Constructivist coding: Learning from selective feedback

ABSTRACT—Although much learning in real-life environ-ments relies on highly selective feedback about outcomes, virtually all cognitive models of learning, judgment, and categorization assume complete and representative feed-back. We investigated empirically the effect of selective feedback on decision making and how people code ex-perience with selective feedback. The results showed that, in contrast to a commonly raised concern, performance was not impaired following learning with selective and biased feedback. Furthermore, even in a simple decision task, the experience that people acquired was not a mere recording of the observed outcomes, but rather a recon-struction from general task knowledge. A personnel manager will know the job performance only of applicants who are employed. Likewise, in many real-life en-vironments, people receive selective feedback that is not only absent on certain trials, but contingent on prior decisions, pro-viding a biased sample of all outcomes (Hogarth, 2005). Still, a vast majority of laboratory studies on learning include complete, unambiguous feedback on every trial, and few studies have in-vestigated empirically how selective feedback about outcomes affects learning. Although occasionally selective feedback has demonstrated beneficial effects on decision making (Grosskopf