Choice-Free Rationality: A Positive Theory of Political Behavior

Introduction

Introduction The temptation to anthropomorphize is, fortunately, at a historic low. Rain, sunshine, lightning, and tornadoes are no longer attributed to purposeful gods and spirits lurking behind or inside these ordinary physical phenomena. Science has succeeded in removing choice and free will from nearly all the machinery of the universe. In the future, cognitive psychology, allied with computer science, may complete this mission where anthropomorphizing first started, in our heads. When this happens, it will be unreasonable to anthropomorphize even human beings. In the meantime, science must proceed despite its uncertainty about how much of anthropomorphism will ultimately survive. This book rests on the assumption that, given the history of science, anthropomorphism is a bad bet and therefore it constitutes a mistake even in the study of politics, humanity's apparently most self-conscious form ofpurposeful and controlled social interaction . Human behavior is behavior. It is the immediate consequence of what goes on inside the brain, not an expression of a special will or free choice executed by a distinct entity, a human self, that happens to have a brain. Politics, similarly, is not the expression of a hidden spiritual world lodged somewhere within human actors. The agent traditionally conceived-the agent who appears in standard rational choice theory and, in particular, expected utility theory-not only believes and desires but as a decision maker is the source of his beliefs and desires. He is the source, a careful examination of the standard theory will show, because he is assumed to have the special capacity to cause himself to believe, prefer, deliberate, decide, and act. Hence when the individual's behavior is explained in terms of his beliefs and desires, this pronoun marks a distinct cause of what the organism embodying those psychological states does. The subject somehow acts on both its biological inheritance and its environment. Political science is better off without positing this free-willed subject. Here I propose to avoid it by populating political models with individuals who maximize their conditional expected utility. There are other, considerably more ambitious strategies-an approach rooted in biology comes to mind-but the rational choice alternative is much less speculative. Ultimately, of course, the 2 Choice-Free Rationality proof of the conditional expected utility hypothesis is in its applications, a number ofwhich will be presented in this book. Yet a full defense also requires us to recognize its more profound and controversial ramifications. For as I noted in the preface, critics of this approach charge that these applications suborn irrationality. In focusing on the outcomes their acts statistically portend and not what they cause, the criticism goes, maximizers of conditional expected utility are being fooled in various ways avoided by successful students of introductory statistics. This criticism is my main reason for briefly delving in chapter I into the philosophical ramifications of rationality. It is only at this level that one can see the deep and, I think, important motivation for models assuming conditional expected utility maximization. Although we will be exploring the definition of conditional expected utility maximization at length, in brief an agent who is rational according to this definition acts in such a way that her expected utility, conditional on her act, is as large as it can be. This means that the utility associated with each of the outcomes consistent with her acting a particular way is multiplied by its probability of occurring given the act. The conditional expected utility of an act, in turn, is the sum of each of these outcome-times-probability multiplications. The rational agent's act is the one associated with the highest such sum. Formally, given a set of actions {A I' A2 , ... ,Ak , ... ,All} the rational act Ak is the one for which Li U(O)Prob(Oi I Ak ) is maximized, where prob(Oi IAk ) is the probability of outcome j conditional on action Ak' Consider the important case of voting in mass elections. In deciding whether to vote, a conditional expected utility maximizer will compare the likelihood of a victory for her preferred candidate given her own participation with the likelihood of a victory given her abstention. When there is statistical dependence between the agent's decision and the decisions of others, we will see, this probability difference multiplied by the net utility of victory can be sufficiently large to motivate the agent to vote even though voting is costly. Thus suppose that Sally Smith's preferred candidate in a presidential election will lose unless he receives...