Numerically Varied S-R Mapping Disorder

Abstract

Qualitative support for the effects on learning of various degrees of S-R mapping disorder was cited to show that the way stimulus and response classes are connected or mapped together may have a strong effect on learning rate. Then to study quantitatively the S-R mapping problem a paradigm based on discrimination learning procedures was constructed. It made possible the numerical variation of S-R mapping on an order-disorder basis. The order-disorder dimension was related a priori to a linear variable for number of different correct choices and to a quadratic variable for conditional mapping uncertainty, U.(R). Mean errors were significantly related to only the quadratic component. Other results include: a closer relation between U.(R) and SDs than between U.(R) and means; unique patterns of errors within groups related to mapping structure; and a correspondence between post-experimental subjective awareness and both task structure and difficulty. Finally, among post hoc explanations of mapping effects one phrased in terms of mapping uncertainty and hypothesis storage, sampling, and generation gave the best account of obtained results.