Comportamento de escolha em humanos: um estudo da diferença entre dizer e fazer.

Abstract

The aims of this study were to verify if response distributions with human subjects occurs according to the description of the Matching Law, if they would be capable to formulate rules describing their behavior, and the conditions controlling this behavior. To achieve these objectives three experiments were developed. In Experiment 1, 12 undergraduate students were set in four groups with different instructions. All were exposed to five different experimental conditions with independent concurrent variable interval-variable interval schedules. The results showed a high level of giving up. In relation to response distributions, exclusive preference for one of the two response keys was observed, and about the report, the criteria utilized did not assure exposition to all parts of the contingency, before the report requirement. These results led to modifications in the procedures that resulted in a new study. In the Experiment 2, four undergraduate students were exposed to other five experimental conditions. The sessions ended after 80 reinforcements, and the criteria to change each condition were the same used in the Experiment 1. The report was required in the session after every eight reinforcements. The results of this study showed exclusive preference for one of the response keys or continuous alternation between the keys, not allowing a stable baseline to analyze the correspondence between the studied repertories. In Experiment 3, four undergraduate students were exposed to five different conditions of non-independent concurrent variable interval-variable interval schedules, and in each condition, five sessions were made. A 10” delay was programmed: five seconds for time out, and 5” for COD. The report was required through the session after every five reinforced responses. The results analyzed by Baum’s equation (1974) revealed R² between 0,78 and 0,98, and all expoent values (a) < 0,90, demonstrating undermatching for all participants. About the rule-formulating behavior, despite the high correlation between response / time ratios and reinforcement ratio, none of the participants showed correspondence between choice behavior and description behavior. Thus, the data obtained in the present study indicated independence between both repertories, showing that response distributions, unlike demonstrated in the literature, are not a function of rule-governed behavior.