The Pigeon and the Spectrum and other Perplexities

Abstract

Recent developments in the study of stimulus generalization by means of operant conditioning methods require a re-examination of the problem of stimulus generalization and a review of the role of this concept in description and analysis of behavior. No final view of the outlines of learning theory can be achieved until more complete and fundamental information about generalization has been obtained. It has been established that in the pigeon orderly and reproducible bidirectional gradients can be obtained for individual Ss along the light-wavelength continuum. A dilemma is posed by the finding that these gradients are orderly on a physical scale; but not on an equal-discriminability scale. The form of the generalization function raises questions as the interpretation and meaning of the just-noticeable-difference concept. The problem of multidimensional stimulus generalization is discussed. The understanding of stimulus relationships in the generalization phenomenon is proposed as an issue of greater significance and difficulty than that of the control of generalization by such conventionally more prominent variables as drive and reinforcement.