-
3. A General Model of Preference and Belief Formation
- University of Michigan Press
- Chapter
- Additional Information
Chapter 3 A General Model of Preference and BeliefFormation In this chapter, I present a general model of preference and belief formation. It is complementary to the rational optimization model of decision making, and is integrated with it to form the basis for a revised general model of action. In the model, choice of action isseen as the result of a decision-making process aimed at maximizing preferences given beliefs, but also triggers changes in these preferences and beliefs. The next section of the chapter defines the concepts of regret and coherence that are central to the model, then discuss the model's basic assumptions. The remaining sections analyze preference and belief formation, examining static, dynamic, and collective choice contexts. These sections include implications of the model for particular empirical phenomena and theories about them. The basic ideas contained in the model are hardly new-they are instead an integration of existing ideas that have been drawn from a variety of academic fields. The ideas are associated with work on individual attitudes, identity, and agency but tie them to a work on collective ideology, culture, and norms. The theoretical literatures involved span all the major social science disciplines and philosophy, and the discussion often juxtaposes the work of academic communities that rarely engage in dialogue. I hope to show that the model does more than stack an eclectic bunch of concepts and theories on top of one another, that it also provides a way to tie their common ideas together, to isolate and address their areas of disagreement, and to extend their implications. I do not claim that this model reflects anything close to the entire reality of human motivation and cognition. Nonetheless, I try to show that it can provide a general and parsimonious way to account for preferences and beliefs and that its predictions are usefully determinate and conform with broad patterns of empirical evidence. I also try to show that it ties empirical patterns to deeper underlying processes and generates novel hypotheses for future investigation. In chapters 4 through 6, I try to show that the model of action that results from integrating the model with a rational optimization model of decision making provides a useful tool for analysis and theory-building on substantive issues that existing rational choice and nonrational choice theories have failed to address adequately. I should note that the model does not specify the bounds that limited or biased information processing places on an individual's ability to optimize. This does not reflect any opposition on my part to bounded rationality models. Instead, I argue that boundedness is consistent with and in some ways even necessary for optimization. In the absence of knowledge that is both infallible and complete, optima can be defined only with respect to constraints that are subjectively constructed around past experiences. Not only will consideration of alternatives be inherently limited, an expansion 81 82 Choosing an Identity in one's choice framework that is unnecessary to accomodate new experiences may hinder rather than facilitate optimization. Even so, I do not pretend to provide a fully adequate model of boundedness, since the main emphasis in this chapter is presenting new assumptions about preference and belief formation, and specifying the location of other important decision-making bounds on top of this would be an unrealistic task. In chapter 7, there is a brief discussion of how the preference and belief model presented here can be integrated with satisficing models and models of nonexpected utility maximization. A Coherence Model of Preference and Belief Formation Formal specification of the model is based on a sUbjective probability framework, though one quite modified from that originated by Savage and widely developed elsewhere. l The notation is hybrid, imparting a flavor of formal logic to traditional choice-theoretic symbols. Some reasons for selecting the notation will be provided as the model is presented, and clarifications are provided when interpretation might be ambiguous. An individual's preferences and beliefs are represented by a rule base P consisting of a set of rules p' = (s· b·) where s' is a , '1, '1" '1, , 1, deterministic statement regarding the state of reality and bi is a probability weight in [0, 1J. A constant within a rule base is the name of a discrete object, an ordered or unordered class of objects, a cartesian product of classes, or a mapping between classes. A variable is an description over a class, i.e., an expression whose denotation is within the class...