Variation in the amount of neurotransmitter released from one trial to the next limits the reliability with which synapses transfer information. We have investigated this at an excitatory synapse made between two types of neuron in the ocellar system of the locust, Schistocerca gregaria. In the seven large second-order neurons of a lateral ocellus (‘L-neurons’), small, graded changes in membrane potential convey information about changes in light and regulate release of neurotransmitter at output synapses. Small, rebound spikes, which are graded in amplitude, enhance responses to rapid decreases in light. Neurons L1Ð3 make excitatory synapses with large, third-order neurons and L-neurons L4Ð5 (Simmons, 1982). At these synapses, a regenerative response in the postsynaptic neuron normally hides the postsynaptic potential mediated by a rebound spike in a presynaptic neuron. We have used a two-electrode voltage clamp in order to measure the amplitudes of postsynaptic currents (PSCs) at the synapses made by L1Ð3 onto L4Ð5. Rebound spikes were elicited in the presynaptic neuron at the ends of pulses of injected hyperpolarising current. Over a range of presynaptic rebound spike amplitudes between 4 and 20 mV, there is a linear relationship between the amplitudes of a presynaptic spike and the PSC it mediates. Residuals either side of a regression line plotting this relationship have a standard deviation about twice as great as the standard deviation of background noise (means from three experiments, ±0.20 mV for residuals, ±0.13 mV for background noise). This indicates that there is considerable noise intrinsic to the process of transmission at these synapses. This intrinsic noise is constant in amplitude throughout the operating range of the synapse, which indicates that presynaptic potential does not regulate the rate at which neurotransmitter is released by simply altering the probability of release of individual vesicles.

This work was supported by BBSRC.