The attenuation of rod signals by backgrounds.

1. The paper which precedes this investigated the nerve interaction between two flashes, lambda at centre (Fig. 1a) and varphi on the surround region (but not on the centre). The size of the inhibitory nerve signal V generated by varphi is given by V = varphi(varphi + sigma), where sigma is the semi-saturation constant.2. A former paper (Alpern & Rushton, 1967) had shown that when the flash varphi falls upon a steady background theta, V suffers attenuation in the G-box (Fig. 1b) down to the fraction theta(D)/(theta(D) + theta) where theta(D) is the eigengrau or receptor noise. Thus, in general, the nerve signal N is given by [Formula: see text].3. This formula had only been established for a moderate range of values. In this paper we use extreme values to explore the limits of its validity. We find the equation to be true over the entire intensity range where N is measurable.4. Six different types of experiment have been performed to test various features of the equation. For instance, if log N is plotted against log varphi for various fixed values of theta, the curve is always the same with simply a vertical shift. And the shift is equal to log(1 + theta/theta(D)) for all values both of theta and of varphi.5. The most interesting curve is the plot of log varphi against log theta for fixed N. This is similar to the Weber-Fechner increment threshold but the criterion is not that varphi be strong enough to be detected, but strong enough to generate an N signal just sufficient to inhibit some fixed lambda flash. These curves (below the onset of saturation) are all the same except for vertical separation, and prove that the condition for flash detection is that a fixed signal, N(0), is generated of size 10(-5) of the maximum signal obtainable (i.e. with varphi large and theta zero).6. With strong backgrounds the curves of (5) above exhibit a marked saturation of the Aguilar & Stiles' type (1954). The family of curves each with a fixed N value shows a remarkable symmetry (Fig. 8) which in fact follows from the equation in (2) above. It has nothing to do with bleached pigment, but follows from the equation in (1) above. V there cannot exceed unity, thus when scaled by the G-box below the criterion level, further increase in varphi will not bring improvement.