Influence of Repetition through Limited Recall

TL;DR

This paper models how imperfect recall in decision makers affects learning from social media feeds, revealing inefficiencies like double-counting, and demonstrates interference effects through experiments, impacting information diffusion and truth discernment.

Contribution

It introduces a stylized model of learning with imperfect recall, analyzing interference effects and their impact on information aggregation in social networks.

Findings

Imperfect recall causes double-counting and extreme opinions.

Interference weakens among participants with differing opinions.

Learning accuracy diminishes as the number of senders increases.

Abstract

Decision makers who receive many signals are subject to imperfect recall. This is especially important when learning from feeds that aggregate messages from many senders on social media platforms. In this paper, we study a stylized model of learning from feeds and highlight the inefficiencies that arise due to imperfect recall. In our model, failure to recall a specific message comes from the accumulation of messages which creates interference. We characterize the influence of each sender according to the rate at which she sends messages and to the strength of interference. Our analysis indicates that imperfect recall not only leads to double-counting and extreme opinions in finite populations, but also impedes the ability of the receiver to learn the true state as the population of the senders increases. We estimate the strength of interference in an online experiment where participants are exposed to (non-informative) repeated messages and they need to estimate the opinion of others. Results show that interference plays a significant role and is weaker among participants who disagree with each other. Our work has implication for the diffusion of information in networks, especially when it is false because it is shared and repeated more than true information.