Behavioral Operations Research: Bridging Decision Theory and Human Psychology

Today I attended Professor Xie's lecture titled “Decision Theory and Methods Based on Behavioral Operations Research,” which was highly inspiring.

Behavioral Operations Research (BOR) extends traditional operations research models by incorporating irrational human factors and psychological preferences, making analyses of complex systems more realistic.

Traditional OR assumes fully rational decision makers who base choices on deterministic or probabilistic information to select the optimal solution. For example, in a simple linear programming problem, decision variables, profit coefficients, constraint matrices, and resource limits are defined, assuming known objective functions, linear preferences, and predictable futures.

However, real human behavior often deviates from these rational assumptions, making predictions inaccurate if psychological factors are ignored.

Behavioral Operations Research integrates several key elements:

Bounded Rationality

Cognitive Biases

Affective Influences

Social Preferences

For instance, in inventory control modeling, the classic (s, S) policy may lead to overstocking due to “order inflation” behavior. A behavioral function can adjust the model to account for this tendency.

Professor Xie also discussed the classic newsvendor problem. The optimal order quantity in the traditional model is derived assuming known demand distribution, ordering cost, selling price, and salvage value.

Experimental work by M. E. Schweitzer and G. P. Cachon (Management Science, 2000) involved 34 Duke University students making ordering decisions under varying cost and price settings. The findings showed a “pull‑to‑center” effect: under high profit margins, participants ordered less than optimal; under low profit margins, they ordered more than optimal.

In high‑profit scenarios, average choices were below the optimal order quantity; in low‑profit scenarios, average choices exceeded the optimal order quantity.

This phenomenon is explained by a regret‑based model, while risk attitude alone cannot account for it.

Beyond the newsvendor problem, many real‑world and industrial phenomena—such as queuing behavior and mental accounting—require a behavioral OR approach that combines operations research with psychology.

Professor Xie concluded by suggesting future research directions: finding interesting examples, mechanism design, and integrating machine learning with behavioral OR.

“Behavioral Operations Research” does not reject traditional models; it enriches them by considering human factors, leading to predictions that better reflect reality.

Images from the lecture illustrate these concepts.