Effect of meaningfulness on judgments of computability

The University of Michigan, Ann Arbor, Michigan 48104 After memorizing a set. of four equations, subjects were given problem frames specifying an unknown variable and two, three, or four given variables. The task on each item was to judge whether the presented variables corresponded to a problem that could be computed. (No values were given and no computation was carried out.) The main finding was a strong interaction: With formulas consisting of uninterpreted letters, judgments of uncomputable items were much slower than judgments of computable items; however, the difference between positive and negative items was much smaller when formulas consisted of meaningful concepts. Attempts to remove the interaction by presenting the nonsense formulas as an integrated network or by using a format allowing easy chunking of the variables in each formula did not succeed. At this point it appears most promising to investigate effects of meaningfulness on basic component processes, such as decisions about whether pairs of items come from the same formula.