The Effects of Redundancy in Multiple-Cue Probability Learning

Abstract

The 80 subjects had a task of a two-cue probability learning at a +.8, +.4,.O,-.4, or-.8 level of stimulus redundancy. An analysis of 50 lcarning trials revealed a significant main cffect of redundancy. Performance was higher whcn redundancy was lower. Thcse findings are interpreted in terms of task characteristics that accompany redundancy, and the implications for the learning of complex inference tasks are discussed. In their study of the role of redundant information in inference tasks, Knowles, Hammond, Stewart, and Summers (1971) found the rate of learning to be influenced by both the sign (positive or negative) and the magnitude of the intercorrelation of two cues. Their finding that learning varies with cue intercorrelation, in the order +.8> +.4>-.4>-3, is not conclusive, however, because of the inevitable confounding of certain parameters of the multiple-cue training task with redundancy. Specifically, Knowles et al. used four training tasks with four different levels of cue intercorrelation (redundancy). They fixed the ratio of the regression deights (bet) in each task at 2: 1; the squared multiple correlation

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 11 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.