Are Monkeys Sensitive to the Regularity of Pay-off?

Abstract

Animals commonly face fluctuations in their environment and resources. To maximize their benefits, they need to integrate the risks attached to potential pay-offs. We do not know, however, to what extent individuals account for irregularity in the latter. We tested the sensitivity of monkeys (Cebus apella, Macaca tonkeana, M. fascicularis) to the irregularity of pay-offs in two different tasks. In a first experiment, the subjects were given an exchange task where the reward probability varied between different conditions, but yielded the same average pay-off. There was no evidence of subjects favoring either condition, meaning that they behaved in accordance with the predictions of the classical decision theory (Expected Utility Theory). In a second experiment, we offered to subjects a choice between two options involving different pay-off regularity. In this case, a wide range of inter-individual variation was found in the choices of individuals. Whereas monkeys accepted irregular pay-off in a rational way, there were individual biases in their preferences. These results indicate that the preferences of animals in a risky situation were not unequivocally shaped by the environment in which species have evolved. Individuals commonly face fluctuations in their environment and resources. As assumed by behavioral ecology theory, natural selection should favor decision mechanisms producing optimal strategies, that is, maximizing the net rate of energy intake while minimizing time investment (Schoener, 1971). In the foraging context, two main variables influence decision-making, namely the expected amount of food and the time needed to obtain it. Regarding delay, animals prefer the variable delay to the fixed delay when choosing between two options; however, they prefer a fixed amount of food to a variable amount (Kacelnik & Bateson, 1996). In the context of the Expected Utility Theory (EUT), Von Neumann and Morgenstern (1944) argue that risky choices can be specified both in terms of outcomes and probabilities of these outcomes, and that how individuals pick one option rather than another is based on maximizing expected utility. This behavior is referred to rationality in classical economic decision-making. Thereafter, any rational, risk-neutral individual should be indifferent when faced with two options having identical expected outcomes. For example, in animals facing variable and fixed delays with identical expected delay for the two options, risk neutrality would prevail if animals do not exhibit preference for either option. In other words, as subjects prefer the variable option to the fixed one, they are risk-prone regarding delays and risk-averse regarding food amounts. These results are very interesting according to the EUT, because it is precisely because animals are risk-averse for food amounts that they care about risk when seeking food resources.