1648 



HWDHOllK 111 I'llSSIl II < u.Y 



Ml ROPHYSIOLOGY III 



Efferent Potlern 

 {oculomotor impulse) 



Perception 



Pattern 



hi; 31 Diagram to indicate postulated interaction between 

 afferent, efferent and re-afferent processes in perception. 

 Self-induced motion of the eye (efferent pattern) causes 

 peripheral motor effects and, concurrently, a central discharge 

 back into the appropriate sensory system (corollary dischargi 

 which normally matches (i.e. cancels) the sensations produced 

 by the active movement (re-afferent pattern). Under normal 

 conditions, corollary discharges and re-afferent patterns 

 balance, so that signals from the environment are perceived 

 (rq motion of external objects 1 and distinguished from 

 relative motions due to the perceiver's own movements. When 

 the visual field is inverted, the same feedback becomes positive 

 and maladaptive, causing subjective spinning of the visual 

 field and (in lower species) forced movements of the body. 

 See Sperry (438) and MacKay (335). [Adapted from von 

 Hoist (504) and von Hoist & Mittelstaedt (505).] 



back continues uncorrected in lower forms hut can be 

 readjusted apparently in man. 



The hypothesis jusi developed is in essence a re- 

 statement of older views regarding a systematic 

 (compensator) I shifting oi retinal space values during 

 eve movement (201, j 1 4, 215). This hypothesis has 

 been recast in feed-hack terms, independently and 

 simultaneously, b> von Holsl & Mittelstaedt (505) 

 and hv Sperrv (458). Although entirely conjectural 

 at this point, the hypothesis is attractive, since it 

 can subsume normal and abnormal phenomena, and 

 can perhaps he elaborated into a more general 

 theorv oi constancies and illusions, and ol perceptual 

 identification. 



CONS1 UN lis, II 1 1 SIONS AND FIOURAL AFTEREFFECTS 



' onstancies: Examples and Measurement 



Within limits, sizes and shapes ol obje< is maintain 

 approximate constancy as their distance and orienta- 

 tion changes. A man moving awa) from an observer 



is seen to recede but does mil appear lo shrink until a 



certain distance is exceeded; this is constancy of size. 

 Neither does the apparent speed of his movements 

 change as they take him further away; this is con- 

 stancy of velocity (see above, p. 1644). A circle rotated 

 out of the frontal plane is usually seen as a circle-at- 

 a-slant and remains distinguishable from an ellipse 

 produced by the corresponding geometric projection; 

 this is constancy ol shape. 



These "constancies' can be quantified by assuming 

 that the actual perception departs more or less from 

 the law of geometrical optics which predicts re- 

 sponses in accord with computed variations in the 

 physical properties of the retinal image — hence 'law 

 of the retinal image.' For the case of the circle-at-a- 

 slant, the law of the retinal image predicts that the 

 perceiver would match it with an ellipse with the 

 same ratio of minor to major axis as that found in the 

 geometrical projection of the slanted circle onto the 

 observer's frontal plane. If the perceiver matches the 

 slanted circle instead with an identical circle appear- 

 ing in his frontal plane, we say that he exhibits com- 

 plete 'constancy. 1 In actual experiments the choices 

 usually fall between the prediction made by the 'law 

 of constancy' and that made by the "law of the retinal 

 image." (Very occasionally, there are instances of 

 'overconstancy,' however.) The incomplete constancy 

 under these conditions has been variously called 

 perceptual "compromise' by Brunswik (73, 76), or 

 'partial regression to the real object' by Thouless 

 (484, 485). Both investigators introduced simple ex- 

 pressions to describe such partial constancies in 

 terms of ratios. The Brunswik ratio (75), BR, is the 

 one used most frequently. 



BR = 



R - S 

 C - S 



where A' = a measure of the response (e.g. subject's 

 choice of a matching shape or size, etc.); 5" = pre- 

 dicted response according to the 'law of the retinal 

 image', (.' = predicted response according to the 

 'law of constancy.' 



This ratio (X too) gives a per eenl value of the 

 extent to which the experimental results agree with 

 the law of constancy. Thus, a Brunswik ratio of 1.00 

 indie. ties complete agreement the observer matches 

 circle with circle A ratio of indicates thai there is 

 no constancy, and that the circle-at-a-slant is matched 

 with the particular ellipse that corresponds to its 

 frontal projection. 



The Thouless ratio I I Ri is the same as (he Bruns- 



