PERCEPTION 



l60I 



all differentiations of which a receptor structure is 

 capable ought to be preserved in later stages within 

 the nervous system; differentiation made possible at 

 the periphery may be lost at some subsequent stage. 

 The absence of color discrimination in the cat may- 

 be a case in point, since the cat retina appears to 

 make color discriminations even though behavioral 

 evidence for such discriminations on the part of the 

 'whole cat' is lacking (see Chapter LX by Neff in 

 this volume). 



If sensory discriminations, according to the views 

 developed here, no longer bear a simple one-to-one 

 relation to the stimulus dimensions, then one of the 

 stronger nineteenth-century arguments for separating 

 sensation and perception is lost. In the new 'dimen- 

 sional 9 psychophysics, sensations have a complexity 

 not unlike that of perception, even though the earlier 

 attempts at separation have not been entirely aban- 

 doned. Consider the classical psychophysical pro- 

 cedures: the matching of a monochromatic color with 

 a dichromatic mixture, the determination of just- 

 discriminable differences along a particular stimulus 

 dimension, the discovery of absolute thresholds. In 

 all these procedures, the hoped-for simplicity ol re- 

 sults was marred by 'errors'; matches between stimuli 

 differed, depending on whether the standard stimulus 

 was placed to the left or right of the variable ("space 

 errors'); stimuli were over- or underestimated, de- 

 pending on whether they were given first or second 

 in a successive comparison ('time errors'); judgments 

 of weights differed markedly, depending on the 

 presence or absence of much lighter or much heavier 

 weights in the series submitted for comparison Cintra- 

 serial effects'). All these phenomena, though interest- 

 ing in their own right, tended to be treated as sources 

 of 'error' in strict psychophysical experiments; the) 

 were to be canceled by balancing the stimulus presen- 

 tation so that effects of spatial position, or temporal 

 order, or of constitution of the series could be elimi- 

 nated. There is, of course, nothing illegitimate about 

 such attempts at controlled experimentation; but the 

 distinction between the intended results as 'primary,' 

 or 'sensory,' and the space or time effects as 'errors,' 

 produces rather than proves the elementary nature 

 of sensation. At the same time the study of percep- 

 tion becomes an exclusive concern with these 'errors' 

 which are left over, as it were, as mere residuals after 

 the more serious business of sensorv psvehophysics 

 has been accomplished. 



Actually, intraserial effects and related phenomena 

 can be dealt with just as rigorously as the supposedly 

 more elementary threshold phenomena. Helson's 



work on 'adaptation levels' (197) shows that the 

 subjective magnitude of a singly presented stimulus 

 depends upon the weighted geometric mean of the 

 series of stimuli that the subject has worked with; 

 but, as in all psychophysical procedures, explicit or 

 implicit verbal instructions are crucial since they 

 determine which stimuli are perceived by the sub- 

 ject as forming part of the series. [However, Stevens' 

 protest (452) should be noted.] That this is so, can 

 be demonstrated very simply. Ask a subject to move 

 a particular weight out of the way (with the excuse 

 that it is cluttering up the table), and this particular 

 weight will not enter into the formation of the series; 

 it will have no effect on the subjective magnitude of 

 the other weights within the test [see Brown (65); 

 also Bruner (71)]. The purity of the psychophysical 

 experiment thus depends largely on the perceptual 

 set (71) adopted by the subject, and the distinction 

 between sensory and perceptual aspects of the task 

 is spurious. 



Beyond the consideration of space and time errors, 

 and of intraserial effects, perception studies in the 

 later nineteenth century were primarily concerned 

 with two sets of phenomena, the so-called "illusions' 

 and 'constancies.' To this day, these have remained 

 central problems in perception. Every textbook shows 

 samples oi some geometrical optical illusions — the 

 famous Muller-Lyer pattern, for instance (see fig. 

 I 1 I .uul most texts list the speri.il theories developed 

 to account for these effects [cf. Boring (53, pp. 238- 

 2 1 y, Woodworth & Schlosberg ( J40J, pp. 417-423)]. 

 Yet to speak of illusions .is speei.il cases — curiosa of 

 perception, as it were is tendentious. As soon as one 

 admits that perception lacks simple one-to-one corre- 

 spondence to physical stimuli, the explanation for 

 perceptual illusion will be sought among the general 

 laws of perception. Once these laws are known, the 

 illusions themselves will be understood (53). 



The same may be said lor the so-called constancies, 

 e.g. the relative constancy of perceived brightness, 

 size or shape, with varying illumination, tilt or dis- 

 tance. The classical treatment of these effects as re- 

 sults of "unconscious inference' by von Helmholtz 

 (501) implies that we ought to be able to perceive 

 the "image' on our retina, and then elaborate our 

 perception by recourse to some judgmental factors 

 into a corrected or interpreted view of the distal ob- 

 ject to which the image refers. But to assume that we 

 ought to 'see' the retinal image may be just as naive 

 an assumption as the belief that in hearing a given 

 pitch, we ought to perceive a frequency. The very 

 notion of a retinal 'image' is a curiously anthropo- 



