Figure 3.

Temporal Precision of Spikes in Retinal Bipolar Cells

(A) Nine responses of a bipolar cell to a stimulus in which light intensity was modulated as a sinusoid (100% contrast) and the frequency swept linearly with time from 0 to 20 Hz over a 5 s period and then back (bottom trace). Responses varied from trial-to-trial and the probability of spike generation is a function of frequency.

(B) Comparison of responses in an OFF cell (top) and ON cell (bottom). Stimulus as in (A), but only the first half is shown. From top: raster-plot of spikes times, superimposed voltage responses (25 trials in the OFF cell and 11 trials in the ON), and averaged generator potential when hyperpolarized to prevent spiking (blue traces, −4 pA). The jitter in spike time is larger when light is modulated at lower frequencies.

(C) The phase shift of each spike plotted as a function of stimulus frequency for the ON cell (green) and OFF cell (red) in (B). Phase shift was measured relative to the first zero crossing of the stimulus cycle in which the spike occurred.

(D) The time deviation of each spike compared to a noise-free linear system responding with the same delay, represented by the line fits.

(E) Standard deviation of spike times (jitter) as a function of stimulus frequency, averaged across bins of 4 Hz (filled circles). Corresponding values of vector strength are shown by the open symbols.

(F) Temporal jitter as a function of stimulus frequency displayed as a log-log plot. The line fitting the points has a slope of −0.72.

All error bars represent one S.E.M.