Individual choices often depend on the order in which the decisions are made. In this paper, we expose a general theory of measurable systems (an example of which is an individual's preferences) allowing for incompatible (non-commuting) measurements.The basic concepts are illustrated in an example of non-classical rational choice. We conclude with a discussion of some of the basic properties of non-classical systems in the context of social sciences. In particular, we argue that the distinctive feature of non-classical systems translates into a formulation of bounded rationality.
JEL: D80, C65, B41Keywords: non-classical system, incompatible measurements, orthospace, state, properties, bounded rationality
IntroductionIn economics, an agent is defined by her preferences and beliefs, in psychology by her values, attitudes and feelings. One also talks about "eliciting" or "revealing" preferences and attitudes. This tacitly presumes that those properties are sufficiently well-defined (determined) and stable. In particular, it is assumed that the mere fact of subjecting a person to an elicitation procedure, i.e., to "measure" her taste does not affect the taste. Yet, psychologists are well aware that simply answering a question about a feeling may modify a person's state * The first author gratefully acknowledge the support of the grant NSh-6417.2006.6 (1993) show in an experiment that simply asking people to state the subjective probability they assign to some event affects the way they make subsequent decisions. The so-called "disjunction effect" (Tversky and Shafir (1992)) may also be viewed in this perspective. In a well-known experiment, the authors find that significantly more students report they would buy a non-refundable Hawai vacation if they knew whether they passed the exam or failed compared to when they don't know the outcome of the examination. In the case they passed, some buy the vacation to reward themselves. In the case they failed, some purchase the vacation to console themselves. When they don't know, a seemingly inconsistent behavior is observed: fewer vacations are being purchased than in any one of the two possible events.In the examples above, the mere fact of subjecting an agent to a procedure that reveals her feeling, preferences or beliefs seems to affect her. In this paper, we propose to adopt a measurement theoretical approach to behavior: actual behavior reveals preferences (or beliefs) in the sense of being the outcome of a measurement of those preferences. Interestingly, Kahneman and A. Tversky explicitly discuss some behavioral anomalies in terms of measurement theory: "Analogously, -to classical physical measurement -the classical theory of preference assumes that each individual has a well-defined preference order and that different methods of elicitation produce the same ordering of options". But, "In these situations -of violation of procedural invariance -observed preferences are not simply read off from some master list; they are actually constructed in the elicitatio...