PERCEPTION 



1607 



fig. 5. Figure and ground. Left, white goblet or black faces (Rubin's vase-figure), right, black 

 claw or white protrusions. [From Rubin 14m, 



their universality lias been questioned; their validity 

 might seem to be limited to line drawings or similar 

 visual patterns. There is, however, some evidence 

 that these principles transcend the situations and 

 sense modality from which they were derived. A 

 fundamental aspect of figure perception relative 

 ease of 'tranposition' — was first described lor auditory 

 patterns; a melody can be recognized whether sung 

 by a bass or soprano, or transposed from one ke\ in 

 another (496). Similarly, a shape remains recogniz- 

 able, within limits, in spite of variations in overall 

 dimensions, color and background, or whether pre- 

 sented to the sense of sight or touch. 



The early work of the Gestaltists has stressed these 

 facts ol transposition rather than their limit--, but 

 these limits merit more systematic study. Melodies 

 become unrecognizable when played backwards, and 

 there seem to be serious restrictions in intennodal 

 transfer, e.g. in identification of visual with tactile 

 patterns (292). 



The most urgent objection, however, to applica- 

 tion of Gestalt principles in perception is their non- 

 objective and nonquantitative character which makes 

 specific predictions of perceptual events so difficult. 

 There is, however, no obvious reason why this dif- 

 ficulty should not be overcome. Rigorous psvcho- 

 physical methods can be applied, as has been done by 

 Bobbin (50) in dealing with the principle of closure, 

 or by Heise & Miller (191) in dealing with auditory 

 patterning. It is here that information theory may be 

 profitably applied, particularly in specifying the 



structure of the stimulus array. Thus, Attneave (11, 

 1 2 ) has given an ingenious method of serial guessing 

 of the component parts of visual patterns in analog) 

 to Shannon's technique (422) for estimating the re- 

 dundancy in message sequences A figure is the "bet- 

 ter," the smaller the number of sequential guesses 

 needed to specify it. 



A final objection to Gestalt principles in percep- 

 tion, as they are usually stated, comes from those, 

 such as Hebb (188), who believe that form percep- 

 tion is learned. Gestalt psychologists, by stressing the 

 ubiquity of the principles of grouping on different 

 ontogenetic and phylogenetic levels, have tried to 

 show that these principles antedate learning (560); 

 in fact, Kohler (267) rejected their designation as 

 'innate,' because he believed that these principles 

 reflect basic physical aspects of the brain processes 

 which correspond to perception. If the regularities 

 revealed by these principles are simple physical laws, 

 then they are prior to, and independent of, the de- 

 velopment of organic form 



Bv contrast, modern empiricists have tried to show- 

 that perceptual patterns have to be acquired labori- 

 ously during normal ontogenetic development. For 

 Hebb (188), perceived forms have some "primitive 

 unity' (his term for figure-ground articulation) prior 

 to learning, but this unit formation does not, he 

 believes, suffice in mediating shapes. Shape is acquired 

 in earlv infancy through successive scanning of con- 

 tours by eye movements which are guided reflexly 

 along predominant brightness gradients in the visual 



