# SFB 303 Discussion Paper No. B - 365

**Author**: Schlag, Karl H.

**Title**: Which one should I imitate?

**Abstract**: We consider the model of social learning by Schlag (1996). Individuals must repeatedly choose an action in a
multi-armed bandit. We assume that each indivdiual observes the outcomes of two other individuals' choices
before her own next choice must be made -- the original model only allows for one observation. Selection of
optimal behavior yields a variant of the proportional imitation rule -- the optimal rule based on one
observation. When each individual uses this rule then the adaptation of actions in an infinite population follows
an aggregate monotone dynamic.

**Keywords**: social learning, multi-armed bandit, imitation, payoff increasing, proportional imitaiton rule, aggregate monotone dynamic.

**JEL-Classification-Number**: C72, C79

**Creation-Date**: March 1996

**URL**: ../1996/b/bonnsfb365.pdf
SFB 303 Homepage

28.04.1998, Webmaster