B R I E F C O M M U N I C AT I O N S

Sleep-dependent learning: a nap the morning on day 1, tested at 19:00 that evening, and then retested at
9:00 the next morning. In the task, a target screen is briefly flashed,
© 2003 Nature Publishing Group [Link]

is as good as a night consisting of a 19 × 19 grid of horizontal bars containing three diago-

nal bars and a central fixation target. Then, after a variable-length
Sara Mednick1, Ken Nakayama1 & Robert Stickgold2 interstimulus interval (ISI), a mask consisting of both horizontal and
diagonal bars appears. The diagonal bars, located in the lower-left
The learning of perceptual skills has been shown in some cases visual quadrant, form either a horizontal row or vertical column. After
to depend on the plasticity of the visual cortex1 and to require each trial, subjects report whether the diagonal bar array was in a row
post-training nocturnal sleep2. We now report that sleep- or column arrangement, as well as whether the central fixation target
dependent learning of a texture discrimination task can be was an "L" or "T." Performance is defined as the threshold ISI required
accomplished in humans by brief (60– 90 min) naps containing for a subject to be 80% accurate in the row/column discrimination.
both slow-wave sleep (SWS) and rapid eye movement (REM) Two experimental groups took naps at 14:00 (group 1: total sleep time,
sleep. This nap-dependent learning closely resembled that 59.3 ± 6.4 (mean ± s.d.); SWS, 20.2 ± 2.0; REM, 4.2 ± 2.2 min; group 2:
previously reported for an 8-h night of sleep in terms of total, 96.3 ± 6.3; SWS, 47.2 ± 5.8; REM, 25.6 ± 4.1 min), while ‘no-nap’
magnitude, sleep-stage dependency and retinotopic specificity, control subjects went about their normal day without midday sleep.
and it was additive to subsequent sleep-dependent All subjects reported sleeping an average of 7.6 ± 2.1 h on the night
improvement, such that performance over 24 h showed as much before day 1 and 7.5 ± 1.2 h on the night before day 2. Improvement
learning as is normally seen after twice that length of time. Thus, was measured as a decrease in threshold ISI from each subject’s base-
from the perspective of behavioral improvement, a nap is as line threshold (at 9:00 on day 1)9.
good as a night of sleep for learning on this perceptual task. No-nap control subjects showed the expected deterioration in per-
formance at 19:00 (13.7 ms longer ISI threshold, P = 0.06, Fig. 1), and
A wide variety of learning processes in both humans and animals performed significantly worse than the nap groups (P = 0.02). The
requires extended, post-training sleep (for review, see ref. 3). deterioration in performance from training to first retest in this group,
Depending on the nature of the task, differing stages of sleep con- measured over an 8-h interval, was similar to that seen in our prior
tribute to this sleep-dependent consolidation process. For the most
part, studies in humans have investigated the benefits of nocturnal
sleep3. Research on behavioral effects of napping has found improve-
15
ment in alertness, productivity and mood4,5, as well as restoration of No-nap
Threshold improvement (ms)

perceptual deterioration6. But whether relatively brief daytime naps 60-min nap
10 90-min nap
can produce learning is not known.
Improvement on visual perceptual tasks can occur over periods of
5
minutes to hours7, as well as over extended periods of days2, and
remain stable for months8. For a visual texture-discrimination task in (n = 28) (n = 13) (n = 2)
which subjects must rapidly discriminate the orientation of a target 0
(n = 13) (n = 17)
embedded in distractors9, initial improvement is seen over the first few
minutes of training10, and slower improvement is seen over subse- –5
quent nights of sleep2,11.
Both SWS12 and REM13 are implicated in a two-stage model of noc- –10
turnal consolidation, with overnight improvement being retinotopi-
cally specific and highly correlated with the product of the amount of –15
–SWS +SWS +SWS
early-night SWS and late-night REM sleep11. Repeated testing within a –REM –REM +REM
day on this task leads to a retinotopically specific deterioration in per-
formance6, which is reversed by 60-min midday naps rich in SWS. But Figure 1 Same-day improvement in no-nap, 60-min nap and 90-min nap
these naps have relatively little REM sleep, and fail to produce signifi- groups, with and without REM and SWS. Left, no-nap group shows
cant improvement over baseline performance. Similar deterioration deterioration at 19:00 from baseline test at 9:00. Center, performance
without sleep is also seen at night12. after naps with SWS but without REM shows neither deterioration nor
improvement. Right, naps with SWS and REM led to significant
We now report that 60- and 90-min naps containing both SWS and
improvement. Only two subjects in the 90-min nap group showed no REM.
REM facilitate learning on this texture-discrimination task, in a man- Experiments were approved by the Committee on the Use of Human
ner similar to that seen after nocturnal sleep. Subjects of Harvard University. Informed written consent was obtained from
Subjects were trained on the texture discrimination task at 9:00 in all subjects.

1Psychology Department, Harvard University, 33 Kirkland Street, Cambridge, Massachusetts 02138, USA. 2Department of Psychiatry at Massachusetts Mental Health

Center, Harvard Medical School, Boston, Massachusetts 02115, USA. Correspondence should be addressed to S.M. (smednick@[Link]).

NATURE NEUROSCIENCE VOLUME 6 | NUMBER 7 | JULY 2003 697

 B R I E F C O M M U N I C AT I O N S

Nap-dependent improvement was not at the expense of subsequent

Threshold improvement (ms)

25
nocturnal improvement. On the contrary, when subjects were retested
Day 2 Day 3
the next morning, the 90-min nap group showed an additional 9.7 ms
20 of improvement (24-h total, 18.1 ms, P < 0.0001), and greater
15 improvement than the no-nap group (Fig. 2, P = 0.03).
10
The no-nap group showed deterioration at 19:00, but normal levels
of improvement on day 2 (7.8 ms; Fig. 2, no-nap). This improvement
5 did not differ significantly from a second (24-h) control group trained
0 at 9:00 and retested 24 h later without an intervening test on the
No-nap 24-h Nap 48-h
evening of day 1 (P > 0.4; Fig. 2, 24-h control).
© 2003 Nature Publishing Group [Link]

Control Control
The nap group actually showed 50% more improvement over a
period of 24 h than the 24-h control group (18.1 vs. 11.8 ms, P = 0.07).
Figure 2 Improvement for nap and no-nap groups. Left, improvements 24 h
Indeed, 24-h improvement in the nap group was as great as that previ-
after training for the no-nap group’s second retest, the 24-h control group’s
first retest, and the 90-min nap group’s second retest, all at 9:00 on day 2. ously reported2 after two nights of sleep (Fig. 2, 48-h control; 18.1 ms
Dotted line shows nap group’s improvement on day 1. Right, improvement vs. 17.5 ms; P > 0.99). Taken together, these findings indicate that a 90-
48 h post-training with no nap. Data for 48-h controls are from ref. 2. Data min nap can produce as much improvement as a night of sleep, and a
for 24-h controls were combined with previously published data2, which nap followed by a night of sleep provides as much benefit as two nights
were not significantly different from the present sample. of sleep.
These findings show that naps can lead to improved performance
on a texture discrimination task similar to previously reported learn-
ing after a full night of sleep2, in terms of magnitude, retinotopic
study, which showed 13.8 ms deterioration over a 2-h interval6. This specificity and dependence on both SWS and REM. Similar improve-
suggests that stimulus exposure rather than inter-test interval pro- ment has been reported after 192 min of early-night sleep12, where
duces deterioration in texture discrimination, and that sleep, rather subjects averaged 74 min of SWS and 24 min of REM—an amount of
than time, is required to reverse this perceptual deterioration6. REM similar to the 25.6 min found in our 90-min nap group.
When tested at 19:00, the 90-min nap group showed significant Finally, napping can significantly enhance the improvement that
improvement (8.4 ms, P = 0.008), whereas the 60-min nap group develops over 24 h. Thus, a nap can not only ameliorate experience-
showed marginal improvement (4.4 ms, P = 0.07). Mindful of our dependent perceptual deterioration, but can also facilitate the learning
hypothesis that both SWS and REM may be necessary for learning in process that results from an hour spent training on a visual texture dis-
naps, we divided the 60-min nap group, all of whom had SWS, into crimination task.
subjects with and without REM and found that 60-min naps with both
ACKNOWLEDGMENTS
SWS and REM produced significant improvement (10.0 ms, P = 0.004,
This research was supported by grants from the US National Institutes of Health
Fig. 1 right). In contrast, 60-min naps with SWS but not REM showed (MH 48832, DA 11744 and NS 26985) and the Air Force Office of Scientific
no improvement (–1.1 ms, P = 0.72, Fig. 1 center) and significantly less Research (83-0320).
than seen in the SWS + REM group (P = 0.01). The 90-min nap group
COMPETING INTERESTS STATEMENT
showed similar results (Fig. 1, white bars). When subjects with SWS +
The authors declare that they have no competing financial interests.
REM naps from the 60- and 90-min groups were combined, improve-
ment correlated significantly with the product of the amount of SWS Received 18 March; accepted 16 May 2003
and REM sleep (stepwise regression, r = 0.37, P = 0.01), similar to find- Published online 22 June 2003; doi:10.1038/nn1078
ings for nocturnal sleep11. The additive effects of SWS and REM seen
in these naps are also similar to those for early and late night sleep 1. Zohary, E., Celebrini, S., Britten, K.H. & Newsome, W.T. Neuronal plasticity that
underlies improvement in perceptual performance. Science 263, 1289–1292
reported previously12. In addition, the amount of improvement did (1994).
not differ significantly from that seen in our previous study of 2. Stickgold, R., James, L. & Hobson, J.A. Visual discrimination learning requires post-
training sleep. Nat. Neurosci. 2, 1237–1238 (2000).
overnight improvement2 (9.7 vs. 11.9 ms; P = 0.5). Thus, nap-
3. Maquet, P. The role of sleep in learning and memory. Science 294, 1048–1052
dependent improvement showed the same magnitude and sleep-stage (2001).
dependency as did overnight improvement2,11 (although of lower 4. Takahashi, M. & Arito, H. Maintenance of alertness and performance by a brief nap
after lunch under prior sleep deficit. Sleep 23, 813–819 (2000).
magnitude than that reported for overnight improvement by oth- 5. Dinges, D.F. & Broughton, R.J. (eds.) Sleep and Alertness: Chronobiological,
ers12,13). Naps with SWS but not REM reversed the deterioration but Behavioral, and Medical Aspects of Napping (Raven Press, New York, 1989).
did not produce actual improvement, whereas naps with both SWS 6. Mednick, S.C. et al. The restorative effect of naps on perceptual deterioration. Nat.
Neurosci. 5, 677–681 (2002).
and REM did both, suggesting that SWS may serve to stabilize per- 7. Fahle, M., Edelman, S. & Poggio, T. Vision Res. 35, 3003–3013 (1995).
formance, and REM may actually facilitate performance improve- 8. Watanabe, T. et al. Greater plasticity in lower-level than higher-level visual motion
processing in a passive perceptual learning task. Nat. Neurosci. 10, 1003–1009
ment. (2002).
We tested the retinotopic specificity of nap-dependent learning by 9. Karni, A. & Sagi, D. Where practice makes perfect in texture discrimination: evidence
training subjects at 9:00 in one visual quadrant (lower left or lower for primary visual cortex plasticity. Proc. Natl. Acad. Sci. USA 88, 4966–4970
(1991).
right, counterbalanced) and then retesting them in the contralateral 10. Karni, A. & Sagi, D. The time course of learning a visual skill. Nature 365, 250–252
quadrant at 19:00, with a 90-min nap at 14:00. Training plus a nap had (1993).
11. Stickgold, R., Whidbee, D., Schirmer, B., Patel, V. & Hobson, J.A. Visual discrimina-
no significant effect on the untrained quadrant (P > 0.2), indicating tion task improvement: a multi-step process occurring during sleep. J. Cogn.
that nap-dependent learning has a retinotopic specificity similar to Neurosci. 12, 246–254 (2000).
that reported for overnight improvement9, as well as for same-day 12. Gais, S. et al. Early sleep triggers memory for early visual discrimination skills. Nat.
Neurosci. 3, 1335–1339 (2000).
deterioration6, and suggesting that the beneficial effects of a 90-min 13. Karni, A. et al. Dependence on REM Sleep of overnight improvement of a perceptual
midday nap are localized to primary visual cortex9. skill. Science 265, 679–682 (1994).

698 VOLUME 6 | NUMBER 7 | JULY 2003 NATURE NEUROSCIENCE