Consistency in nervous systems: Trial-to-trial and animal-to-animal variations in the responses to repeated applications of a sensory stimulus in Aplysia

Abstract

What is the internal noise in a nervous system? We studied this question by determining the trial-to-trial consistency of the neuronal response in the abdominal ganglion of Aplysia californica. Because our voltage-sensitive dye recordings detected the spike activity from a large fraction of the neurons in the ganglion, these results provide a reasonably complete characterization of the consistency of the response to a sensory stimulus. The consistency of each neurons was evaluated by the number and timing of spikes in the response. The variability in the spike count was described using the coefficient of variation. The spike count variations follow a Poisson distribution, indicating that most of these variations were the result of a random process. For each neuron the reliability of the response to touch was measured in two ways; both measures indicated a broad distribution of reliabilities within the neuron population. The time of the maximum response also varied substantially in some animals. These timing variations were in part due to random processes and in part due to systematic effects (changes in activity of many neurons that were highly correlated). The time course of the activity of individual neurons was compared with the time course of the gill withdrawal. In some animals the activity of individual neurons was only poorly correlated with the behavior; in contrast, the summed activity of groups of neurons matched the behavior quite well. This implies that the behavioral output of the system may be a distributed combination of the activity of many neurons. The differences between animals were substantially larger than the trial-to-trial differences in one animal. The responses made by different preparations differed along many dimensions.