doi:10.1093/brain/awh647
Brain (2006), 129, 108–114
Rhinal–hippocampal connectivity determines
memory formation during sleep
Juergen Fell,1 Guillén Fernández,1,3,4 Martin T. Lutz,1 Edgar Kockelmann,1 Wieland Burr,1
Carlo Schaller,2 Christian E. Elger1 and Christoph Helmstaedter1
Departments of 1Epileptology and 2Neurosurgery, University of Bonn, Bonn, Germany, 3F.C. Donders Center for
Cognitive Neuroimaging and 4Department of Neurology, Radboud University, Nijmegen, The Netherlands
Correspondence to: Dr Juergen Fell, University of Bonn, Department of Epileptology, Sigmund-Freud-Strasse 25,
D-53105 Bonn, Germany
E-mail: [email protected]
Compared with waking state attention, volition and semantic processing play a minor role during sleep. Thus,
investigating declarative memory formation during sleep may allow us to isolate mnemonic core processes.
The most feasible approach to memory formation during sleep is the analysis of dream memories. Lesion
and imaging studies have demonstrated that encoding of declarative memories, i.e. consciously accessible
events and facts, depends on operations within the rhinal cortex and the hippocampus, two substructures
of the medial temporal lobe. Successful memory formation is accompanied by a transient rhinal–hippocampal
interaction. Consequently, the ability to memorize dreams may be related to mediotemporal connectivity.
Therefore, we recorded EEG during sleep from rhinal and hippocampal depth electrodes implanted in 12
epilepsy patients (eight women, mean age 41.1 6 6.4 years). They were awakened during rapid eye movement
sleep (REM) and asked to recall their dream. Via coherence analyses we show that rhinal–hippocampal connectivity values are approximately twice as large for patients with good dream recall versus those patients with
poor recall. This suggests that rhinal–hippocampal connectivity is a key factor in determining declarative
memory formation.
Keywords: EEG; memory formation; sleep; epilepsy; medial temporal lobe
Abbreviations: MTL = medial temporal lobe; REM = rapid eye movement sleep; SWS = slow wave sleep
Received April 5, 2005. Revised August 24, 2005. Accepted August 30, 2005. Advance Access publication October 26, 2005
Introduction
Lesion studies have firmly linked the medial temporal lobe
(MTL) to the declarative memory system, which enables us to
consciously remember events and facts (Eichenbaum, 2000;
Squire et al., 2004). Functional investigations have shown that
MTL activations and transient rhinal–hippocampal interactions are associated with successful declarative memory
formation (Brewer et al., 1998; Wagner et al., 1998;
Fernández et al., 1999; Fell et al., 2001, 2003b; Otten et al.,
2001). However, these operations may be masked by less
controllable factors, such as attention or semantic processing,
which indirectly support declarative memory formation
(Brewer et al., 1998; Wagner et al., 1998; Otten et al.,
2001). Such support operations presumably play a minor
role during sleep (Hobson et al., 2000). Thus, investigating
declarative memory formation during sleep, as assessed
by dream recall, may allow us to isolate a core mnemonic
process.
#
The ability to recall dreams varies greatly among subjects
(e.g. Goodenough, 1991). The frequency and quality of dream
recall are related to factors like different personality types,
sleep habits and one’s attitude towards dreams (Schredl
and Montasser, 1996). However, in a recent extensive study
with 444 participants, only 8.4% of the interindividual variance of dream recall frequency could be explained by those
factors (Schredl et al., 2003). Although dreams occurring
during rapid eye movement sleep (REM) are dominated by
visual imagery (Snyder, 1970; Hobson et al., 2000), no significant correlation was found between daytime visual memory
performance and dream recall frequency. Thus, dream
recall variations are probably largely determined by memory
formation during sleep, rather than by daytime memory
capability.
Are there any neurophysiological factors that predict dream
recall? The formation of declarative memories is accompanied
The Author (2005). Published by Oxford University Press on behalf of the Guarantors of Brain. All rights reserved. For Permissions, please email: [email protected]
Rhinal–hippocampal coherence in memory formation
by the phase synchronization of rhinal–hippocampal EEG
activity in the gamma range and an increased EEG coherence
in the low-frequency range (Fell et al., 2001; 2003b). It has
been suggested that declarative memory is disfacilitated
during sleep causing dreams to be more difficult to
remember than those experiences made during the waking
state (Hobson et al., 2000; Hobson and Pace-Schott, 2002).
When comparing the sleep state with the waking state, indeed
a reduced rhinal–hippocampal and intrahippocampal EEG
coherence has been observed, particularly within the
gamma range (Fell et al., 2003a). Therefore, we hypothesized
that variations in mediotemporal connectivity may account
for the interindividual differences in one’s ability to build
declarative memories during sleep.
We recorded the sleep EEG from 12 patients with pharmacoresistant MTL epilepsies, subsequent to one adaptation
night. Multicontact depth electrodes had been implanted
stereotactically along the longitudinal axis of each MTL
(Van Roost et al., 1998) during the presurgical evaluation
because the seizure onset zone could not be precisely
determined with non-invasive investigations. Each subject
had at least one electrode contact which could be localized
unequivocally within the anterior parahippocampal gyrus.
Another electrode was localized within the anterior
hippocampus (anterior third), and a third was localized
within the posterior hippocampus (posterior third). To
reduce the possibility of introducing uncontrolled variables
brought about by the epileptic process, we analysed only those
EEG recordings that were taken from the MTL contralateral
to the zone of seizure origin (Puce et al., 1989; Grunwald
et al., 1995).
During the night, depth and scalp EEG, electrooculogram
and electromyogram were recorded and monitored online
(see Methods). The patients were awakened upon the onset
of the second and the following REM phases, which were
positively identified using Rechtschaffen/Kales criteria
(Rechtschaffen and Kales, 1968). The latency period of the
awakenings with respect to REM onset was 5 min. Immediately upon awakening, the patients were asked whether they
remembered having had a dream and whether they could
recall any specific content.
Brain (2006), 129, 108–114
109
EEG recording and sleep staging
Sleep EEG was recorded subsequent to one adaptation night.
Arrangement of an additional adaptation night to experience the
modalities of awakening and reporting was not possible due to
the constraints given by the clinical schedule patients had to pass.
Thus, the estimates of the frequency of REM-dreams were expected
to be less accurate than those possible in multiple-night studies,
where subjects also preliminary experience the modality of awakening and reporting dream contents. Informed consent had been
obtained from all patients (eight women, mean age: 41.1 6
6.4 years) and the local institutional ethics committee approved
the study. Electrode contact placement was ascertained by examining
MRIs acquired in the sagittal, axial and coronal planes, which were
adjusted to the longitudinal axis of the hippocampus. Electrode
contacts were mapped by transferring their positions from MRIs
to standardized anatomical drawings (Jackson and Duncan, 1996;
Van Roost et al., 1998). Each patient was diagnosed with a unilateral
seizure origin within one MTL (six patients—right; six
patients—left). EEG data obtained from the non-pathological
MTL in patients with a unilateral seizure origin have been shown
to be qualitatively unchanged when compared with the invasive
EEGs recorded in monkeys (Paller et al., 1992).
Depth electroencephalograms were referenced to linked mastoids,
bandpass-filtered [0.01 Hz (6 dB/octave) to 70 Hz (12 dB/octave)]
and recorded with a sampling rate of 200 Hz. Scalp EEG was recorded
from positions C3, C4 and O1 (10–20 system). Electroocular activity
was registered at the outer canthi of both eyes and submental
electromyographic activity was acquired with electrodes attached
to the chin. Interelectrode impedances were all below 5 kV. Two
experienced raters (J.F. and W.B.) performed a visual analysis based
on Rechtschaffen/Kales criteria (Rechtschaffen and Kales, 1968). All
EEG epochs were inspected twice for movement artefacts and epileptiform activity. Artefact segments were discarded irrespective
of the duration of artefacts. For instance, epochs containing an epileptiform spike were rejected because the spikes would contribute
spurious power to the higher frequency ranges. Moreover, recordings
from the reference electrodes were inspected for artefacts or high
voltage EEG contamination and respective epochs were discarded. In
total, 41.8% of all EEG epochs were excluded from further analysis.
Power and coherence values were calculated for the remaining 11 110
epochs, of which 2477 corresponded to the waking state; 632 to Stage
1; 1651 to Stage 2; 1410 to slow wave sleep (SWS) (Stages 3 and 4)
and 963 to REM sleep.
Power and coherence analysis
Methods
Patients
To ensure sufficient comprehension of instructions and general verbal ability, only patients with a vocabulary IQ of at least 85 (i.e. falling
in the unimpaired range) were included. Before the recording night,
patients had been given a self-generated questionnaire addressing
general sleeping and dreaming behaviour. Statistical analyses of the
questionnaire results indicated that the groups with good versus poor
dream memory during the laboratory night did not differ with regard
to overall quality of sleep, frequency of dream recall or occurrence of
nightmares (P = 0.84, P = 0.79, respectively P = 0.29, Chi-Square).
However, the group with good (laboratory) dream memory
exhibited significantly shorter sleep durations (6.63 versus 8.25 h,
P = 0.047).
We partitioned every 20 s EEG epoch into 16 non-overlapping
subsegments, each of which lasted 1.25 s. Based on the calculation
of fast Fourier transforms for these subsegments (cosine tapering/
windowing), we determined the average power and spectral coherence for each epoch. This was done for the following frequency
bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta1
(12–16 Hz), beta2 (16–20 Hz), gamma1 (20–28 Hz), gamma2
(28–36 Hz) and gamma3 (36–44 Hz). Spectral coherence quantifies
the frequency specific degree of linear relationship between two
signals (Challis and Kitney, 1991). It is given by the average cross
power spectrum of two signals divided by the average powers of
both signals. The values of spectral coherence range between zero
and one. Prior to statistical analysis, the coherence values were
Fisher-z-transformed. For each patient power and coherence values
were averaged for each of the combinations of sleep stage
110
Brain (2006), 129, 108–114
J. Fell et al.
Table 1 Analysis of dream reports for the groups with good versus poor dream memory
Group
Good dream recall
Poor dream recall
Patient No.
Number of
awakenings
Dream-like experiences
Thought-like experiences
No reports
No.
No.
TU
DFS
BC
No.
TU
DFS
BC
2
1
4
2
2
2
1
1
2
2
2
2
(50%)
(100%)
(50%)
(100%)
(100%)
(100%)
4
6
1;1
3;5
2;9
2;3
7
7
5;5
7;8
6;7
7;7
0
0
0;0
1;3
2;1
0;0
1
0
1
0
0
0
0
–
0
–
–
–
3
–
3
–
–
–
0
–
0
–
–
–
Group
n=6
1
2
3
4
5
6
13
10
0
0
0
0
0
2
(0%)
(0%)
(0%)
(0%)
(0%)
(40%)
M = 6.6
SD = 1.0
–
–
–
–
–
7;7
M = 0.7
SD = 1.1
–
–
–
–
–
0;0
2
4
2
1
3
4
5
M = 3.6
SD = 2.5
–
–
–
–
–
3;2
Group
n=6
19
2
1
2
3
4
5
6
1
1
0
0
1
0
3
(50%)
(0%)
(25%)
(0%)
(0%)
(0%)
0
0
1
0
0
0
(0%)
(0%)
(25%)
(0%)
(0%)
(0%)
1
(25%)
(50%)
(0%)
(0%)
(25%)
(0%)
0
0
–
–
0
–
3
3
–
–
3
–
0
0
–
–
0
–
3
1
1
3
3
3
(75%)
(50%)
(100%)
(100%)
(75%)
(60%)
14
TU, temporal units (a temporal unit is defined as ‘whatever activities could have occurred synchronously and were not described as having
occurred successively’) (Foulkes and Schmidt, 1983); DFS, Dream-like Fantasy Scale; 1, Mind was blank; 2, Experienced something but forgot
what it was; 3, Conceptual (no imagery), everyday content; 4, Conceptual (no imagery), bizarre content; 5, Non-hallucinatory (sensory)
imagery, everyday content; 6, Non-hallucinatory (sensory) imagery, bizarre content; 7, Hallucinatory (sensory) imagery, everyday content;
8, Hallucinatory (sensory) imagery, bizarre content; BC, Bizarre elements (number). Generally, objects/actions/persons etc. that do not
exist or are impossible in waking-life reality are defined as bizarre (Schredl and Erlacher, 2003). No., Number; M, mean; SD, standard
deviation.
[Awake, Stage 1, Stage 2, SWS (Stages 3 and 4), REM], frequency
band and electrode position (power) or electrode pair (coherence).
REM awakenings and dream reports
Sleep EEG was monitored online by an experienced rater (E.K.).
The patients were immediately awakened during REM phases
(average number of awakenings: 2.8 6 1.4), and asked whether
they remembered having had a dream. If so, they were instructed
to give a detailed narrative description. In order to explore their
dream memory in further detail, we also conducted a structured
interview which included 15 questions that addressed specific
dream content. Only reports with characteristics like story-like
development, self-participation and presence of perceptual features
indicating a vivid and/or bizarre episode were counted as dreams.
More thought-like mentation processes, which often occur in early
sleep and NREM-stages, but also during REM sleep, were not rated as
dreams (Cicogna et al., 2000). Key qualitative and quantitative
content characteristics of the dream reports are given in Table 1.
The variables dream length (in temporal units, Foulkes and Schmidt,
1983), score in the ‘Dream-like Fantasy Scale’ DFS (Foulkes, 1970)
and bizarreness of dream content (Schredl and Erlacher, 2003) were
quantified (see notes to Table 1 for more detailed information). The
three analyses of dream reports were carried out by one scorer, who
was blind as to the aim and design of the study. These measures
indicate that subjects in the group with good dream memory actually
remembered dream-like experiences, and not just thought-like
mentations.
We divided the patients into two equally-sized groups according
to their dream recollection ability: six subjects remembered that they
had a dream and could recall specific dream content at least 50% of
the time (average recall rate: 79.2 6 24.6%); five subjects could not
recall any dreams and one could remember his dreams in two out of
five cases (average recall rate: 6.7 6 16.3%). Five subjects of the first
group (good dream memory) always delivered a dream or mentation
report and one subject gave a report in three out of four cases.
Subjects of the second group (poor dream memory) gave no report
in at least 50% of the cases (see Table 1).
Memory tests
Prior to electrode insertion, we assessed memory performance by
administering two list-learning tests. Verbal material memory was
quantified by administering a German adaptation of the Rey
Auditory Verbal Learning test (Helmstaedter et al., 2001). This test
requires learning and immediate free recall from a list of 15 words
in five trials, and a delayed free recall of this list after distraction
(learning a second list in one trial) with a 30 min delay. The patient
must then identify the target items from orally presented alternatives.
Learning capacity (total correct responses over five immediate freerecall tests) and performance on delayed free recall (correct responses
following distraction after 30 min) were standardized (mean = 100,
SD = 10) with respect to normative test data. The measure of interest
was the total verbal memory score given by the sum of the individual
scores divided by two. The DCS-R, a revised version of the DCS
(Diagnosticum für Cerebralschädigung; Lamberti and Weidlich,
1999) was chosen to assess visual/figural memory performance.
The DCS-R requires repetitive learning and immediate free reproduction from a set of nine abstract designs in six learning trials.
The measure of interest was the learning capacity, i.e. the number
of correctly reproduced designs at the final test. This score was also
standardized using normative data of healthy controls.
Rhinal–hippocampal coherence in memory formation
Brain (2006), 129, 108–114
111
Statistical analysis
To analyse power and coherence changes we conducted a four-way
analysis of variance (ANOVA) with BAND (eight frequency bands)
and POSITION (rhinal cortex, anterior hippocampus, posterior hippocampus) or PAIR (rhinal cortex–anterior hippocampus, rhinal
cortex–posterior hippocampus, anterior hippocampus–posterior
hippocampus) as repeated measures and dream MEMORY (good,
poor) and STAGE (Awake, Stage 1, Stage 2, SWS and REM) as
independent variables. For the coherence values we additionally calculated two-way ANOVAs with the factors MEMORY and STAGE
separately for each electrode pair and frequency band. Comparisons
of daytime memory scores between the groups with poor and good
dream memory were carried out with two-tailed t-tests.
Results
To eliminate the possibility that the amount and quality of
dream recall were determined by memory performance during the waking state and not by memory formation during
sleep, we assessed the patients’ verbal and figurative memory
by administering list-learning paradigms during the daytime
(see Methods). The group with poor dream recall and the
group with good dream recall showed no significant differences in memory performance during the waking state. No
significant differences were noted with regard to figurative
memory (83.3 6 12.0 versus 76.8 6 19.3; two-tailed t-test:
T10 = 0.70; P > 0.4) and verbal memory (90.1 6 4.4 versus
92.3 6 6.4; T10 = 0.68; P > 0.5).
We found no significant statistical differences in spectral
power (no main effect of MEMORY [F(1,50) = 0.04] or twoway or higher-order interactions of the other factors with
MEMORY) for the groups with good dream recall versus
those with poor dream recall. However, we did detect
very prominent spectral coherence differences related to
dream recall (main effect of MEMORY [F(1,50) = 27.87;
P < 10 5]). The coherence level between the rhinal cortex
and the anterior hippocampus (Fig. 1) was 2.5 times higher
for the group with good dream recall (average factor across
all frequency bands and stages: 2.46 6 0.43). Furthermore,
the coherence level between the rhinal cortex and posterior
hippocampus (Fig. 2) was about twice as high (average
factor: 1.91 6 0.41) and the intrahippocampal coherence
level (Fig. 3) was 1.5 times higher (average factor: 1.45 6
0.28) for those subjects with good dream recall (significant
MEMORY · PAIR interaction [F(2,100) = 6.77; P < 0.01;
Huynh-Feldt e = 0.910].
The interindividual differences for dream memory
depended upon the frequency band (MEMORY · BAND;
[F(7,350) = 10.81; P < 10 11; e = 0.376] and MEMORY ·
BAND · PAIR [F(14,700) = 3.21; P < 0.05; e = 0.154] interaction). Rhinal–hippocampal coherence effects were most
significant for the theta, alpha and lower beta band
[main effects of MEMORY: rhinal–anterior hippocampal,
F(1,50) 14.91; each P < 0.0005; rhinal–posterior hippocampal, F(1,50) 18.59; each P < 0.0001]. Intrahippocampal
coherence effects were most significant for the alpha, lower
Fig. 1 Localization of MTL depth electrodes. Left: the shaded
areas in the standardized drawings indicate the approximate
location of the MTL electrode contacts which were selected for
EEG analysis (Am, amygdala; Cs, collateral sulcus; Hi, hippocampus; Pg, parahippocampal gyrus; Rc, rhinal cortex). Right: two
coronal MRIs of a patient with bilateral depth electrodes in situ
(radiological convention, L, left). White arrows indicate MTL
electrode contacts within the rhinal cortex (above) and within the
hippocampus (below).
Fig. 2 Connectivity between rhinal cortex and anterior
hippocampus. Shown are the grand averages of EEG coherence
while awake and during the various sleep stages (Stage 1, Stage 2,
SWS and REM) for the group with poor dream recall (left) and
with good dream recall (right). Both the mean and standard
error of the mean (SEM) are plotted.
beta and lower gamma band [F(1,50) 8.73; each P < 0.005].
Dream recall related effects for rhinal–posterior hippocampal
coherence in the upper gamma bands [F(1,50) = 3.22] and
intrahippocampal coherence in the delta [F(1,50) = 1.71], the
middle gamma [F(1,50) = 3.44] and upper gamma band
[F(1,50) = 1.61] were not significant.
112
Brain (2006), 129, 108–114
Fig. 3 Connectivity between rhinal cortex and posterior hippocampus. Shown are the grand averages of EEG coherence while
awake and during the various sleep stages (Stage 1, Stage 2, SWS
and REM) for the group with poor dream recall (left) and with
good dream recall (right). Both the mean and standard error of
mean (SEM) are plotted.
Rhinal–hippocampal coherence values are, generally,
larger than intrahippocampal coherence values [average
factor: 1.60 6 0.42; main effect of PAIR, F(2,100) = 16.10,
P < 10 6, e = 0.910]. The coherence values decrease from the
waking state towards Stage 2, SWS and REM sleep [significant
contrast of waking state against Stage 2, SWS and REM:
F(1,54) = 4.11; P < 0.05], which has been described elsewhere
(Fell et al., 2003a). However, the effect of dream recall does
not depend upon the stage classified for the EEG segment (no
significant MEMORY · STAGE interaction [F(4,50) = 0.25]
or higher-order interactions involving the factors MEMORY
and STAGE).
Discussion
Following the idea that investigating memory formation in
the sleeping state may allow us to isolate mnemonic core
processes, the present study is the first that examined the
interrelationship between declarative memory formation during sleep and mediotemporal EEG characteristics. Our data
suggest that dream memorization is closely associated with
rhinal–hippocampal and intrahippocampal connectivity.
Admittedly, the present investigation has some methodological limitations. As dream recall technique an interview
without ad hoc scoring of the perceptual and narrative features
of the dream contents has been applied. However, later quantification of the variables dream length, dream-like fantasies
and bizarreness of the reports (Foulkes, 1970; Foulkes and
Schmidt, 1983; Schredl and Erlacher, 2003) verified that
subjects in the group with good dream memory actually
J. Fell et al.
remembered dream-like experiences. It has been demonstrated that dreams can be recalled also after awakening
from NREM sleep, in normal subjects with a rate of 50%
(Nielsen et al., 2000). Thus, an interesting question would be
to compare the mediotemporal EEG characteristics in case of
successful memorization of dreams occurring during NREM
sleep with dreams during REM sleep. Addressing this question, which has not been carried out in the present study,
could provide a reliable and complementary test for the hypothesis that successful dream memorization is associated with
increased mediotemporal connectivity.
The frequency of dream reports following awakenings during REM sleep in our patients group deviates from the values
reported by other studies. In healthy volunteers, there is a
robust finding of an average dream recall rate following REM
awakenings of 80% (e.g. Nielsen 2000). Also in a sample of
patients with temporal lobe epilepsies successful dream recall
was found in >80% of awakenings from REM sleep (Cipolli
et al., 2004). But there are important differences in the clinical
characteristics of both groups. Our preoperative sample
consisted of a homogeneous group of lesional patients with
medically intractable MTL epilepsies. In these patients
pathology is largely confined to mediotemporal structures,
including severe morphological damage such as hippocampal
sclerosis or atrophy. Hence, brain structures that are known
to be important for declarative memory formation are
heavily affected (Eichenbaum, 2000; Squire et al., 2004). In
contrast, the patient group investigated by Cipolli et al. (2004)
represented temporal lobe epilepsies, where no lesions were
detectable at a CT scan. Thus, the reduced frequency of dream
recall in our patients group may originate from their specific
clinical characteristics. Furthermore, a methodological bias
resulting from the strong clinical constraints of the examination of patients with depth electrodes cannot be ruled out.
One key limitation may be the absence of an additional
adaptation night to experience the modalities of awakening
and reporting, so that the retrieval strategy applied by our
patients could have been less adequate than that usually
applied by normal subjects in experimental studies on
dream recall. This methodological constraint may represent
another reason for the low frequency of dream recall observed
in our patient group.
Nevertheless, the present study provides evidence indicating that successful memorization of dreams is accompanied
by an enhanced rhinal–hippocampal and intrahippocampal
EEG coherence. These findings corroborate and expand upon
previous results and suggest that an enhanced connectivity,
not an EEG power increase within mediotemporal structures,
is the crucial factor supporting declarative memory formation (Fell et al., 2001, 2003b). The unambiguous separation of
power and coherence effects was accomplished by recording
intracranial EEGs. This would not have been possible with
surface EEGs due to the anatomical proximity of the inspected
areas (Bullock et al., 1995). Our results indicate that models
(Buzsáki, 1996; Fernández and Tendolkar, 2001) which
propose that building new declarative memories requires a
Rhinal–hippocampal coherence in memory formation
direct cooperation between the rhinal cortex and hippocampus apply not only to the waking state, but also to the
sleep state.
Our data do not strictly exclude a third pacemaker site
driving coherent activity in rhinal cortex and hippocampus
or in anterior and posterior hippocampus independently
from each other. For instance, the thalamus or the basal
forebrain might represent such pacemaker sites for the rhinal
cortex and the hippocampus (Witter et al., 1989). However,
the strong anatomical connection between both structures
supports the hypothesis of a direct rhinal–hippocampal
interaction underlying the coherence differences (Witter
et al., 1989; Amaral and Insausti, 1990). Also for anterior
and posterior hippocampus a common pacemaker is not
plausible, since their afferent connections are channelled
through different sections of the entorhinal, perirhinal and
post-rhinal cortices (e.g. Moser and Moser 1998).
Rhinal–hippocampal coherence values were, generally,
found to be larger than intrahippocampal coherence values.
This result emphasizes the existence of a strong anatomical
and functional interaction between the rhinal cortex and the
hippocampus (Amaral and Insausti, 1990) and supports the
hypothesis of a functional dissociation within the hippocampus (Moser and Moser, 1998). The differences in mediotemporal coherence between subjects with good versus poor
dream memory did not depend upon the stage classified for
the EEG segment. With the reservations outlined above this
finding suggests that the mediotemporal connectivity during
the waking state and non-REM stages and the connectivity
during REM sleep are equally predictive for dream recall
(Fig. 4).
The observed coherence effects cannot be explained by
systematic differences in the signal of the reference electrodes
(linked mastoids). References were inspected for artefacts and
high voltage EEG contamination and respective epochs were
discarded. Generally, surface potentials are much smaller than
intracranial potentials. Furthermore, the lack of systematic
differences in spectral power between the group with poor
versus good dream memory speaks against a reference influence. Since the differences in average rhinal–hippocampal
and intrahippocampal coherence values between the group
with good versus poor dream memory did appear across all
frequency bands, this effect is equivalent to an increased correlation between the EEG signals for the group with good
dream memory compared with the group with poor dream
memory.
Attentional top–down modulation, volitional control or
semantic processing presumably play a minor role during
sleep (Hartmann, 1966; Goodenough, 1991). Therefore,
dream memory formation is more closely related to a core
operation of declarative memory formation because, unlike
those experimental situations defined by conventional paradigms (e.g. those utilizing the subsequent memory effect), it
occurs spontaneously and unintentionally (Brewer et al.,
1998; Wagner et al., 1998; Fernández et al., 1999; Otten
et al., 2001). Thus, our approach not only addresses memory
Brain (2006), 129, 108–114
113
Fig. 4 Connectivity between anterior and posterior hippocampus.
Shown are the grand averages of EEG coherence while awake and
during the various sleep stages (Stage 1, Stage 2, SWS and REM)
for the group with poor dream recall (left) and with good dream
recall (right). Both the mean and standard error of mean (SEM)
are plotted.
formation during sleep, but may also allow us to isolate a basic
process of declarative memory formation that is not masked
by indirect support operations. With the reservations
described above the pronounced difference in rhinal–
hippocampal and intrahippocampal coherence among subjects with good dream recall versus those with poor recall
suggests that mediotemporal connectivity is a key factor in
determining successful memory formation.
The lack of significant differences in daytime memory
scores for the groups with good versus poor dream memory
weakens the possibility of a parasitic effect of retrieval on the
dream recall results. During the waking state, in situations
where memorization is intended, low mediotemporal connectivity is possibly compensated by other factors such as
semantic elaboration. Moreover, a transient increase of mediotemporal connectivity may be evoked by attention (Fell et al.,
2001). Since coherence values during the waking state, nonREM stages and REM sleep are equally predictive for dream
recall, mediotemporal connectivity may represent a factor
in declarative memory formation, which is modulated on a
long-term scale, or even a trait variable, which might have a
genetic basis (De Quervain et al., 2003; Heinz et al., 2004).
Acknowledgements
This research was supported by the Deutsche Forschungsgemeinschaft (DFG El-122-8). The authors would like to
thank Nikolai Axmacher, Peter Klaver, Klaus Lehnertz,
Karen Woermann and two anonymous referees for helpful
comments and suggestions.
114
Brain (2006), 129, 108–114
References
Amaral DG, Insausti R. Hippocampal formation. In: Paxinos G, editor. The
human nervous system. San Diego: Academic Press; 1990. p. 711–55.
Brewer JB, Zhao Z, Desmond JE, Glover GH, Gabrieli JDE. Making memories:
Brain activity that predicts how well visual experience will be remembered.
Science 1998; 281: 1185–7.
Bullock TH, McClune MC, Achimowicz JZ, Iragui-Madoz VJ, Duckrow RB,
Spencer SS. EEG coherence has structure in the millimeter domain:
subdural and hippocampal recordings from epileptic patients. Electroencephalogr Clin Neurophysiol 1995; 95: 161–77.
Buzsáki G. The hippocampo-neocortical dialogue. Cereb Cortex 1996; 6:
81–92.
Challis RE, Kitney RI. Biomedical signal processing. Part 3. The power
spectrum and coherence function. Med Biol Eng Comput 1991; 29: 225–41.
Cicogna P, Natale V, Occhionero M, Bosinelli M. Slow wave and REM sleep
mentation. Sleep Res Online 2000; 3: 67–72.
Cipolli C, Bonanni E, Maestri M, Mazzetti M, Murri L. Dream experience
during REM and NREM sleep of patients with complex partial seizures.
Brain Res Bull 2004; 63: 407–13.
De Quervain DJ-F, Henke K, Aerni A, Coluccia D, Wollmer MA, Hock C,
et al. A functional genetic variation of the 5-HT2a receptor affects human
memory. Nat Neurosci 2003; 6: 1141–2.
Eichenbaum H. A cortical-hippocampal system for declarative memory.
Nat Rev Neurosci 2000; 1: 41–50.
Fell J, Klaver P, Lehnertz K, Grunwald T, Schaller C, Elger CE, et al. Human
memory formation is accompanied by rhinal–hippocampal coupling and
decoupling. Nat Neurosci 2001; 4: 1259–64.
Fell J, Staedtgen M, Burr W, Kockelmann E, Helmstaedter C, Schaller C,
et al. Rhinal–hippocampal EEG coherence is reduced during human
sleep. Eur J Neurosci 2003a; 18: 1711–6.
Fell J, Klaver P, Elfadil H, Schaller C, Elger CE, Fernández G. Rhinal–
hippocampal theta coherence during declarative memory formation:
interaction with gamma-synchronization? Eur J Neurosci 2003b; 17:
1082–8.
Fernández G, Effern A, Grunwald T, Pezer N, Lehnertz K, Dümpelmann M,
et al. Real-time tracking of memory formation in the human rhinal cortex
and hippocampus. Science 1999; 285: 1582–5.
Fernández G, Tendolkar I. Integrated brain activity in medial temporal
and prefrontal areas predicts subsequent memory performance: human
declarative memory formation at the system level. Brain Res Bull 2001;
55: 1–9.
Foulkes D, Schmidt M. Temporal sequence and unit composition in dream
reports from different stages of sleep. Sleep 1983; 6: 265–80.
Foulkes D. The Dreamlike Fantasy Scale: a rating manual. Psychophysiology
1970; 7: 335–6.
Goodenough DR. Dream recall: history and current status of the field.
In: Ellman SJ, Antrobus JS, editors. The mind in sleep: psychology and
psychophysiology. New York: Wiley & Sons; 1991.
Grunwald T, Elger CE, Lehnertz K, Van Roost D, Heinze HJ. Alterations
of intrahippocampal cognitive potentials in temporal lobe epilepsy.
Electroencephalogr Clin Neurophysiol 1995; 95: 53–62.
Hartmann E. The psychophysiology of free will: an example of vertical
research. In: Lowenstein R, Newman L, Schur M, Solnit A, editors.
Psychoanalysis: a general psychology. New York: International University
Press; 1966. p. 521–36.
J. Fell et al.
Heinz A, Braus DF, Smolka MN, Wrase J, Puls I, Hermann D, et al.
Amygdala-prefrontal coupling depends on a genetic variation of the
serotonin transporter. Nat Neurosci 2004; 8: 20–1.
Helmstaedter C, Lendt M, Lux S. VLMT Verbaler Lern- und
Merkfähigkeitstest. Göttingen: Beltz Test GmbH; 2001.
Hobson JA, Pace-Schott EF. The cognitive neuroscience of sleep: neuronal
systems, consciousness and learning. Nat Rev Neurosci 2002; 3: 679–93.
Hobson JA, Pace-Schott EF, Stickgold R. Dreaming and the brain: toward a
cognitive neuroscience of conscious states. Behav Brain Sci 2000; 23:
793–842.
Jackson GD, Duncan JS. MRI neuroanatomy. New York: Churchill
Livingstone; 1996.
Lamberti G, Weidlich S. A visual learning and memory test for neuropsychological assessment. Seattle: Hogrefe and Huber; 1999.
Moser M-B, Moser EI. Functional differentiation in the hippocampus.
Hippocampus 1998; 8: 608–19.
Nielsen TA. A review of mentation in REM and NREM sleep: ‘covert’ REM
sleep as a possible reconciliation of two opposing models. Behav Brain Sci
2000; 23:851–66.
Otten LJ, Henson RNA, Rugg MD. Depth of processing effects on neural
correlates of memory encoding. Brain 2001; 124: 399–412.
Paller KA, Mc Carthy G, Roessler E, Allison T, Wood CC. Potentials evoked
in human and monkey medial temporal lobe during auditory and visual
oddball paradigms. Electroencephalogr Clin Neurophysiol 1992; 84:
269–79.
Puce A, Kalnins RM, Berkovic SF, Donnan GA, Bladin PF. Limbic P3 potentials, seizure localization, and surgical pathology in temporal lobe epilepsy.
Ann Neurol 1989; 26: 377–85.
Rechtschaffen A, Kales A. A manual of standardized terminology, techniques,
and scoring system for sleep stages of human subjects. Washington D.C.:
Public Health Service. NIH Publication No. 204, U.S. Government Printing
Office; 1968.
Schredl M, Montasser A. Dream recall: state or trait variable? Part I. Model,
theories, methodology and trait factors. Imag Cogn Pers 1996–97; 16:
181–210.
Schredl M, Erlacher D. The problem of dream content analysis: validity as
shown by a bizarreness scale. Sleep Hypn 2003; 5: 129–35.
Schredl M, Wittmann L, Ciric P, Götz S. Factors of home dream recall:
a structural equation model. J Sleep Res 2003; 12: 133–41.
Snyder F. The phenomenology of dreaming. In: Madow L, Stone LH, editors.
The psychodynamic implications of the physiological studies on dreams.
Charles C. Thomas: Springfield; 1970. pp. 124–51.
Squire LR, Stark CE, Clark RE. The medial temporal lobe. Annu Rev Neurosci
2004; 27: 279–306.
Van Roost D, Solymosi L, Schramm J, Van Oosterwyck B, Elger CE.
Depth electrode implantation in the length axis of the hippocampus for
the presurgical evaluation of medial temporal lobe epilepsy: a computed
tomography-based stereotactic insertion technique and its accuracy.
Neurosurgery 1998; 43: 819–26.
Wagner AD, Schacter DL, Rotte M, Koutstaal W, Maril A, Dale AM, et al.
Building memories: remembering and forgetting of verbal experiences as
predicted by brain activity. Science 1998; 281: 1188–91.
Witter MP, Groenewegen HJ, Lopes da Silva FH, Lohman AHM. Functional
organization of the extrinsic and intrinsic circuitry of the parahippocampal
region. Prog Neurobiol 1989; 33: 161–253.