given the MAACL Today Mood Inventory for the third time. and a short questionnaire. After completing these forms. all Ss were debriefed and paid. Rrsults When 2 individuals acting in concert received inescapable noise, their concerted performance on a new. but escapable, test was disrupted. When 2 individuals acting in concert first received escapable noise. their test performance was facilitated.
Learned
helplessness
617
in groups
20
Gr - Gr
18 16 14
SD
Condition
Fig. 2. Mean number
of escapes,
400
6.66
4.67
S
SDS and significance level for Experiment 0 groups.
1 (Gr-Gr)
for the S, U and
Mean escapes Analysis of variance of mean number of escapes (Escapes x Condition) revealed a significant main effect, F(2, 33) = 10.39, P < 0.0001. Newman-Keuls* tests of mean number of escapes shows that the S condition performed significantly better than the U condition, P < 0.01. The S group performed marginally better than the 0 condition (P < 0.06). The U condition escaped significantly worse than the 0 condition, (P < 0.05). The latency, trial to criterion and conditional probability variables all revealed similar results to the escape variable. Recall of correct response by S’s self-reporting After the test task Ss were asked to write down what they believed the correct response was to the shuttle-box task. All Ss attempted some form of an answer. We accepted more complex solutions as correct as long as they contained the proper solution within them. In the S condition, 19 out of 24 Ss knew the correct response, significantly more than the 11 out of 24 and the 10 out of 24 in the 0 and U conditions, respectively, who knew the correct response. MAACL
scores for depression
Condition x Sex analysis of MAACL, 1 (at Time l), depression revealed that conditions did not differ on mean Depression score upon starting the experiment (P > 0.25). This analysis was followed by a Condition x Sex x MAACL, 1, 2, 3 (repeated measures) analysis. MAACL Depression scores did not change significantly within or between conditions. EXPERIMENT
2 (In-Gr)
Method Subjects As in Experiment 1 (Gr-Gr), Ss were 36 male and 36 female University of Pennsylvania undergraduates, recruited from class solicitation and from advertisements asking for volunteers for a group-problem solving experiment that would pay $2.50 for approx 1 hr of participation. Assignment to experimental condition proceeded as in Experiment 1 until each condition consisted of 6 male and 6 female groups. The three experimental conditions were: (a) solvable (S); *All paired
comparisons
are done
by the Newman-Keuls
procedure.
(b) unsolvable (c) ;t practice Thcrc
(U): and ~rctreatn~etlt
were no significant
(0) control
differences
group.
of mean
BDI score between
the 3 conditions.
(P > 0.25).
P~t,t~~~~it~~~~,~~t t(r,di (~~l~/i~.~f~u~ti). The apparatus is the same as described in Experiment I with the f~~llowing changes. The white communication lights remained in place but they were shut off. Thus. during the pretreatment phase. the Ss could not communicate and were not aware of the responses of their ‘partner’. The signal lights served the same functions as they did in Experiment 1 (Gr-Gr) except that the feedback was now independent, that is one Ss red light could be illuminated while the other’s green light was illuminated. Ss were not able to observe the response of their ‘partner’. The aversive stimulus was the white noise delivered as in Experiment I, except that the noise to 1 S could be terminated independently of the other. Test tusks. The apparatus was exactly as in the test phase of Experiment 1.
Each S was seated with the apparatus on a desk in front of him. A barrier prevented direct visual ~ornn~uni~~ition between S’s and they were instructed that they were to use no form of communication other than the apparatus that was given them. The BDI. MAACL and a consent form were tilled out. Pwtwutm~tlt pl~rrsc~.S ~td U cotditions. All Ss in these 2 conditions received the same instructions. These instructions were similar to the pretreatment instructions in Experiment I except that the Ss’ escape was not said to be interdependent, nor was it. The FR-3 response was correct for the S condition and it had to be performed within 5 sec. An individual performing the correct response within the 5 set terminated only his noise and only he received a green light. ,Ss were allowed a minimum of 30 trials and a maximum of 50 trials in which to achieve IO consecutive correct responses. If 1 S reached his criterion before his ‘partner‘. he simply received no more trials. The LI Ss were yoked to members of the S group, with one member of the U group yoked to one member of the S group. Each S in the U group received a red light after termination of noise, regardless of his response. Prtrt,/ic,cJ pr-c~tr~~utnlcnt. The S and U conditions in this experiment followed the procedures of the conventional conditions in the helplessness literature except that 2 Ss are in the room at one time. As in Experiment I the 0 condition is a practice, noncommunication condition. Ss were instructed as follows: “You have been randomly chosen to be in a group that will participate in a short game. As you can see, there is a set of buttons in front of you. The yellow light will come on, we would like each of you to begin pushing buttons. The only requirement is that you press buttons in a sequence. Now there are no correct or incorrect responses, that is no success or failure. We merely want you to practice diti’erent sequences of button pushes. We are interested in seeing what different types of sequences you come up with. The yellow light will remain on for 5 set at a time, and plcasc only respond while the yellow light is lit. There will be a randomly predetermined number of trials.” Twenty trials in which the yellow light was lit then took place. After the pi-etreatl~lent task, Ss in all conditions were administered the MAACL for a second time. Test p/rtr.w. The test phase followed the same procedure as in Experiment I with the correct response requiring the joint efTort as described above. Thus. in this experiment the Ss received individual prctrcatment and were tested in groups (In-Gr). The 5 dependent measures were again: (a) number of successes (escapes); (b) mean response latencies: (c) trials to criterion: (d) Ss’ subjective recall of correct response: ie) ~ondition~~l
Learned
helplessness
20
In - Gr
SD
Condition
Fig. 3. Mean
number
in groups
of escapes
620
502
532
S
0
U
and, SDS for Experiment
2 (In-Gr)
for the S, U and 0 groups.
probability of a correct response given a correct response on the previous trial; and (f) mean MAACL scores for depression. After 20 trials, all Ss got their third MAACL and a short questionnaire. After completing these Ss were debriefed and paid. Results The results indicate
that individual
pretreatment
does not affect concerted
group
performance.
Mean escupes Individual pretreatment led to no significant differences in mean number of escapes conditions when testing was done as a group [F(2, 33) = 0.17, P > 0.251. Similar results the rest of the measures.
between hold for
Mean MA ACL scores ,for depression A Condition x Sex analysis of mean MAACL Depression scores revealed an inadvertant, but significant main effect for Condition and a possibly more interesting effect for the Condition x Sex interaction (P < 0.05). The 0 group was more depressed at the start of the experiment (P < 0.05). Further analysis showed that males differed across conditions on Depression score (P < 0.01) but females did not (P > 0.05). Because of this, a Condition x MAACL, 1, 2, 3 Depression (repeated measures) was done for females only (Table I). The Trial x Condition interaction was significant [F(4,66) = 4.06. P < 0.011. The mean Depression score increased significantly after pretreatment for the U group (P < 0.01) but not for the other 2 groups. So, individually-induced helplessness may have made females more depressed.
EXPERIMENT
3 (Gr-In)
Method Su@ects Ss were 18 male and 18 female University of Pennsylvania undergraduates, recruited and paid as in Experiments 1 and 2. Ss were assigned randomly to each of the 3 experimental conditions until there were 6 males and 6 females in each condition. Ss were run in same-sex pairs resulting in 6 groups per condition (3 male groups, 3 female groups per condition). The experimental conditions are the same as in the previous 2 experiments. The mean BDI scores for the conditions did not differ (P > 0.25).
620 Table 1. Means and SDS of MAACL depression after training (2) and after test (3) for Experiment S
scores before (I), 2 (In-Gr; N = 12)
U
0 P > 0.05
,
r
P
kk 10.67 5.38
&
P < 0.05 14.33 5.16 16.42 4.46 15.167 4.71
x SD _ $
Il.33 5.57 9.41 s.33
18.75 4.33 18.50 4.42 16.50 4.275
MAACL,
S;
MAACL,
St
14.83 4.45 14.67 4.89
14.08 6.67 14.25 4.41
9.75 7.747 14.917 6.16
J
MAACL,
St
15.42 4.03
14.58 6.19
12.33 6.91
>
MAACL, MAACL,
FWil&S P < 0.01 P < 0.05
Apparatus Pretreatment task. The apparatus and details of the aversive stimulus (white noise) are exactly as described in Experiment 1. Test [ask. The apparatus is the same as described in Experiment 1 (Gr-Gr) but it was modified in an analogous way to the pretreatment apparatus modifications in Experiment 2. The white communication lights above the shuttle box remained in place but were removed from the circuit. Thus, no communication took place between Ss. The white noise and the colored signal lights could be independently presented to the Ss.
Pretreatment phase. The pretreatment procedure is exactly as outlined in Experiment 1 (Gr-Gr). Tesr phase. This test phase was the same for all conditions. Ss performed independently and had 5 set on each trial to perform the correct response (the correct response required the S to move the shuttle handle all the way to the right end of his or her shuttle box and then to the left end). In this experiment Ss received their pretreatment in groups and were tested individually (Gr-In), so the 6 dependent measures are measures of individual performance, not group performance, as in the first 2 experiments.
-
r
20
Gr In
16 2 g
13.50
14
13 33
:: m
12-
';; IO2
8-
E
6-
p
42-
I
Fig. 4. Mean
number
SD
6 30
Condltlon
S
of escapes
565
5 96
0
U
and SDS for Experiment
3 (Gr-In)
for the S, U and 0 groups.
Learned helplessness
in groups
621
Results Group
helplessness
did not have strong
effects on individual
escape performance.
The S and 0 groups had virtually the same number of mean escapes while the U group had fewer than both, but this difference did not reach significance (P > 0.1). Similar results hold for the other dependent variables.
An analysis of variance revealed no significant differences between conditions for Depression scores upon entering the experiment. The mean Depression score and Condition x Sex x MAACL I, 2, 3 Depression analysis of this data showed no systematic changes across the experiment.
To assess whether experiencing the pretreatment in a group vs individually affected later group performance. the results of Experiment 1 (Gr-Gr) were compared to Experiment 2 (In-Gr) on 4 of the dependent measures: (a) mean escapes; (b) mean response latencies; (c) trials to criterion; (d) conditional probability of a correct response given a correct response of the prior trial. The Condition x Experiment interaction was significant [F(2,66) = 5.31, P -c 0.011. Ss pretreated with unsolvable noise as a group had ,f&er mean escapes when later tested as groups than those pretreated individually when later tested as groups (P < 0.01). The other groups did not differ significantly across the experiments. Similar findings were obtained with the other measures.
DISCUSSlOb!
Learned helplessness can be produced in groups. When 2 individuals acting in concert learned that their joint efforts were to no avail in turning off noise (group pretreatment), they later escaped poorly in concert when noise was actually escapable (Experiment 1). Such group helplessness training, while impairing performance, did not make the individual feel more depressed, however, as it commonly does when individuals are made helpless with inescapable noise (Miller and Norman, 1979). There is no parallel debilitation when individuals are first made helpless singly and then concerted performance is tested in pairs (In-Gr, Experiment 2). Although their group performance was unaffected by individual helplessness pretreatment, they probably did become more depressed, however. Similarly. there appears to be little transfer of helplessness from group to individual testing. When 2 individuals acting in concert had their joint effort rendered useless, they performed only slightly and not significantly worse than controls when tested singly (Gr-In, Experiment 3). We tentatively suggest that group helplessness is mediated by the same mechanism as individual helplessness. When an individual acting in concert with another finds that noise goes off independently of their concerted responses, each forms an expectation that future concerted responses with that other person will be ineffective. Consequently, when placed with that other person in a new task in which concerted action is called for. each individual shows two deficits: (I) a lnotivational deficit, with the probability decreased of each initiating voluntary responses to obtain the desired outcome jointly: (2) a cognitive deficit, with each individual having more difficulty learning that the groups’ concerted action succeeds, even when it actually does. Consequently. the concerted performance of the group will show both fewer escape responses executed and more difficulty learning to escape, even when successful responses are made, What about transfer of helplessness from groups to individuals and from individuals to groups? We suggest that the domain of the expectation that responding will be futile depends on what unit of response was in fact trained to be helpless. That is, if group action was futile, the individual comes to expect that future group action will again be futile, but not necessarily that his own individual action again will be futile (e.g. “My partner is incompetent, but I’m O.K.“). Similarly if an individual learns that he, singly. was helpless, he need not expect that later concerted action with another will be futile (e.g. “I’m incompetent. but my new partner may salvage us.“).
622
DAVIIJ K. SIMKIU cl cl/
This interpretation can be cast into attributional terms (Abramson, Seligman and Teasdale, 1978). During typical individual helplessness training, an individual often attributes his failure to a relatively internal cause, e.g. his own incompetence. While more external attributions can be made (e.g. the experimenter’s unfairness), they are probably less frequent (Abramson, 1978). During group helplessness training, the presence of a coworker explicitly allows more room for a more external attribution (e.g. my coworker’s incompetence or our joint incompetence). So, we speculate that group helplessness training produces more external attributions for failure than does individual helplessness training. Now consider the implications of such attributions if subjects are tested either as a group or individually. After group helplessness training, what will happen when the same group is faced with a new task’? Attributions to either the coworker’s or joint incompetence will produce an expectation So, group that joint responding will be in vain, which will in turn, lower response initiation. helplessness training should produce group helplessness later. In contrast, if the individuals are tested alone after group helplessness training, individual helplessness should be less likely. If the cause of the group failure was external-my coworker’s or our joint incompetence--this does not imply individual incompetence (e.g. he may be stupid or we may not be able to solve problems as a team. but this doesn’t mean I can’t solve them alone). Similar logic holds for individual helplessness training followed by group testing: the individual having failed alone makes a relatively internal attribution. Now, confronted with a task that has a coworker to help out, an acquired internal attribution for failure will not be as damaging (e.g. even if I am incompetent, we might still succeed because my coworker might be extremely competent, or we might work well as a team). So. on attributional grounds, group helplessness training is likely to produce later failure as a group. In contrast. group helplessness training is less likely to produce later individual failure, just as individual helplessness training is less likely to produce later group failure, since in both cases, the attributions for failure made in training are less relevant to the testing situation. This account remains speculative, however, since we did not query attributions after helplessness training. Mood, unlike performance, may not be governed in this way. Group helplessness debilitated group performance but did not depress individual mood (Experiments 1 and 3). while individual helplessness seemed to lower mood, but did not worsen group performance (Experiment 2). Depressed mood, we believe, stems directly from expectations about individual helplessness and only derivatively from expectations about group helplessness. In conclusion, 2 individuals acting in concert. whose actions are ineffective in turning off noise show later deficits at escaping noise together in a new situation when noise is actually escapable. They do not show such helplessness, however. when tested alone. Individuals, whose lone responses are ineffective in escaping noise, do not show later escape deficits when acting together in pairs. We suggest that individuals form expectations.about the inefiectiveness of their joint responses, as well as about their individual responses. \Ye further suggest that an expectation that joint action will be ineffective lowers the probability that joint action will be initiated. and makes it more difficult for these individuals to learn that their joint action has succeeded. .Ickrlo~~./e~~~~~mc,n/s~We wish to thank Lyn Abramson. Lauren Alloy, Paul Eelen, Amy Semmel. Joe Volpicelli and the helplcssncss seminar for guidance and comments. This research was supported by a PHS Grant. MH19604 to Martin E. P. Seligman, a Guggenheim Fellowshtp to Martin E. P. Scligman and NSF Grant No. BNS 76-22943.A02 to the Center for Advanced Study in the Behavioral Sciences, Stanford, Calif. REFERENCES Abramson Abramson abnorm.
L. (1978) Unpublished Doctoral Dissertation. Univ. of Pennsylvania, Philadelphia, Penn. L., Seligman M. E. P. and Teasdale J. (1978) Learned helplessness in humans: critique and reformulation. Psvchol.
81, 32-48. Depression: Clinical,
J.
Beck A. T. (1967) Experimental. and Theoretical Aspects. Hoeber, New York. Garber J. and Seligman M. E. P. (Eds) (1981) Human Helolessness: Theory und Atukations. Academic Press, New York. Hiroto D. S. and Skligman M. E. P. (1975) Generality of learned helplessness in man. J. Person. sec. Ps_vchol. 31, 31 l-327. Miller I. and Norman W. (1979) Learned helplessness in humans. Psychol. Bull. 86, 93-l 18. Sacco W. P. and Hokanson J. E. (1978) Expectations of success and anagram performance of depressives in public and private settings. J. abnorm. Psychoi. 87, 122-130. Turner L. and Solomon R. L. (1962) Human traumatic avoidance learning: theory and experiments on the operantrespondent distinction and failures to learn. P.YLY-hol. Monogr. 76, 40 (Whole No. 599). Zuckerman M. and Lubin B. (1965) Manual for the Multiple Affkct Adjective Chech List. EDITS. San Diego. Calif.
LEARNED
0005-7967:x3 53.00 + 0.00
Pergamon Press Ltd
HELPLESSNESS
DAVID K. SIMKIN, JAN P. LEDERER Department of Psychology,
University
and MARTIN E. P. SELIGMAN*
of Pennsylvania. U.S.A. (Rewired
IN GROUPS
3815 Walnut
Street.
Philadelphia,
PA 19104,
12 Fehruar~ 1983)
Summary-Group learned helplessness 1s demonstrated in Experiment I. Groups of 2 tried to turn off noise by their joint action. In the solvable group (S), noise ofTset was contingent on their sequence of button pushing. In the yoked, unsolvable group (U), noise ofTset was independent of all sequences of button pushes they produced. In a practice group (0). subjects practiced coordinated sequences of button pushes with their partners, but heard no noise. Later, all 3 groups were tested in pairs in a shuttlebox which required coordinated joint responding to turn off noise. The unsolvable group escaped more poorly than the other 2 groups, paralleling helplessness effects with individuals. Experiment 2 and 3 found no transfer from individual helplessness training to group testing and no transfer from group helplessness training to individual testing. We suggest that the same mechanism, the expectation of response ineffectiveness. may mediate both individual and group lcarncd helplessness.
INTRODUCTION
In principle not only can all the voluntary actions of 1 individual be futile, but so also can the concerted actions of 2 or more individuals. Will such a group learn to expect that its future concerted actions will be in vain? Can a group learn to become helpless? To date, the entire helplessness experimental literature involves single subjects, be they human or animal. Yet speculation about what learned helplessness might explain has occasionally extended to interpersonal relations, families, races and even nations (e.g. Sacco and Hokanson, 1978, 1982; Garber and Seligman, 1981. We set out to determine if the standard helplessness procedure used with individuals (i.e. Hiroto and Seligman, 1974) could be applied with parallel results to 2 people acting in concert? This study reports the results of 3 experiments. In the first experiment pairs or subjects received pretreatment as a group with escapable noise, inescapable noise or no noise followed by a test for helplessness as a group (Gr-Gr). In this test, only concerted action by both individuals produced termination of loud noise. In the second experiment subjects received pretreatment as individuals followed by a test for helplessness as a group (In-Gr). The third experiment used a group pretreatment and individual testing (Gr-In).
EXPERIMENT
1 (Gr-Gr)
Method
Ss were 36 male and 36 female University of Pennsylvania undergraduates. They were recruited from class solicitation and advertisements asking for volunteers for a group problem-solving experiment which would pay $2.50? for approx 1 hr of participation. This pool consisted predominantly of freshman and sophomore students enrolled in non-psychology courses. Ss were recruited until 12 males and 12 females were obtained for each of 3 experimental conditions. Assignment to these groups was random. Ss were run in same-sex pairs resulting 12 groups per condition (6 male groups, 6 female groups per condition).
*To whom all correspondence should be addressed tRun before 1978. You should be so lucky! 613
The 3 experimental conditions were: (a) solvable problems (S); (b) unsolvable problems (U); and (c) a practice pretreatment control group
(0).
After assignment. Ss were administered the Beck Depression Inventory (BDI), a test of affective, cognitive and somatic symptoms of depression (Beck, 1967). There were no significant differences in mean depression ratings among the group upon their arrival at the laboratory (P > 0. I ).
Prrtvcwtmmt tmk. The apparatus consisted of a stimulussresponse panel mounted in front of each S‘s chair and a Pioneer headset (model SE-205) for each S (See Fig. 1). The horizontal panel consisted of 3 black spring-loaded buttons mounted on a wooden block 12 x 3 x I!” and arranged in a horizontal line spaced 3” apart. Above each button was a white light mounted on a wooden backboard 12” in height which corresponded to each of the 3 buttons of the partners’ push-button apparatus. It was by this arrangement that Ss were able to communicate. that is when Sl pushed the left-most button, the left-most white light on the panel of S2’s was lit; when S2 pushed the middle button, the middle white light on S l’s panel was lit etc. This allowed each S to know both his own movements as well as those of his partner. Above these white lights were located a series of colored lights, I red, 1 yellow and 1 green. These lights served as special signals from the experimenters to the Ss and will be further explained in the Procedure section. The aversive stimulus used was white noise played through Pioneer headsets. emanating from a Testan Deluxe Audio Generator set at 90 dB. Noise onset was controlled by a tape timer which allowed a 5-set presentation of the aversive stimulus according to a preset random interval schedule [average inter-trial interval (ITI) of 7 XC]. Experimenters were informed of all activities by lights located in an adjacent equipment room. Test /u.sli. The apparatus in the testing phase was a modified Manipulandutn Type S test originally designed by Turner and Solomon (I 962) and used by Hiroto (1974) as a human analog to the two-way shuttle box used in animal studies. The manipulandum was 23 x 5 x 6” with a 4” Plexiglas knob protruding from the top. The knob slid along a 12 x I” straight channel on the cover of the box. Attached to the knob on the underside of the channel was a Plexiglas disc which made contact with the 6 hidden switches (see Fig. 1).
Colored
slgnal
lights
White communication
\
\
(A) Pretreatment button
(B) task
-
Test
push
Fig. I
task
-
\ shuttle
box
Learned
helplessness
615
in groups
Once again communication between the Ss was by way of white lights (6 lights), located on a wooden backboard 18y in height, which were connected with the 6 hidden switches of the partner’s shuttle. The 6 lights traced the movements of the partner’s shuttle handle; e.g. when S 1 moved his shuttle handle from the extreme left to the extreme right. the white lights of Ss panel lit one at a time showing the position of the partner’s handle. This allowed each S to know the movements of his partner. Above these communication lights were the series of colored signal lights: 1 red, 1 yellow and 1 green. The aversive stimulus was the white noise played through the headsets at 90dB.
Ss were introduced to the experimenter and seated. Each S’s apparatus was located on a desk in front of him. Between the two desks was a barrier that prevented the Ss from seeing each other. The Ss were instructed that they were to use no form of communication other than the apparatus that was given them (no talking, hand signals etc.). The Ss then filled out the BDI, a Multiple Affect Adjective Check List (MAACL) and a consent form. The MAACL (Zuckerman and Lubin, 1965), which indexes depression, anxiety and hostility, was also administered after the pretreatment and after the test task. Pretwmtmrnt phase. All Ss in these 2 conditions received the same instructions. They were told:
“During this part of the experiment you will both be wearing these headsets that are to the left of you, and from time to time a loud noise will come on for a while. When that noise comes on, there is something that you two as a team can do to stop it. There is some predetermined, alternating sequence of button pushes that you and your partner can perform that will stop the noise. By alternating, we mean that each button push of one subject must be followed by a button push of the other subject. It is your collective task to try to figure out the solution to how the noise can be turned off. By finding the solution to the problem quickly and consistently you can minimize the amount of unpleasant noise that you will receive. There will be a randomly predetermined number of trials in which you are to solve one sequence, and this sequence will remain constant throughout the entire set of trials. The noise is scheduled to come on at predetermined random intervals. Now there are several lights located on the base in front of you. The three white lights are synchronized with the set of buttons that your partner operates. When he pushes his right button, your right light goes on etc. (demonstration). This will enable each of you to know what button you and your partner are pushing. On the panel in front of you. there is also a set of colored lights. These lights will serve as signals for you. If the red light goes on yfter the noise stops, then your response did not turn off the noise, but rather the noise has stopped automatically according to a preprogrammed schedule. If, on the other hand, the green light comes on after the noise stops, then your response has turned off the noise and you have found the solution to the problem. You should attempt to have as little aversive noise as possible, so that by consistently turning off the noise you will suffer from the noise as little as possible. Taking the headphones off or dismantling the apparatus in any way are not acceptable ways of turning off the noise. Only responding while the noise is on counts. O.K. are are there any questions before we begin? Before we get started, I would like to give you a sample of what the noise will be like, so that you can decide whether or not you wish to continue (one presentation of white noise at 90 dB). At the sound of the next noise the first trial will begin.” A sequence of three button pushes (FR-3) was set up for the S condition. required to push his/her left-most button, followed by the other S pushing finally the first S pushing his middle button. This alternating sequence of to be performed without any other responses intervening, but responses sequence did not detract. This alternating sequence was required to occur
The S on his left-most three button made before within 5 set
the left was button and pushes had the correct to terminate
the noise and obtain a green signal light which indicated that they had found the correct sequence. If the correct response was not performed. the noise stopped automatically at 5 set and the S’s’ received a red signal light. indicating that their response did not terminate the noise. Response latencies were measured by experimenters with a hand-held stopwatch. Ss were allowed a minimum of 30 trials and a maximum of 50 trials in which to achieve IO consecutive correct responses. This was to ensure that all groups learned the task and ended with IO consecutive correct responses. All Ss met this criterion. Groups in the U condition were yoked to groups in the S condition. Here the amount of aversive noise the Ss received was determined by the recorded response latencies of the matched S group. There was no correct response for the U groups so Ss always received the red signal light at the termination of the noise. The 0- conditon is not the conventional control condition used in helplessness research. Usually, the control condition receives no pretreatment. but in the S and U group prior experience with our means of communication was a considerable advantage. So, some form of practice was in order. The Ss were instructed as follows: “You have been randomly chosen to be in a group that will practice in sort of a game. As you can see, there is a set of buttons and lights in front of you. The yellow light will come on at predetermined random intervals. Each time the yellow light does come on. we would like both you and your partner to begin pushing buttons. The only requirement is that you press buttons in what we call an alternating sequence. By alternating. we mean that each button push of one subject must be followed by a button push of the other subject. Now there are no correct or incorrect responses. that is no success or failure. We merely want you to practice different sequences of button pushes with your partner. We are interested in seeing what different types of sequences you both can come up with. The yellow light will remain on for five seconds at a time. Please respond only while the yellow light is lit. There will be a randomly predetermined number of trials. Now you can see there are three white lights located on the base in front of you and these three white lights are synchronized with the set of buttons your partner operates. When he pushes his right button your right light goes on etc. (demonstration). This will enable each of you to know what buttons both you and your partner are pushing.” Twenty trials in which the yellow light was lit took place. After the pretreatment task. Ss in all conditions were again administered the MAACL Form and were told that they were to “check off each and every adjective that applies to how you are feeling at this moment”. Test phrc~. The test phase was the same for all conditions. The shuttle-box task was used and the operation of the tracking lights was demonstrated. Instructions were similar to the S and U pretreatment instructions and the operation of the signal lights was explained. Ss were told that there would be 20 trials Again, Ss were given 5 set on each trial in which to perform the correct response. The correct response was for 1 S to move the shuttle handle to one end of his shuttle box. and his partner to move the handle to the opposite end. Thus, there were actually four correct patterns. Six dependent measures were obtained for each group: (a) number of successes (escapes); (b) mean response latencies (with a maximum of 5 set per trial); (c) trials to criterion. defined as the trial number of the fifth successive trial with a response latency of less than 5 set: (d) Ss subjective recall of correct response; (e) conditional probability of performing a correct rcsponsc given a correct response on the previous trial; (f) MAACL scores. After 20 trials were completed all Ss were given the MAACL Today Mood Inventory for the third time. and a short questionnaire. After completing these forms. all Ss were debriefed and paid. Rrsults When 2 individuals acting in concert received inescapable noise, their concerted performance on a new. but escapable, test was disrupted. When 2 individuals acting in concert first received escapable noise. their test performance was facilitated.
Learned
helplessness
617
in groups
20
Gr - Gr
18 16 14
SD
Condition
Fig. 2. Mean number
of escapes,
400
6.66
4.67
S
SDS and significance level for Experiment 0 groups.
1 (Gr-Gr)
for the S, U and
Mean escapes Analysis of variance of mean number of escapes (Escapes x Condition) revealed a significant main effect, F(2, 33) = 10.39, P < 0.0001. Newman-Keuls* tests of mean number of escapes shows that the S condition performed significantly better than the U condition, P < 0.01. The S group performed marginally better than the 0 condition (P < 0.06). The U condition escaped significantly worse than the 0 condition, (P < 0.05). The latency, trial to criterion and conditional probability variables all revealed similar results to the escape variable. Recall of correct response by S’s self-reporting After the test task Ss were asked to write down what they believed the correct response was to the shuttle-box task. All Ss attempted some form of an answer. We accepted more complex solutions as correct as long as they contained the proper solution within them. In the S condition, 19 out of 24 Ss knew the correct response, significantly more than the 11 out of 24 and the 10 out of 24 in the 0 and U conditions, respectively, who knew the correct response. MAACL
scores for depression
Condition x Sex analysis of MAACL, 1 (at Time l), depression revealed that conditions did not differ on mean Depression score upon starting the experiment (P > 0.25). This analysis was followed by a Condition x Sex x MAACL, 1, 2, 3 (repeated measures) analysis. MAACL Depression scores did not change significantly within or between conditions. EXPERIMENT
2 (In-Gr)
Method Subjects As in Experiment 1 (Gr-Gr), Ss were 36 male and 36 female University of Pennsylvania undergraduates, recruited from class solicitation and from advertisements asking for volunteers for a group-problem solving experiment that would pay $2.50 for approx 1 hr of participation. Assignment to experimental condition proceeded as in Experiment 1 until each condition consisted of 6 male and 6 female groups. The three experimental conditions were: (a) solvable (S); *All paired
comparisons
are done
by the Newman-Keuls
procedure.
(b) unsolvable (c) ;t practice Thcrc
(U): and ~rctreatn~etlt
were no significant
(0) control
differences
group.
of mean
BDI score between
the 3 conditions.
(P > 0.25).
P~t,t~~~~it~~~~,~~t t(r,di (~~l~/i~.~f~u~ti). The apparatus is the same as described in Experiment I with the f~~llowing changes. The white communication lights remained in place but they were shut off. Thus. during the pretreatment phase. the Ss could not communicate and were not aware of the responses of their ‘partner’. The signal lights served the same functions as they did in Experiment 1 (Gr-Gr) except that the feedback was now independent, that is one Ss red light could be illuminated while the other’s green light was illuminated. Ss were not able to observe the response of their ‘partner’. The aversive stimulus was the white noise delivered as in Experiment I, except that the noise to 1 S could be terminated independently of the other. Test tusks. The apparatus was exactly as in the test phase of Experiment 1.
Each S was seated with the apparatus on a desk in front of him. A barrier prevented direct visual ~ornn~uni~~ition between S’s and they were instructed that they were to use no form of communication other than the apparatus that was given them. The BDI. MAACL and a consent form were tilled out. Pwtwutm~tlt pl~rrsc~.S ~td U cotditions. All Ss in these 2 conditions received the same instructions. These instructions were similar to the pretreatment instructions in Experiment I except that the Ss’ escape was not said to be interdependent, nor was it. The FR-3 response was correct for the S condition and it had to be performed within 5 sec. An individual performing the correct response within the 5 set terminated only his noise and only he received a green light. ,Ss were allowed a minimum of 30 trials and a maximum of 50 trials in which to achieve IO consecutive correct responses. If 1 S reached his criterion before his ‘partner‘. he simply received no more trials. The LI Ss were yoked to members of the S group, with one member of the U group yoked to one member of the S group. Each S in the U group received a red light after termination of noise, regardless of his response. Prtrt,/ic,cJ pr-c~tr~~utnlcnt. The S and U conditions in this experiment followed the procedures of the conventional conditions in the helplessness literature except that 2 Ss are in the room at one time. As in Experiment I the 0 condition is a practice, noncommunication condition. Ss were instructed as follows: “You have been randomly chosen to be in a group that will participate in a short game. As you can see, there is a set of buttons in front of you. The yellow light will come on, we would like each of you to begin pushing buttons. The only requirement is that you press buttons in a sequence. Now there are no correct or incorrect responses, that is no success or failure. We merely want you to practice diti’erent sequences of button pushes. We are interested in seeing what different types of sequences you come up with. The yellow light will remain on for 5 set at a time, and plcasc only respond while the yellow light is lit. There will be a randomly predetermined number of trials.” Twenty trials in which the yellow light was lit then took place. After the pi-etreatl~lent task, Ss in all conditions were administered the MAACL for a second time. Test p/rtr.w. The test phase followed the same procedure as in Experiment I with the correct response requiring the joint efTort as described above. Thus. in this experiment the Ss received individual prctrcatment and were tested in groups (In-Gr). The 5 dependent measures were again: (a) number of successes (escapes); (b) mean response latencies: (c) trials to criterion: (d) Ss’ subjective recall of correct response: ie) ~ondition~~l
Learned
helplessness
20
In - Gr
SD
Condition
Fig. 3. Mean
number
in groups
of escapes
620
502
532
S
0
U
and, SDS for Experiment
2 (In-Gr)
for the S, U and 0 groups.
probability of a correct response given a correct response on the previous trial; and (f) mean MAACL scores for depression. After 20 trials, all Ss got their third MAACL and a short questionnaire. After completing these Ss were debriefed and paid. Results The results indicate
that individual
pretreatment
does not affect concerted
group
performance.
Mean escupes Individual pretreatment led to no significant differences in mean number of escapes conditions when testing was done as a group [F(2, 33) = 0.17, P > 0.251. Similar results the rest of the measures.
between hold for
Mean MA ACL scores ,for depression A Condition x Sex analysis of mean MAACL Depression scores revealed an inadvertant, but significant main effect for Condition and a possibly more interesting effect for the Condition x Sex interaction (P < 0.05). The 0 group was more depressed at the start of the experiment (P < 0.05). Further analysis showed that males differed across conditions on Depression score (P < 0.01) but females did not (P > 0.05). Because of this, a Condition x MAACL, 1, 2, 3 Depression (repeated measures) was done for females only (Table I). The Trial x Condition interaction was significant [F(4,66) = 4.06. P < 0.011. The mean Depression score increased significantly after pretreatment for the U group (P < 0.01) but not for the other 2 groups. So, individually-induced helplessness may have made females more depressed.
EXPERIMENT
3 (Gr-In)
Method Su@ects Ss were 18 male and 18 female University of Pennsylvania undergraduates, recruited and paid as in Experiments 1 and 2. Ss were assigned randomly to each of the 3 experimental conditions until there were 6 males and 6 females in each condition. Ss were run in same-sex pairs resulting in 6 groups per condition (3 male groups, 3 female groups per condition). The experimental conditions are the same as in the previous 2 experiments. The mean BDI scores for the conditions did not differ (P > 0.25).
620 Table 1. Means and SDS of MAACL depression after training (2) and after test (3) for Experiment S
scores before (I), 2 (In-Gr; N = 12)
U
0 P > 0.05
,
r
P
kk 10.67 5.38
&
P < 0.05 14.33 5.16 16.42 4.46 15.167 4.71
x SD _ $
Il.33 5.57 9.41 s.33
18.75 4.33 18.50 4.42 16.50 4.275
MAACL,
S;
MAACL,
St
14.83 4.45 14.67 4.89
14.08 6.67 14.25 4.41
9.75 7.747 14.917 6.16
J
MAACL,
St
15.42 4.03
14.58 6.19
12.33 6.91
>
MAACL, MAACL,
FWil&S P < 0.01 P < 0.05
Apparatus Pretreatment task. The apparatus and details of the aversive stimulus (white noise) are exactly as described in Experiment 1. Test [ask. The apparatus is the same as described in Experiment 1 (Gr-Gr) but it was modified in an analogous way to the pretreatment apparatus modifications in Experiment 2. The white communication lights above the shuttle box remained in place but were removed from the circuit. Thus, no communication took place between Ss. The white noise and the colored signal lights could be independently presented to the Ss.
Pretreatment phase. The pretreatment procedure is exactly as outlined in Experiment 1 (Gr-Gr). Tesr phase. This test phase was the same for all conditions. Ss performed independently and had 5 set on each trial to perform the correct response (the correct response required the S to move the shuttle handle all the way to the right end of his or her shuttle box and then to the left end). In this experiment Ss received their pretreatment in groups and were tested individually (Gr-In), so the 6 dependent measures are measures of individual performance, not group performance, as in the first 2 experiments.
-
r
20
Gr In
16 2 g
13.50
14
13 33
:: m
12-
';; IO2
8-
E
6-
p
42-
I
Fig. 4. Mean
number
SD
6 30
Condltlon
S
of escapes
565
5 96
0
U
and SDS for Experiment
3 (Gr-In)
for the S, U and 0 groups.
Learned helplessness
in groups
621
Results Group
helplessness
did not have strong
effects on individual
escape performance.
The S and 0 groups had virtually the same number of mean escapes while the U group had fewer than both, but this difference did not reach significance (P > 0.1). Similar results hold for the other dependent variables.
An analysis of variance revealed no significant differences between conditions for Depression scores upon entering the experiment. The mean Depression score and Condition x Sex x MAACL I, 2, 3 Depression analysis of this data showed no systematic changes across the experiment.
To assess whether experiencing the pretreatment in a group vs individually affected later group performance. the results of Experiment 1 (Gr-Gr) were compared to Experiment 2 (In-Gr) on 4 of the dependent measures: (a) mean escapes; (b) mean response latencies; (c) trials to criterion; (d) conditional probability of a correct response given a correct response of the prior trial. The Condition x Experiment interaction was significant [F(2,66) = 5.31, P -c 0.011. Ss pretreated with unsolvable noise as a group had ,f&er mean escapes when later tested as groups than those pretreated individually when later tested as groups (P < 0.01). The other groups did not differ significantly across the experiments. Similar findings were obtained with the other measures.
DISCUSSlOb!
Learned helplessness can be produced in groups. When 2 individuals acting in concert learned that their joint efforts were to no avail in turning off noise (group pretreatment), they later escaped poorly in concert when noise was actually escapable (Experiment 1). Such group helplessness training, while impairing performance, did not make the individual feel more depressed, however, as it commonly does when individuals are made helpless with inescapable noise (Miller and Norman, 1979). There is no parallel debilitation when individuals are first made helpless singly and then concerted performance is tested in pairs (In-Gr, Experiment 2). Although their group performance was unaffected by individual helplessness pretreatment, they probably did become more depressed, however. Similarly. there appears to be little transfer of helplessness from group to individual testing. When 2 individuals acting in concert had their joint effort rendered useless, they performed only slightly and not significantly worse than controls when tested singly (Gr-In, Experiment 3). We tentatively suggest that group helplessness is mediated by the same mechanism as individual helplessness. When an individual acting in concert with another finds that noise goes off independently of their concerted responses, each forms an expectation that future concerted responses with that other person will be ineffective. Consequently, when placed with that other person in a new task in which concerted action is called for. each individual shows two deficits: (I) a lnotivational deficit, with the probability decreased of each initiating voluntary responses to obtain the desired outcome jointly: (2) a cognitive deficit, with each individual having more difficulty learning that the groups’ concerted action succeeds, even when it actually does. Consequently. the concerted performance of the group will show both fewer escape responses executed and more difficulty learning to escape, even when successful responses are made, What about transfer of helplessness from groups to individuals and from individuals to groups? We suggest that the domain of the expectation that responding will be futile depends on what unit of response was in fact trained to be helpless. That is, if group action was futile, the individual comes to expect that future group action will again be futile, but not necessarily that his own individual action again will be futile (e.g. “My partner is incompetent, but I’m O.K.“). Similarly if an individual learns that he, singly. was helpless, he need not expect that later concerted action with another will be futile (e.g. “I’m incompetent. but my new partner may salvage us.“).
622
DAVIIJ K. SIMKIU cl cl/
This interpretation can be cast into attributional terms (Abramson, Seligman and Teasdale, 1978). During typical individual helplessness training, an individual often attributes his failure to a relatively internal cause, e.g. his own incompetence. While more external attributions can be made (e.g. the experimenter’s unfairness), they are probably less frequent (Abramson, 1978). During group helplessness training, the presence of a coworker explicitly allows more room for a more external attribution (e.g. my coworker’s incompetence or our joint incompetence). So, we speculate that group helplessness training produces more external attributions for failure than does individual helplessness training. Now consider the implications of such attributions if subjects are tested either as a group or individually. After group helplessness training, what will happen when the same group is faced with a new task’? Attributions to either the coworker’s or joint incompetence will produce an expectation So, group that joint responding will be in vain, which will in turn, lower response initiation. helplessness training should produce group helplessness later. In contrast, if the individuals are tested alone after group helplessness training, individual helplessness should be less likely. If the cause of the group failure was external-my coworker’s or our joint incompetence--this does not imply individual incompetence (e.g. he may be stupid or we may not be able to solve problems as a team. but this doesn’t mean I can’t solve them alone). Similar logic holds for individual helplessness training followed by group testing: the individual having failed alone makes a relatively internal attribution. Now, confronted with a task that has a coworker to help out, an acquired internal attribution for failure will not be as damaging (e.g. even if I am incompetent, we might still succeed because my coworker might be extremely competent, or we might work well as a team). So. on attributional grounds, group helplessness training is likely to produce later failure as a group. In contrast. group helplessness training is less likely to produce later individual failure, just as individual helplessness training is less likely to produce later group failure, since in both cases, the attributions for failure made in training are less relevant to the testing situation. This account remains speculative, however, since we did not query attributions after helplessness training. Mood, unlike performance, may not be governed in this way. Group helplessness debilitated group performance but did not depress individual mood (Experiments 1 and 3). while individual helplessness seemed to lower mood, but did not worsen group performance (Experiment 2). Depressed mood, we believe, stems directly from expectations about individual helplessness and only derivatively from expectations about group helplessness. In conclusion, 2 individuals acting in concert. whose actions are ineffective in turning off noise show later deficits at escaping noise together in a new situation when noise is actually escapable. They do not show such helplessness, however. when tested alone. Individuals, whose lone responses are ineffective in escaping noise, do not show later escape deficits when acting together in pairs. We suggest that individuals form expectations.about the inefiectiveness of their joint responses, as well as about their individual responses. \Ye further suggest that an expectation that joint action will be ineffective lowers the probability that joint action will be initiated. and makes it more difficult for these individuals to learn that their joint action has succeeded. .Ickrlo~~./e~~~~~mc,n/s~We wish to thank Lyn Abramson. Lauren Alloy, Paul Eelen, Amy Semmel. Joe Volpicelli and the helplcssncss seminar for guidance and comments. This research was supported by a PHS Grant. MH19604 to Martin E. P. Seligman, a Guggenheim Fellowshtp to Martin E. P. Scligman and NSF Grant No. BNS 76-22943.A02 to the Center for Advanced Study in the Behavioral Sciences, Stanford, Calif. REFERENCES Abramson Abramson abnorm.
L. (1978) Unpublished Doctoral Dissertation. Univ. of Pennsylvania, Philadelphia, Penn. L., Seligman M. E. P. and Teasdale J. (1978) Learned helplessness in humans: critique and reformulation. Psvchol.
81, 32-48. Depression: Clinical,
J.
Beck A. T. (1967) Experimental. and Theoretical Aspects. Hoeber, New York. Garber J. and Seligman M. E. P. (Eds) (1981) Human Helolessness: Theory und Atukations. Academic Press, New York. Hiroto D. S. and Skligman M. E. P. (1975) Generality of learned helplessness in man. J. Person. sec. Ps_vchol. 31, 31 l-327. Miller I. and Norman W. (1979) Learned helplessness in humans. Psychol. Bull. 86, 93-l 18. Sacco W. P. and Hokanson J. E. (1978) Expectations of success and anagram performance of depressives in public and private settings. J. abnorm. Psychoi. 87, 122-130. Turner L. and Solomon R. L. (1962) Human traumatic avoidance learning: theory and experiments on the operantrespondent distinction and failures to learn. P.YLY-hol. Monogr. 76, 40 (Whole No. 599). Zuckerman M. and Lubin B. (1965) Manual for the Multiple Affkct Adjective Chech List. EDITS. San Diego. Calif.