Probability and work as determiners of multichoice behavior.

Abstract

This study is concerned with effects which certain features of problem-solving tasks have upon behavior. The type of problem-solving situation studied is one in which an indication of malfunction has a probabilistic relationship to several possible defective components, each of which differs from the other in the probability with which it is defective and in the amount of work that is required to check it. The results indicate that (1) the probability variable weighed more heavily than the work variable in determining subjects'' (Ss'') choices, (2) the increase in the efficiency of low-work choices and the decrease in the efficiency of high-probability choices resulted in parallel changes in the frequency of occurrence of these choices, and (3) while the low-work choice became more popular when it was made the most efficient choice, 11 of 12 Ss did not adopt the simple, most efficient strategy of always selecting the low-work alternative. The Ss'' use of mixed strategies in the task where it was most efficient always to select the low-work alternative resulted in that group doing 8.5% excess work. Only one S in that group correctly identified the optimum checking sequence, while 7 of the 12 Ss actually named the least effective sequence as best. Had each S used only the checking sequence he reported to be best, the group would have done 12.5% excess work. Results suggest that in many tasks men do not discover efficient behavior sequences as a consequence of performing the task. A method was outlined for achieving more nearly optimum performance in such situations.