Cognitively-Constrained Learning from Neighbors
45 Pages Posted: 4 Dec 2019
Date Written: July 18, 2019
Agents in a network want to learn the state of the world from their own signals and their neighbors’ reports. But they are cognitively-constrained: they have finite and heterogeneous cognitive abilities. We model cognitive ability as a measure of their sophistication when they reason on behalf of a chain of neighbors. Using a tractable learning rule, agents identify old information and extract new signals from their neighbors to the best of their cognitive abilities. We characterize a cutoff level of cognitive ability for each agent, which depends only on the network structure. Using this property, we show that agents need very moderate levels of cognitive abilities to learn correctly in some environments. But an agent mislearns if her cognitive ability falls short of her cutoff.
Keywords: cognitively-constrained learning rule, mislearning in networks, depth of reasoning
JEL Classification: D03, D83, D85
Suggested Citation: Suggested Citation