Behavioral economics and drug choice: effects of unit price on cocaine self-administration by monkeys

Abstract

The application of microeconomic theory to the experimental analysis of behavior has been termed behavioral economics. There has been an increasing interest in applying the concepts of behavioral economics to the study of drug self-administration. In a previously published experiment (Nader and Woolverton, 1992), rhesus monkeys ( N = 3) were trained in a discrete-trials choice procedure and allowed to choose between intravenous injections of cocaine (0.03–1.0 mg/kg/injection) and food presentation (1 or 4 pellets; 1 g/pellet) during daily 7-h experimental sessions. When cocaine or food was available under a fixed-ratio (FR) 30 schedule, cocaine intake increased in a dose-related manner for all monkeys. When the response requirement (FR) for cocaine was differentially increased by doubling or quadrupling, the frequency of cocaine choice decreased, shifting the cocaine dose-response function to the right. The present paper is a reanalysis of data from that experiment. Several mathematical models, differentially incorporating the effects of FR, dose and number of food pellets, were compared. When cocaine consumption was analyzed using a multiple linear regression analysis with FR, dose and number of pellets as separate main effects (model I), the R 2 was 0.82. When FR and dose were combined into one factor, unit price (UP, responses/mg/kg), and cocaine consumption was analyzed as a linear function of UP (model IIA), the R 2 was 0.54. When cocaine consumption was analyzed as a curvilinear, negatively accelerated function of UP (model IIB), the R 2 was 0.53. The difference between models I and IIA was statistically significant while models IIA and IIB were not different. Since the choice was mutually exclusive, cocaine consumption was analyzed as a function of the ratio of the UP of cocaine and the UP of food, the reinforcer that was lost when drug was chosen. The R 2 for this analysis was 0.58. Multiplying that ratio by the sum of the UPs for cocaine and food (Bickel, personal communication) resulted in an increased R 2 of 0.64. Eliminating data from low dose conditions (Bickel, personal communication) improved the fit of model IIB but not of I or IIA. Thus, a UP analysis of cocaine consumption under the present conditions significantly decreased the proportion of the variance that was accounted for by the multiple linear regression model. These results suggest that changes in FR and dose do not necessarily have the same functional effect on drug consumption and that the applicability of the UP model to drug self-administration may be limited.