A well-known solution of the diffusion equation gives an exponential square-root function as the frequency response for a one-dimensional diffusion or transmission process. When two or more such processes are cascaded, the result is still an exponential square-root characteristic, but with a longer time constant. This seems to explain why flicker thresholds obey the Kelly-Veringa diffusion model at high frequencies, even though the psychophysically inferred diffusion process is much slower than the first stage of visual transduction measured by, for example, late receptor potentials. Two such stages in tandem are sufficient to account for the psychophysical data, because the psychophysical time constant is proportional to the square of the number of stages involved. In addition, the nonlinear behavior of flicker thresholds under intense light adaptation can be explained if the loss factor in the first stage is proportional to the amount of the photopigment bleached. Apparently the flicker thresholds are governed by first- and second-order retinal neurons.