《new challenges to the rationality assumption》由会员分享，可在线阅读，更多相关《new challenges to the rationality assumption（20页珍藏版）》请在金锄头文库上搜索。
1、New Challenges to the Rationality Assumption Author(s): Daniel Kahneman Reviewed work(s): Source: Journal of Institutional and Theoretical Economics (JITE) / Zeitschrift fr die gesamte Staatswissenschaft, Vol. 150, No. 1, Symposium on The New Institutional Economics Bounded Rationality and the Analy
2、sis of State and Society (March 1994), pp. 18-36 Published by: Mohr Siebeck GmbH the first footnote of his book was properly apologetic about the poor fit of the word “utility“ to pleasure and pain, but he found no better alternative. The decision utility of an outcome, as in modern usage, is the we
3、ight assigned to that outcome in a decision. The distinction between experienced utility and decision utility opens new avenues for the study of rationality. In addition to the syntactic criterion of consistency, we can now hope to develop a substantive/hedonic criterion for the rationality of a dec
4、ision: does it maximize the expectation of experi- enced utility? Of course, this criterion is not exhaustive, and its adoption implies no commitment to a hedonistic philosophy. As Sen has often pointed out (e.g., Sen 1987), the maximization of (experienced) utility is not always “what people are tr
5、ying to achieve.“ It is surely the case, however, that people sometimes do try to maximize pleasure and minimize pain, and it may be instructive to drop the assumption that they perform this optimization task flawlessly. Errors in the assignment of decision utility to anticipated outcomes can arise
6、from inaccurate forecasting of future hedonic experience. Correct pre- diction of future tastes is therefore one of the requirements of rational decision making (March 1978). Kahneman and Snell 1990 defined the predicted utility of an outcome as the individuals beliefs about its experienced utility
7、at some future time. Two sets of empirical questions arise: (i) How much do people know about their future tastes? Is it likely that an objective observer (or a government) could make more accurate pre- dictions than individuals would make on their own behalf? (ii) Do people adequately consider the
8、uncertainty of their future tastes in making decisions? Are decision utilities adequately informed by reasoned beliefs about experi- enced utility? Additional issues arise because of possible disparities between memory and actual hedonic experience. Outcomes are commonly extended over time, and glob
9、al evaluations of such outcomes are necessarily retrospective - and there- fore subject to errors. Examples of substantial discrepancies between retrospec- tive utility and real-time utility are discussed below. The restoration of Benthams notion of utility as an object of study evi- dently sets a l
10、arge agenda for theoretical and empirical investigation. The following sections summarize highlights of what has been learned in early explorations of this agenda. Decision utility, predicted utility, and the rela- tions between real-time and retrospective utility are discussed in turn. The final se
11、ction reviews possible implications of the findings for the rationality debate. This content downloaded by the authorized user from 192.168.52.68 on Wed, 21 Nov 2012 04:26:08 AM All use subject to JSTOR Terms and Conditions22 Daniel Kahneman JHTITE 3. Some Characteristics of Decision Utility Decisio
12、n utility has long been a topic of study, and much is known about it. The following discussion selectively addresses three research conclusions that are of particular relevance to the issue of rationality, as it is construed in this paper, (i) Carriers of utility. The main carriers of decision utili
13、ty are events, not states; in particular, utility is assigned to gains or losses relative to a reference point which is often the status quo (Kahneman and Tversky 1979). (ii) Loss aver- sion. Losses loom larger than corresponding gains (Kahneman and Tversky 1979; Tversky and Kahneman 1991). (iii) Fr
14、aming effects. The same objec- tive outcomes can be evaluated as gains or as losses, depending on the framing of the reference state (Tversky and Kahneman 1986). An early observation that illustrates points (i) and (iii) above was labeled the isolation effect (Tversky and Kahneman 1986). Problem . A
15、ssume yourself richer by $300 than you are today. You have to choose between a sure gain of $ 100 50% chance to gain $200 and 50% chance to gain nothing Problem 2. Assume yourself richer by $ 500 than you are today. You have to choose between a sure loss of $100 50% chance to lose nothing and 50% ch
16、ance to lose $200. It is easily seen that the two problems are extensionally equivalent in terms of wealth: both offer a choice between a state in which wealth is increased by $400 and a gamble with equal chances to increase current wealth by $300 or by $ 500. If people spontaneously evaluate options in these terms they will choose the same option in the two problems - but observed preferences favor the sure thing
