For pt.I see ibid., vol.35, no.11, p.941-7 (1988). The objective of this research is the noninvasive prediction of retinal cell activity; more specifically, the goal is to assess the healthiness of discrete regions of the retina. The authors describe a multiple-stimulus approach which involves a single-potential measurement of the cornea as is commonly done clinically at present, but which multiple-flash stimuli. The technique is ultimately dependent on the patient's ability to focus on the external flash stimulus pattern. The individual components of the ERG (electroretinogram), which are directly associated with specific parts of the stimulus pattern (and thus, localized regions on the retina), are stripped off of the composite ERG using signal-correlation techniques. At the heart of this approach is the requirement that the isolated stimuli are each individually uncorrelated to one another, i.e., they are random in time. An analytical ERG is constructed to test the efficacy of the use of single- and double-kernel correlations for predicting retinal impulse responses.<>