PERCEPTION 



1603 



lish whether the defense was a mere withholding of 

 unpopular responses or a truly perceptual effect. 



The need for converging operations is particularly 

 pressing in work on perception in young children or 

 in animals below man. In the absence of verbal re- 

 port we have to depend entirely on differential motor 

 reactions on the subject's part, but this does not mean 

 that any given set of these reactions is sufficient to 

 reach conclusions regarding the subject's perceptual 

 repertoire. Earthworms will pull pine needles into 

 their burrows by grasping them at the base (where 

 the pair of needles hang together). Conversely, they 

 pull leaves into their burrows by grasping the tip, 

 never the stem, so that the leaf rolls itself into a tube 

 as it disappears into the burrow. This highly adaptive 

 behavior looks like form perception, but the basis of 

 discrimination is gustation [cf. Mangold (338)]. The 

 worms avoid powdered extracts of the base of leaves 

 and ingest powdered extracts of the tips, and con- 

 versely for pine needles. 



Thus, converging operations are needed to define 

 the sensory basis of discrimination; moreover, any 

 tendency to reduce all perception to discriminatory 

 responses would reintroduce the classical sensory bias 

 and force us to omit all those problems from con- 

 sideration which are crucial in developing a physio- 

 logic theory of perception. 



SOME CENTRAL PROBLEMS FOR A 

 THEORY OF PERCEPTION 



What are those perceptual phenomena which the 

 traditional physiology of the senses has left out? It is 

 difficult to make a complete list, but we can enumerate 

 half a dozen aspects of perception which anv physio- 

 logic theory would base to take into account. These 

 aspects are, briefly: a) patterning, b) selectivity, r) 

 reaction to relations or ratios of stimulation, d) reac- 

 tion to similarity, e) apprehension of serial (temporal) 

 order, and f) equivalent reactions to certain spatial 

 and temporal sequences [cf. Lashley (308)]. 



Patterning 



Perception is patterned, since some parts of a 

 stimulus array are always perceived as belonging 

 together while others are not [cf. Schumann (415)]. 

 We tend to see (or feel) 'things,' and not the holes 

 between them [cf. Koffka (284)]. Temporal sequences 

 are analogously structured into events [cf. Johanssen 

 (241)]. Such patterning would imply, as its neural 



counterpart, an interaction of concomitant and suc- 

 cessive processes in afferent systems. Interaction be- 

 tween different sense modalities is perhaps a special 

 form of this general patterning of perception. The 

 difficulty here is not the interaction as such, but its 

 organization — the how and why of interaction — in 

 direct analogy to the problems of coordinated move- 

 ment. 



Selectivity 



Perception is selective. Although the structuring of 

 what we perceive is largely determined by the actual 

 distribution of stimuli in an array, we can selectively 

 attend to one part as against others (see fig. 14). 

 Such selectivity, in neural terms, would probably 

 amount to some preliminary priming or sensitization 

 (308), favoring one part of a neural system over others 

 and hence one way of structuring a complex sensory 

 input over its alternatives. Another, and currently 

 attractive, view would assign a special role in this 

 selection or filtering to extralemniscal afferents (245, 

 324), or to efferent pathways to sense organs (139, 

 140, 168, 203, 204 



Rem lion in Relations 



Perception is relational, that is the organism reacts 

 to ratios of excitation rather than absolute amounts. 

 A classic instance of this is the case of 'transposition' 

 of learned reaction to size differences [cf. Kohler 

 (266)]. Once trained to select the larger of two stimuli 

 (e.g. two squares, a, 10 cm and b, 20 cm in height), 

 the subject readily selects the larger member of 

 another pair of stimuli in which all dimensions have 

 been doubled (thus, he will choose b', when a' = 20 

 cm and b' = 40 cm). There is little difficulty in con- 

 structing models of central nervous activity that 

 would incorporate this feature, but most speculations 

 about the neural basis of discrimination (see Chapter 

 LX by Neff in this volume) either have ignored it, or 

 have tried to reinterpret these phenomena in Pav- 

 lovian terms by invoking principles of conditioning 

 and of "primary stimulus generalization' (374). Con- 

 siderable ingenuity has been displayed in these re- 

 interpretations [cf. Spence (434, 435) and Hull (226)], 

 but the attempts have not been convincing [for cri- 

 tique, see Kliiver (259), Lashley (304), Hunter (227, 

 228), Rudel (407, 408) and others] and are contra- 

 dicted by some experimental evidence (162, 464). 

 Thus, a child or animal (chick, monkey or chim- 

 panzee) can be trained to select an object of inter- 



