Properties of human sleep-wake cycles: parameters of internally synchronized free-running rhythms.

Abstract

Sleep-wake alternations are governed by circadian regularities. In order to evaluate these regularities without interference from social constraints or behavioral influences, we conducted experiments under constant conditions, excluding all external time references. The experiments were conducted for approximately 1 month to ensure that the rhythms were at a steady state. A homogeneous sample of 27 human subjects with free-running and internally synchronized rhythms was analyzed with regard to numerous sleep-wake parameters. There was no temporal trend during the entire period or in individual wake or sleep episodes. The onset of sleep was consistently and by far the most variable reference phase within the sleep-wake cycle. The results of negative serial correlations within the sleep-wake rhythm were relevant. Essentially, every distortion in the duration of a cycle is followed, with high probability, by a deviation in the duration of the following cycles in the opposite direction; i.e., any chance variation in the duration of a cycle is corrected with the next, and to a smaller amount with the next but one, cycle. Hence, an intrinsic stabilizing mechanism of the underlying pacemaker is in effect. Secondarily, there are negative serial correlations among adjacent wake and sleep episodes. Every deviation of an episode from the long-term mean results in an opposite deviation of the following episode. In other words, a wake episode determines the duration of the following sleep, and a sleep episode determines the duration of the following wake. All these negative serial correlations are highly significant interindividually. Another relevant result concerns the difference between females and males. The mean sleep-wake cycle is significantly shorter in females than in males, on the average by 28 min. Even more significant is the sex difference in the fraction of sleep. On the average, the wake episode is shorter by 1 h 49 min and the sleep episode is longer by 1 h 21 min in females than in males; i.e., the fraction of sleep is larger for 18% in females than males. There are indications that the established sex difference concerns, within the human multioscillator system, only that oscillator which is predominantly responsible for sleep-wake rhythm, but not the other oscillator, which is predominantly responsible, for instance, for deep body temperature. On the other hand, no parameter describing variabilities for the period or the separate episodes shows a sex difference, either in amount or in the temporal sequence of the variations as expressed in the serial correlations.