As neurophysiological investigations of sleep cycle control have provided an increasingly detailed picture of events at the cellular level, the concept that the sleep cycle is generated by the interaction of multiple, anatomically distributed sets of neurons has gradually replaced the hypothesis that sleep is generated by a single, highly localized neuronal oscillator.

Cell groups that discharge during rapid-eye-movement (REM) sleep (REM-on) and neurons that slow or cease firing during REM sleep (REM-off) have long been thought to comprise at least two neurochemically distinct populations. The fact that putatively cholinoceptive and/or cholinergic (REM-on) and putatively aminergic (REM-off) cell populations discharge reciprocally over the sleep cycle suggests a causal interdependence.

In some brain stem areas these cell groups are not anatomically segregated and may instead be neurochemically mixed (interpenetrated). This finding raises important theoretical and practical issues not anticipated in the original reciprocal-interaction model. The electrophysiological evidence concerning the REM-on and REM-off cell groups suggests a gradient of sleep-dependent membrane excitability changes that may be a function of the connectivity strength within an anatomically distributed neuronal network. The connectivity strength may be influenced by the degree of neurochemical interpenetration between the REM-on and REM-offcells. Recognition of these complexities forces us to revise the reciprocal-interaction model and to seek new methods to test its tenets.

Cholinergic microinjection experiments indicate that some populations of REM-on cells can execute specific portions of the REM sleep syndrome or block the generation of REM sleep. This observation suggests that the order of activation within the anatomically distributed generator populations may be critical in determining behavioral outcome. Support for the cholinergic tenets of the reciprocal-interaction model has been reinforced by observations from sleep-disorders medicine.

Specific predictions of the reciprocal-interaction model and suggestions for testing these predictions are enumerated for future experimental programs that aim to understand the cellular and molecular basis of the mammalian sleep cycle.