1. Simultaneous intracellular recordings were made from two neighboring N amacrine cells, one an ON amacrine (NA) cell and the other an OFF amacrine (NB) cell. Extrinsic current was injected into one amacrine cell, and the resulting intracellular responses were recorded from the other amacrine cell. Test signals included 1) a single-frequency sinusoid, 2) a depolarizing or hyperpolarizing pulse, or 3) a white-noise modulated current. In some cell pairs, membrane noise was measured in the dark as well as under a steady background illumination. 2. Current pulses injected into a NA cell evoked a damped oscillation from a NB cell. The first-order kernel derived by cross-correlating the white-noise current injected into a NA cell against the evoked response from a NB cell was a large depolarization followed by a damped oscillation. The frequency of oscillations varied slightly from pair to pair but averaged 35 Hz. 3. Current pulses injected into a NB cell evoked a sign-inverting response (hyperpolarization) of very small amplitude from a NA cell. Similarly, the first-order kernel was a hyperpolarization of very small amplitude. 4. The power spectrum of the membrane noise recorded from NA and NB cells in the dark or during steady illumination often showed a peak at 35 Hz. Such membrane noise synchronizes synergistically among NA cells and among NB cells in the dark. In addition, the membrane fluctuations seen in NA and NB cells in the dark were out of phase. 5. Transmission between NA and NB cells was largely accounted for by a linear component; however, a very small but significant second- and third-order nonlinearity was also generated. 6. These results show that the interactions occurring between amacrine cells of opposite response polarity are much more complex than those between cells of the same response polarity and that the neural circuitry in the inner retina actively controls interactions between ON and OFF channels in the dark as well as in the presence of light stimuli.
Comments on the analysis of membrane noise by the "integral spectrum" procedure
