132 Neural Aspects of Vision \1 : 4 



stimulates A and observes a firing rate, it can be reduced by simul- 

 taneously stimulating B. If now C is also stimulated, the firing rate 

 of B will be reduced, thereby permitting the rate of A to rise toward 

 its original value. Thus, the response of any receptor, although affected 

 directly only by its neighbors, depends in a complicated manner on the 

 responses of all the other receptors. 



Similar mutual inhibitions have been observed in vertebrate eyes 

 between the receptors exciting one ganglion cell. It is tempting to 

 hypothesize that in the limulus these mutual interactions are the result 

 of direct interfiber synapses, but in the human retina they are mediated 

 by h and k type cells. This mutual inhibition of neighboring receptors 

 serves to increase acuity by decreasing the effects of glare and of scatter- 

 ing within the eye. It also makes the threshold much higher near a 

 bright object. Thus, gradations at the edge of a bright light appear 

 much sharper to the eye than to a series of independent photocells. 



Sharpening effects of this type are well known in psychophysical 

 studies. They support the idea that neighboring receptors do inhibit 

 each other. Psychophysical evidence, however, cannot clarify whether 

 these effects in human vision occur at the receptors themselves or at the 

 first set of neurons with which the rod and cone fibers synapse. It is 

 even possible that a major portion of the sharpening in human vision 

 occurs within the central nervous system. 



A different type of neural analysis has been demonstrated by Land 

 and his associates. They found that, although the description given 

 previously in this chapter for color discrimination was valid for large 

 patches of color or for one or two colors in the visual field, it was very 

 misleading for color vision as it normally occurs. To show this, they 

 used two photographic slides, one exposed in the short wavelength 

 region of visible light and the other in the long wavelength region. 

 When these were used simultaneously but illuminated with two different 

 broad bands of light, the natural color sensations were reproduced. 

 Similar experiments with narrow bands of light (that is, monochromatic 

 lights) produced about two-thirds of the possible colors. The effective 

 colors depended only on the per cent of the maximum (or average) of 

 each light transmitted and not on their absolute intensities. It further 

 depended on a random (or gaussian) distribution of small patches of 

 colors such as occur in the normal visual field. 



These results can be brought into accord with the model in Figure 2 

 by very slightly modifying the assumptions made. One notes in that 

 model that although three receptors are excited, essentially two ratios 

 are used for color vision under photoptic conditions. These are the 

 ratios of the responses of the m and p type cells to that of the s type cells. 

 It is clear that only the two ratios can be important and not the absolute 



