2 : 4/ Light and the Eye 



45 



intensity / can be measured. Therefore, a plot of the fraction of number 

 of correct responses when the light was perceived by the subject against 

 the log / should have the same shape as one of the curves shown in 

 Figure 11. By adding an arbitrary constant to log/, it should be 

 possible to show that the experimental points correspond best to one 

 value of n. 



This experiment satisfies the criterion of not needing to measure the 

 constant k in Equation 6 and gives unique data for the determination 

 for any individual value of the integer n in Equation 8. The value 

 for this constant for some human subjects indicates that n is as high as 

 eight. For other subjects, consistent values as low as one or two have 



05 



Figure I I . P n versus log b for quantum threshold calculation. 

 In this graph, P n is the Poisson distribution probability for n 

 or more events occurring, and b is the average number of events 

 occurring. 



been found for the number of photons necessary to elicit a visual response. 

 In spite of these individual variations, the human data support the idea 

 that the quantum threshold n is a very low number. Most of these 

 measurements are for rod vision, but there is nothing to indicate that 

 the threshold number of photons absorbed is different for cones. 



For the human eye, it is impossible to determine whether the response 

 measured is that of a single receptor. It is possible in experiments using 

 invertebrate eyes, such as those of the king crab, limulus. These eyes 

 have only rod-like receptors called ornmatidia. There is one receptor per 

 nerve fiber. For threshold experiments, the eye, with the optic nerve 

 attached, is removed from the animal. The nerve is then dissected until 

 only one nerve fiber remains intact. It then becomes possible to 



