J.— PSYCHOLOGY 209 



of its validity. Other evidence, however, points in the same direction. 

 Beyrl found that children showed a greater tendency to see objects in 

 their real sizes from two to ten years and that this tendency was even 

 larger in an adult group. Brunswik found an increase in the tendency to 

 see real whiteness through childhood at any rate up to the age of fourteen 

 years. 



Against these findings, we have Katz's denial of an increase with age 

 based on experiments of Burzlaff who found that it was possible to devise 

 an experiment on size perception in which objects were seen as their real 

 sizes at very early years. I do not think this experiment is relevant to 

 the question. If the size experiment can be so arranged as to give coni- 

 plete phenomenal regression for early years, this way of doing the experi- 

 ment is unsuitable for determining whether or not the effect increases 

 with age. A real increase may be masked by an unsuitably designed 

 experiment. 



On the whole, it looks as if the tendency to phenomenal regression does 

 increase with age. This suggests that the tendency is to some extent 

 plastic to experience. This suggestion is considerably strengthened by 

 the observation that although the tendency to see things in their real 

 shapes and sizes was not absent in a group of artists, it was significantly 

 less than in a control group of corresponding age. This might be ex- 

 plained not as due to the artists' acquired habit of reacting to the stimulus 

 characters of objects but as the result of selection, those of high pheno- 

 menal regression not being likely to become artists since this characteristic 

 handicaps them in learning to draw in perspective. The second explana- 

 tion seems, however, to be ruled out by the further observation that there 

 is no such difference to be observed between a group of art students and 

 a control group of university students. The decisive factor in lowering 

 the tendency to phenomenal regression in the artist group must, therefore, 

 be the greater length of time during which they have formed the habit of 

 reacting to the stimulus characters of objects. 



It is, nevertheless, very improbable that the tendency to phenomenal 

 regression is wholly the product of experience. Kohler has shown that 

 it is, at any rate, found amongst animals with less highly organised nervous 

 systems than our own, since both chimpanzees and hens could be trained 

 to take food from the whiter of two greys even when it was so much less 

 illuminated than the other that its stimulus intensity was lower than that 

 of the darker grey. Phenomenal regression to real colour is also reported 

 to have been observed in fishes and in chicks of three months old. 



These facts have been claimed to disprove an empiricist theory of the 

 origin of those properties of perceived space which determine phenomenal 

 regression (e.g. by Koffka). This claim can hardly be maintained, since 

 the empiricist may retort that we know nothing of the possible speed of 

 learning of spatial relations or of the level of nervous development at 

 which this learning is possible. We can, however, agree that these facts 

 render the empiricist theory somewhat improbable. They do not, 

 however, even offer an argument against the view that the tendency to 

 phenomenal regression may be influenced though not originally produced 

 by experience. The hypothesis that I would suggest is that a phenomenal 

 space so organised in its properties as to produce the phenomenal regression 



