WHEN WILL THE MILK SPOIL - EVERYDAY INDUCTION IN HUMAN INTELLIGENCE

We sought to understand (a) the mental processes underlying everyday p redictive and postdictive judgments involving objects undergoing eithe r change or no change of state over time and (b) the relation of these processes to intelligence. Forty adult (nonstudent) participants were asked to solve 40 induction problems, each presented in four forms, f or a total of 160 test items. Half of the items involved predictions a nd half involved postdictions; within each of these two categories, ha lf of the items involved a state change from present to future (predic tion) or past (postdiction), and half did not. In addition, each parti cipant completed convergent-discriminant psychometric ability tests me asuring inductive reasoning, deductive reasoning, and vocabulary. Pred iction was performed more rapidly than postdiction, but was also more susceptible to errors of judgment. Judgments involving change of stale were more rapid and less error-prone than were judgments involving no change. A single information-processing model was useful for describi ng performance on both prediction and postdiction problems. A quantifi cation of stimulus variables affecting performance via this model (wit h six predictor variables) provided a good account of participants' re sponse latencies. Response latencies showed convergent and discriminan t validity, exhibiting rather high correlations with inductive reasoni ng, but not with deductive reasoning. The paradigm we used thus seems to provide one useful approach to understanding the relationship of ev eryday induction to human intelligence, and to provide a complement to the more abstract kinds of problems-such as analogies, classification s, series completion, and matrix problems-typically seed to test induc tive-reasoning abilities, as an aspect of intelligence.