Abstract
We study a simple model for social learning agents in a restless multiarmed bandit. There are agents, and the bandit has good arms that change to bad with the probability . If the agents do not know a good arm, they look for it by a random search (with the success probability ) or copy the information of other agents' good arms (with the success probability ) with probabilities or , respectively. The distribution of the agents in good arms obeys the Yule distribution with the power-law exponent in the limit and . The system shows a phase transition at . For , the variance of per agent is finite (diverges as with ). There is a threshold value for the system size that scales as . The expected value of the number of the agents with a good arm increases with for . For and , all agents tend to share only one good arm. If the shared arm changes to be bad, it takes a long time for the agents to find another good one. decreases to zero as , which is referred to as the “echo chamber.”
- Received 7 July 2016
DOI:https://doi.org/10.1103/PhysRevE.94.052301
©2016 American Physical Society