Figure 7. Model of clockwork multivesicular release can account for the detection threshold of rod-bipolar cells.

(A) Schematic of a pool of simulated rod photoreceptors providing input to a simulated rod-bipolar cell. Yellow rods indicate cells absorbing a photon; gray rods indicate cells generating only noise. (B) Temporal dynamics of transmitter release illustrating the key parameters of the Erlang release model (see text for details). A flash is delivered 5 s into the trial. (C) Poisson counts underlying the Erlang process. (D) Accumulation of nine Poisson events in (C) triggers an Erlang event. (E) Erlang events in (D) convolved with function mimicking multivesicular release (Equation 1), with additive Gaussian noise. (F–I) Same as (B–E), but for a rod that does not absorb a photon. (J–O) Fraction correct in two-alternative forced-choice detection task as a function of flash strength. Points show output of model, and curves show cumulative Gaussian fits. Blue shows performance of model for parameters estimated from data. Orange and purple show performance for lower and higher model parameters, respectively, for spontaneous release rate (O), Erlang factor (R), rebound amplitude (L), suppression time (M), decay time (N), and rod number in pool (O).