Adaptive Models and the Power of the Incumbent

The spatial model of voting is a benchmark in theories purporting to explain political behavior. Yet, evidence against the accuracy of both the assumptions and predictions of the spatial model is accumulating in the scholarly literature. One long-held result of spatial theory is that incumbents always lose elections to challengers. Despite the fact that empirical results fail to confirm this finding, political science has not been able to explain why there exists such a glaring difference between theory and actual elections. In large part, spatial theory's failure to illuminate problems of this kind stems from its reliance upon an unrealistic model of human cognition: substantive rationality. By assuming that political agents possess complete information, formal theoretic approaches have discarded the central dynamic of political choice. This proposal will offer an alternative approach that seeks to model agents (parties and voters) as limited information processors. Central to this idea is the role of attention in decision making. Simply put, political agents possess limited amounts of attention to allocate to political issues, and do so in a dynamic fashion. Complexity theory provides a framework to model the use of information by political actors and has been utilized in this paper to construct a computational experiment of elections that provides an explanation for why incumbents fare so well on election day. The results not only illuminate the nature of the incumbency, but also argue for a new focuses on the role of information in political choice.