-
JOURNAL OF THE EXPERIMENTAL ANALYSIS OF BEHAVIOR
TAKING AND THE DISRUPTION OF COOPERATION'DAVID R. SCHMITT AND
GERALD MARWELL
UNIVERSITY OF WASHINGTON AND UNIVERSITY OF WISCONSIN
Subjects could either cooperate or respond on a lower-paying
individual task. In Exp. I andII, either subject could make a
response that took $1.00 of the other's earnings wheneversubjects
chose to cooperate. The two experiments differed as to whether
taking responseswere effective continuously or intermittently. Both
experiments showed that the opportu-nity to take disrupted
cooperative behavior. Experiment III indicated that if taking was
pos-sible regardless of whether the subjects cooperated or
responded on the individual task,subjects either cooperated or
terminated the experiment.
Several experiments have shown that humancooperative responses,
like individual behavior,are controlled by their relation to
reinforce-ment (e.g., Azrin and Lindsley, 1956; Cohen,1962;
Lindsley, 1966; Mithaug and Burgess,1968). Marwell, Schmitt, and
Shotola (inpress) have explored the effects of another
re-inforcement variable on rate of cooperative be-havior.
Cooperation yielded greater reinforce-ment than non-cooperation but
also entailedthe risk that either subject could take a speci-fied
amount of the other's earnings. If the alter-native of taking was
available, most subjectsfirst took and then opted not to cooperate
byengaging in an individual task yielding less re-inforcement.The
three experiments reported here specify
further some of the conditions under whichthe opportunity to
take from a partner dis-rupts ongoing cooperative behavior. The
firsttwo experiments compared the effects of differ-ent schedules
under which the taking responsewas effective. In Exp. I, the
response could oc-cur whenever both subjects chose to cooperate.In
Exp. II, the response could occur intermit-tently. Experiment III
investigated the effectof taking when this response was possible
dur-ing either cooperative or individual activity.
'This research was supported by a grant from theNational Science
Foundation (GS-1695). Experiment Iwas part of a larger study in
which Dr. Robert Shotolacollaborated. Reprints may be obtained from
David R.Schmitt, Department of Sociology, University of
Wash-ington, Seattle, Washington 98105; or from GeraldMarwell,
Department of Sociology, University of Wis-consin, Madison,
Wisconsin 53706.
EXPERIMENT I
METHOD
SubjectsFour pairs of college students (three female
and one male) volunteered to work as paidparticipants.
ApparatusEach of the subject rooms contained a table-
mounted panel 9 by 18 in. (22.9 by 45.7 cm)with a plunger
(Lindsley knob), a switch forchoosing to work on a cooperative or
an indi-vidual task, a button for taking money fromthe partner,
stimulus lights, and two add-sub-tract counters. All functions were
labelled. Adiagram of the panel is shown in Fig. 1. Thewhite light
in the lower right illuminated for0.1 sec or for 3 sec whenever the
other subjectmade a response. The green lights next to thecounters
flashed for every reinforcement countregistered. Each count was
worth 0.1 cent.Counters showed accumulated earnings for thecurrent
2-hr session only. For all sessions afterthe first, both subjects'
previous totals werewritten on a blackboard on the wall
directlybehind the panel. A closed-circuit televisionreceiver to
the left of the panel showed theamount of money available to each
subject forwork on each of the tasks.
ProcedureSubjects reported to separate waiting rooms
in the laboratory area and were escorted indi-405
1971, 15, 405-412 NUMBER 3 (MAY')
-
DAVID R. SCHMITT and GERALD MARWELL
10 counts equal 1 cent Panel onYour money Q
If yellow light is on, c 010131514|81pressing this buttonwill
give you $ 1.00of other's money
Other's money Light flashes whenWork with O e other personother
person lol3 514181 pulls knob
() Ot If blue light is onother person has_
Work alone chosen to work with you
plungerFig. 1. Diagram of subject panel.
Table 1Conditions Defining Training and Baseline Segments
ReinforcersCooperation
Segment (Each Subject) Individual Length of SegmentSESSION 1
1 training (cooperation only) 0.64 none 335 cooperative
responses2 training (In this and all sub-
sequent segmnents cooperationor individual responding
waspossible.) 0.14 0.64 135 individual responses
3 training 0.64 0.14 135 cooperative responses4 baseline-no
taking 0.34 0.24 15 min
SESSION 25 baseline-no taking 0.34 0.24 30 min6 taking 0.34 0.24
75 min7 baseline-no taking 0.3a 0.24 15 mi
SESSION 38 baseline-no taking 0.34 0.24 30 mi9 taking 0.34 0.24
60 min10 baseline-no taking 0.34 0.24 15 min
SESSION 411 baseline-no taking 0.34 0.24 30 min12 taking 0.34
0.24 90 min
SESSION 513 baseline-no taking 0.34 0.24 30 min14 taking 0.34
0.24 60 min15 baseline-no taking 0.34 0.24 15 min
r
406
-
TAKING AND COOPERATION
vidually to their experimental rooms. A sub-ject did not meet
his partner at any time dur-ing the experiment.
Before the first segment of the training ses-sion (see Table 1)
subjects were instructed inthe operation of the cooperative task.
Only co-ordinated responses were reinforced. Eithersubject could
pull first, thus illuminating thewhite response light on the
other's panel for 3sec. Reinforcement (a counter advance of
sixcounts) occurred whenever the second subjectpulled his plunger
in the 0.5-sec period afterthe response light went out; this
response il-luminated the response light on the other sub-ject's
panel for 0.1 sec. If the second subjectpulled the knob before or
more than 0.5 secafter the response light went off there was
nocounter advance. Instead, the next responseby either subject
reinitiated the other's re-sponse light for 3 sec. Consecutive
pulls by asubject responding first also initiated the oth-er's
response light for 3 sec. Each reinforce-ment was followed by a
2-sec period duringwhich cooperative responses were not
rein-forced. The red panel light was turned off dur-ing this
period. (See Schmitt and Marwell,1968, for a more complete
discussion of thecharacteristics of the task.)
Before Segment 2, subjects were instructedin the operation of
the individual task and thetask selection switch. On the individual
task,each pull of the plunger advanced the counterand illuminated
the response light on the othersubject's panel for 0.1 sec. To
equate the fre-quency of counter advances in time under thetwo
conditions, each individual response wasfollowed by a 5-sec period
in which the redpanel light was turned off and responses werenot
reinforced. The cooperative response withits initial 3-sec interval
and 2-sec timeout alsorequired a minimum of 5 sec to be
completed.Each subject could choose to work on either
the individual or cooperative task by operatingthe toggle switch
on his panel. The coopera-tive task could be performed only if both
sub-jects chose to work together. If either or bothchose to work
alone, the individual task oper-ated for both subjects. A blue
light on eachsubject's panel was illuminated whenever hispartner
switched to "work with other person''.Thus, each subject was always
informed of hispartner's task choice.
In Segment 2, each subject received 0.6 centfor the individual
task and 0.1 cent for cooper-
ation. In Segment 3, these reinforcer magni-tudes were
reversed.In the final 15 min of the training session
(Segment 4) subjects worked under the base-line condition.
Subjects received 0.3 cent forcooperating and 0.2 cent for working
individu-ally. Subjects working for 0.2 cent on the indi-vidual
task could earn approximately $1.25per hour, an amount previously
found suffi-cient to maintain participation in the experi-ment.
Session 2 began with 30 min of the baselinecondition (Segment 5)
to determine whethersteady-state cooperation would be maintainedby
the 0.1-cent difference (steady-state coopera-tion was defined as
at least 80% of a baselinesegment on the cooperative task). At the
be-ginning of Segment 6, a cover was removedfrom the take button
and the following in-structions were given to the subjects:Each of
you now has a button which youcan press to take money from the
otherperson. Each time you press the buttonyou take one dollar from
the other per-son and give one dollar to yourself. If theother
person has less than one dollar,pressing the button will take away
almostall of his money, down to about one cent.
Each subject pressed the button and observedthe resulting
transfer. Approximately 1 minwas required to transfer $1.00 after
the takebutton was pressed. During this time thepanel light went
off, a buzzer sounded contin-uously, and neither cooperation,
individualresponses, nor taking was reinforced.
Subjects were then instructed in the func-tion of the
task-selection switch in taking:Both of you turn your switches to
"workalone". As you see, the amber light showsthat the "take money"
button is now off.The button is off whenever either one orboth of
you switches to "work alone".It is on only when you both switch
to"work with the other person".The task-selection switch remained
opera-
tive during the transfer of money. Thus, ifeither subject
switched to the individual task,the take button would be
inoperative whenthe transfer was completed.A 75-min take segment
followed the instruc-
tions. In all but Session 4 this segment wasfollowed by a 15-min
baseline segment in
407
-
DAVID R. SCHMITT and GERALD MARWELL
which the take buttons were inoperative andcovered. The take
segment was 90 min in Ses-sion 4 and 60 min in Sessions 3 and 5. At
thebeginning of the third and all subsequent ses-sions, a 30-min
baseline segment precededthe take segment. As in Session 2, this
segmentpermitted each of the subjects to earn approxi-mately $1.00
before any opportunity to take,and in addition determined the
reversibilityof any effects.
RESULTSFigure 2 shows the cooperative and individ-
ual response rates for each pair during totaltime spent under
baseline and taking condi-tions in Sessions 2 through 5. With no
oppor-tunity to take (baseline), pairs spent almost allof their
time cooperating. Individual respond-ing was limited to occasional
single responsesat the beginning of a segment and a number
ofresponses by one pair during the final 15-minbaseline segment in
Session 5. With taking pos-sible, all four pairs of subjects
evidenced sub-stantial disruption of cooperation. Three ofthe pairs
stopped cooperating almost entirelyand the subjects worked
individually.
In the three non-cooperative pairs, takingwas confined to
Sessions 2 and 3. Typically, aswitch to cooperation by one subject
was fol-lowed by a nearly simultaneous switch to co-operation with
taking by the other, who then
w
z
C,)w
U)z
0
Ul)w COOPERATIVE
---- INDIVIDUAL
BASELINE TAKE
Fig. 2. Cooperative and individual response rates dur-ing total
time under baseline and take conditions.
switched immediately back to the individualtask. In two of these
pairs, only one subjecttook: 15 times in one pair and three times
inthe other. In the third pair, one of the subjectstook once in
Session 2. The pair began Session3 by cooperating, but after 13
cooperative re-sponses, the subject who had lost in Session 2took.
The pair did not cooperate for the re-mainder of the take segments.
In these threepairs, cooperation resumed whenever
baselineconditions were reinstituted and the oppor-tunity to take
was absent.The fourth pair cooperated for substantial
periods during each session. Both subjects tookfour times in
Session 2 but cooperated totallythroughout Sessions 3 and 4.
However, duringthe final 8 min of Session 5, one subject againtook
and chose to work individually. Atyp-ically, cooperation did not
reappear during thesubsequent baseline period.
EXPERIMENT IIExperiment I showed that the continuous
availability of a taking response disrupted co-operation.
Experiment II was designed to de-termine if the intermittent
availability of thisresponse would produce similar results. In
thisexperiment, taking was possible at irregularintervals during
the experimental segments.
METHOD
Subjects and ProcedureFive pairs of female college students
worked
under the same procedures as used in Exp.I with the following
exceptions:
1. Reinforcer magnitudes in the baselineand initial take
segments were 0.4 cent for eachcooperative response and 0.3 cent
for each in-dividual response (compared with 0.3 centand 0.2 cent
respectively in Exp. I). Experi-ment II was conducted several
months afterExp. I and the increases were instituted to con-tinue
to make the minimum amounts subjectscould earn comparable to the
increasing scaleof other university work. For pairs C-H andT-K it
was necessary to increase the reinforcerfor cooperation to 0.5 cent
during the baselinesegment in Session 2 to produce almost
totalcooperation. When these pairs proceeded tothe take segment,
the larger reinforcer wasmaintained.
2. In each session, a 30-min baseline pre-
408
-
TAKING AND COOPERATION
ceded a 90-min take segment, and sessionsended after the take
segment.
3. During the take segments in Sessions 2and 3, the opportunity
to take was scheduledintermittently rather than continuously.
Two-minute periods with taking possible werescheduled at irregular
or variable times thataveraged 30 min (VT 30-min schedule).
Three2-min periods occurred during the 90-min seg-ment. The take
button during the 2-min pe-riods was operative (indicated by the
amberlight next to the button) only if both subjectshad switched to
the cooperative task. At othertimes, a switch to cooperation did
not maketaking possible. Thus, for subjects who coop-erated, the
amber light periodically flashed onwithout warning, indicating that
taking wasnow possible. If subjects did not cooperate, nosignal
indicated the occurrence of the take pe-riods.
Subsequent sessions for each pair dependedupon the stability of
the pair's previous be-havior, determined by inspecting the
cumula-tive records. If cooperation did not develop,the magnitude
of the reinforcer for coopera-tion was increased to 0.9 cent. If a
stable pat-tern of partial or total cooperation resultedunder the
VT 30-min schedule, the frequencyof 2-min take periods was
increased to an aver-age of one every 5 min (VT 5-min). Finally,
forpairs that cooperated under the VT 5-minschedule, taking was
made possible continu-ously during the periods of cooperation (as
inExp. I).
RESULTSAs Fig. 3 shows, the VT 30-min schedule,
with taking possible less than 7% of the time,resulted in little
or no cooperation for threeof the five pairs when they received 0.4
or 0.5cent for cooperation. Two of these pairs, L-Gand L-S, never
cooperated or took during twosessions. The third pair, C-H,
cooperated dur-ing the first half of Session 2 before one
subjecttook. The pair then worked individually forthe remainder of
Session 2 and all of Session3. The final two pairs, T-K and H-L,
both co-operated often and took whenever possible.Overall rates of
cooperation, however, werelower than in the baseline periods. Pair
T-Kworked three and H-L four sessions under theVT 30-min
schedule.When the VT 5-min schedule was intro-
duced for the two cooperating pairs, T-K con-tinued to cooperate
and take during the first30 min of two sessions. However, for the
re-mainder of these two sessions and during anentire third session,
the pair worked individu-ally. Pair H-L continued to cooperate and
takewhenever possible during a single session. Inthis pair's final
session, taking was possiblecontinuously. Under this condition,
both co-operation and taking ceased.
In two of the three pairs totally disruptedby the VT 30-min
schedule, increasing the re-inforcer for cooperation from 0.4 or
0.5 cent to0.9 cent substantially increased the rate of
co-operation. For pair L-S, steady-state coopera-tion, broken by
periods of taking, emergedafter two sessions. Pair C-H also
cooperatedfrequently, although both subjects switchedrepeatedly
from one task to the other andtook several times during the two
sessions. Thethird pair, L-G, continued to work totally onthe
individual task for two sessions. One or theother of the subjects
occasionally switched tocooperation, but if her partner did
likewise,the other would return immediately to theindividual
task.When the VT 5-min schedule was intro-
duced for the two newly cooperating pairs,L-S continued to
cooperate and no taking oc-curred. The pair remained totally
cooperativeduring a final session when taking was possi-ble
continuously, and also when the reinforcerwas reduced to 0.4 cent
during the last 30 minof the session. For pair C-H, however, the
VT5-min schedule increased taking and individualresponding by the
end of the first session. Dur-ing a second session, the pair worked
solely onthe individual task.A potentially important constant in
Exp. II
was the 2-min period during which the oppor-tunity to take was
present when the pair wascooperating. This time period was
sufficient topermit more than one take by the subjects. Inmost of
the 2-min periods in which subjectstook, more than one take
occurred. UnlikeExp. I, subjects who took did not routinelyswitch
immediately to the individual task. Asa consequence, retaliatory
taking often oc-curred. Thus, the 2-min take periods fre-quently
ended with no net transfer of money.The failure of either subject
to profit, however,did not appear to facilitate future
cooperationor reduce taking, since cooperation was equallylikely
after two takes by a single subject. In ad-
409
-
DAVID R. SCHMITT and GERALD MARWELL
1098765
4 /4~ or 53 ~ ~ /ifor Coop.
w2
2 O05BASELINE TAKE TAKE TAKE
w VT30MIN VT5 MIN CONTINUOUS0~~~~
a. 9C,)Iw 8 I H-L
7 61/ C-H6 A L-SO L-G
- x COOPERATIVE4 / INDIVIDUAL3 A2 1' 90: for Coop.
0/BASELINE TAKE TAKE TAKE
VT30MIN VT5MIN CONTINUOUSFig. 3. Cooperative and individual
response rates during total time under the various schedules. The
upper
half of the figure shows the response rates under the initial
0.4- or 0.5-cent reinforcement magnitudes for co-operation. The
lower half shows the response rates of the three groups who worked
under the 0.9-cent rein-forcement magnitude for cooperation after
little cooperation under the initial magnitudes.
410
-
TAKING AND COOPERATION
dition, continuous cooperation never emergedin the two groups in
which takes by both sub-jects appeared most frequently, T-K and
H-L.The simplest explanation of the multiple takesis that subjects
were trying to maximize thenumber of times they took during the
shorttimes available.
EXPERIMENT IIIExperiments I and II showed that over time,
subjects typically avoided the higher-payingcooperative task
when taking was possible. Ex-periment III determined how
cooperation wasaffected by the opportunity to take, regardlessof
task choice. A similar condition, run for ashort period of time in
the earlier study (Mar-well et al., in press) produced some
increase incooperation.
METHODSubjects and Procedure
Six pairs of college students (five female andone male) worked
under procedures that werethe same as those for Exp. I, except that
dur-ing the first three take segments, taking waspossible
continuously regardless of task choice.If stable cooperation
occurred, the conditionswere changed so that taking was possible
onlywhen subjects cooperated (as in Exp. I and II).Finally, the
earlier continuous condition wasrepeated. The reinforcement
magnitudes werethose used in Exp. II.
RESULTSAs Fig. 4 shows, the continuous opportunity
to take eventually resulted in cooperation infour of the six
pairs. The cooperating pairsdiffered mainly in the time required
before co-operation emerged. Cooperation began earlyin Session 2
for pairs W-S and P-C, in Session3 for pair E-H, and in Session 4
for pair L-P.Taking occurred in each of these pairs and wasnearly
continuous before cooperation. Duringthis period, few task
responses of either kindwere made. When cooperation began,
takingceased for three of the four pairs. The threecontinued
cooperating during the final sessionwhen taking was possible only
during periodsof cooperation.The fourth pair, P-C, began
cooperating
early in Session 2 and continued thereafter ex-
IOr
w
Mz
i"IU)z
O L-P* W-Sa A-BA C-HU P-C* E-H
- COOPERATIVE---- INDIVIDUAL
\0
BASELINE TAKEFig. 4. Cooperative and individual response rates
dur-
ing total time under baseline and take conditions.
cept for several takes by both subjects at theends of Sessions 3
and 4. When in Session5 taking could occur only when subjects
coop-erated, the pair began by cooperating. How-ever, following
taking by both subjects, Sub-ject P, who at that point had most of
SubjectC's money, began working individually. Afterapproximately 30
min, Subject C walked outof the experiment and refused to come
backfor a previously scheduled sixth session.The remaining two
pairs also refused to con-
tinue in the experiment. Although both pairswere originally
scheduled for six sessions, onesubject in each group quit after
Session 2. Inboth pairs, taking in Session 2 had been contin-uous.
In one of the pairs, the subject who quithad earned slightly more
than her partner,while in the other, her partner had takenmost of
the money.
DISCUSSIONThe three experiments demonstrated that
the opportunity to take generally disrupted oreliminated ongoing
cooperative behavior. Dis-ruption was greatest in Exp. I, when
takingcould occur whenever subjects cooperated. Co-operation was
also disrupted in Exp. II wheretaking was possible only
occasionally duringperiods of cooperation. For pairs that
werelargely cooperative when taking was possibleinfrequently,
increased opportunity to takeeliminated cooperation. Even under the
high-
411
-
412 DAVID R. SCHMITT and GERALD MARWELL
est reinforcement magnitude for cooperation,where subjects could
each earn over $5.00 perhour, two of three pairs stopped
cooperatingwhen taking could occur frequently.
In Exp. I, with taking possible continuouslyduring periods of
cooperation, two pairs nevermade any cooperative responses and a
thirdmade very few. Therefore, cooperation wasrarely or never
reinforced in the presence oftake opportunities, a factor that
might accountfor its absence. In Exp. II, however, four
pairsultimately. cooperated for substantial periodsunder the VT
30-min take schedule (two afterthe reinforcement for cooperation
had beenmarkedly increased). Yet in only one did coop-eration
continue when opportunities to takewere more frequent or
continuous. Thus, theresults further emphasize the dependence
ofcooperation on the frequency with which tak-ing is available
regardless of the reinforcementreceived during periods of
cooperation.
In Exp. III, with taking possible regardlessof task choice, the
disruption of cooperationwas temporary. However, taking continued
oreven increased before subjects began cooperat-ing. No pair was
without some taking and intwo of the pairs, taking was frequent and
pro-longed. The aversive character of taking underthese conditions
is suggested by the escape fromthe experiment itself by subjects in
threepairs.
In conclusion, subjects in the present ex-
periments typically made the response thatprovided the greatest
and most immediate re-inforcement, i.e., taking, despite the
conse-quences to the partner and the subsequent lossof
reinforcement when subjects ceased to co-operate. Since Marwell, et
al. (in press) showedthat subjects will not destroy $1.00 of the
oth-er's money when it gives them nothing, the re-inforcement for
taking appears to be crucial.
REFERENCESAzrin, N. H. and Lindsley, 0. R. The reinforcement
of cooperation between children. Journal of Abnor-mal and Social
Psychology, 1956, 52, 100-102.
Cohen, D. J. Justin and his peers: an experimentalanalysis of a
child's social world. Child Develop-ment, 1962, 33, 697-717.
Lindsley, 0. R. Experimental analysis of cooperationand
competition. In T. Verhave (Ed.), The experi-mental analysis of
behavior. New York: Appleton-Century-Crofts, 1966. Pp. 470-501.
Marwell, G., Schmitt, D. R., and Shotola, R. Coopera-tion and
interpersonal risk. Journal of Personalityand Social Psychology (in
press).
Mithaug, D. E. and Burgess, R. L. The effects of dif-ferent
reinforcement contingencies in the develop-ment of social
cooperation. Journal of ExperimentalChild Psychology, 1968, 6,
402-426.
Schmitt, D. R. and Marwell, G. Stimulus control inthe
experimental study of cooperation. Journal ofthe Experimental
Analysis of Behavior, 1968, 11,571-574.
Received I June 1970.
