268 H. K. HARTLINE, F. RATLIFF AND W. H. MILLER 



It is clear that this treatment may be extended to any number of groups. 

 We will consider the case of three groups, but will have no occasion to go 

 beyond this number. Our experiments have been confined to special cases 

 concerning the separate and combined effects of two retinal regions on a third, 

 from one of whose receptors we recorded optic nerve activity. Our results 

 have been analyzed and pubhshed (Hartline and RatliflF, 1958). The coeffi- 

 cients required in the analysis are group coefficients, but only certain combina- 

 tions are observable experimentally. The experimental results in all cases are 

 interpretable in terms of the theory developed for three idealized groups of 

 receptor units. 



In most of these experiments, we illuminated independently two moder- 

 ately large (1-2 mm in diameter) retinal regions, A and B together with a 

 tliird small region X. One of the ommatidia in the region X served as a "test" 

 receptor; its optic nerve fiber was isolated and the discharge of impulses 

 in it recorded. Unless otherwise noted, we attempted to confine the spot of 

 light to the ommatidium of the test receptor alone; when we wished to 

 accentuate the effects of this third group in the interaction we then enlarged 

 the spot of light to include the immediate neighbors of the test receptor. In 

 some cases we also recorded from the optic nerve fiber of an ommatidium in 

 one of the other regions. We will summarize our main results briefly. 



When A and B were separated by 5 mm or more, and were on opposite 

 sides of X, they interacted little if at all. As we have said earlier in this paper 

 (of. Fig. 14), their combined effect in lowering the frequency of X was then 

 equal to the sum of the effects they produced separately 



(-^XA» ^XB > 0; ^AB» ^BA = 0). 



However, if A and B were close together (^ab' ^ba > 0)' their combined 

 effect was less than the sum of their separate effects, because of their mutual 

 inhibition (cf. Fig. 15). To extend this latter experiment, we held constant 

 the intensity on one, B, and varied the iUumination on A; the combined 

 effect on X of A and B together then increased as a hnear function of the 

 effect of A alone on X. This was true, of course, only in that range of inten- 

 sities for which A was strongly enough excited to respond, in the presence of 

 B, at a level that exceeded its threshold of action on X and on B. Examples 

 of results obtained for various configurations of A, B and X are shown in 

 Fig. 19. For A and B close together and at approximately the same distance 

 from X, the slope of this hnear function was always positive, but less than 1, 

 being smaller the closer together we placed A and B. If A was placed farther 

 from X than B, so as to inhibit X only slightly while stiU affecting B strongly, 

 the action of the combination was principally determined by A's inhibition 

 of B producing an indirect effect on X to release it from the inhibition exerted 

 by B. The slope of the function was then negative. This is the case of dis- 

 inhibition described earlier in this paper. If we placed B farther from X than A 



