Grantee: Pomona College, Claremont, CA, USA
Researcher: Shlomi Sher, Ph.D.
Grant Title: The rational construction of preference
https://doi.org/10.37717/220020427
Program Area: Understanding Human Cognition
Grant Type: Scholar Award
Amount: $600,000
Year Awarded: 2015
Duration: 8 years
A major focus of my research is in the psychology of judgment and decision making. This work falls within a research tradition in which normative models (which specify how ideal actors should behave) are compared with descriptive theories of behavior (which specify how real actors do behave). Over the last forty years, psychologists have compiled a lengthy catalogue of descriptive violations of the traditional theory of rational choice.
The human decision maker (DM) one meets in the psychological literature is capricious and malleable. Subtle changes in framing (how options are described), procedure (how evaluations are elicited), and context (which other options are considered) are shown to trigger dramatic swings in expressed preference. Nonetheless, while preferences are often inconsistent across choice episodes, people strive to maintain consistency within episodes – a phenomenon known as coherent arbitrariness. These findings suggest that preferences are “constructed,” rather than revealed, in the act of choice, and are often seen as decisive demonstrations of irrationality.
My research explores the construction of preference from a different perspective. I argue that an adequate normative analysis depends on subtle descriptive assumptions that are often overlooked. To formulate a normative theory, we need to understand the structure of the DM’s values. To test the theory, we need to understand the DM’s dynamically evolving beliefs about the choice situation. What is rational depends on what is sought and what is known. My goal is to incorporate a richer and more realistic descriptive assessment of values and knowledge into the psychological study of human rationality. I focus on the ramifications of imperfect knowledge and incompletely specified values.
When knowledge is limited, learning and decision making are intimately intertwined. The menu of options serves as a sample from an unfamiliar world, and the rational DM may update her beliefs, and hence her preferences, as new options are sampled. The normative analysis of choice then requires an empirical analysis of the information structure of the environment, the manner in which beliefs are updated as it is explored, and the effects of these inferences on preferences. My research aims to map out this information structure, and to clarify its implications for effects of framing, procedure, and context.
When underlying values are not defined to infinite precision, choice – rational or otherwise – is an inherently constructive activity. Importantly, it can be shown that local consistency is valuable even when individual decisions are globally arbitrary. Furthermore, when memory limits constrain the degree of consistency that can be maintained, a bias to favor the status quo (which is ubiquitous in human decision making and usually regarded as counter-normative) may be advantageous. This analysis suggests a new perspective on the psychology and rationality of coherent arbitrariness.
This program of research explores the unexpectedly complex interplay between normative and descriptive levels of analysis. A rational analysis depends on subtle empirical assumptions concerning the structure of human beliefs and values. In turn, it suggests fruitful new questions and hypotheses for the psychological study of preference construction.