AbstractThis paper analyzes a model of learning by imitation, where besides the decision maker, there is a population of individuals facing the same decision problem. We analyze a property called Absolute Expediency, which requires that the decision maker’s expected payoff increases from one round to the next for every decision problem and every profile of realized actions taken by the population. We give a simple characterization of the expediency property and show that its basic feature is proportional imitation: the change in the probability attached to the played action is proportional to the difference between the received and the sampled payoff (the sampled payoff plays the role of an aspiration level).
[1]
J. Cross.
A Stochastic Learning Model of Economic Behavior
,
1973
.
[2]
Jörgen W. Weibull,et al.
Evolutionary Game Theory
,
1996
.
[3]
K. Schlag.
Why Imitate, and If So, How?, : A Boundedly Rational Approach to Multi-armed Bandits
,
1998
.
[4]
Tilman Börgers,et al.
Learning Through Reinforcement and Replicator Dynamics
,
1997
.
[5]
Kumpati S. Narendra,et al.
Learning automata - an introduction
,
1989
.
[6]
Antonio J. Morales,et al.
Expedient and Monotone Learning Rules
,
2004
.
[7]
D. Fudenberg,et al.
The Theory of Learning in Games
,
1998
.