The present study investigated whether presenting of a record of participant responses and corresponding outcomes within active contingency tasks influenced the probability-of-responding [P(R)] effect. In two experiments, participants (
n=132) decided whether or not they would administer an experimental drug to 50 different patients, presented on a computer screen one-by-one, and, at the end of the session, they estimated drug efficacy. We manipulated the presence or absence of a record of prior responses-outcomes in both experiments and manipulated experienced outcome density between the experiments, contrasting high (Experiment 1) and low (Experiment 2) densities. Regression and partial correlation analyses were conducted on the data. Within the high outcome-density conditions, presenting a record produced a higher P(R) effect than in the no record condition. In the low outcome-density condition, presenting a record produced veridical contingency judgments based on the actual contingency, whereas the no record condition led to non-systematic results. Partial correlation analyses revealed that the simple regression line relating estimated drug efficacy to frequency of administrations may reflect not only the P(R) but also actual contingencies.
View full abstract