| This dissertation includes three papers using quantitative models to sensibly describe what kinds of preferences political actors will or actually do hold when existing theory offers no insight. The first two papers use evolutionary game theory to predict ways in which politicians, artificially selected on the basis of good performance to remain in office, will in the long run diverge from instrumental rationality as ordinarily assumed in game theory. The first sets out a general principle for producing models of preference evolution in games as political models, namely, that the information about opponent preferences necessary for evolution of non-rational preferences comes from opponents' previous plays, and applies it to two simple games. The second uses the same principles in more detail on a bargaining game that models the plea negotiations between a prosecutor and a defense attorney, leading to a conclusion that failure to learn from setbacks during a trial is an evolutionarily favored trait among prosecutors. The third paper addresses the ideological preferences of Supreme Court justices, which existing statistical models do not effectively compare to those of elected officials since the two groups never vote on the same items, by identifying a set of political actors with whom both groups commonly interact: organized interest groups who vote on Supreme Court cases with amicus curiae briefs and on electoral candidates using campaign donations. |
