Figure 1.
Depicted are the weekly laboratory sessions. Note that the CONTROL NIGHT was randomly scheduled either before or after the EXPERIMENTAL SESSIONS (EXPERIMENTAL NIGHT + FOLLOW-UP). FOLLOW-UP recall was only performed after the EXPERIMENTAL NIGHT.
Figure 2.
Automatic detection of slow oscillations.
A, slow oscillations were detected from the time point at which subjects spent at least five minutes of continuous sleep in stage S2 or SWS and ended up to 60 minutes thereafter. B, generic detection of a slow oscillation in sleep stage S3. After low-pass filtering the signal at 4Hz, fixed detection criteria were applied to own-built Matlab routines (MathWorks®, Natick, MA). Each detection was specified by five time points: (13) start of the slow oscillation down-state (23), down-state peak (33), start of the up-state and end of the down-state (43), peak of the up-state (53), end of the up-state. a, down-state length; b, down-state peak amplitude; c, negative-to-positive peak-to-peak amplitude; d, up-state length.
Figure 3.
Learning induced changes in slow oscillations.
A, subjects who increased their memory performance overnight (I+) also increased their down-state (P=0.018) and up-state peak (P=0.082t) amplitude from the CONTROL to the EXPERIMENTAL NIGHT. B, slow oscillations peak amplitudes did not differ between the CONTROL and the EXPERIMENTAL NIGHT in subjects who did not increase their memory performance from pre to post sleep (I-). As shaded, I+ and I- seem to generally differ in their up-state phase length (P=0.076t). t, statistical trend (P<0.10).
Figure 4.
Relation between slow oscillation amplitudes and overnight change in memory performance.
A, shown is a significant (2-tailed) correlation (P= 0.045) between changes in the slow oscillation up-state peak amplitude from the CONTROL to the EXPERIMENTAL NIGHT and the overnight change in recall performance. Note that subjects showing strong increases in their up-state amplitude are the ones showing the strongest overnight memory improvements. B, shown is a relation (P= 0.07t) between changes in the absolute down-state peak amplitude of the slow oscillation from the CONTROL to the EXPERIMENTAL NIGHT and the overnight change in recall performance. Note that subjects showing strong increases in their absolute down-state amplitude are the ones showing the strongest overnight memory improvements. t, statistical trend (P<0.10); MEM-CHANGE, overnight memory change (morning-evening recall performance).
Figure 5.
Relation between slow oscillation phase lengths and overnight change in memory performance.
A, shown is the (2-tailed) correlation (P= 0.058t) between changes in the up-state length of the slow oscillation from the CONTROL to the EXPERIMENTAL NIGHT and the overnight change in recall performance. Note that subjects showing strong increases in their up-state phase length are the ones showing the strongest overnight memory improvements. B, shown is the zero-correlation between changes in the down-state length of the slow oscillation from the CONTROL to the EXPERIMENTAL NIGHT and the overnight change in recall performance. t, statistical trend (P<0.10); MEM-CHANGE, overnight memory change (morning-evening recall performance).