evolutionary ultimatum game with responders dissatisfaction
TRANSCRIPT
Evolutionary Ultimatum Game with Responder's Dissatisfaction
Yougui WangDepartment of Systems Science, School of Management,
Beijing Normal University, Beijing, P. R. China
2008-06-30
Courses
Introduction Standard Game Game with Responder’s Dissatisfaction Discussion and Conclusion
The Ultimatum Game Two people divide a pie or a sum of money:
total sum: m
proposer’s offer: p
responder: accept/reject
proposer responder
accept
reject
(m-p, p)
(0,0)
p
Game Theory:
Responder accept any offer p>0
Proposer offer the smallest share ε
Sub-game perfect equilibrium: (ε, accept)
Experimental Results:
Proposers’ offer : 40%-50% of the “pie”.
When offer was less than 30%, half of the responders reject it.
Explanation of Experimental Results Utility functions
: monetary payoff : other aspects players care about (fairness, envy, altruism, etc) Weakness of this explanation Bounded rationality Ignorance of learning effect Individual purpose
))(,( 1 nxxfUU
ix
Evolutionary Ultimatum Game Repeated games Players interact with each other more than once.
Adaptive Strategy
Players change their strategies
Successful strategy spread
Collective Dynamics
Replicator equation
Evolutionary Ultimatum Game (Nowak, 2000; Page, 2000, 2002, Hardling, 2007).
1. Roles of the players were decided randomly
2. Only monetary payoff was considered
Our work
1.The role of player is fixed. (Heterogeneous)
2. Responder’s dissatisfaction is taken into account
The standard game
The reward of money is standardized to 1 H=“high offer” h, L=“low offer” l. (h>l>0) A=“accept”, R=“reject”.
proposer
H
L
responder
responder
(1-h, h)
A
R
(1-l, l)
(0, 0)
A
Proportion of players with each strategy
Responder Proposer
“accept”: x high offer H: y
“reject”: 1-x low offer L: 1-y
Unsymmetrical replicator dynamics
)( rrxx
)( ppyy
Players change their strategies according to the payoffs of their “parents” or how much they received last time.
Equilibrium of Standard game Differential equations
)]1()1)[(1(
)1)(1(
lxhyyy
lyxxx
The evolutionary stable equilibrium:
Meaning: All proposers willing to offer the low share l; all responders accept it.
The low offer is chosen freely below the high one, so the equilibrium will lead to sub-game perfect: (ε, accept)
)0,1( yx
Responder’s Dissatisfaction When offered l, responder was dissatisfied
proposer
H
L
responder
responder
(1-h, h)
A
R
(1-l, l-c)
(0, 0)
A
The dissatisfaction c will be incorporated into the responder’s choice when she chooses strategies and is assumed the same among all responders.
Analyzing the differential equations
The case of
The evolutionary stable equilibrium:
Meaning: All proposers will offer the high share h and all the responders plan to choose “reject” if they receive the low offer l
l is not enough to compensate dissatisfaction
)]1()1)[(1(
)1)(1(
lxhyyy
xlcxxx
lc )1,0( yx
The case of
The planar steady state solutions:
The line and respectively orientate the evolution of y and x.
Define
lc
)]1()1)[(1(
)1)(1(
lxhyyy
xlcxxx
)0,0( yx )0,/)(( ylclx ]1,0[1 xy
)1/()1( lhx lclx /)(
lcllhk /)()1/()1(
(1) k=0
x
1
y
10L
H
L
CLx
1
1
“origin”
indifferent equilibrium
No Stable Equilibrium
(2) k>0
x
1
y
10L
Hx
1
1
“origin”
“saddle”
L
CLx
indifferent equilibrium
Evolutionary stable Equilibrium: )1,(L
CL
(3) k<0
x
1
y
10L
Hx
1
1
“origin”
“convergence”
L
CLx
indifferent equilibrium
Evolutionary stable Equilibrium: )0,(L
CL
Discussion
Dissatisfaction parameter c is assumed as a constant, high offer is normalized to 0.5
Define
Calculate the identification term
clchllkllf )()1()( 2
)())(( ksignlfsign
4/134)2/1( 22 cccc
Function with different levels of C)(lf
Conlusions The proposer may be more selfish when the
dissatisfaction is small, but the strategy of too small offer can not prevail. Responder will reject the unfair offer with a remarkable proportion.
When the dissatisfaction is large enough, proposer will be afraid of rejection for bringing such feeling to responder.
People would maintain their rights and interests
well if they had strong feelings of unfairness.
Thanks!