Content uploaded by Varda Liberman
Author content
All content in this area was uploaded by Varda Liberman on Jan 04, 2014
Content may be subject to copyright.
10.1177/0146167204264004PERSONALITY AND SOCIAL PSYCHOLOGY BULLETINLiberman et al. / “NAME OF THE GAME” EFFECTS
The Name of the Game: Predictive Power of
Reputations Versus Situational Labels in
Determining Prisoner’s Dilemma Game Moves
Varda Liberman
The Interdisciplinary Center Herzliya, Israel
Steven M. Samuels
U.S. Air Force Academy
Lee Ross
Stanford University
Two experiments, one conducted with American college students
and one with Israeli pilots and their instructors, explored the pre-
dictive power of reputation-based assessments versus the stated
“name of the game” (Wall Street Game vs. Community Game) in
determining players’ responses in an N-move Prisoner’s
Dilemma. The results of these studies showed that the relevant
labeling manipulations exerted far greater impact on the players’
choice to cooperate versus defect—both in the first round and
overall—than anticipated by the individuals who had predicted
their behavior. Reputation-based prediction, by contrast, failed
to discriminate cooperators from defectors. A supplementary
questionnaire study showed the generality of the relevant short
-
coming in naïve psychology. The implications of these findings,
and the potential contribution of the present methodology to the
classic pedagogical strategy of the demonstration experiment, are
discussed.
Keywords: Prisoner’s Dilemma; lay psychology; name of the game;
construal; demonstration experiments
Psychologists have long recognized that understand
-
ing, predicting, and altering human behavior require
that we attend to the manner in which the relevant actors
interpret or “construe” the stimulus situations that con
-
front them (Asch, 1952; Bruner, 1957; Lewin, 1935;
Thomas & Znaniecki, 1918). In this article, we present
three studies that deal with the impact of a labeling
manipulation designed to influence such construals and
with a hypothesized failing in intuitive psychology
(Nisbett & Ross, 1980; Ross & Nisbett, 1991). The first
two studies pit the power of this labeling manipulation
against the predictive power of assessments made about
those actors by people who know them well. A third study
further explores lay beliefs that individuals hold about
the power of situational labels on themselves and their
peers.
The particular context employed in our investigation
was the familiar Prisoner’s Dilemma in which partici-
pants must decide whether to “cooperate” or “defect”
(Luce & Raiffa, 1957). In selecting this context, we
address not only the longstanding dialogue between
social psychology and personality psychology but also a
newer dialogue and seeming tension between the sub
-
jectivist tradition of psychology and the objectivist spirit
of game-theory economics.
Construal and the Prisoner’s Dilemma
The Prisoner’s Dilemma (PD) literature is too exten
-
sive to be reviewed here, but we can note that both altru
-
istic and strategic motives have been offered to explain
why people might choose to cooperate in the N-round
version of the game (see Axelrod, 1984; Dawes, 1980;
Kerr, 1995; Komorita & Parks, 1996; Messick, 1999). An
early finding relevant to our present research is that play
-
ers are more willing to cooperate, even in a single-round
version of the game, when they think that similarly coop
-
1175
Authors’ Note: Correspondence concerning this article should be ad
-
dressed to Lee Ross, Stanford University, Dept. of Psychology, Jordan
Hall, Bldg. 420, Stanford, CA 94305-2130; e-mail: ross@psych.stan
-
ford.edu.
PSPB, Vol. 30 No. 9, September 2004 1175-1185
DOI: 10.1177/0146167204264004
© 2004 by the Society for Personality and Social Psychology, Inc.
erative choices are likely to be forthcoming from their
partner. By the same token, “cooperators” seem to think
that cooperation is a more common and normative
response than do “defectors” (Dawes, McTavish &
Shaklee, 1977; Kelley & Stahelski, 1970).
Even more relevant to our present concerns are previ
-
ous studies that manipulated the way the payoff matrix,
response alternatives, or participants’ roles in various
games or allocation tasks are “framed” (Allison &
Messick, 1985; Dawes, 1980; Samuelson & Allison, 1994;
see reviews by Komorita & Ellis, 1995; Schwartz-Shea &
Simmons, 1995). Studies contrasting conditions in
which the description of the relevant tasks or games
evoked norms related to business dealings versus ethical
dilemmas, or more generally competitive versus cooper
-
ative norms (Allison, Beggan, & Midgley, 1996; Blount &
Larrick, 2000; Larrick & Blount, 1997; Tenbrunsel &
Messick, 1999; van Dijk & Wilke, 2000), are of particular
note.
We expect our present research design to provide fur
-
ther evidence about the malleability of construal pro
-
cesses and to show that influencing such processes can in
turn influence behavioral choices of the sort made in the
PD game. At the same time, we expect to show that
laypeople fail to appreciate the power of such manipula-
tions and instead give undue weight to their perceptions
of individual differences in personal attributes (notably,
cooperativeness vs. competitiveness) that members of
our society infer and rely on in their normal interper-
sonal dealings. Thus, in our first two studies, we employ
an experimental design that allows us to contrast the
power of our specific manipulation—that is, the name
that we attached to the game in describing the relevant
matrix—with the behavioral predictions of people who
knew the individuals well and based their predictions on
that knowledge.
For half of the participants in the studies we shall
describe, the relevant label was the Wall Street Game (or,
in the case of the Hebrew-speaking Israelis in Study 2,
the Bursa Game). For the other half of our sample, the
label in question was the Community Game (or, in
Hebrew, the Kommuna Game). The former label, of
course, connotes rugged individualism, concern with
self-interest, and contexts in which competitive or
exploitative norms are likely to operate. The latter label,
by contrast, connotes interdependence, collective inter
-
est, and contexts wherein cooperative norms are likely to
operate. (No explicit mention of social norms was made,
of course, and attention was not directed to the signifi
-
cance of the relevant name.) Our primary hypothesis
was that the responses of our research participants
would defy both the dispositionist expectations of the
individuals who predicted those responses and also the
dispositionist notions of social science colleagues and of
future psychology students who learn about our
findings.
STUDY 1: REPUTATIONS VERSUS
SITUATIONAL LABELS IN THE PD GAME
Participants
Forty-eight Stanford undergraduate men, selected as
described below, were recruited by telephone. They
were told that they had been selected to participate in a
study on negotiation in which they could “earn money”
but nothing about the criterion for that selection.
Procedure
The study was run in two phases. In Phase 1, dormi
-
tory resident assistants (RAs) heard a detailed account of
the instructions (including two mentions of the name of
the game) and payoff matrix to be used in our study.
They then were asked (a) to nominate the students they
deemed “most likely to play C” (cooperate) and “most
likely to play D” (defect) on the first round of the game
and (b) to indicate the relevant likelihoods. They also
were asked to reassess the relevant likelihood if the stipu-
lated name of the game changed from the Wall Street
Game to the Community Game, or vice versa. In Phase 2,
the nominees participated, face to face, in a seven-trial
PD game, knowing that they and their counterparts
would be paid in accord with the following payoff matrix,
which was presented both on a large poster board at the
front of the room and on a sheet of paper given to each
participant:
Player 2’s Choice
CD
C Player 1 + 40 cents Player 1 – 20 cents
Player 2 + 40 cents Player 2 + 80 cents
Player 1’s Choice
D Player 1 + 80 cents Player 1 zero cents
Player 2 – 20 cents Player 2 zero cents
All participants chosen for this second phase were
individuals whose nominators had assessed them as at
least 85% likely to cooperate or 85% likely to defect—
not only given the originally stipulated name for the
game but also if the alternative name had been stipu
-
lated. Furthermore, the two participants paired together
to comprise a given dyad always had a similar nomina
-
tion status—either most likely to cooperate or most
likely to defect.
The instructions participants received were the same
ones previously presented to the nominators, including
a pair of verbal references to either the Wall Street Game
1176 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
or the Community Game. Aside from this difference in
the name of the game, all nominees were treated identi
-
cally. Before play started, the experimenter made certain
that each player understood the relevant matrix by hav
-
ing them specify the payoff each player would receive
given each combination of moves. Participants were told
at the outset that the game would continue for seven
rounds and that after specifying their own move for a
given round they would learn their counterpart’s move,
and hence what their respective payoffs would be.
The game then commenced and continued until the
seventh round had been completed. Afterward, partici
-
pants completed some additional measures (not dis
-
cussed in this report), were thanked for their efforts, and
were paid the sum they had accumulated over the seven
rounds.
Results and Discussion
Results both for the first round and the seven rounds
overall are presented in Figure 1. It is immediately
apparent that the participants’ nomination status as
most likely to cooperate versus most likely to defect had
no predictive power at all. It is equally clear that the
name of the game exerted a considerable effect on the
participants’ choices. When playing the Community
Game, 67% of the most likely to cooperate nominees
and 75% of the most likely to defect nominees cooper-
ated on the first round. When playing the Wall Street
Game, 33% of participants with each nomination status
cooperated, chi-square (collapsing across nomination
status) = 6.76, p < .01 (Fisher exact p = .02).
Choices over the entire seven rounds presented
essentially the same picture (see Figure 1). This picture
reflects the fact that participants who both cooperated
and received cooperation on the first round generally
continued to cooperate over subsequent rounds,
whereas participants who had defected and/or faced
defection on the first round tended to defect thereafter
(again, without regard to their nomination status). In
fact, individuals nominated as most likely to cooperate
proved to be slightly less cooperative overall (a mean of
3.25 cooperative responses per dyad over the seven
rounds) than participants nominated as most likely to
defect (a mean of 3.58 cooperative responses), F < 1.0.
Furthermore, the name of the game presented to partici
-
pants at the outset of the game continued to make its
impact felt. Individuals playing the Community Game
(collapsing across the nomination status) offered a
mean of 4.63 cooperative responses (i.e., a 66.1% rate of
cooperation). Individuals playing the Wall Street Game
(again collapsing across nomination status) offered a
mean of 2.21 cooperative responses (i.e., a 31.5% rate of
cooperation), F(1, 20) = 11.83, p < .005.
More detailed inspection of the responses of individ-
ual dyads over the seven rounds revealed that in the
Community Game, mutual cooperation occurred on 43
of 84 possible occasions, whereas mutual defection
occurred on 16 occasions. By contrast, in the Wall Street
Game, mutual cooperation occurred on only 11 of 84
possible occasions, whereas mutual defection occurred
on 42 occasions. Significance testing of these differences
is problematic because responses across successive
rounds obviously are not independent. However, we can
note that the mean number of cases of mutual coopera
-
tion per dyad over the seven rounds was greater in the
Community Game (M = 3.58) than in the Wall Street
Game (M = 0.92), t(22) = 3.20, p < .01. This pattern, inci
-
dentally, was apparent even in the final round, where the
strategic demands of the situation might be expected to
induce defection among erstwhile cooperators.
Whereas 9 of 12 Wall Street Game dyads showed mutual
defection on that round, only 3 of 12 Community Game
dyads did so, Fisher exact p = .02.
Our nominators’ confident designations of likely
cooperators and defectors, and their conviction that the
name of the game would exert little impact on their
nominees’ moves, were thus belied by the data. However,
because we utilized only nominees whose nominators
thought they would be relatively unaffected by the name
of the game manipulation, our test of nominators’ accu
-
racy was obviously biased. Accordingly, in Study 2, we
opted to deal with predictions about all of the individu
-
als in a given population of PD players—again, predic
-
tions made by people who knew them well. We also
attempted to determine the effects of the name of the
game on the participants’ expectations about the moves
Liberman et al. / “NAME OF THE GAME” EFFECTS 1177
20
40
10
First Round Cooperation Overall Cooperation
50
30
Wall Street
Game
C
ommunity
Game
Wall Street
Game
C
ommunity
Game
Percent Cooperate
80
70
60
Nomination Status
“Most Likely Cooperators”
“Most Likely Defectors”
Figure 1 First round and overall cooperation in the Community
Game versus Wall Street Game by nominated “most likely
cooperators” and “most likely defectors” (Study 1).
that would be chosen by their counterparts and the asso
-
ciation of such expectations with their own choices.
STUDY 2: REPLICATION AND EXTENSION WITH
ISRAELI PILOTS AND FLIGHT INSTRUCTORS
In this study, we contrasted the PD responses of elite
Israeli pilot trainees with predictions made earlier by
their flight instructors—individuals who interacted with
them extensively and were responsible for assessing their
suitability for particular assignments (including assign
-
ments requiring successful independent or interdepen
-
dent functioning). The design of the study also included
an additional source of predictions—the participants’
self-ratings of potentially relevant traits and similar
rating by their instructors.
Overview
Instructors and trainees in the Israeli Air Force Train
-
ing School participated in this experiment, which was
conducted in Hebrew. The trainees played five rounds of
a PD game, this time featuring a payoff matrix that
offered both players 6 points for mutual cooperation, –6
points for mutual defection, and 8 points and –8 points,
respectively, for trainees who defected in the face of
cooperation or cooperated in the face of defection.
1
For
half of the participants, the relevant game was labeled
the Bursa Game (connoting the competitive norms of
the stock market), whereas for the other half of our par-
ticipants, it was labeled the Kommuna Game (connoting
more cooperative and interdependent norms). The
moves the trainees chose in the face of these two labels
were assessed in light of both prior likelihood assess
-
ments by the trainees’ instructors and of prior trait rat
-
ings made by the instructors and the trainees themselves.
Procedure
Several weeks before our participants played the PD
game, during a routine paperwork task, they were asked
to complete a questionnaire in which they used 7-point
scales to evaluate themselves, relative to the other train
-
ees in their squadron, in terms of 11 trait descriptors,
including competitiveness. Prior to the experimental
sessions, the trainees in each squadron were divided into
two groups, one of which would later play the Bursa
Game and one of which would later play the Kommuna
Game. Half of the instructors made predictions about
the trainees they knew best who would play the Bursa
Game and half made predictions about the trainees they
knew best who would play the Kommuna Game.
Before making their predictions, the instructors
received exactly the same information about the game
that would later be presented to the players themselves.
In addition to making predictions about whether spe
-
cific trainees would choose C or D on the first round,
instructors also were asked to estimate the percentage of
all trainees in the school that would choose C on the first
round and to rate the traits (including cooperativeness),
skills, and habits of the trainees they knew best. Finally,
they were asked to estimate the likelihood of coopera
-
tion versus defection by the trainees if the alternative
name of the game were presented.
During the experimental session itself, trainees were
assigned to play the Bursa Game (20 trainees) or
Kommuna Game (20 trainees) in accordance with their
predetermined experimental condition. The study was
run during an in-class meeting in which participants
from all squadrons were represented. Trainees were
seated in two rows, facing each other, knowing only that
their counterpart was in the row facing them. At the
front of the room, they could see the matrix of the game
clearly visible, with the appropriate name for their ses
-
sion (Bursa or Kommuna) provided. They then were
presented with a PowerPoint presentation about the
rules of the game, emphasizing that they would not know
their counterpart’s identity but that they would learn his
payoff after each round in accord with the relevant
matrix.
They were further informed that they would be play-
ing the game several times with the specific number of
rounds determined by the throw of a die (with the game
concluding when a six was thrown). “The object of the
game,” they were told, is to “accumulate, personally, as
many points as you can.” Participants each got a sheet
presenting the payoff matrix, with the relevant name
(Bursa or Kommuna) written above it.
The game began with the participants writing down
their first move (C or D) on a questionnaire containing
an identification number known only to the experi
-
menter. They also were asked what first-round move they
had expected from the other player. Then they passed
the questionnaires to the experimenter, who shortly
thereafter returned the questionnaires to the partici
-
pants with the other player’s move filled in. At that point,
trainees recorded both their payoff and their counter
-
part’s payoff for the round and (after the experimenter
indicated that he had not thrown a six) the game contin
-
ued until five rounds of the game had been completed
(at which point the experimenter informed the partici
-
pants that the previous round had been their last). A
final questionnaire item then asked participants, “What
percentage of all trainees in the school do you think
would choose move C in the first round of the game?”
Results and Discussion
Our results (shown in Table 1) reveal that both on
Round 1 and on subsequent rounds of the game, the
label Bursa Game prompted higher rates of defection
than the label Kommuna Game. Whereas only 4 of the
1178 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
20 Bursa Game participants opted to cooperate on their
first move, 11 of the 20 Kommuna Game participants did
so (Fisher exact p = .024). Also, the mean number of Cs
per dyad over the five rounds of the game was 4.50 (SD =
3.08) in the Kommuna Game versus 2.50 (SD = 2.23) in
the Bursa Game, t(18) = 1.42, p = 0.086 (one-tailed).
2
Moreover, Kommuna Game dyads achieved mutual
cooperation on a mean of 1.6 rounds per dyad over the
five rounds of the game, whereas Bursa Game dyads
achieved mutual cooperation on a mean of only 0.3 such
rounds per dyad, t(18) = 1.89, p = .037.
Expectations Regarding Counterpart
In the Bursa Game, 10 trainees thought their counter-
part had chosen move D, whereas 9 thought he had cho-
sen C (1 trainee failed to answer the question.) By con-
trast, only 5 of the 20 Kommuna Game trainees thought
their counterparts had chosen move D, whereas 15
thought he had chosen C, Fisher exact p = .07. Similarly,
Bursa Game trainees estimated that only 24.1% of their
peers would opt for first-round cooperation, whereas
Kommuna Game trainees estimated that 45.8% would
do so, t(38) = 3.36, p = .004. Furthermore, there was a
clear association between these percentage estimates
and participants’ own moves (r = .56, p < .02, for trainees
in the Kommuna Game; r = .63, p < .005, for trainees in
the Bursa Game). In this regard, however, there was also
a notable between-condition difference. Not surpris
-
ingly, all participants (in both conditions) who antici
-
pated defection from their counterpart opted to defect
themselves. But whereas 6 of the 9 Bursa Game trainees
who anticipated cooperation from their counterpart
chose D, only 4 of the 15 Kommuna Game trainees who
anticipated cooperation chose D, Fisher exact p = .067.
Instructors’ Predictions
Our findings regarding the impact of the name of the
game on trainees’ first-round PD choices take on partic
-
ular significance in light of the relevant likelihood esti
-
mates made by the trainees’ instructors. As Table 2
makes apparent, the mean likelihood estimates offered
regarding first-round cooperation by trainees were just
as high when the instructors thought they would be play
-
ing the Bursa Game (48.7%) as when they thought they
would be playing the Kommuna Game (48.2%).
What is equally clear from Table 2 is that the instruc-
tors’ likelihood estimates for individual trainees (similar
to nomination status of participants in Study 1) had no
predictive utility; that is, overall, the estimated likeli-
hood of cooperation for the trainees in Study 2 who did
in fact choose C (43.3%) was no higher—in fact, it was
somewhat but not significantly lower—than the esti-
mated likelihood of cooperation for trainees who chose
D (51.6%). Furthermore, this lack of association
between actual and predicted choices was equally evi
-
dent with respect to predictions about Bursa Game train
-
ees and those about Kommuna Game trainees.
Another indication that instructors gave insufficient
weight to the name of the game was provided by their
estimates about the responses of “all trainees in the
school.” The mean estimate made by instructors who
heard the label Kommuna Game was 47.4%, whereas the
mean estimate by instructors who heard the label Bursa
Game was 46.4%. These estimates, it should be noted,
stand in marked contrast to the corresponding estimates
made by the trainees who had actually participated in
the two experimental conditions. Trainees in the
Kommuna Game condition, it will be recalled, predicted
a 45.8% first-round cooperation rate for other trainees
in the school, whereas those in the Bursa Game condi
-
tion predicted a 24.1% rate. One interpretation of this
result is that trainees used their own responses as a proxy
in estimating what their peers would do under the same
circumstances, and their predictions accordingly
reflected the impact of the name-of-the-game manipula
-
tion with reasonable accuracy. Instructors, having no
Liberman et al. / “NAME OF THE GAME” EFFECTS 1179
TABLE 1: Trainees’ Decision to Choose C (Cooperation) Versus D (Defection) in Playing the Bursa Versus Kommuna Game (Study 2)
First Round Decisions
Stipulated Name of the Game
Bursa Kommuna
Chose C 4 (20%) 11 (55%)
Chose D 16 (80%) 9 (45%)
Decisions Over Five Rounds of the Game
Total Proportion of Total Proportion of Mean # of Cs Mean # of Mean # of
Cs Overall Ds Overall per Dyad CCs per Dyad DDs per Dyad
Bursa game 25/100 75/100 2.50 0.3 2.8
Kommuna game 45/100 55/100 4.50 1.6 2.1
such proxy because they had not actually played the
game (and probably did not consider what move they
personally would make), disregarded the stipulated
name and based their estimate on their impressions
about overall trainee cooperativeness.
3
An alternative interpretation is that the name of the
game altered the participants’ perception of what consti
-
tuted normative play in the game and, hence, both how
they played and how they expected the majority of their
peers to play (but did not do so in the case of instructors
who focused on the traits of their trainees rather than on
the relevant situational norms).
The instructors’ response to a stipulated change in the
name of the game was further revealing. Although the
mention of an alternative name may have focused their
attention on behavioral norms, it did not lead them to
make adequate allowance for the relevant difference.
They estimated a mean change in likelihood of coopera-
tion of 12.9 points—again, much less than the actual 35
percentage point effect (55% for Kommuna Game and
20% for the Bursa Game) that was found.
Predictive Utility of Instructors’
and Trainees’ Trait Ratings
Neither the trait ratings offered by the instructors nor
the trainees’ self-ratings proved useful in predicting
either first round or overall PD play. Notably, the mean
instructor ratings of trainee cooperativeness were actu
-
ally marginally lower for trainees who cooperated in the
first round of the game (M = 3.91) than for trainees who
defected on that first round (M = 4.75), t(38) = 1.68, p =
.09. Similarly, the mean self-ratings of competitiveness
for trainees who defected on the first round of the game
(M = 3.85) were actually a little lower (although not sig
-
nificantly so) than the mean competitiveness self-ratings
for trainees who cooperated on that first round (M =
4.23). Furthermore, the self-characterizations by the
trainees proved to be virtually uncorrelated both with
the instructors’ ratings of the trainees’ cooperativeness
and with instructors’ specific predictions about their first
round choice of moves.
The findings of our first two studies suggest that
laypeople give little weight to the name of the game (or,
by implication, other external factors that influence the
way a particular situation is likely to be construed by the
actor responding to that situation). Our third and final
study explores this failure in lay psychology in more
detail. In particular, it asks individuals to make predic
-
tions about the impact of the name-of-the-game manipu
-
lation not only on peers in general and on likely cooper
-
ators and defectors but also on themselves and their best
friend. The design of Study 3 also allows us to contrast
the weight given to the name of the game when that
name is embedded within a general description of the
procedure to be employed and the weight given to it
when the two alternative names are explicitly contrasted.
STUDY 3: LAY PREDICTIONS ABOUT THE IMPACT OF
THE NAME OF THE GAME ON SELF AND OTHERS
Method
In Study 3, a group of students again made likelihood
estimates about the first-round PD game move for a peer
they nominated as most likely to cooperate and for a
peer they nominated as most likely to defect. They also
made similar estimates for self, for best friend, and for
Stanford students in general. In half of the cases, the
name of the game stipulated again was the Wall Street
Game, and in half of the cases, the name of the game stip
-
ulated was the Community Game. (In this study, how
-
ever, because our concern lay solely with the nature of
the relevant predictions, we did not proceed to have the
relevant nominees actually play the game in question.)
The participants in Study 3 later made a second set of
predictions about their “nominees,” this time with a stip
-
ulated change in the name of the game either from the
Wall Street Game to the Community Game, or vice versa.
In half of the cases, predictions were made about all five
target individuals with a given name of the game stipu
-
lated and then about the same individuals with the other
name of the game stipulated; in half of the cases, the
change in the name of the game was stipulated with
respect to each target in turn. The order in which partici
-
pants were asked about the relevant targets (“most likely
cooperators” and “most likely defectors” before self, best
friend, and Stanford students in general vs. “most likely
1180 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
TABLE 2: Instructors’ Likelihood Estimates That Trainees’ First Round Would Be C Rather Than D (Study 2)
Bursa Game Kommuna Game Combined
Mean Likelihood Mean Likelihood Mean Likelihood
Estimates Regarding N Estimate Re C N Estimate Re C N Estimate Re C
Trainees who chose C 4 40.0% 11 44.5% 15 43.3%
Trainees who chose D 16 50.9% 9 52.8% 25 51.6%
All trainees 20 48.7% 20 48.2%
NOTE: C = cooperation, D = defection.
cooperators” and “most likely defectors” after self, best
friend, and Stanford students in general) was also varied.
The result was a 2 × 2 × 2 design that allowed us to exam
-
ine both the impact of the initially stipulated name of the
game on the relevant likelihood estimates and the
degree to which those estimates changed when the rater
was asked about the effect of a change in the name of the
game. The former thus involved a between-subject com
-
parison, whereas the latter involved a within-subject
comparison.
Results and Discussion
Statistics summarizing our participants’ predictions
regarding first-round choices of moves are shown in
Tables 3 and 4 below. Table 3 pertains to differences in
likelihood estimates given the initially stipulated names
of the game. Table 4 pertains to changes in estimates par
-
ticipants made when asked about the effect of a change
in the name of the game from the Wall Street Game to
the Community Game, or vice versa.
The data presented in Table 3 suggest that partici-
pants gave relatively little weight to the initially stipulated
name of the game. The relevant between-condition dif-
ference in estimates was about 11 percentage points in
the case of estimates for best friends and for Stanford
undergrads in general and about 7 percentage points for
estimates about “most likely defectors” (p < .05 in all
three cases). The corresponding difference was 4 per-
centage points in the case of estimates for “most likely
cooperators” and slightly less than 7 percentage points
in estimates for self (p > .10). The actual choices made by
the nominated “most likely cooperators” and “most
likely defectors” in Study 1, it will be recalled, had pro
-
duced a much greater between-condition difference (34
percentage points and 43 percentage points,
respectively).
Making the difference in the stipulated name of the
game salient and explicit (i.e., by asking participants first
for predictions given one name of the game and then for
predictions given the alternative name of the game)
increased the weight given to the name of the game (see
Table 4). Overall (combining across the order in which
the two names were stipulated), the differences in esti
-
mates for the Wall Street versus Community Game were
17.0 percentage points for estimates regarding self, 17.4
percentage points for estimates regarding best friends,
and 18.7 percentage points for estimates regarding stu
-
dents in general. The differences in estimates were
smaller but still highly significant for estimates regard
-
ing “most likely cooperators” (13.6 percentage points)
and “most likely defectors” (11.8 points).
Given our present concerns, however, the most
important thing to note about the means summarized in
Table 4 involves the relationship between the estimated
and actual impact of the name of the game. Even with
the difference in name of the game highlighted by
explicit mention of the alternatives, the allowance for
that difference made by our Study 3 respondents (simi
-
lar to that made by our Study 2 nominators) continued
to be too small. In particular, it continued to be consider
-
ably smaller than the relevant 34 and 43 percentage-
point differences in actual percentages of cooperation
versus defection suggested by our Study 1 results.
GENERAL DISCUSSION
The most important result of our two experiments is
the discrepancy between the actual effects of our “name
of the game” manipulation and the effects anticipated by
those predicting the responses on the basis of extensive
contact with the players. Indeed, predictions about indi
-
vidual players proved valueless, as did trait ratings—
including trait ratings made by the players themselves.
These studies also provide evidence about the mediating
role played by the participants’ expectations about the
Liberman et al. / “NAME OF THE GAME” EFFECTS 1181
TABLE 3: Likelihood Estimates Regarding First Round Cooperation
of Stipulated Players in Wall Street Game Versus Commu
-
nity Game When Names of Game Embedded (Study 3)
Community Game Wall Street Game Estimated
(n = 48) Mean (n = 48) Mean Difference
ML cooperators 81.75 77.50 4.25
ML defectors 23.02 15.98 7.04*
Diff (MLC – MLD) 58.73 61.52**
Self 63.19 56.33 6.86
Best friend 66.62 55.94 10.68*
Stanford undergrads 52.44 41.04 11.40*
NOTE: ML = most likely, MLC = most likely cooperators, MLD = most
likely defectors.
*p < .05.
TABLE 4:
Likelihood Estimates Regarding First Round Cooperation
of Various Stipulated Players When Alternative Names of
Game Explicitly Contrasted
a
(Study 3)
Community Wall Street Estimated
Game Mean Game Mean Difference
ML cooperator 84.52 70.95 13.57***
ML defector 28.23 16.40 11.83***
Diff (MLC – MLD) 56.29 54.55
Self 67.80 50.80 17.00***
Best friend 70.06 52.61 17.45***
Stanford students 56.51 37.82 18.69***
NOTE: ML = most likely, MLC = most likely cooperators, MLD = most
likely defectors.
a. Both orders for originally stipulated and alternative name of game
combined (n = 96).
***p < .001.
play of their counterparts. When told they were playing
the Bursa Game, participants expected defection to be
the most likely response; when told they were playing the
Kommuna Game, they expected cooperation to be the
most likely response. Furthermore, regardless of which
game they were playing, participants who expected
defection from their counterpart overwhelmingly opted
to reciprocate that defection. Expectations of coopera
-
tion, however, had a less straightforward consequence.
Whereas Kommuna Game participants expecting coop
-
eration generally opted to cooperate in return, Bursa
Game participants expecting cooperation generally
opted to exploit that cooperation by defecting. In other
words, the effect of expectations regarding other’s
choices on own choices depended on the name of the
game, and thus on the way the participants construed
the game.
The result of these tendencies over successive rounds,
in which defection begat defection and cooperation was
sustained only when it was mutual, was inevitable
(Axelrod, 1984); that is, first-round responses tended
to dictate later-round responses, and as a consequence,
overall rates of cooperation—especially mutual coop-
eration—were significantly higher in the Community/
Kommuna Game than in the Wall Street/Bursa Game.
Person Versus Situation and the Didactic Strategy
of the Demonstration Experiment
Although our studies obviously speak to the tradi-
tional issue of the “person versus the situation” (see Ross
& Nisbett, 1991), a note of clarification is in order. The
results of our studies illustrated the power of the situa-
tion, or at least the power of situational labels and
construals, but they did not prove (nor were they
designed to prove) that individual differences have no
predictive power in the Prisoner’s Dilemma or other
games presenting players with similar choices and trade
-
offs. Indeed, a search of the literature reveals a number
of studies showing significant cross-situational consis
-
tency in responses to non-zero-sum games that oblige
individuals to choose among the goals of maximizing
joint outcome, maximizing own outcome, or maximiz
-
ing advantage of self over others (e.g., Bem & Lord,
1979
4
; Bennett & Carbonari, 1976; Kuhlman &
Marshello, 1975).
There are even some studies showing significant cor
-
relations between PD play and paper-and-pencil mea
-
sures of specific cognitive abilities (Pincus & Bixenstine,
1979) and/or measures of specific traits, including
locus of control, self-monitoring, Type-A behavior, and
sensation-seeking (Boone, De Brabander, & van
Witteloostuijn, 1999); adherence to Protestant Ethic val
-
ues (Furnham & Quilley, 1989); both Factor G (which
involves moral values and concerns) and Factor E (which
deals with dominance-submissiveness) on the 16 PF
inventory (Gillis & Woods, 1971); as well as various mea
-
sures of cooperative, accommodative, or prosocial (vs.
competitive, egoistic, or exploitative) personal motiva
-
tions and orientations (e.g., Houston, Kinnie, Lupo,
Terry, & Ho, 2000; Parks & Rumble, 2001; Vinacke,
1974). However, it is worth noting that to produce such
statistical significance, some aggregation of the relevant
choice measures typically has been required (whereas in
both of the present experiments the name-of-the-game
manipulation proved powerful enough to produce sig
-
nificant between-condition differences in response on a
single occasion, that is, the first round of the game).
What the present studies dealt with was not the (lack
of) predictive value of individual differences relative to
that of a situational manipulation but rather the (lack
of) predictive value of the type of lay assessments that
people make about their peers—even peers they think
they know well and about whom they are willing to make
relatively confident predictions vis-à-vis play in the Pris
-
oner’s Dilemma. In this context, it is worth noting the
relevance of our present study to the traditional didactic
or pedagogic strategy of the “demonstration experi-
ment” (see Devine & Brodish, in press; Ellsworth &
Gonzalez, in press). A list of classic experiments in this
tradition would include Milgram’s (1963, 1974) obedi-
ence studies, Asch’s (1951, 1956) conformity studies,
Rosenthal and Jacobson’s (1968) studies of self-fulfilling
expectations in the classroom, Freedman and Fraser’s
(1966) foot-in-the-door study, Darley and Batson’s
(1973) bystander intervention studies, and other crown
jewels of experimental social psychology. These studies
did not test falsifiable general propositions about
human behavior, and they did not link the relevant out
-
comes to specific cognitive or motivational mediators.
Instead, they illustrated the power of particular situa
-
tional influences that proved more powerful than our
students or other laypeople (and many of our colleagues
in psychology, as well as those in other disciplines) had
heretofore appreciated. In so doing, they taught us to
make less dispositionist attributions about actors who
are responding to the demands and constraints of their
social contexts (see Nisbett & Ross, 1980; Ross, Greene,
& House, 1977; Ross & Nisbett, 1991).
The studies noted above did not offer this lesson
explicitly. None included measures of lay expectations as
part of their original design (although Milgram,
famously, did collect predictions from colleagues—who
consistently underestimated the relevant level of obedi
-
ence). Instead, the investigators invited their readers to
weigh the relevant findings against their own expecta
-
tions and underlying assumptions about human psychol
-
ogy and then to accommodate those views to the relevant
empirical lessons.
5
The present studies, by contrast, were
1182 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
designed to offer the same lessons (both to future stu
-
dents of social psychology and to our colleagues in eco
-
nomics and other social science disciplines) more
explicitly—that is, by contrasting actual effects of the
manipulations in question with the effects anticipated by
individuals who know the actors well and base their pre
-
dictions and likelihood assessments on that knowledge.
In a sense, what we have provided is an empirical para
-
ble. This parable dealt with the importance of attending
to subjective interpretations or construals, the folly of
overemphasizing the role of objective economic out
-
comes in human decision making, and the dubious
value of impressions about personal dispositions—espe
-
cially when one is called on to make predictions about
the behavior of actors whom one has never observed in
the relevant situations.
Implications for Everyday Attributions and
Searches for Common Ground
In the case of trained social scientists, suggestions
about the importance of subjective construal and the
possible impact of manipulations of such construal are
hardly necessary—at least not when we are perusing the
pages of a scholarly journal. Indeed, when reminded of
earlier classic studies and the situationist and subjectivist
traditions of our field and asked to venture a prediction
about the outcome of the present experiments, many of
our professional peers would predict a significant effect
of the name of the game, and some would even antici-
pate the limited success of the lay personologists who
offered assessments in our studies. Rather, such sugges-
tion becomes most relevant when people, laypeople and
social scientists alike, encounter each other in our every
-
day lives, discuss issues of social policy, or read the daily
news.
Our present findings, moreover, pertain as much to
attributions as they do to social predictions. They imply
that upon observing behavior that challenges our expec
-
tations about a given actor or group we should avoid
making broad dispositional inferences. Instead, we
should entertain the possibility that we have failed to
appreciate the structural demands and constraints of the
situation at hand (Blount & Larrick, 2000; Morris,
Larrick, & Su, 1999) and/or failed to take into account
the way those demands and constraints were interpreted
and linked to preexisting cognitive structures and situa
-
tional norms by the relevant actors. These findings in no
way negate the fact that different individuals may inter
-
pret particular types of situations differently—that the
same discussion may be an invigorating exercise for one
party and an identity-threatening debate for another, or
that one person’s jolly coed softball game may be
another’s test of athletic prowess. Rather, the findings of
our present research suggest that most of us are
equipped with a range of different cognitive schemas
that could be applied to a given situation and that our
behavior may depend on whatever factors determine
which schema happens to be recruited by chance or by
subtle features of the context at hand.
A study coauthored by the third author of this article
(Kay & Ross, 2003)—one directly inspired by the first of
the three studies we reported here—illustrates this
point. It showed that a simple priming manipulation
(having participants unscramble words to form sen
-
tences) could influence the way individuals construed
the norms of the PD game, and hence, the responses
they subsequently made, in the same way as our labeling
manipulation. (Unfortunately, however, no attempt was
made in that study to compare these actual effects with
lay presumptions about the power of the relevant
manipulation.)
An obvious question raised by the present studies is
their relevance to contexts in which participants
respond not to a simple payoff matrix but to a more com
-
plex and realistic task involving social interaction and
negotiation. Can one change the types of offers negotia
-
tors make, and the responses they make, to the offers of
their counterparts, by changing the manner in which the
task at hand is presented (e.g., as a search for justice in
the light of conflicting claims and entitlement vs. a
search for mutual advance over the status quo)?
Although it is difficult to do rigorous random assign-
ment experiments to answer such questions in conse-
quential negotiations that occur outside the laboratory,
the memoirs of skilled mediators (e.g., Saunders, 1991)
suggest the importance of changing perceived norms
and thus the meaning of compromise. There is also
encouraging evidence from within the laboratory tradi
-
tion that different framings of Ultimatum Game and
other resource allocation tasks—both via explicit labels
and descriptions of the task or the player’s roles (e.g.,
Blount & Larrick, 2000; Larrick & Blount, 1997) or by
implicit priming (Kay, Bargh, Wheeler, & Ross, in press)
cues—can make individuals more generous in their
treatment of other players. Let us hope that collabora
-
tion between practitioners and researchers in this
domain will lead both to greater sophistication and even
some new insights about how to produce “communities
of cooperators” (Axelrod, 1984) who thrive and
encourage others to follow their example.
NOTES
1. This payoff matrix was decided on after pretesting indicated that,
in this context, for this particular population, the payoff matrix that
had been used in Study 1 (with no label attached to it) yielded much
more defection than cooperation.
2. Results reflecting the name of the game on choices of coopera
-
tion versus defection constitute a direct replication of Study 1 (neces
-
sarily with smaller Ns because only members of the relevant pilot train
-
Liberman et al. / “NAME OF THE GAME” EFFECTS 1183
ing program could be tested). Accordingly, here and wherever else we
present Study 2 findings relevant to the name-of-the-game effect, the
data were subjected to one-tailed instead of two-tailed tests of signifi
-
cance. It further should be noted that in the case of predictions regard
-
ing play over the entire game, as opposed to those pertaining only to
Round 1, the number of degrees of freedom is halved (because dyads
rather than individuals necessarily become the unit of analysis), which
of course reduces the significance level of any between-condition
differences.
3. One of our anonymous reviewers observed that it would be inter
-
esting to investigate the corresponding attributions made by the play
-
ers, that is, whether, the experience of having played the Prisoner’s
Dilemma (PD) game with a given label protects the players from mak
-
ing unwarranted inferences about other actors who have opted to
cooperate versus defect in the face of the relevant Community Game or
Wall Street Game label—unwarranted inferences to which mere
observers of the relevant behavior would presumably fall prey (see
Morris, Larrick, & Su, 1999).
4. The Bem and Lord study is one investigation that did investigate,
albeit somewhat indirectly, the value of peer assessments in predicting
PD play. The investigators first had a group of five graduate student rat
-
ers produce Q-sorts for the traits they thought might characterize play
-
ers with particular goals (i.e., maximizing own payoff vs. maximizing
joint payoff vs. maximizing the difference of payoff for self relative to
other) in PD-like, non-zero sum games. They then identified a subset of
players who showed high consistency in pursuing those particular goals
when presented with relevant sets of outcome matrices for self and
other. Finally, they showed significant commonalities between the Q-
sorts for those consistent individuals that had been obtained from
peers who knew them well and the Q-sorts produced by the graduate
student raters. The methodology employed and results of this study
offered a thoughtful contribution to then-contemporary debate
between Bem and his critics about the use of more “idiographic” strate-
gies for discovering cross-situation consistency in behavior. But the
authors neither showed nor claimed to show that laypeople could
directly predict the responses of their peers in a given PD game—
certainly not well enough to justify the levels of confidence shown by
the individuals who made predictions in our present studies. (In addi-
tion, there is no reason to imagine that the authors would claim that
the predictive value of the Q-sort procedure they devised was as great as
the power of our present labeling manipulation.)
5. A few subsequent investigators examined whether laypeople
underestimate the power of the particular situational manipulations
involved in these classic studies and accordingly make unwarranted
dispositional inferences. Safer (1980) and Bierbrauer (1979) provided
such evidence with respect to the Milgram situation and Pietromonaco
and Nisbett (1982) did so with respect to the Darley and Batson
manipulation.
REFERENCES
Allison, S. T., Beggan, J. K., & Midgley, E. H. (1996). The quest for
“similar instances” and “simultaneous possibilities”: Metaphors in
social dilemma research. Journal of Personality and Social Psychology,
71, 479-497.
Allison, S. T., & Messick, D. M. (1985). Behaviors or non-behaviors as the
cooperative or defecting responses in two types of social dilemmas. Unpub
-
lished manuscript.
Asch, S. E. (1951). Effects of group pressures upon the modification
and distortion of judgment. In H. Guetzkow (Ed.), Groups, leader
-
ship, and men.Pittsburgh,PA:CarnegiePress.
Asch, S. E. (1952). Social psychology. New York: Prentice Hall.
Asch, S. E. (1956). Studies of independence and conformity: A minor
-
ity of one against a unanimous majority. Psychological Monographs,
70(9, Whole No. 416).
Axelrod, R. (1984). The evolution of cooperation. New York: Basic Books.
Bem, D. J., & Lord, C. G. (1979). Template matching: A proposal for
probing the ecological validity of experimental settings in social
psychology. Journal of Personality and Social Psychology, 37, 833-846.
Bennett, R. P., & Carbonari, J. P. (1976). Personality patterns related
to own-, joint-, and relative-gain maximizing behaviors. Journal of
Personality and Social Psychology, 34, 1127-1134.
Bierbrauer, G. (1979). Why did he do it? Attribution of obedience
and the phenomenon of dispositional bias. European Journal of
Social Psychology, 9, 67-84.
Blount, S., & Larrick, R. P. (2000). Framing the game: Examining
frame choice in bargaining. Organizational Behavior and Human
Decision Processes, 81, 43-71.
Boone, C., De Brabander, B., & van Witteloostuijn, A. (1999). The
impact of personality on behavior in five Prisoner’s Dilemma
games. Journal of Economic Psychology, 20, 343-377.
Bruner, J. (1957). Contemporary approaches to cognition.Cambridge,
MA: Harvard University Press.
Darley, J. M., & Batson, C. D. (1973). From Jerusalem to Jericho: A
study of situational and dispositional variables in helping behav
-
ior. Journal of Personality and Social Psychology, 27, 100-119.
Dawes, R. M. (1980). Social dilemmas. Annual Review of Psychology, 31,
169-193.
Dawes, R. M., McTavish, J., & Shaklee, H. (1977). Behavior, communi
-
cation, and assumptions about people’s behavior in a common
dilemma situation. Journal of Personality and Social Psychology, 35,
1-11.
Devine,P.G.,&Brodish,A.B.(inpress).Modernclassicsinsocialpsy
-
chology. Psychological Inquiry.
Ellsworth,P.C.,&Gonzalez,R.(inpress).Questions,comparisons,
and preparation: Methods of research in social psychology. In M.
Hogg & J. Cooper (Eds.), Sage handbook of social psychology.
Freedman, J. L., & Fraser, S. C. (1966). Compliance without pressure:
The foot-in-the-door technique. Journal of Personality and Social Psy
-
chology, 4, 195-202.
Furnham, A., & Quilley, R. (1989). The protestant work ethic and the
Prisoner’S Dilemma game. British Journal of Social Psychology, 28,
79-87.
Gillis, J. S., & Woods, G. T. (1971). The 16 PF as an indicator of perfor-
mance in the Prisoner’s Dilemma game. Journal of Conflict Resolu-
tion, 15, 393-402.
Houston, J. M., Kinnie, J., Lupo, B., Terry, C., & Ho, S. S. (2000). Com-
petitiveness and conflict behavior in simulation of a social
dilemma. Psychological Reports, 86, 1219-1225.
Kay,A.C.,Bargh,J.A.,Wheeler,J.S.,&Ross,L.(inpress).Material
priming: The influence of mundane physical objects on situa-
tional construal and competitive behavioral choice. Organizational
Behavior and Human Decision Processes.
Kay, A. C., & Ross, L. (2003). Nonconscious influences and conscious
reflection: The interplay of implicit cues and explicit situational
construals in the Prisoner’s Dilemma. Journal of Experimental Social
Psychology, 39, 634-643.
Kelley, H. H., & Stahelski, A. J. (1970). Social interaction basis of
cooperators’ and competitors’ beliefs about others. Journal of Per
-
sonality and Social Psychology, 16, 66-91.
Kerr, N. L. (1995). Norms in social dilemmas. In D. A. Schroeder
(Ed.), Social dilemmas: Perspectives on individuals and groups.
Westport, CT: Praeger.
Komorita, S. S., & Ellis, A. L. (1995). Reward structure and coopera
-
tion in the N-personPrisoner’sDilemma.InD.A.Schroeder
(Ed.), Social dilemmas: Perspectives on individuals and groups (pp. 15-
30). Westport, CT: Praeger.
Komorita, S. S., & Parks, C. D. (1996). Social dilemmas.Boulder,CO:
Westview.
Kuhlman, D. M., & Marshello, A. (1975). Individual differences in the
game motives of own, relative, and joint gain. Journal of Research in
Personality, 9, 240-251.
Larrick, R. P., & Blount, S. (1997). The claiming effect: Why players
are more generous in social dilemmas than in ultimatum games.
Journal of Personality and Social Psychology, 72, 810-825.
Lewin, K. (1935). Dynamic theory of personality. New York: McGraw-Hill.
Luce, R. D., & Raiffa, H. (1957). Games and decisions: Introduction and
critical survey.NewYork:JohnWiley.
Messick, D. M. (1999). Alternative logics for decision making in social
settings. Journal of Economic Behavior and Organization, 39, 11-28.
Milgram, S. (1963). Behavioral study of obedience. Journal of Abnor
-
mal and Social Psychology, 67, 371-378.
Milgram, S. (1974). Obedience to authority: An experimental view.New
York: Harper & Row.
1184 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
Morris, M., Larrick, R. P., & Su, S. K. (1999). Misperceiving negotia
-
tion counterparts: When situationally determined bargaining
behaviors are attributed to personality traits. Journal of Personality
and Social Psychology, 77, 52-67.
Nisbett, R. E., & Ross, L. (1980). Human inference: Strategies and short
-
comings of social judgment. Englewood Cliffs, NJ: Prentice Hall.
Parks, C. D., & Rumble, A. C. (2001). Elements of reciprocity and
social value orientation. Personality and Social Psychology Bulletin,
27, 1301-1309.
Pietromonaco, P., & Nisbett, R. E. (1982). Swimming upstream
against the fundamental attribution error: Subject’s weak general
-
izations from the Darley and Batson study. Social Behavior, 20,1-4.
Pincus, J., & Bixenstine, V. E. (1979). Cognitive factors and coopera
-
tion in the Prisoner’s Dilemma game. Psychological Record, 29, 463-
471.
Rosenthal, R., & Jacobson, L. (1968). Pygmalion in the classroom:
Teacher expectation and pupils’ intellectual development.NewYork:
Holt.
Ross, L., Greene, D., & House, P. (1977). The false consensus effect:
An egocentric bias in social perception and attribution processes.
Journal of Experimental Social Psychology, 13, 279-301.
Ross, L., & Nisbett, R. E. (1991). The person and the situation: Perspectives
of social psychology. New York: McGraw-Hill.
Safer, M. A. (1980). Attributing evil to the subject, not the situation:
Student reaction to Milgram’s film on obedience. Personality and
Social Psychology Bulletin, 6, 205-209.
Samuelson, C. D., & Allison, S. T. (1994). Cognitive factors affecting
the use of social decision heuristics in resource-sharing tasks.
Organizational Behavior and Human Decision Processes, 58, 1-27.
Saunders, H. H. (1991). The other walls: The Arab-Israeli peace process in a
global perspective. Princeton, NJ: Princeton University Press.
Schwartz-Shea, P., & Simmons, R. T. (1995). Social dilemmas and per-
ceptions: Experiments on framing and inconsequentiality. In D. A.
Schroeder (Ed.), Social dilemmas: Perspectives on individuals and
groups (pp. 87-103). Westport, CT: Praeger.
Tenbrunsel, A. E., & Messick, D. M. (1999). Sanctioning systems, deci
-
sion frames, and cooperation. Administrative Science Quarterly, 44,
684-707.
Thomas, W. I., & Znaniecki, F. (1918). The Polish peasant in Europe and
America. Chicago: University of Chicago Press.
van Dijk, E., & Wilke, H. (2000). Decision-induced focusing in social
dilemmas: Give-some, keep-some, take-some, and leave-some
dilemmas. Journal of Personality and Social Psychology, 78, 92-104.
Vinacke, W. E. (1974). Accommodative strategy and communication
in a three-person matrix game. Journal of Personality and Social Psy
-
chology, 29, 509-525.
Received March 19, 2003
Revision accepted December 1, 2003
Liberman et al. / “NAME OF THE GAME” EFFECTS 1185
