This paper investigates the effect of voters’ short-term memory on political outcomes by considering politics as a collective learning process. We find that short-term memory may lead to cycles of polarization and consensus across parties’ platforms. Following periods of party consensus, short-term memory implies that there is little variation in voters’ data and therefore limited information about the true state of the world. This in turn allows parties to further their own interests and hence polarize by offering different policies. In contrast, periods of polarization and turnover involve sufficient variation in the data that allows voters to be confident about what the correct policy is, forcing both parties to offer this policy.

In this paper we analyze elections when voters underestimate the correlation between their information sources (“correlation neglect”). We find that this cognitive bias can improve political outcomes. We show that the extreme beliefs which result from correlation neglect induce some voters to base their vote on information rather than on political preferences. We characterize conditions on the distribution of preferences under which this induces higher vote shares for the optimal policies and better information aggregation.

We develop a dynamic model of political competition between two groups that differ in their subjective model of the data generating process for a common outcome. One group has a simpler model than the other group as they ignore some relevant policy variables. We show that policy cycles must arise and that simple world views—which can be interpreted as populist world views—imply extreme policy choices. Periods in which those with a more complex model govern increase the specification error of the simpler world view, leading the latter to overestimate the positive impact of a few extreme policy actions.