Creating gratings by optical inference offers two advantages for investigating human temporal vision: (1) it allows use of spatial frequencies high enough that antagonistic surrounds or lateral processes are not modulated, and so only center mechanisms are tested; and (2) it allows one to separate some of the serial processes that determine the overall temporal properties of the system. We overcame the difficulty of producing counterphase gratings with interferometers by using ferroelectric crystals to reverse the phase in one arm of the interferometer during each half cycle of modulation. First we replicated classical temporal contrast sensitivity curves at varying spatial frequencies to ensure that these stimuli are comparable to conventional stimuli. Then we presented pairs of interference gratings either as counterphasing or moving gratings. Superimposing a flickering grating on a steady grating creates (one might say 'injects') a flickering distortion product at the site of a nonlinear process within the retina, whereas moving a pair of interference gratings along the bars of the nearly perpendicular distortion grating modulates the signal that produces the distortion grating without modulating the distortion grating itself. With this array of stimuli we measured the temporal transmission preceding and following the nonlinear stage at 5 and 30 c/deg. However, we found that when the spatial frequencies were increased above those than can be imaged on the retina with conventional techniques, frequencies that are too high to allow lateral interactions, temporal transmission can be described by a nonlinear stage sandwiched by two lowpass filters with identical time constants of about 10 msec. The selective attenuation of low temporal frequencies that gives the temporal contrast sensitivity function its characteristic bandpass shape, which occurs only when the spatial frequency is low, is due entirely to attenuation beyond the nonlinear stage.