THE VARIABILITY OF CENTRAL NEURAL ACTIVITY IN A SENSORY SYSTEM, AND ITS IMPLICATIONS FOR THE CENTRAL REFLECTION OF SENSORY EVENTS

Abstract

The time intervals between consecutive discharges of single ventrobasal thalamic neurons, both when active in the absence of intentional peripheral stimuli and when driven by steady sensory stimulation, were measured in unanesthetized macaque monkeys. Some statistical parameters of these internal sequences were estimated, with the hope of determining which quantitative aspects of neural impulse trains may be of importance for the central neural representation of sensory events, and for discrimination. The standard deviation is a constant percentage of the mean interval for populations of discharge intervals during that activity of thalamic neurons evoked by steady peripheral stimuli. This relation holds over a wide range of mean intervals; the regression slope is about 0.6. For that activity of certain thalamic neurons which goes on in the absence of intentional peripheral stimuli, however, this is not the case, for here the standard deviation is an irregular and frequently decreasing proportion of the mean interval, as the latter shortens. How this relation might be of importance for the central neural mechanisms for sensory discrimination was explored from the point of view of a hypotheticated decision process, one based upon the statistical properties of trains of neural discharges rather than one dealing with thresholds and absolute differences in levels of activity. This exploration was carried out with the aid of computational simulation of neuronal activity. The linear relation between mean interval and standard deviation would allow, on this theory, discrimination between populations with much smaller differences in mean intervals than could be accomplished if, for example, the standard deviation were constant, and independent of the mean interval. In about 50% of the trains of activity studied an autocorrelation analysis revealed the presence of periodic fluctuations which could only very rarely be related to other measured cyclic physiological events, such as the respiratory or heart rates, changes in end-tidal CO2, changes in blood pressure, etc. These periodicities can, we believe, contribute a major fraction to the error with which a mean response to repeatedly applied stimuli of the same intensity can be estimated. This finding is thought to emphasize the need to view the central reflection of sensory events in terms of the response characteristics of neuron pools, for if these variations are asynchronous they may be cancelled and thus not appear in the activity of the pool, viewed as an ensemble. On the other hand, the possibility should not be dismissed that these slow variations are themselves of value for the integrative mechanisms of neuronal populations.