Dynamics of respiration and respiratory neuronal activity in REM sleep in the cat
The purpose of this study was to determine the dynamical behavior of respiration and medullary respiratory neurons during REM sleep. Past descriptions of respiration in that state as irregular or erratic are imprecise and unsatisfactory. Thirty-six respiratory neurons from four cats were recorded in wakefulness, non- REM, and REM sleep. Intratracheal pressures, a measure of breathing, and neuronal firing rates were analyzed to determine the mutual information between them. The dimension, entropy, and Lyapunov exponents of each signal were calculated. The maximum firing rate of each neuron during a breath was also analyzed. The mutual information measure showed that measurement of neuronal acdvity at a resolution of 12 bits gave 0.5 to 2.2 bits of information about the co-occurring intratracheal pressure. There was no significant change in mutual information for individual cells across states of consciousness. This indicates that the relationship between respiratory neuronal activity and respiration is maintained during the irregular breathing of REM sleep-a result that does not support the idea that behavioral influences bypass the medullary automatic system in that state. Respiratory muscle activity, particularly upper airway muscle activity, is depressed in REM sleep. This could be the result of depressed central neuronal respiratory activity. However, maximum neuronal firing rate of neurons with high mutual information increased in REM sleep relative to wakefulness-indicating that reduced drive from central respiratory neurons may not be the cause of the atonia. The intratracheal pressure during REM sleep has an average dimension of 7.29, average entropy of 1.12, and average Lyapunov exponent of 1.00. The neuronal activity during REM sleep has an average dimension of 9.05 and average entropy of 0.51. An entropy or Lyapunov exponent greater than zero but less than infinity indicates that a nonlinear dynamic system with sensitive dependence on initial conditions (chaos) produced the signal. These results show (1) that the medullary neurons with high mutual information are not depressed during REM sleep, (2) that the respkatory system is organized as a deterministic, chaotic system, and (3) that the dimension of breathing and respiratory neuronal activity is low relative to the many definable influences on respiration and neuronal activity, such as chemoreceptors, lung stretch receptors, and many types of neurons and ion channels-indicating significant self-organization in the system.