SUMMARY Evidence for circasemidian sleep/wake regulation is briefly reviewed with respect t o protocols used t o quantify sleep propensity. Existing models of sleep/wake regulation are examined i n view of their ability t o accommodate d a t a which demonstrate an afternoon sleep period. Finally, a modelling approach is briefly outlined which emphasizes t h e maintenance of t h e phase and amplitude characteristics of t h e circadian rhythm of body ( a n d brain) t e m p e r a t u r e and predicts t h e circasemidian p h e n o m e n a .KEYWORDS circasemidian, m o d e l , polyphasic, sleep/wake regulation This paper derives from a request by the organizers for the senior author to review at the workshop the current evidence for a 2 per day (circasemidian, hemidian) regulation of sleep propensity and the extent to which existing models accommodate evidence for polyphasic sleep. Models of sleep/wake regulation have primarily addressed monophasic sleep placement in circadian time and the within sleep ultradian NREM/REM alternation. In general they have de-emphasized, or totally ignored, evidence for the inherently polyphasic nature of human sleep. The present paper briefly reviews evidence for a 2 per day regulation of sleep and the status of this process in current models. It closes with an outline of a relevant new modelling approach we have been developing.
EVIDENCE FOR CIRCASEMIDIAN SLEEP/WAKE REGULATIONThe experimental evidence supporting an endogenous 2 per day biorhythmic modulatjon of sleep probability is compelling (reviewed in Broughton 1989). The first proposal of such a control mechanism (Broughton 1975) postulated that it represented a harmonic of the fundamental circadian sleep distribution and might reflect a quasi-12 h rhythm of sws.The ubiquity of napping, which spans cultural and age groups, supports sleep's temporal complexity. During early childhood there is an evolution from a polyphasic pattern to a nocturnal sleep period with several naps and then a single daytime nap taken in the afternoon. Students, who have flexible hours, are frequently afternoon nappers. After retirement, when social restrictions are lifted, napping is common. Naps are again regularly selected in the mid-afternoon period rather than randomly as a nonrhythmic, exclusively recuperative, homeostatic mechanism would predict.Performance studies since the initial report of Blake (1967), have repeatedly shown a 'post-lunch' dip in task variables sensitive to lowered vigilance (sleepiness) which is, moreover, not explicable by food intake.Sleep latency as a measure of ability to get to sleep is a particularly important measure: in order for the (undoubtedly multiple) functions of sleep to occur, the individual must be able to fall asleep. Sleep latency measures by MSLT have typically shown a mid-afternoon decrease indicative of greater ease of wake/sleep transitions at that time but with considerable variability in magnitude and timing in various reports. The latter may be related to several factors: differences involving control...