Tilburg Institute for Behavioral Economics Research

"Can There Ever Be Too Many Flavors of Häagen-Dazs? Anticipatory Beliefs and Choice Overload Behavior"
A growing body of research in psychology and economics has attempted to demonstrate that people can suffer from “choice overload” from too many choices. This large and growing literature was initiated by Iyengar & Lepper’s (2000) (IL’s 2000) field experiment which showed that people were less likely to purchase when faced with more variety. Though intuitively appealing, attempts at replication have yielded mixed results. A 2010 meta-analysis concluded that there was as yet no sufficient. We hypothesized that choice overload behavior was driven by uninformed consumers’ anticipatory beliefs about surplus from sampling. We first surveyed subjects for possible “disgust” in 6 product categories. We then randomly chose 4 among these and secretly observed consumers after we switched between high and low varieties. As predicted, we found that choice overload behavior was an increasing function of surveyed disgust. Hence, surveyed disgust could be a sufficient condition. To our knowledge, this is the first data to separate psychological theories of choice overload behavior (starting with IL 2000), which predicts choice overload behavior to be increasing on the number of options, and contextual inference theory of Kamenica (2009), in which consumers should infer decreasing average surplus for increasing variety. We extend this theory by allowing consumers who like a product more to be more to be tolerant of disappointment in sampling. Now, consistent with the meta-analysis and our data, choice overload or choice loving behavior would be predicted depending upon anticipated surplus.

Ralph Bayer

"Facing man or machine -- when do humans reason better?"
In this paper we design a logical reasoning task that can be represented as a game with other players or as a decision problem, where automata take the role of the other players. The task is designed such that strategic uncertainty should not play a role, and hence the quality of reasoning should be independent of framing. Varying the difficulty of the task (i.e. iteration steps required), we find that for difficult tasks subjects perform much better if they play with computers, while the opposite is true for simple problems. These surprising results have important implications for the expected quality of human reasoning in games.

"Building Synthetic Economies"

Robert Bloomfield will describe the goals and challenges of his Synthetic Economy Research Engine (SERE), a software platform being developed to allow researchers to construct experimental settings in which participants manage virtual firms with realistic production processes, endogenous competitive pressures, flexible contractual arrangements and varied financial reports.