Download Intelligence Personality and Gains from Cooperation in Repeated Interactions PDF

TitleIntelligence Personality and Gains from Cooperation in Repeated Interactions
LanguageEnglish
File Size2.7 MB
Total Pages144
Document Text Contents
Page 1

INTELLIGENCE, PERSONALITY AND GAINS FROM

COOPERATION

IN REPEATED INTERACTIONS

EUGENIO PROTO, ALDO RUSTICHINI, AND ANDIS SOFIANOS

Abstract: Intelligence and personality significantly affect social outcomes of indi-
viduals. We study how and why these traits affect the outcome of groups, looking
specifically at how these characteristics operate in repeated interactions providing
opportunity for profitable cooperation. Our experimental method creates two groups
of subjects who are similar but have different levels of certain traits, such as higher
or lower levels of intelligence, Conscientiousness and Agreeableness. We find that
intelligence has a large and positive long-run effect on cooperative behavior when
there is a conflict between short-run gains and long-run losses. Initially similar,
cooperation rates for groups with different intelligence levels diverge, declining in
groups of lower intelligence, and increasing to reach almost full cooperation levels in
groups of higher intelligence. Cooperation levels exhibited by more intelligent sub-
jects are payoff sensitive, and not unconditional. Personality traits have a natural,
significant although transitory effect on cooperation rates.
JEL classification: C73, C91, C92, B83
Keywords: Repeated Prisoner’s Dilemma, Cooperation, Intelligence

1. Introduction

Intelligence and personality affect individual behavior in social environments.

With no interaction, as in single-agent decision problems, the relationship is straight-

forward. For example, when the task involves generosity to others, and the trait is

the Altruism facet of Agreeableness, we can expect generous behavior to increase

with the Altruism score; when the task is a cognitive task (as in a parlour game) we

can expect behavior to be closer to optimal in individuals with higher intelligence.

Date: September 27, 2016.
The authors thank several co-authors and colleagues for discussions on this and related re-

search, especially Gary Charness, Pedro Dal Bó, Drew Fudenberg, Guillaume Fréchette, Gianluca
Grimalda, John Kagel, David Levine, Josh Miller, Mahnaz Nazneen, Andrew Oswald, Antonio
Penta, Doris Pischedda, Louis Putterman, Carlo Reverberi, Andrew Shotter. and the participants
of the NBER Economics of Culture and Institutions 2015 Meeting in Boston. We thank CAGE
(The Center for Competitive Advantage in the Global Economy) and the Behavioural Science Global
Research Priority for generous funding. AR thanks the NSF, grant SES-1357877. AS thanks the
support by the Economic and Social Research Council [grant number ES/J500203/1].

Page 2

2 EUGENIO PROTO, ALDO RUSTICHINI, AND ANDIS SOFIANOS

When the interaction is strategic, instead, the link may be complex. If the interac-

tion is simple, as in strictly competitive games, the link might be the one we have

seen for single-agent problems. But in games where equilibrium does not provide

us with a unique solution, a completely new approach is needed. This is what we

develop here. First, consider the matter of intelligence. We use here the widely

accepted definition proposed in a 1996 report by a Task Force created by the Board

of Scientific Affairs of the American Psychological Association (Neisser et al., 1996).

There, intelligence is defined as “the ability to understand complex ideas, to adapt

effectively to the environment, to learn from experience, to engage in various forms

of reasoning, to overcome obstacles by taking thought.” If this definition is adopted,

the relationship between intelligence and outcomes for a single individual is natural

and clear. Higher intelligence functions, everything else being equal, as a technolog-

ical factor; it allows larger, faster and better levels of production. This prediction

is natural and is also supported by extensive research in psychology and economics

(Neal and Johnson, 1996; Gottfredson, 1997; Bowles et al., 2001; Heckman et al.,

2006; Jones and Schneider, 2010)

The relationship between intelligence and outcomes is less clear when one instead

considers the link between intelligence and strategic behavior, and when one seeks

to explain how the outcomes of groups are influenced. The technological factor

becomes less important, since social outcomes depend less on skill and more on the

behavior of others. A conceptual link is missing.

Choice as a cognitive task.A possible conceptual link between intelligence and be-

havior in social situations is to view choice in economic and social interactions as a

cognitive task; the link follows as a corollary. This produces the general view that

intelligence reduces behavioral biases (e.g. Frederick, 2005; Oechssler et al., 2009;

Dohmen et al., 2010; Benjamin et al., 2013). For example, higher intelligence may

reduce violations of transitivity; or, in choice under uncertainty, the behavior of

subjects with higher intelligence is better described by expected subjective utility.

When we apply this intuition to behavior in strategic environments, we are led to

the conjecture that more intelligent individuals in real life - and more intelligent

subjects in an experiment - will exhibit a behavior closer to the game theoretic

predictions. When refinements of the Nash concept are relevant, particularly sub-

game perfection, behavior more in line with the prediction of the refinement for the

individual is expected in subjects of higher intelligence.

This prediction finds some support when games are strictly competitive (such as

the Hit 15 game in Burks et al. (2009)). Palacios-Huerta and Volij (2009) show

Page 72

A-14 EUGENIO PROTO, ALDO RUSTICHINI, AND ANDIS SOFIANOS

session of the C-split treatment. The top row of figure A.2 presents the distribu-

tions of the Agreeableness scores in the A-split treatment and the bottom row the

distributions of Conscientiousness scores in the C-split treatment (tables A.27 up

to A.30 present a description of the main data in different separated sessions and

tables A.38 and A.39 show the correlations among individual characteristics).

Table A.45 contrasts the main characteristics of the participants across high- and

low-A sessions. Overall, we can say that the two pairs of sessions are very similar

in all characteristics and have a very similar representation of the different culture

groupings as seen in the bottom panel of table A.45. There is slight difference in

the gender composition in the two groups for which we account for in the statisti-

cal analysis of the data. Table A.46 contrasts the main participant characteristics

between the high- and low-C sessions. It’s apparent that few characteristics have

significant differences between the two groups. This could be due to a relationship

between Conscientiousness the trait and the characteristics listed here. Nevertheless,

we control for these characteristics in the statistical analysis of the data.

C.6. Timeline of the Experiment.

Day One.

(1) Participants were assigned a number indicating session number and specific

ID number. The specific ID number corresponded to a computer terminal

in the lab. For example, the participant on computer number 13 in session

4 received the number: 4.13.

(2) Participants sat at their corresponding computer terminals, which were in

individual cubicles.

(3) Instructions about the Raven task were read together with an explanation

on how the task would be paid.

(4) The Raven test was administered (30 matrices for 30 seconds each matrix).

Three randomly chosen matrices out of 30 tables were paid at the rate of 1

GBP per correct answer.

(5) The Holt-Laury task was explained on a white board with an example, as

well as the payment for the task.

(6) The Holt-Laury choice task was completed by the participants (10 lottery

choices). One randomly chosen lottery out of 10 played out and paid (Sub-

jects in sessions 1 & 2 of the high continuation probability treatment and

sessions 1-2 of low continuation probability treatment did NOT have this).

(7) The questionnaire was presented and filled out by the participants.

Page 73

INTELLIGENCE AND PERSONALITY IN REPEATED INTERACTIONS A-15

Between Day One and Two.

(1) Allocation to high and low groups made. An email was sent out to all

participants listing their allocation according to the number they received

before starting Day One.

Day Two.

(1) Participants arrived and were given a new ID corresponding to the ID they

received in Day One. The new ID indicated their new computer terminal

number at which they were sat.

(2) The game that would be played was explained on a white-board (in the

Minnesota sessions no white-board was available so the game was explained

by using examples on the participants’ screens), as was the way the matching

between partners, the continuation probability and how the payment would

be made.

(3) The infinitely repeated game was played. Each experimental unit earned

corresponded to 0.004 GBP.

(4) In the combined treatment participants completed a decoding task and a

one-shot dictator game.

(5) A de-briefing questionnaire was administered.

(6) Calculation of payment was made and subjects were paid accordingly.

C.7. Dates and Details. Tables A.1 up to A.7 below illustrate the dates and

timings of each session across all treatments. In the top panels the total number

of subjects that participated in Day 1 of the experiment is listed and by comparing

with the corresponding ’Total Returned’ column from the bottom panels it becomes

apparent that there is relatively small attrition between Day 1 and Day 2. For

example, for the high delta treatment, only 10 subjects out of 140 did not return on

Day 2.

Page 143

INTELLIGENCE AND PERSONALITY IN REPEATED INTERACTIONS A-85

T
a
b
l
e
A
.6
3
.

In
d

iv
id

u
a
l

D
e
c
la

r
e
d

P
r
e
fe

r
r
e
d

S
tr

a
te

g
ie

s
T

h
is

ta
b

le
is

b
a
se

d
o
n

th
e

q
u

es
ti

o
n

s:


If
m

y
o
p
p
o
n

e
n

t
d
o
e
s

n
o
t

co
o
p
e
ra

te
w

it
h

m
e

th
is

m
a
n

y
ti

m
e
s,

th
e
n

I
w

il
l

n
o
t

co
o
p
e
ra

te
e
it

h
e
r:

1
;

2
;

3
;

4
;

m
o
re

(5
)”

a
n

d


If
I

st
o
p

co
o
p
e
ra

ti
n

g
,

th
is

is
:

F
o
r

e
v
e
r

(=
0
);

U
n

ti
l

o
th

e
rs

st
a
rt

co
o
p
e
ra

ti
n

g
a
g
a
in

(=
1
)”

δ
=

0
.7

5
δ

=
0
.5

0
δ

=
0
.7

5
IQ

G
ro

u
p
/
P

a
rt

it
io

n
H

ig
h

L
o
w

D
iff

.
H

ig
h

L
o
w

D
iff

.
C

o
m

b
in

e
d

H
ig

h
C

o
m

b
in

e
d

L
o
w

D
iff

.

S
tr

a
te

g
y

A
lw

a
y
s

C
o
o
p

e
ra

te
.0

9
0
9

.0
7
8
1

.0
1
2
8

.0
7
4
1

.0
4
2
8

.0
3
1
3

.0
5
6
6

.1
4
0
3

-.
0
8
3
7

G
ri

m
a
ft

e
r

1
D

.0
3
0
3

.0
3
1
2

-.
0
0
0
9

.0
1
8
5

.0
4
2
8

-.
0
2
4
3

.0
5
6
6

.0
1
7
5

.3
9
1

T
it

fo
r

T
a
t

.1
5
1
5

.0
9
3
7

.0
5
7
8

.2
0
3
7

.1
2
8
6

.0
7
5
1

.2
4
5
3

.1
9
3
0

.0
5
2
3

G
ri

m
w

e
a
k

.0
4
5
4

.0
7
8
1

-.
0
3
2
7

.0
7
4
1

.0
5
7
1

.0
1
7
0

.0
9
4
3

.0
5
2
6

.0
4
1
7

T
it

F
o
r

T
a
t

w
e
a
k

.6
5
1
5

.5
9
3
7

.0
5
7
8

.5
5
5
6

.5
.0

5
5
6

.8
4
9
0

.8
0
7
0

.0
4
2
0

N
.

S
u
b

je
c
ts

6
6

6
4

5
4

5
6

5
3

5
7

L
e
g
e
n
d

A
lw

a
y
s

C
o
o
p

e
ra

te
=

st
o
p

c
o
o
p

e
ra

ti
n
g

a
ft

e
r

m
o
re

th
a
n

4
D

e
fe

c
ts

;
G

ri
m

a
ft

e
r

1
D

=
st

o
p

c
o
o
p

e
ra

ti
n
g

fo
r

a
ft

e
r

1
D

e
fe

c
t

A
N

D
fo

r
e
v
e
r;

T
it

fo
r

T
a
t

=
st

o
p

c
o
o
p

e
ra

ti
n
g

fo
r

a
ft

e
r

1
D

e
fe

c
t

A
N

D
c
o
o
p

e
ra

te
a
g
a
in

a
ft

e
r

C
o
o
p

e
ra

ti
o
n
;

G
ri

m
w

e
a
k
=

st
o
p

c
o
o
p

e
ra

ti
n
g

a
ft

e
r

1
,2

,3
o
r

4
D

e
fe

c
ts

A
N

D
fo

r
e
v
e
r;

T
it

fo
r

T
a
t

w
e
a
k

=
st

o
p

c
o
o
p

e
ra

ti
n
g

fo
r

a
ft

e
r

1
,2

,3
o
r

4
D

e
fe

c
ts

A
N

D
c
o
o
p

e
ra

te
a
g
a
in

a
ft

e
r

C
o
o
p

e
ra

ti
o
n

N
o
te

:

p


v
a
lu
e
<

0
.1

,


p


v
a
lu
e
<

0
.0

5
,



p


v
a
lu
e
<

0
.0

1

Page 144

A-86 EUGENIO PROTO, ALDO RUSTICHINI, AND ANDIS SOFIANOS

(Eugenio Proto) Department of Economics, University of Warwick,

E-mail address: [email protected]

(Aldo Rustichini) Department of Economics, University of Minnesota, 1925 4th Street

South 4-101, Hanson Hall, Hanson Hall, Minneapolis, MN, 55455

E-mail address: [email protected]

(Andis Sofianos) Department of Economics, University of Warwick,

E-mail address: [email protected]

Similer Documents