Effects of task predictability and cue validity on interpersonal learning of inference tasks involving both linear and nonlinear relations

Abstract

Pairs of subjects learned two-cue inference tasks with one linear and one nonlinear cue. Task predictability and the distribution of the validities of the cues were varied in a factorial design. The subjects adjusted their utilization of the cues to the magnitudes of the validities of the cues. As a consequence, relative achievement was lower when task predictability was law. Nonlinear cues received lower weights than linear cues, and suffered a further reduction when they were used in conjunction with linear cues. Therefore, achievement was lower when only the nonlinear cue was relevant than when only the linear cue was relevant, and lower when both cues were relevant than when only one cue was relevant. When subjects who started out using only the valid cue were paired with subjects who used a nonvalid cue, they started to use the nonvalid cue. This suggests that interpersonal learning may introduce bias in the utilization of the cues.