Figure 8. Changes in rod signal kinetics: tradeoffs between timing and SNR.

(A) Integration time of retinal output measured from On parasol RGCs (n=8) as a function of background luminance. Integration time corresponds to the integral of the response (up to the zero-crossing) divided by the peak amplitude. Inset traces show normalized PSTHs of spike responses from On parasol RGCs, with three luminance levels (i.e. 0.3, 30 and 900 R*/rod/s) highlighted. (B) Model simulating the SNR for photon capture for a range of integration times, and for various levels of mean luminance (i.e. photon flux). Longer integration times improve the reliability of retinal encoding. Solid lines represent the SNRs experienced by individual rods (left axis) and individual RGCs (right axis) as a function of integration time (taken from A) for a range of photon capture rates (i.e. luminance). Black markers represent integration time measurements taken from On parasol RGC spike recordings. White markers represent a simulated scenario in which integration time does not increase as luminance decreases (i.e. integration time observed at the highest background tested is held constant).