Neurobiology of Learning and Memory 162 (2019) 1–8

Contents lists available at ScienceDirect

Neurobiology of Learning and Memory

journal homepage: www.elsevier.com/locate/ynlme

Sleep accelerates re-stabilization of human declarative memories T

a b c a,⁎
Malen D. Moyano , Susanne Diekelmann , María E. Pedreira , Cecilia Forcato
a
Unidad Ejecutora de Estudios de Neurociencias y Sistemas Complejos, CONICET, Universidad Nacional Arturo Jauretche, Hospital de Alta Complejidad en Red El Cruce
“Néstor Kirchner”, Florencio Varela, Argentina
b
Institute of Medical Psychology and Behavioral Neurobiology, University of Tübingen, Tübingen, Germany
c
Laboratorio de Neurobiología de la Memoria, Departamento de Fisiología y Biología Molecular y Celular, IFIBYNE-CONICET, Facultad de Ciencias Exactas y Naturales,
Universidad de Buenos Aires, Pab. II, 2do piso, Buenos Aires, Argentina

ARTICLE INFO ABSTRACT

Keywords: Consolidated memories can return to a labile state upon presentation of a reminder, followed by a period of re-
Reconsolidation stabilization known as reconsolidation. This period can take several hours, and if an amnesic agent (e.g. new
Declarative memory learning) is administered inside the time window of reconsolidation (when the memory is still labile) the
Interference memory is impaired, whereas the memory remains unaffected if the amnesic agent is administered outside this
Sleep
time window. Sleep plays a fundamental role in the consolidation and integration of new memories, and recently
Slow wave activity
sleep has also been implicated in memory reconsolidation. Here, we studied the role of sleep in accelerating the
reconsolidation time window. On day 1, participants learned a list of syllable-pairs (List 1). On day 2, they
received a reminder, followed by interference learning (List 2) administered either after 90 min of wakefulness,
after 90 min of sleep, or after 10 h of wakefulness. On day 3, participants had to recall List 1 first, followed by
List 2, and we assessed the Retrieval-Induced-Forgetting Effect (RIF) on List 2 as a measure of List 1 memory
stability. We found that the 90 min sleep group showed an intact RIF effect similar to the 10 h wake group,
reflecting stable List 1 memory after 90 min of sleep and after 10 h of wakefulness. However, the RIF effect was
absent after 90 min of wakefulness, suggesting that the List 1 memory was still labile at that time. Moreover, the
RIF effect in the 90 min sleep group was associated with power density in the slow oscillation frequency band
(0.5–1 Hz) during SWS and S2. These findings suggest that 90 min of sleep accelerate memory re-stabilization
after reminder presentation, shortening the reconsolidation time window and protecting the memory against
subsequent interference. This rapid memory re-stabilization may depend on slow oscillation activity during
NREM sleep.

1. Introduction inhibitor, new learning, beta-blocker) is presented inside the re-

After acquisition, new memories are in a labile state and they need
to stabilize in order to persist, a process known as consolidation. This
process involves modifications of the synapses concerning the engram
(synaptic consolidation) as well as a redistribution of information to
long-term storage areas (system consolidation) (Dudai, Karni, & Born,
2015). However, once consolidated, memories are not fixed. After the
presentation of a reminder, stored memories can return to a labile state,
and become susceptible to enhancement or impairment again (Nader,
Schafe, & Le Doux, 2000). These memories then require reconsolidation
in order to persist, which involves the re-stabilization of the memory
trace taking time to be accomplished (around 6 h, Nader et al., 2000;
Walker, Brakefield, Hobson, & Stickgold, 2003; Forcato et al., 2007;
Herszage & Censor, 2017). If an amnesic agent (e.g. protein synthesis
consolidation time window, the re-stabilization process and subsequent
memory performance are impaired. However, if the amnesic agent is
presented outside the reconsolidation time window (after more than
6 h, Nader et al., 2000; Walker et al., 2003; Forcato et al., 2007;
Herszage & Censor, 2017), the memory trace remains intact. This
process of labilization and reconsolidation has been demonstrated in
different species and types of memories including declarative memories
in humans (Forcato et al., 2007; Hupbach, Gomez, Hardt, & Nadel,
2007; Nader et al., 2000) and represents a unique opportunity to
modify stored information (Forcato, Fernández, & Pedreira, 2014).
Forcato et al. (2007) showed that a consolidated declarative
memory (list of syllable pairs, List 1) can be reactivated by the pre-
sentation of a cue-reminder and a second learning session (of another
list of syllable pairs, List 2) interferes with its re-stabilization when it is

⁎
Corresponding author at: Unidad Ejecutora de Estudios de Neurociencias y Sistemas Complejos, CONICET, Universidad Nacional Arturo Jauretche, Hospital de
Alta Complejidad en Red El Cruce “Néstor Kirchner”, Av. Calchaquí 5400, 1888 Florencio Varela, Buenos Aires, Argentina.
E-mail address: cecilia.forcato@gmail.com (C. Forcato).

https://doi.org/10.1016/j.nlm.2019.04.012
Received 4 December 2018; Received in revised form 25 March 2019; Accepted 24 April 2019
Available online 25 April 2019
1074-7427/ © 2019 Elsevier Inc. All rights reserved.
M.D. Moyano, et al.
presented 5 min or 6 h after the reminder (inside the time window of
reconsolidation) but not 10 h later when the memory is already re-
stabilized (outside the time window). Impaired re-stabilization of the
target memory inside the time window of reconsolidation is typically
indicated by the absence of a Retrieval-Induced Forgetting effect
(Forcato et al., 2007; Forcato, Argibay, Pedreira, & Maldonado, 2009;
Forcato, Fernández, & Pedreira, 2013). Retrieval-Induced Forgetting
(RIF, Anderson, Bjork, & Bjork, 1994; MacLeod & Macrae, 2001) allows
for an assessment of the integrity of the memory trace, i.e. whether a
stored memory is intact or impaired, beyond simple retrieval perfor-
mance. According to the RIF effect, retrieval of one memory (e.g. List 1)
can temporarily block the retrieval of a related memory (e.g. List 2,
Forcato et al., 2007). Importantly, this RIF effect is only observed when
the memory trace that is recalled first is intact, whereas the RIF effect is
not observed when the first recalled memory trace in impaired. That is,
if the stored memory of List 1 is intact, its retrieval blocks the sub-
sequent retrieval of the memory of List 2 (evidenced as intact RIF ef-
fect). On the other hand, if the stored memory of List 1 is impaired, its
retrieval does not block the subsequent retrieval of the List 2 memory
(evidenced by no RIF effect; Forcato et al., 2007, 2009, 2013). Apart
from the RIF effect, memory performance at retrieval may also be in-
fluenced by simultaneous retrieval interference. If memories of both
List 1 and List 2 are recalled during the same session at retrieval, they
may interfere with each other leading to acute performance decre-
ments, which are however, independent of the integrity of the under-
lying memory trace (Forcato et al., 2007, 2009).
Sleep plays a key role in the consolidation of new information
(Rasch & Born, 2013). For declarative memories, sleep strengthens and
stabilizes newly encoded representations and integrates them into pre-
existing long-term stores (Diekelmann & Born, 2010; Diekelmann,
Büchel, Born, & Rasch, 2011; Dudai et al., 2015). This positive effect of
sleep is not only observed after a full night of sleep but also after short
naps of 6–90 min (Alger, Lau, & Fishbein, 2012; Diekelmann, Biggel,
Rasch, & Born, 2012; Lahl, Wispel, Willigens, & Pietrowsky, 2008).
Recent evidence suggests that sleep also benefits memory reconsolida-
tion (Walker et al., 2003; Klinzing, Rasch, Born, & Diekelmann, 2016;
Kindt & Soeter, 2018; but see Hardwicke, Taqi, & Shanks, 2016). Slow
wave sleep (SWS) and slow wave activity (0.5–4 Hz) in particular have
been linked to both consolidation and reconsolidation of declarative
memories (Alger et al., 2012; Diekelmann et al., 2012; Klinzing et al.,
2016; Macdonald & Cote, 2016; Wilhelm et al., 2011).
Here, we asked whether a diurnal 90 min-nap accelerates memory re-
consolidation by shortening the time window during which labile mem-
ories are susceptible to interference. To test this question, participants
learned a list of syllable pairs on day 1 (List 1) and received a reminder on
day 2. Reminder presentation was followed by an interference task
(learning of a second list of syllable pairs, List 2) either after 90 min of
wakefulness, after a 90 min nap, or after 10 h of wakefulness. Participants
were then tested on both lists on day 3, specifically assessing the RIF effect
on List 2. We expected that after 90 min of wakefulness, the memory is still
labile (inside the reconsolidation window) and thus impaired by inter-
ference learning, as evidenced by the absence of a RIF effect on List 2. After
10 h of wakefulness, we expected the memory to be already re-stabilized
(outside the reconsolidation window) and thus not to be affected by in-
terference learning, evidenced by an intact RIF effect on List 2. Importantly,
we hypothesized that 90 min of diurnal sleep following reminder pre-
sentation accelerates memory reconsolidation (i.e. shortens the re-
consolidation window) such that the memory would be protected against
interference learning already after 90 min and this effect would be revealed
by an intact RIF effect on List 2.
2. Materials and methods
2.1. Participants
86 undergraduate and graduate students volunteered for the study
2
Neurobiology of Learning and Memory 162 (2019) 1–8
(50 women, 36 men). None of the participants reported ongoing med-
ication, health problems, medical interventions, or history of psychia-
tric, neurological, or sleep disorders. Subjects that reached at least 60%
of correct responses during the last four training trials were included in
the analysis. 14 subjects were excluded from the analysis because they
did not reach the learning criterion (9), did not sleep (2), did not un-
derstand the instruction (1), did not complete all experimental sessions
(1), or rehearsed the syllables during the retention interval (1). The
final sample included 72 participants, with ages ranging from 18 to 36
(23.87 ± 0.50 years).
All the experiments were carried out in accordance with the Code of
Ethics of the World Medical Association (Declaration of Helsinki).
Before their participation in the experiment, subjects signed a written
informed consent that had been approved by the “Comité de Ética del
Hospital en Alta Complejidad en Red El Cruce, Dr. Néstor Kirchner”.
2.2. Experimental groups
Participants were randomly assigned to one of five groups. All
groups performed the List 1 training on day 1 and were tested on day 3
(Fig. 1A). The groups differed in the treatment they received on day 2.
The “Reactivation-90minwake-Interference” group (n = 15) received
the reminder and 90 min later the interference task, within the re-
consolidation time window (Forcato et al., 2007). The “Reactivation-
10hwake-Interference” group (n = 14) received a reminder and 10 h
later the interference task, outside the reconsolidation time window
(Forcato et al., 2007). The “Reactivation-90minsleep-Interference”
group (n = 13) received a reminder, and learned the interference task
after sleeping for 90 min. Two more control groups were assessed. The
control List 1 group (“CTL-L1”, n = 19) received the reminder on day 2
but did not learn the interference task. Participants in the control List 2
group (“CTL-L2”, n = 11) skipped List 1 training on day 1, learned the
interference task on day 2 without any reminders and were tested on
day 3.
2.3. Procedure
For the “Reactivation-90minsleep-Interference” group, all subjects
spent an adaptation day in the sleep lab including placement of elec-
trodes before the experimental day. On the adaptation day, participants
arrived at the lab between 12:00 and 14:00 h. First, they signed the
informed consent form and filled out the personal data questionnaire.
Thirty minutes later they were prepared for polysomnographical re-
cordings and were then allowed to sleep in a quiet, darkened room.
They were awakened after 90 min of sleep and electrodes were re-
moved. The adaptation day and the experimental day 1 were separated
by at least 48 h. For the experiment proper, participants returned to the
lab on three consecutive days. On day 1, they arrived at the lab between
12:00 h and 15:00 h and had to complete the Stanford Sleepiness Scale
first (SSS, 7-point scale, ranging from 1 “feeling active, vital, alert, or
wide awake” to 7 “no longer fighting sleep, sleep onset soon, having
dream-like thoughts”). After that, they learned the first list of syllable
pairs (List 1) on the computer and then left the lab. On day 2, between
12:00 and 14:00 h, the participants returned to the lab and the elec-
trodes were placed. The procedure took twenty minutes. After that,
they completed the SSS, received the reminder and were then allowed
to sleep for 90 min. The experimenter counted 90 min from the first
spindle or K-Complex detected online. After being awakened, partici-
pants waited 30 min to avoid sleep inertia. During this time, electrodes
were removed and participants were allowed to freshen up. Then they
performed the SSS again, and after that learned the interference task
(List 2). They were then allowed to leave the lab and returned on the
next day. On day 3, they completed the SSS again and were tested for
retrieval on both lists.
The participants in the “Reactivation-90minwake-Interference”
group arrived at the lab on day 1 between 12:00 h and 15:00 h. First,
M.D. Moyano, et al.
they signed the informed consent form, filled out the personal data
questionnaire and completed the SSS. After that, they learned the first
list of syllable-pairs (List 1) and then left the lab. On day 2, between
12:00 and 14:00 h, the participants returned to the lab, completed the
SSS and received the reminder. Following the reminder presentation,
they were allowed to leave the lab and returned 90 min later. During
the 90 min period, they were allowed to continue with their normal
activities but they were instructed not to fall asleep. Upon their return
to the lab, they completed the SSS again and learned the interference
task (List 2). They were then allowed to leave the lab and returned on
the next day, day 3, between 12:00 and 14:00 h. They completed the
SSS again and were tested for retrieval on both lists.
The “Reactivation-10hwake-Interference” group arrived at the lab
on day 1 between 7:00 h and 9:00 h, signed the informed consent form,
filled out the personal data questionnaire, completed the SSS, and
learned the first list of syllable-pairs (List 1). They returned to the lab on
day 2 between 7:00 h and 9:00 h and followed the same procedure as
the “Reactivation-90minwake-Interference” group, except that they had
a 10 h period between the reminder and the interference task. During
this period they were allowed to leave the lab and continue with their
normal activities. They were instructed not to sleep during this period.
They were tested for retrieval on day 3, at the same time they were
trained on the first day.
The procedures for the control groups were similar to the experi-
mental wake groups but they only learned one list of syllables. The
“CTL-L1” group arrived at the lab on day 1 between 12:00 h and
15:00 h. First, they signed the informed consent form, filled out the
personal data questionnaire and completed the SSS. After that, they
learned the List 1 and then left the lab. On day 2, between 12:00 and
14:00 h, the participants returned to the lab and completed the SSS.
After that, they received the reminder and then left the lab. They re-
turned on day 3, between 12:00 and 14:00 h, completed the SSS again
and were then tested for retrieval on List 1. The “CTL-L2” group started
the experiment directly on day 2. They arrived at the lab between
12:00 h and 15:00 h. First, they signed the informed consent form, filled
out the personal data questionnaire and completed the SSS. After that,
they learned the List 2 and then left the lab. On day 3, between 12:00
and 14:00 h, the participants returned to the lab, completed the SSS and
were tested for retrieval on List 2.
2.4. The task
The task consisted of memorizing five pairs of nonsense syllables,
associated with a context formed by a background color on the com-
puter screen, an image and music, presented through headphones. The
syllables were formed by three letters (Fig. 1B). We have previously
shown that including a context associated to the list of syllables im-
proves memory retention (Forcato et al., 2007).
2.4.1. List 1 training session
List 1 was formed by five pairs of nonsense cue-response syllables:
ITE-OBN, ASP-UOD, FLI-AIO, NEB-FOT, COS-GLE (bold type: cue-
syllable; regular type: response-syllable, Fig. 1B). The training session
of List 1 consisted of 10 runs, associated to a context consisting of a blue
background color, an image of an Italian coast and a tarantella melody
(Fig. 1C). Each run was formed by five trials, including the presentation
of the entire list of five pairs of syllables. In the first run, the entire list
of all five syllable pairs was presented once. The first run started with
the presentation of the context: first, the image of the Italian coast with
blue background color appeared on the screen for 4 s. Then, a tarantella
melody played along with the image and background color for another
4 s. After that, the context continued while the syllables were presented.
First, one cue-syllable appeared at the left top side of the monitor’s
screen and an empty response box was displayed on the right top. Next,
the corresponding response-syllable appeared in the response box and
stayed there for 4 s. Immediately thereafter, the syllable pair
3
Neurobiology of Learning and Memory 162 (2019) 1–8
disappeared and another cue-syllable was shown one line below and the
process was repeated until the list was complete. Each cue-syllable was
taken at random and successively from the list of five until the run was
complete. Thus, in the first run the participants observed how all five
pairs were completed once. In the following nine runs, participants had
to complete the syllable-pairs themselves. Each of those nine runs began
with the presentation of one cue-syllable at the left top side of the
monitor’s screen and an empty response box on the right top. Subjects
were required to write down the corresponding response-syllable
within 5 s. Depending on the subject’s response, there were three pos-
sible outcomes: first, if no syllable was written down, the correct one
was shown for 4 s; second, if an incorrect syllable was written, it was
replaced by the correct one and it was shown for 4 s; and third, if the
correct response was given, it stayed for a further 4 s. Immediately
thereafter, the syllable pair disappeared and another cue-syllable was
shown one line below and the process was repeated until the list was
complete. Each cue-syllable was taken at random and successively from
the list of five until the run was complete. After the whole list was
presented, a black background was shown for 3 s and the procedure was
repeated until the entire list of syllable-pairs was completed nine times.
This training procedure (with syllable pairs being presented in-
dividually and consecutively for 4 s) was based on previous studies
showing that this procedure resulted in an optimal level of memory
strength at encoding (Forcato et al., 2007; Stickgold, 2009). The List 1
training session took about 10 min.
2.4.2. Reminder
The reminder included the context of List 1 (i.e. blue background
color, an image of an Italian coast and a tarantella melody) followed by
one cue-syllable taken at random, at the left top side of the monitor’s
screen and an empty response box on the top right. After 2 s, a notice
was displayed on the monitor announcing that the session was inter-
rupted, thus not allowing the subject to write down the response-syl-
lable (Fig. 1D). It has been shown that this procedure triggers memory
reconsolidation (Forcato et al., 2007).
2.4.3. Interference task
Learning of the interference task was similar to the List 1 training
session, but with a different context (red background colour, the image
of a forest and classical music). The interference task consisted of the
learning of another list of syllable pairs, List 2, which was formed by
five different pairs of nonsense cue-response syllables: OEN-SRO, DRI-
CRE, AIC-POA, TIU-PLA, KEC-CLO (Fig. 1B). Like the List 1 training
session, it was formed by 10 runs. In the first run, participants observed
how the five syllable pairs were completed once and in the successive
nine runs they had to write down the corresponding response-syllables.
Feedback procedures were as for List 1 training. The interference task
took about 10 min.
2.4.4. Testing session
On day 3, List 1 and List 2 memory was tested. List 1 memory was
always tested first before List 2 memory. For both lists, testing was
formed by two runs each, similar to the training session, i.e. including
recall of each of the syllable pairs twice. Cue-syllables were taken at
random and successively from the list of five. Subjects were required to
write down the corresponding response-syllable within 5 s. The testing
session took about 4 min (2 min per List).
The written responses were registered and an error was considered
when a participant wrote an incorrect response-syllable or if they did
not answer at all. We took the total errors made in the two runs and
calculated the percentage of errors for each subject for List 1 and List 2
testing. Furthermore, we classified the errors into four categories:
“void”, when no response was written down; “intralist”, when a wrong
response-syllable from the same list was written down; “intrusion”,
when a syllable from the other list was written down; and “confusion”,
when the written response-syllable was not included in any of the lists.
M.D. Moyano, et al. Neurobiology of Learning and Memory 162 (2019) 1–8

Fig. 1. Experimental design and memory task. A. Protocol. The experiment was run on three consecutive days. On day 1, all groups received the List 1 training
(except for the “CTL-L2” group) and were tested on day 3. The groups differed with regard to the treatment on day 2. The “CTL-L1” group received only the reminder.
The “Reactivation-90minwake-Interference” group received the reminder and 90 min later the interference task (List 2). The “Reactivation-10hwake-Interference”
group received the reminder and 10 h later the interference task. The “Reactivation-90minsleep-Interference” group received the reminder, slept for 90 min and
received the interference task after being awakened. The “CTL-L2” group only learned the interference task on day 2 and was tested on day 3. B. Lists. List 1 (L1) and
List 2 (L2, interference task) differed with regard to the cue-response syllable pairs. Each list contained five pairs of nonsense cue syllables (indicated in gray boxes)
and response syllables (indicated in white boxes). C. Training session. Each list was associated with a specific context, i.e. a background color, an image and specific
music (for List 1: blue background, image of an Italian coast, tarantella music). The training session consisted of 10 runs. Each run started with the presentation of the
context, i.e. the background color and the image for 4 s, followed by the same stimuli accompanied by the music for another 4 s. After that, the first cue-syllable
appeared on the left side of the screen, with an empty response box on the right. During the first training run, the response-syllable was then presented automatically
in the response box on the right. After that, the next cue-syllable appeared one line below and the procedure continued until all five syllable pairs were presented
once. After the first run, a black screen was displayed for 3 s before the next run started. The following nine runs were identical to the first run, except that the
response-syllables were no longer presented automatically but participants had to write down the response-syllables in the response box on the right. D. Reminder.
The reminder session was similar to one training trial of List 1, with presentation of the context (background color, image, music) and one cue-syllable. However, the
trial was interrupted 2 s after the presentation of the cue-syllable without allowing subjects to complete the trial with the response-syllable. (For interpretation of the
references to colour in this figure legend, the reader is referred to the web version of this article.)

4
M.D. Moyano, et al.
2.4.5. Demo
Before the List 1 training session, participants were presented with a
demo program to receive all the instructions and to make sure that all
participants understood the task. The demo program consisted of two
trials, similar in structure as the training session but with another
context and two different pairs of nonsense syllables.
2.5. Sleep recordings
2.5.1. Polysomnography
Standard polysomnographic recordings were obtained using
SOMNOscreen™ (SOMNOmedics), including electroencephalography
(EEG), electromyography (EMG), and electrooculography (EOG). Six
electrodes were used for EEG, placed at F3, F4, C3, C4, P3 and P4,
according to the International 10–20 system, referenced to electrodes
attached to the mastoids. Data were recorded at a sampling frequency
of 256 Hz and bandpass filtered between 0.2 and 30 Hz. Recordings
were scored according to standard criteria (Rechtschaffen & Kales,
1968). Analysis was based on epochs free of visually identified EEG
artifacts.
2.5.2. Power spectral analysis
Power density was calculated separately for all sleep epochs of sleep
stage 2, SWS (sleep stages 3 and 4) and Rapid Eye Movement sleep
(REM). Only artifact-free intervals were analyzed by Fast Fourier
Transformations (FFT) with an adapted Hanning window applied to
subsequent blocks of 2048 data points (∼8 s), with 205 data points
overlap (∼0.8 s). Individual mean power density was averaged across
all electrodes in the following frequency bands: slow oscillations
(0.5–1 Hz), delta (1–4 Hz), theta (4–8 Hz), slow spindle (9–12 Hz), fast
spindle (12–15 Hz), beta (15–30 Hz), and alpha (8–13 Hz). One parti-
cipant was excluded from the power spectral analysis because of too
much noise in the EEG signal.
2.6. Data analysis and statistics
We analyzed two different measures to assess retrieval and the
stability of the target memory (List 1). First, we analyzed the Retrieval-
Induced Forgetting (RIF) effect (Anderson et al., 1994; MacLeod &
Macrae, 2001) by calculating the percentage of List 2 errors, indicating
to what extent the prior retrieval of the target memory (List 1) blocks
the retrieval of related memories (List 2). The presence of the RIF effect,
i.e. reduced List 2 retrieval, speaks for a stable List 1 target memory,
whereas the absence of the RIF effect, i.e. unaffected List 2 retrieval, is
an indicator of an impaired List 1 target memory (Forcato et al., 2007,
2009). Second, we analyzed the retrieval performance of the List 1
target memory by calculating the percentage of List 1 errors. This
measure indicates to what extent the retrieval of List 1 is subject to
simultaneous retrieval interference at the time of recall (due to si-
multaneous recall of List 1 and List 2 in the same session), an effect that
is independent of the integrity of the stored List 1 memory trace
(Forcato et al., 2007, 2009).
The percentage of errors at training of List 1 and List 2 were ana-
lyzed with repeated measures ANOVA with “group” as a between
subjects factor and “run” as a within subjects factor. The percentage of
total errors at List 1 and List 2 testing was analyzed with one-way
ANOVA with “group” as a between subjects factor followed by LSD
Post-hoc comparisons. Results of the LSD Post-hoc tests are reported
uncorrected; but note that the main results remain essentially un-
changed when Bonferroni correction for multiple comparisons is ap-
plied (with a corrected alpha of 0.05/12 = 0.0042). Pearson correlation
analysis was performed to determine the correlation between power
density during sleep and the percentage of errors in List 1 and List 2
5
Neurobiology of Learning and Memory 162 (2019) 1–8
memory at testing. All correlations are reported uncorrected, but note
that none of the correlations remained significant after Bonferroni
correction for multiple comparisons (with a corrected alpha of 0.05/
16 = 0.003). The percentage of “void”, “intralist”, “confusion” and
“intrusion” type of errors was analyzed with independent one-way
ANOVAs. For all analyses, we applied a significance threshold of
p = 0.05.
3. Results
3.1. Sleep accelerates memory re-stabilization
Initial learning performance of List 1 and List 2 was similar between
groups (Fig. 2, insets; List 1: Fgroup(3,56) = 2.13, p = 0.11; List 2:
Fgroup(3,49) = 0.74, p = 0.53). All groups showed successful learning
across the learning trials of both List 1 (Frun(8,448) = 78.70,
p < 0.001; Fgroupxrun(24,448) = 0.95, p = 0.54) and List 2
(Frun(8,392) = 90.35, p < 0.001; Fgroupxrun(24,392) = 0.77, p = 0.76).
However, striking differences between groups developed at testing on
day 3 when examining the RIF effect on List 2 (Fig. 2A, ANOVA, F
(3,49) = 5.20, p = 0.003). As expected, the “Reactivation-90minwake-
Interference” group showed a similar performance than the
“CTL-L2” group that was only trained and tested on List 2 (List 2
errors: 26.0 ± 5.1% vs. 12.7 ± 0.5%, respectively, p = 0.105), evi-
dencing no RIF effect, whereas the “Reactivation-10hwake-
Interference” group showed a significantly higher percentage of List 2
errors than the “CTL-L2” group, evidencing an intact RIF effect (List 2
errors: 37.9 ± 6.4% vs. 12.7 ± 0.5%, p = 0.003). Thus, the inter-
ference task presented 90 min after the reminder interfered with the re-
stabilization of the List 1 memory (evidenced by no RIF effect on List 2).
However, when the interference task was presented 10 h later, when the
reconsolidation time window was already closed, it did not impair List
1 memory re-stabilization (evidenced by an intact RIF effect).
Regarding the sleep group, we observed that the “Reactivation-
90minsleep-Interference” group showed similar performance than the
“Reactivation-10hwake-Interference” group. Participants in the
“Reactivation-90minsleep-Interference” group had a significantly
higher percentage of List 2 errors at testing than the “CTL-L2” group
(List 2 errors: 42.3 ± 5.4% vs. 12.7 ± 0.5%, p = 0.001), showing an
intact RIF effect (Fig. 2A). The RIF effect on List 2 was comparable
between the “Reactivation-90minsleep-Interference” group and the
“Reactivation-10hwake-Interference” group (p = 0.57), whereas the
“Reactivation-90minsleep-Interference” group differed significantly
from the “Reactivation-90minwake-Interference” group (p = 0.04).
However, the difference between the “Reactivation-10hwake-Inter-
ference” group and the “Reactivation-90minwake-Interference” group
did not reach significance (p = 0.12).
When examining the percentage of errors at testing of the memory
of List 1, all three experimental groups showed higher List 1 errors than
the “CTL-L1” group that was only trained and tested on List 1 and re-
ceived the reminder on day 2 (Fig. 2B, ANOVA F(3,57) = 5.44,
p = 0.002, LSD comparisons all p < 0.032), with no differences be-
tween the single experimental groups (all p > 0.19). These findings
suggest that in addition to the intact or absent RIF effect on List 2,
retrieval of List 1 was subject to acute simultaneous retrieval inter-
ferences (Forcato et al., 2007).
Furthermore, when analyzing the percentage of type of errors, the
main differences we observed stem from “void” errors (and fewer “in-
tralist” and “confusion” errors), but almost no “intrusions”. There was a
significant difference in the percentage of List 2 “void” type errors
between groups (ANOVA F(3,49) = 4.57, p = 0.005). There were no
significant differences between groups for “intralist”, “confusion”, or
“intrusion” type errors for List 2 (all Fs(3,49) > 0.53, p > 0.14). For
M.D. Moyano, et al. Neurobiology of Learning and Memory 162 (2019) 1–8

Fig. 2. Sleep accelerates memory re-stabilization. A. Retrieval-Induced Forgetting (RIF) effect. The mean percentage of errors ( ± SEM) at List 2 testing on day 3 is
presented. The “Reactivation-10hwake-Interference” and “Reactivation-90minsleep-Interference” groups show intact RIF effects at List 2 recall, indicating stable List
1 memory. The “Reactivation-90minwake-Interference” group does not show a RIF effect, indicating labile List 1 memory. B. Retrieval of List 1 Memory. The mean
percentage of errors ( ± SEM) at List 1 testing on day 3 is presented. All groups show simultaneous retrieval interference effects, i.e. reduced recall of List 1 memory.
Significances indicate planned post-hoc comparisons between experimental groups and the respective control groups (CTL-L2 in A, CTL-L1 in B). Insets show learning
performance during the training session for List 2 (in A) and List 1 (in B), respectively.*, p < 0.05; **, p < 0.01; ***, p < 0.001; NS, p > 0.10.

List 1 testing, there was a significant difference between groups for the
“confusion” type errors (F(3,57) = 2.76, p = 0.05). There were no
significant differences between groups for any of the other types of
errors (all Fs(3,57) > 1.55, p > 0.08).
3.2. Slow wave activity is involved in memory re-stabilization
with List 1 errors). Participants in the “Reactivation-90minsleep-Inter-
ference” group slept on average for 81.6 ± 5.7 min, with
4.3 ± 1.2 min in wakefulness, 17.1 ± 3.5 min in Stage 1,
42.0 ± 4.5 min in Stage 2, 7.1 ± 2.5 min in Stage 3, 0.2 ± 0.1 min in
Stage 4 and 11.0 ± 2.4 min in REM sleep. Mean power density in S2,
SWS and REM sleep are shown in Table 1.

3.3. Control measures

When examining the correlation between power density during
sleep in the “Reactivation-90minsleep-Interference” group and the RIF
There were no significant differences for the Stanford Sleepiness
effect, i.e. List 2 errors at testing, we observed a significant positive
Scale between groups during any of the sessions (Table 2, all p ≥ 0.12).
correlation with power density of slow oscillations (0.5–1 Hz) during S2
as well as during SWS (Fig. 3, r = 0.65, p = 0.02; r = 0.66, p = 0.02,
respectively). We also found a positive correlation between the RIF 4. Discussion
effect and power density of delta waves (1–4 Hz) during S2 (r = 0.67,
p = 0.02). Thus, the higher the power density in slow oscillations and Here we show that sleep plays a fundamental role in the re-stabi-
delta during the 90 min nap on day 2, the stronger the RIF effect on day lization of consolidated memories, shortening the reconsolidation time
3, an indirect measure of List 1 memory stabilization. No other corre- window triggered by the presentation of a reminder. First of all, we
lations reached significance (all p > 0.06, including any correlations replicated previous results in the wake state, showing that a reminder

Fig. 3. Association between slow oscillations and the Retrieval-Induced Forgetting (RIF) effect. Graphs display correlations between the RIF effect (i.e. percentage of
List 2 errors at testing) and slow oscillation power density (0.5–1 Hz) in A. stage 2 sleep (S2) and B. slow wave sleep (SWS).

6
M.D. Moyano, et al. Neurobiology of Learning and Memory 162 (2019) 1–8

Table 1 conditions. On the contrary, the RIF effect was absent after 90 min of
Power density (µV2/Hz). wakefulness, suggesting that List 1 memory was not yet re-stabilized at
S2 this time and sensitive to disruption by interference learning. Thus, the
specific pattern of intact and absent RIF effects in the different ex-
Slow Oscillations (0.5–1 Hz) 94.72 ± 14.61 perimental groups allows drawing conclusions about the integrity of the
Delta (1–4 Hz) 27.07 ± 2.86
underlying target memory traces. Specifically, we conclude that after
Slow Spindle (9–12 Hz) 1.64 ± 0.19
Fast Spindle (12–15 Hz) 1.89 ± 0.34
reactivation, the target memory returns to a labile state and requires
Alpha (8–13 Hz) 1.79 ± 0.18 time to re-stabilize, which takes up to 10 h during wakefulness. How-
SWS
ever, intervening sleep speeds up the re-stabilization process, with the
Slow Oscillations (0.5–1 Hz) 253.40 ± 43.98 target memory already being re-stabilized after 90 min. It can be
Delta (1–4 Hz) 57.37 ± 9.33 speculated that the reconsolidation of declarative memories during
Slow Spindle (9–12 Hz) 1.26 ± 0.22 sleep relies on spontaneous ‘replay’ of the recently reactivated memory
Fast Spindle (12–15 Hz) 1.39 ± 0.27
traces, similar to the replay of newly acquired memories during initial
Alpha (8–13 Hz) 1.65 ± 0.31
sleep-dependent consolidation, promoting memory re-stabilization and
REM
re-distribution (Diekelmann & Born, 2010; Eschenko, Ramadan, Mölle,
Theta (4–8 Hz) 3.07 ± 0.70
Beta (15–30 Hz) 0.70 ± 0.36
Born, & Sara, 2008; Rasch & Born, 2013).
In addition to the RIF effect, we also evaluated memory perfor-
Mean power density ± SEM during the 90 min nap in the mance of List 1. This is important to differentiate actual storage im-
“Reactivation-90minsleep-Interference” group. S2 (stage 2 sleep), SWS pairments of the target memory from simple retrieval failures. When
(slow wave sleep, combined stage 3 and stage 4 sleep), REM (rapid eye testing two related memories at the same time in a retrieval test,
movement sleep). memory performance can be reduced due to simultaneous retrieval
interferences. Considering that memories are stored in complex asso-
Table 2 ciative networks, one memory cue can activate different traces si-
Stanford sleepiness scale. multaneously (e.g. List 1 and List 2). In this case, memories can com-
Day 1 Day 2 Day 2 (List 2 Day 3 (Testing) pete at retrieval leading to impaired recall performance despite the
(Training (Reminder) Interference stored memory being intact (Forcato et al., 2007). For example, when
List 1) task) retrieving List 1 memory, items of List 2 can compete for retrieval re-
CTL-L1 2.22 ± 0.43 2.87 ± 0.51 – 2.78 ± 0.46 sulting in reduced List 1 recall at testing. This is indeed what we ob-
R-90minwake-Int 1.93 ± 0.21 2.40 ± 0.35 2.20 ± 0.22 2.13 ± 0.31 served in all experimental groups in the present study. Specifically, we
R-10hwake-Int 1.64 ± 0.20 2.07 ± 0.22 1.86 ± 0.21 1.79 ± 0.19 found reduced recall (i.e. more errors) in retrieval of List 1 in all re-
R-90minsleep-Int 2.62 ± 0.38 2.92 ± 0.33 2.46 ± 0.24 2.15 ± 0.42 activation groups, i.e. in the groups that were required to recall two sets
CTL-L2 – – 2.64 ± 0.43 2.73 ± 0.38
of memories simultaneously at testing (List 1 and List 2), when com-
p-value p = 0.12 p = 0.25 p = 0.23 p = 0.25
pared to the control group that recalled only one set of memories (CTL-
Mean ratings in the Stanford Sleepiness Scale ± SEM and p-values for one- L1). These findings show that simultaneous retrieval interferences were
way ANOVAs separately for each session (i.e. Training List 1, Reminder, List 2 present at testing of List 1, yet these acute effects were independent of
Interference task, Testing). the long-term integrity of the underlying memory trace, as evidenced
by differential outcomes for the experimental groups in the RIF effect.
presented 24 h after learning, labilized the consolidated List 1 memory, Regarding the type of errors, we observed for both lists that the
and a second learning task (List 2, interference task) impaired List 1 re- majority of errors stem from “void” errors, i.e. no answer given, with
stabilization when presented inside the time window of reconsolidation only a few “intralist” and “confusion” errors and almost no “intrusions”.
(i.e. 90 min after reactivation in the “Reactivation-90minwake- From previous research, it might be expected that reconsolidation ef-
Interference” group). No impairment was observed when the inter- fects mainly result from intrusion types of errors. Hupbach et al. (2007)
ference task was presented outside the reconsolidation time window used a list of objects task and found that reconsolidation of List 1 was
10 h after reactivation (“Reactivation-10hwake-Interference” group) revealed by an increment of intrusions from List 2. These divergent
(Forcato et al., 2007). This result indicates that 90 min after reactiva- results with regard to intrusions could be due to differences in the
tion, the memory was still in a labile state, while 10 h later it was al- learning paradigm and design. While we applied learning lists of pairs
ready re-stabilized and insensitive to the action of amnesic agents. of nonsense syllables, Hupbach and colleagues used concrete objects in
However, we further show that if we allowed participants to sleep after a basket. Moreover, we assessed retrieval in a cued recall test, whereas
reactivation, the interference learning of List 2 did not impair List 1 re- Hupbach and colleagues used free recall testing. Future studies will
stabilization 90 min after reactivation (in the “Reactivation-90min- have to test these factors and their role in memory reconsolidation more
sleep-Interference” group). This finding suggests that the intervening systematically.
sleep period accelerated the re-stabilization process of the labilized List When examining potential sleep features that may be involved in
1 memory such that the reconsolidation window was already closed the observed acceleration of memory re-stabilization processes during
after 90 min, which is normally only observed after 10 h of wakefulness. sleep, we observed an association with slow wave activity (slow oscil-
We used the Retrieval-Induced Forgetting effect (RIF) as a tool to lations and delta activity) during SWS and S2. We found significant
demonstrate memory reconsolidation impairment of List 1. According correlations between the RIF effect and power density of slow oscilla-
to this effect, the retrieval of a target memory can temporarily block tions (0.5–1 Hz) in S2 as well as in SWS in the 90 min sleep group. The
subsequent retrieval of a related memory (Anderson et al., 1994; correlation between power density of delta waves (1–4 Hz) and the RIF
Forcato et al., 2007). Thus, if List 1 memory storage is intact, its re- effect was significant for S2 but not for SWS. Thus, higher power den-
trieval can interfere with subsequent List 2 memory retrieval (intact sity in slow wave activity (slow oscillations and delta waves) during
RIF). On the other hand, if List 1 memory storage is impaired, its re- NREM sleep may be associated with a stronger RIF effect as an indicator
trieval does not interfere with List 2 retrieval (absent RIF, Forcato et al., of greater re-stabilization of the List 1 memory. These findings are in
2007). Hence, we evaluated impairment of List 1 memory by analyzing keeping with previous reports of an involvement of NREM sleep and
the presence of the RIF effect on List 2. As expected, we observed an slow wave activity in declarative memory consolidation and re-
intact RIF effect after 10 h of wakefulness as well as after 90 min of consolidation (Alger et al., 2012; Diekelmann et al., 2012; Klinzing
sleep, indicating successful re-stabilization of List 1 memory in both et al., 2016; Macdonald & Cote, 2016; Wilhelm et al., 2011). However,

7
M.D. Moyano, et al.
the observed correlations were no longer significant after correction for
multiple comparisons, and therefore, should be interpreted with cau-
tion.
Recently, Klinzing et al. (2016) examined whether sleep supports
memory reconsolidation of remote memories triggered by a reminder
and how this process differs from sleep-dependent consolidation. Par-
ticipants were trained on a visuo-spatial task in the presence of an odor.
The “reconsolidation” group learned the task on day 1, and on day 2,
they received a reminder (odor) and a mock recall test before a 40-
minute sleep or wake period. The “remote consolidation” group fol-
lowed the same procedure but did not receive reactivation on day 2.
Participants in the “recent consolidation” group skipped the procedure
on day 1 and learned the task immediately before the sleep or wake
period. When the memory was tested shortly after the sleep or wake
interval, the results showed that the 40-min sleep period facilitated the
reconsolidation of reactivated memories, whereas the consolidation of
non-reactivated remote memories was less affected. Recently encoded
memories did not benefit from sleep at all. Klinzing et al. (2016) also
found that the time spent in SWS was associated with subsequent
memory performance selectively in the “reconsolidation” group. Our
results confirm and extend these findings in suggesting that even after
longer re-stabilization periods of 24 h (compared to only 40 min in
Klinzing et al., 2016), sleep and particularly slow wave activity may be
involved in the re-stabilization of reactivated memories.
In another study on the role of sleep in memory reconsolidation and
future relevance, Macdonald and Cote (2016) trained participants on
two blocks of syllable pairs with the instruction that only one of the
blocks would be tested at retrieval. 24 h later, one group of subjects
received reactivation with reminders that labilized the memories (“la-
bile” group), whereas another group of subjects received reactivation
with reminders that did not labilize the memories (“stable” group).
Although the authors did not observe any differences between groups in
memory performance at testing, they found that slow wave power
(0.5–4 Hz) during S2 as well as theta power (4–8 Hz) during REM sleep
was associated with the retention of future-relevant memories selec-
tively in the “labile” group. The associations with slow wave power are
well in line with the correlations we found for power density in the
same frequency bands (0.5–1 Hz and 1–4 Hz) in the present study, while
we did not observe any correlations with power density in REM sleep.
These differences could be due to differences in the experimental de-
sign. Most importantly, in the present study we applied a second
learning task to interfere with the re-stabilization process after re-
activation, whereas Macdonald and Cote did not use any amnesic
agents but simply assessed memory re-stabilization without any inter-
ventions. Despite slightly different outcomes, these findings together
lend further support to the view that slow wave activity is involved in
the reprocessing and re-stabilization of reactivated memories during
sleep.
Financial disclosures
The funders had no role in study design, data collection and ana-
lysis, decision to publish, or preparation of the manuscript. The authors
have declared that no competing interests exist.
Acknowledgment
We thank Camila Jorge for her help with data collection. This work
was supported by AGENCIA (PICT 2014 N° 2143 and PICT 2016 N°
0229 to CF).
Neurobiology of Learning and Memory 162 (2019) 1–8
References
Alger, S. E., Lau, H., & Fishbein, W. (2012). Slow wave sleep during a daytime nap is
necessary for protection from subsequent interference and long-term retention.
Neurobiology of Learning and Memory, 98(2), 188–196. https://doi.org/10.1016/j.
nlm.2012.06.003.
Anderson, M. C., Bjork, R. A., & Bjork, E. L. (1994). Remembering can cause forgetting:
Retrieval dynamics in long-term memory. Journal of Experimental Psychology.
Learning, Memory, and Cognition, 20(5), 1063–1087.
Diekelmann, S., Biggel, S., Rasch, B., & Born, J. (2012). Offline consolidation of memory
varies with time in slow wave sleep and can be accelerated by cuing memory re-
activations. Neurobiology of Learning and Memory, 98(2), 103–111. https://doi.org/
10.1016/j.nlm.2012.07.002.
Diekelmann, S., & Born, J. (2010). The memory function of sleep. Nature Reviews
Neuroscience, 11(2), 114–126. https://doi.org/10.1038/nrn2762.
Diekelmann, S., Büchel, C., Born, J., & Rasch, B. (2011). Labile or stable: Opposing
consequences for memory when reactivated during waking and sleep. Nature
Neuroscience, 14(3), 381–386. https://doi.org/10.1038/nn.2744.
Dudai, Y., Karni, A., & Born, J. (2015). The consolidation and transformation of memory.
Neuron, 88, 20–32. https://doi.org/10.1016/j.neuron.2015.09.004.
Eschenko, O., Ramadan, W., Mölle, M., Born, J., & Sara, S. J. (2008). Sustained increase in
hippocampal sharp-wave ripple activity during slow-wave sleep after learning.
Learning & Memory, 15(4), 222–228. https://doi.org/10.1101/lm.726008.
Forcato, C., Argibay, P. F., Pedreira, M. E., & Maldonado, H. (2009). Human re-
consolidation does not always occur when a memory is retrieved: The relevance of
the reminder structure. Neurobiology of Learning and Memory, 91(1), 50–57. https://
doi.org/10.1016/j.nlm.2008.09.011.
Forcato, C., Burgos, V. L., Argibay, P. F., Molina, V. A., Pedreira, M. E., & Maldonado, H.
(2007). Reconsolidation of declarative memory in humans. Learning & Memory, 14(4),
295–303. https://doi.org/10.1101/lm.486107.
Forcato, C., Fernández, R. S., & Pedreira, M. E. (2013). The role and dynamic of
strengthening in the reconsolidation process in a human declarative memory: What
decides the fate of recent and older memories? PLoS ONE, 8(4), e61688. https://doi.
org/10.1371/journal.pone.0061688.
Forcato, C., Fernández, R. S., & Pedreira, M. E. (2014). Strengthening a consolidated
memory: The key role of the reconsolidation process. Journal of Physiology - Paris,
108, 323–333. https://doi.org/10.1016/j.jphysparis.2014.09.001.
Hardwicke, T. E., Taqi, M., & Shanks, D. R. (2016). Postretrieval new learning does not
reliably induce human memory updating via reconsolidation. Proceedings of the
National Academy of Sciences of the United States of America, 113(19), 5206–5211.
https://doi.org/10.1073/pnas.1601440113.
Herszage, J., & Censor, N. (2017). Memory reactivation enables long-term prevention of
interference. Current Biology, 27(10), 1529–1534.e2. https://doi.org/10.1016/j.cub.
2017.04.025.
Hupbach, A., Gomez, R., Hardt, O., & Nadel, L. (2007). Reconsolidation of episodic
memories: A subtle reminder triggers integration of new information. Learning &
Memory, 14(1–2), 47–53. https://doi.org/10.1101/lm.365707.
Kindt, M., & Soeter, M. (2018). Pharmacologically induced amnesia for learned fear is
time and sleep dependent. Nature Communications, 9, 1316. https://doi.org/10.1038/
s41467-018-03659-1.
Klinzing, J. G., Rasch, B., Born, J., & Diekelmann, S. (2016). Sleep's role in the re-
consolidation of declarative memories. Neurobiology of Learning and Memory, 136,
166–173. https://doi.org/10.1016/j.nlm.2016.10.004.
Lahl, O., Wispel, C., Willigens, B., & Pietrowsky, R. (2008). An ultra short episode of sleep
is sufficient to promote declarative memory performance. Journal of Sleep Research,
17(1), 3–10. https://doi.org/10.1111/j.1365-2869.2008.00622.x.
Macdonald, K. J., & Cote, K. A. (2016). Sleep physiology predicts memory retention after
reactivation. Journal of Sleep Research, 25, 655–663. https://doi.org/10.1111/jsr.
12423.
MacLeod, M. D., & Macrae, C. N. (2001). Gone but not forgotten: The transient nature of
retrieval-induced forgetting. Psychological Science, 12(2), 148–152. https://doi.org/
10.1111/1467-9280.00325.
Nader, K., Schafe, G. E., & Le Doux, J. E. (2000). Fear memories require protein synthesis
in the amygdala for reconsolidation after retrieval. Nature, 406(6797), 722–726.
https://doi.org/10.1038/35021052.
Rasch, B., & Born, J. (2013). About sleep’s role in memory. Physiological Reviews, 93(2),
681–766. https://doi.org/10.1152/physrev.00032.2012.
Rechtschaffen, A., & Kales, A. (1968). A manual of standardized terminology, technique and
scoring system for sleep stages of human sleep. Los Angeles, CA, USA: Brain Information
Service, Brain Information Institute, UCLA.
Stickgold, R. (2009). How do I remember? Let me count the ways. Sleep Medicine Reviews,
13(5), 305–308. https://doi.org/10.1016/j.smrv.2009.05.004.
Walker, M. P., Brakefield, T., Hobson, J. A., & Stickgold, R. (2003). Dissociable stages of
human memory consolidation and reconsolidation. Nature, 425(6958), 616–620.
https://doi.org/10.1038/nature01930.
Wilhelm, I., Diekelmann, S., Molzow, I., Ayoub, A., Mölle, M., & Born, J. (2011). Sleep
selectively enhances memory expected to be of future relevance. Journal of
Neuroscience, 31(5), 1563–1569. https://doi.org/10.1523/JNEUROSCI.3575-10.
2011.