The effects of changing levels of arousal on the spontaneous activity of cortical neurones: I. Sleep and wakefulness.

Abstract

In the preceding paper (Burns & Webb 1976), it was shown that the interval distributions derived from the activity of single cortical neurones can be described by log-normal curves. This description proved satisfactory for cells in visual parietal and auditory cortex. Thus, two parameters – a modal interval and a geometric standard deviation (g. s. d.) – are sufficient to define the whole temporal pattern of discharge for neurones that fire faster than 2.5/s. The same two parameters may be used to describe the first parts of the interval distributions of cells firing less frequently. The purpose of the present paper is to find out whether the values of these two parameters vary systematically with an animal’s state of alertness. Records have therefore been made from single neurones in various parts of the cerebral cortex of unrestrained male cats, when the animals were awake and when they were sleeping. A cat was said to be asleep when he lay with his head supported by some part of the apparatus, with eyes shut, and pinnae unresponsive to laboratory noises. R. e. m. sleep was identified by jerky movements of limbs and eyes. If one records from the same neurone when the cat is awake and when he is asleep, the values of both mode and g. s. d. change with the onset of sleep. Thus either parameter will provide a comparative measure of the animal’s state of alertness. On average the modal interval shortens by a factor of three when an animal falls asleep. This coincides with an increase of 42 % in the size of the g. s. d. The geometric coefficient of variation, which is a dimensionless measure of scatter about the mode - g. c. v. = [log (g. s. d. )]/[log (modal interval)] - also showed systematic changes. On average the g. c. v. increased by a factor of 2.4 when an animal fell asleep. The animal’s state of arousal could also be assessed by examining a single train of action potentials. Interval distributions with modal intervals which are shorter than 20 ms appear to be characteristic of neural activity recorded from a sleeping cat. This rule offers an 88 % chance of successfully classifying a single interval distribution. The size of the g. c. v. can also serve as an efficient ‘test’ of arousal. If one assumes that g. c. vs larger than 0.32 are diagnostic of records taken from animals which are asleep, one’s chance of making an accurate classification is also 88 %. No similar distinction could be made between quiet and r. e. m. sleep.