An Integrated Learning Theory of Descriptive and Experience-Based Decisions

Carnegie Mellon University, Pittsburgh PA

Investigators

Abstract

This research program will unify currently fragmented decision theories leading to a greater understanding of general decision making process beyond the traditional distinction of judgment and choice. Additionally, this research will provide insights into how decision options are evaluated over different attributes, and how they are ultimately selected under different amounts and types of information. Given that our decisions are often informed by some combination of explicitly described information and previous experience, this research is expected to have large impact in the multiple contexts in which we make decision every day, including financial, health, energy, and other contexts. The research may impact communication policies, the design of information and decision support systems, marketing strategies, and many other areas by clarifying what impact different amount, type, and format of information has on risk taking and decision making. This project will go beyond descriptions of phenomena by constructing behavior prediction algorithms. Because we will deliver computational implementations of the mental processes involved in making these decisions, this research has the potential to enhance and improve traditional technology of decision support and big data approaches used to predict decisions. People make decisions by using information that varies in format, completeness, accuracy, certainty, and many other factors. Often, we rely on descriptive information (explicit definition of risks, outcomes, probabilities) and on experiential information (implicit collection of past outcomes and probabilities through feedback). Decisions that use both types of information are ubiquitous in both trivial, low-risk situations and highly consequential and risky situations, but most research deals with decisions made from descriptions or from experience exclusively; not a combination of both. This research undertakes a comprehensive integration of decision making theories and models, and proposes to generate a novel theory of descriptive and experience-based decisions. The research begins with performing a conceptual integration of the key psychological processes involved in judgment and choice in descriptive and experience-based decisions; second, establishes behavioral phenomena of the integration of these two types of information and their effects on choice; and third, proposes new computational formalizations to extend current models of learning to include the processes discovered through experimentation.