Effects of test cost and cognitive demand on hypothesis-testing behaviors: Three experiments

Abstract

Both positive and diagnostic tests of hypotheses have apparent survival value when viewed within the context of the everyday social environment. It was predicted, however, that commonly observed preferences for both types of tests would be sensitive to a number of identifiable situational factors. Three experiments were conducted in which changes in hypothesis-testing behaviors were measured as a function of the cost of conducting tests and the cognitive demands of tests with regard to processing negatives. All experiments employed a fully computerized procedure in which subjects explored ancient temples with the goal of discovering who built them. Test-cost was manipulated by making some tests time-consuming and tedious to conduct, and cognitive demand was manipulated by controlling the negative or positive nature of the initial given evidence. Subjects' preferences for positive and diagnostic tests were shown to change predictably whenever these factors conflicted with each other or with test-cost considerations. The ability to test diagnostically was also shown to be powerfully affected by the nature of the presented evidence. In one experiment, subjects were nearly unanimous in choosing diagnostic tests whenever the presented evidence was positive, whereas these same subjects did not even exhibit a preference for diagnostic tests when given equally informative negative evidence. Overall, the results argue strongly for the need to qualify descriptions of human hypothesis-testing behaviors by referring to the relevant task parameters under which the behaviors in question are exhibited.