Multi-stage mental process for economic choice in capuchins

Abstract

We studied economic choice behavior in capuchin monkeys by offering them to choose between two different foods available in variable amounts. When monkeys selected between familiar foods, their choice patterns were well-described in terms of relative value of the two foods. A leading view in economics and biology is that such behavior results from stimulus-response associations acquired through experience. According to this view, values are not psychologically real; they can only be defined a posteriori. One prediction of this associative model is that animals faced for the first time with a new pair of foods learn to choose between them gradually. We tested this prediction. Surprisingly, we find that monkeys choose as effectively between new pairs of foods as they choose between familiar pairs of foods. We therefore, propose a cognitive model in which economic choice results from a two-stage mental process of value-assignment and decision-making. In a follow-up experiment, we find that the relative value assigned to three foods in sessions in which we tested them against each other combine according to transitivity.

Introduction

In recent years, an increasing scientific interest has developed apropos the psychology of economic choice. A large body of evidence shows how humans engaged in various economic behaviors are subject to a variety of ‘fallacies’. Well-documented phenomena include hyperbolic discounting, loss aversion, framing effects, the Allais paradox, and fairness constraints (Ainslie, 1992, Camerer, 2003, Camerer et al., 2003, Kagel and Roth, 1995, Kahneman and Tversky, 2000). The growing consensus is that a satisfactory account of economic behavior will entail a deeper understanding of the underlying mental processes. One area likely to contribute to this endeavor is the study of economic choices in other species, particularly in monkeys. Indeed, recent work has undertaken this approach (Brosnan and De Waal, 2003, de Waal and Davis, 2003, Hauser et al., 2003).

In general, behavioral economics has been most concerned with relatively complex choices. For example, a number of preference reversals and inconsistencies are observed in the presence of uncertainty, or in social contexts. But, arguably, a satisfactory psychological model is still lacking even for much simpler economic choices. For example, consider an animal choosing between two pieces of food, both equally and immediately available. Are its choices consistent across trials? And if so, what mental operations do they require?

Here, we present an experimental paradigm designed to help address these questions. In a first experiment, we offered monkeys two foods in variable amounts. We found that their choice patterns were well-described in terms of relative values of the two foods. In literature, animal food choices have been often described invoking theories of optimality, such as standard economics (Samuelson, 1947) or optimal foraging (Kamil, Krebs, & Pulliam, 1987). Our results were not inconsistent with these theories. However, these frameworks essentially imply an associative psychological model of choice (Skinner, 1953, Skinner, 1981) (see Section 2.3). In a second experiment, we tested the major prediction of this model, namely that offer-choice associations develop through trial and error. To the contrary, we found that monkeys choose effectively even between foods that they have never encountered together before. As an alternative to the associative model, we propose a cognitive model according to which economic choice results from a two-stage process of value-assignment and decision-making.

Parts of these results have been previously presented in abstract form (Padoa-Schioppa, Jandolo, & Visalberghi, 2003)