It is commonly believed that memory consolidation during sleep takes place
primarily during REM sleep (REM - 'rapid eye movement'). Here we present da
ta and methodological considerations indicating that this concept should be
replaced in favor:of a more differentiated view on sleep associated memory
formation, which aside from the different sleep stages (REM sleep versus s
low wave sleep) takes into account the type of memory (declarative versus n
on-declarative) and the release of cortisol from the hypothalamo-pituitary-
adrenal (HPA) system.
Most of REM sleep occurs during the second half of nocturnal sleep whereas
the first half is dominated-by extensive periods of slow wave sleep (SWS).
The comparison of memory performance after periods of early and late retent
ion sleep indicates that consolidation benefits more from the first half of
sleep dominated by SWS than by the second half of sleep dominated by REM-s
leep. However, this is true only for declarative memories mediated via hipp
ocampal structures. In contrast, non-declarative procedural memories appear
to benefit more from the second than first half of sleep. SWS in the first
half of sleep is associated with a distinct inhibition of cortisol release
whereas cortisol blood concentrations are strongly increased during the se
cond half of sleep. Cortisol directly regulates hippocampal neuronal activi
ty via glucocorticoid receptors (GR) and mineralicorticoid receptors (MR)-
The improving effect of early sleep on consolidation of declarative memorie
s can be completely blocked by experimentally enhancing blood cortisol conc
entrations during this period of sleep. A comparable blockade can be induce
d by administration of dexamethason which is a selective GR agonist.
Results show that declarative memories are consolidated primarily during pe
riods of early sleep associated with SWS. The inhibition of cortisol releas
e leading to an inactivation of hippocampal GR during this early sleep peri
od, represents a necessary prerequisite for the consolidation process.