Cue selection in hypothesis generation: Reading habits, consistency checking, and diagnostic scanning

Abstract

Two experiments investigated retrieval cue selection in hypothesis generation. In both experiments, subjects completed six hypothesis generation problems. Then a surprise recall test was given to measure which data a subject attended while generating a hypothesis. Subject's patterns of data recall were categorized as to whether (a) reading habits, (b) reading habits and consistency checking, or (c) diagnostic scanning was used to select cues during hypothesis generation. In Experiment 1, subjects were instructed to generate hypotheses that were consistent with all problem data. Experiment 2 required subjects to report the first hypothesis suggested by problem data. Subjects in both experiments generally used reading habits and consistency checking to select cues to retrieve hypotheses from memory. These results provide converging evidence concerning the existence of consistency checking, a predecisional process first described by S. D. Fisher, C. F. Gettys, C. Manning, T. Mehle, and S. Baca(1983, Organizational Behavior and Human Performance, 31, 233–254). Of the three cue selection strategies, diagnostic scanning, although seldom used, resulted in the highest probability of generating correct hypotheses. Although this finding is impressive, little evidence was found concerning the general use of diagnostic scanning in hypothesis generation.