1. Cognitive neuroscience and neuropsychology 1
Modulation of theta phase synchronization in the human
electroencephalogram during a recognition memory task
Sung-Phil Kima, Jae-Hwan Kanga, Seong-Hyun Choea, Ji Woon Jeongb,
Hyun Taek Kimb, Kyongsik Yunc, Jaeseung Jeongc and Seung-Hwan Leed
To the extent that recognition memory relies on synchronized between the frontal and the left parietal areas
interactions among widely distributed neural assemblies during the recognition of previously viewed objects. These
across the brain, phase synchronization between brain results suggest that the recognition memory process
rhythms may play an important role in meditating those may involve an interaction between the frontal and
interactions. As the theta rhythm is known to modulate the left parietal cortical regions mediated by theta phase
in power during the recognition memory process, we aimed synchronization. NeuroReport 00:000–000
2012 Wolters
c
to determine how the phase synchronization of the theta Kluwer Health | Lippincott Williams & Wilkins.
rhythms across the brain changes with recognition NeuroReport 2012, 00:000–000
memory. Fourteen human participants performed a visual
object recognition task in a virtual reality environment. Keywords: electroencephalogram, phase synchronization,
recognition memory, theta oscillations, virtual reality
Electroencephalograms were recorded from the scalp
of the participants while they either recognized objects that Departments of aBrain and Cognitive Engineering, bPsychology, Korea University,
Seongbuk-gu, Seoul, cDepartment of Bio and Brain Engineering, Korea
had been presented previously or identified new objects. Advanced Institute of Science and Technology (KAIST), Daejeon and
d
From the electroencephalogram recordings, we analyzed Department of Psychiatry, Ilsan Paik Hospital, Inje University, Goyang,
Gyeonggi, Republic of Korea
the phase-locking value of the theta rhythms, which
indicates the degree of phase synchronization. We found Correspondence to Seung-Hwan Lee, MD, PhD, Department of Psychiatry,
Ilsan Paik Hospital, Inje University, 2240 Daehwa-dong, Ilsan seo-gu, Goyang,
that the overall phase-locking value recorded during Gyeonggi 411-706, Republic of Korea
the recognition of previously viewed objects was greater Tel: + 82 319 107 260; fax: + 82 319 199 776;
e-mail: lshspss@hanmail.net
than that recorded during the identification of new objects.
Specifically, the theta rhythms became strongly Received 26 March 2012 accepted 10 April 2012
Introduction retrieval of recognition memories [7,8]. The cross-correla-
Recognition memory, which is a complex cognitive func- tion of theta and gamma oscillations between the frontal
tion, requires communication between neural assemblies and the parietal cortical regions became stronger with
over the brain [1]. This neural communication induces recognition memory [9]. Increases in frontoparietal coher-
local and global temporal alignments of the firing activity ence in gamma oscillations were also induced by recogni-
of neural assemblies [2,3]. It also induces the modulation of tion memory [10].
the brain oscillations reflecting synchronous activity of a
Phase synchronization in the human EEG has also been
neural assembly; the amplitude of an oscillation indicates
associated with recognition memory [2,11]. However,
the modulation of a local neural assembly, whereas phase
although many studies have reported the modulation of
synchronization between oscillations reflects synchronous
gamma phase synchronization between the frontal and
firing activity between assemblies [3]. Especially, there is
the parietal cortical regions in the context of recognition
substantial evidence that phase synchronization is a key
memory [4,10], little is known about how theta phase
mechanism underlying neural communication [2].
synchronization is modulated relative to recognition
The modulation of phase synchronization across distrib- memory.
uted brain regions has been associated with many cogni-
Therefore, we aim to investigate the patterns of theta
tive functions [3] as well as memory processes [4]. Phase
phase synchronization in the human EEG during a recogni-
synchronization during memory retrieval has primarily
tion memory task. The previous findings lead us to predict
been examined with theta and gamma oscillations. The
that theta phase synchronization across the frontal and
role of gamma oscillations is to bind diverse perceptual
parietal cortical regions increases during recognition memory.
information into memory, whereas the role of theta
In agreement with a previous study [12], we also expect the
oscillations is to control the temporal order of individual
left parietal area to be a focal point in theta synchronization
memory representations [1,5,6].
networks.
A number of studies have reported theta and gamma
oscillations in the human electroencephalogram (EEG) in Participants and methods
the context of recognition memory [1]. The power of Fourteen healthy individuals (five men, nine women,
theta and gamma oscillations increased with the successful 29.2±6.8 years old) participated in the study. All of the
0959-4965
2012 Wolters Kluwer Health | Lippincott Williams & Wilkins
c DOI: 10.1097/WNR.0b013e328354afed
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

2. 2 NeuroReport 2012, Vol 00 No 00
participants provided written informed consent. The EEG recording and acquisition were performed using the
experimental procedure was approved by the Ethics Neuroscan SynAmps 64-channel amplifier and Quickcap
Committee of Inje University (IB-0802-006). All the electrodes (Compumedics Neuroscan Inc., El Paso, Texas,
participants had normal or corrected-to-normal vision and USA). The vertical electrooculogram was monitored using
no history of neurological disease. The participants per- two electrodes, one placed near the outer canthus of the
formed a recognition memory task designed in a virtual left eye and the other beneath the eye. All the scalp
environment (VE) [13]. The VE consisted of four locations electrodes were referred to linked electrodes placed
(office, library, lounge, and conference room) and each on the left and right earlobes. The impedances were
location contained 15 different office items (e.g. a maintained below 10 kO. The analog EEG signals were
vending machine in the lounge). The levels of familiarity, sampled at 1 kHz and filtered through both a band-pass
emotional valence, and arousal of these items were filter (0.1–100 Hz) and a (60 Hz) notch filter. As we
evaluated in a previous study [14]. examined whole-brain EEG synchronizations, we selected
19 channels according to the 10–20 international system
Figure 1 illustrates the overall experimental procedure. for analysis.
The experiment included a navigation (encoding) session
We analyzed 1700 ms segments from each trial: from
and a retrieval session. In the navigation session, the par-
500 ms before to 1200 ms after stimulus onset. A total of
ticipants passively navigated through the four locations in
520 REC and 458 CR trials from all the participants were
a random order. The participants were instructed to
included in the analysis, with 37.1±10.5 (average±SD)
remember the items in each location. Each object was
REC and 32.7±3.5 CR trials per participant. We reduced
presented for 2000 ms with an interstimulus interval of
noise by removing hidden noise sources using an inde-
5000 ms. In the retrieval session, the participants perfor-
pendent component analysis [15].
med a recognition memory task. The participants were
instructed to press one button if they recognized an item The EEG signal in each trial was band-pass filtered
as having been presented during navigation in the same (4–8 Hz) to extract theta activity using a finite-impulse
location (‘old’ items) or the other button if they did not response filter (length: 300 ms; bandwidth: 2 Hz). The
recognize the item as having been presented previously instantaneous phase and amplitude of a theta rhythm
(‘new’ items). A total of 40 new and 60 old objects were were estimated using the Hilbert transform [16]. The
presented for 500 ms each, with randomly varying inter- phase-locking value (PLV) was computed between theta
trial intervals (2000–4000 ms). The EEG was recorded rhythms from every pair of EEG channels [17]. The PLV
during the retrieval session. We selected those retrieval was the normalized length of the vector sum of unit
trials in which the participants correctly recognized old vectors, where the angle of each unit vector represented a
items [hereafter called recognition (REC) trials] or cor- phase difference between two rhythms at a given time
rectly rejected new items [correct rejection (CR) trials] instant in the corresponding trial. A larger PLV indicates
for further analysis. stronger phase synchronization.
Fig. 1
(a) Navigation (b) Retrieval
Old item
Fixation OR Response
period New item period
Location Item
+
2000−4000 ms 500 ms 2000 ms
Illustration of the experimental procedure. (a) The virtual environment, with an item in a location as presented during the navigation session (top).
During this session, participants passively navigated through a virtual environment containing four locations (library, office, lounge, and meeting room)
(bottom). In each of the four locations, 15 items were presented for 2 s each, with an interstimulus interval of 5 s. (b) The retrieval session procedure.
In each trial of this session, participants fixed their gaze at a ‘ + ’ mark in the center of the screen for 2-4 s. An item was then presented in one of the
four locations for 0.5 s. This item had either been presented in the preceding navigation session (old items) or was newly presented (new items).
Participants responded to each item by pressing one of two buttons to identify that item as old or new.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

3. Theta phase synchrony for recognition Kim et al. 3
The PLV difference between REC and CR was calculated those for CR from 400 to 1200 ms after stimulus [t(170) >
by subtracting the CR PLVs from the REC PLVs and by 2.8, P < 0.01] (Fig. 2b).
time-averaging those differences in each nonoverlapping
Figure 3 shows the EEG pairs that showed significant
poststimulus window of 100 ms. From all channel pairs,
increases in PLVs for either REC or CR (P < 0.01,
this yielded 171 PLV difference values for each window.
Bonferroni’s) after stimulus onset. The number of pairs
We statistically evaluated whether this set of differences
indicating greater PLVs for REC than for CR began to
was significantly different from zero using a t-test. Fol-
increase at 400 ms after stimulus and peaked in the
lowing the same procedure, but replacing PLV with the
900–1000 ms window (Fig. 3a). In contrast, only a few
trial-average theta amplitude, we also evaluated the dif-
electrode pairs indicated greater PLVs for CR than for
ference in theta amplitude between REC and CR.
REC in any of the poststimulus windows.
We next investigated pair-wise event-related changes in To further analyze inter-regional phase synchronization
the PLV difference. Given a pair of the EEG channels, we patterns from the above results, we selected four regional
compared the 100 PLV difference values in each post- groups of EEG channels: left frontal (LF: FP1, F7, F3),
stimulus window with a baseline. The baseline difference right frontal (RF: FP2, F8, F4), left posterior (LP: P7, P3,
value was obtained from the 500 ms segment recorded O1), and right posterior (RP: P8, P4, O2). Then, we
before stimulus onset. Using the same procedure as des- calculated the average number of synchronized pairs
cribed above, we obtained 500 baseline PLV difference between two regions for 0–500 ms after stimulus (before
samples. The adjusted confidence intervals (CIs) of the stimulus offset) and for 500–1200 ms after stimulus (after
baseline and the poststimulus window were estimated stimulus offset). We found the strongest phase synchroni-
using the bootstrap method (significance level of 0.01) zation between the right frontal and left posterior regions
[18]. If the lower bound of the poststimulus CI was for REC during the second period, with an average of
greater than the upper bound of the baseline CI, we 2.14 synchronized pairs (Fig. 4a). In contrast, less syn-
considered the REC PLV to be significantly greater than chronized pairs were observed for CR (Fig. 4b).
the CR PLV. In the opposite case, the CR PLV was
considered to be significantly greater than the REC PLV. Discussion
Recognition in episodic memory involves the retrieval of
Results an event along with other contextual aspects of the event,
Of all 14 participants, the average correct response rates including information about time and space, self-reference,
were 72.9±14.8% for old items and 95.9±3.5% for new emotional experience, and personal significance [19]. How-
items. The reaction times after stimulus onset were ever, conventional stimuli used in laboratory studies of
1376.1±165.9 ms for REC and 1050.9±150.0 ms for CR. episodic memory do not provide rich contexts for this
complex form of recognition memory [20]. The present
Overall, the REC PLV was greater than the CR PLV
study partially addressed this problem by using a VE with
(Fig. 2a). Specifically, the REC PLV was significantly
realistic stimuli where participants could experience a
greater than CR PLV during a period of 400–1100 ms after
sensation of immersion [13].
stimulus [t(170) > 2.8, P < 0.01]. The CR PLV was greater
only for 200–300 ms after stimulus [t(170) < – 2.8, P < Our observation of strong theta phase synchronization
0.01]. The theta amplitudes for REC were greater than between the right frontal and left parietal areas is in
Fig. 2
(a) Phase (b) Magnitude
0.08 0.4
0.06 0.3
Power difference
PLV difference
0.04 0.2
0.02 0.1
0 0
−0.02 −0.1
0 500 1000 0 500 1000
Time (ms) Time (ms)
(a) Differences in the theta phase-locking values (PLVs) observed during recognition (REC) and during correct rejection (CR). The average PLV
differences between REC and CR across all 171 electroencephalogram electrode pairs are plotted for each nonoverlapping 100-ms time window
from 0–1200 ms after stimulus. A positive difference indicates that the REC PLV is greater than the CR PLV. (b) Differences in theta power observed
during REC and during CR. The same procedure described above for the PLV was used to analyze theta power.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

4. 4 NeuroReport 2012, Vol 00 No 00
Fig. 3
(a) Recognition (b) Correct rejection
0.0–0.1 s 0.1–0.2 s 0.2–0.3 s 0.3–0.4 s 0.0–0.1 s 0.1–0.2 s 0.2–0.3 s 0.3–0.4 s
0.4–0.5 s 0.5–0.6 s 0.6–0.7 s 0.7–0.8 s 0.4–0.5 s 0.5–0.6 s 0.6–0.7 s 0.7–0.8 s
0.8–0.9 s 0.9–1.0 s 1.0–1.1 s 1.1–1.2 s 0.8–0.9 s 0.9–1.0 s 1.0–1.1 s 1.1–1.2 s
Temporal variation in the distribution of phase-synchronized electroencephalogram electrode pairs. Black lines indicate pairs of electrodes that
showed significantly greater PLVs during the recognition of old items (a) or during the correct rejection of new items (b) in each 100-ms time window
from 0 to 1200 ms after stimulus.
Fig. 4 but also long-range communication of these with frontal
(a) Recognition (b) Correct rejection
neural assemblies.
Gamma and theta phase synchronizations have been shown
LF RF LF RF
to play prominent roles in the encoding and retrieval of
episodic memory [1]. Gamma phase synchronization sup-
0–0.5 s
ports a bottom-up process of local memory representation,
whereas theta phase synchronization supports a top-down
RP
process to organize local assemblies for integrated memory
LP RP LP
representation [1]. Specifically, theta phase synchroniza-
tion between prefrontal and parietal areas supports the ex-
ecutive function for frontal top-down control over posterior
LF RF
regions [4,11,12,21]. Theta oscillations also mediate inter-
LF RF
actions between the hippocampus and the neocortical
regions during long-term memory processes [22]. Coherent
0.5–1.2 s
theta oscillations in a network of prefrontal, mediotempor-
al, and visual cortical regions during recognition memory
LP RP
have been documented using magnetoencephalography [21].
LP RP
The modulation of theta phase synchronization observed in
the present study may support this theoretical role of theta
Inter-regional theta phase synchronization patterns. The phase-locking phase synchronization in frontal top-down control over
value (PLV) was used to measure the connectivity between the left posterior cortical regions during the retrieval of recognition
frontal (LF), right frontal (RF), left parietal (LP), and right parietal (RP)
regions of the scalp during the retrieval of recognition memory of old memory.
items (a) and during the correct rejection of new items (b). Inter-regional
connectivity was analyzed for two poststimulus time segments: Our results indicated that theta phase synchronization and
0–500 ms (while items were visible) and 500–1200 ms (after items had
disappeared). The thickness of each line is proportional to the average amplitude increased during the recognition of old items
number of cross-region pairs of electrodes that showed significant approximately 400 ms after stimulus onset. In previous
increases in PLV. The maximum thickness corresponds to 2.14 pairs
(per 100 ms) and the minimum thickness corresponds to 0.14. ERP studies, this period was associated with the recol-
lection process [23]. It has also been shown that theta
power increases significantly during a recollection period
of 600–1200 ms, indicating that induced theta oscillations
agreement with the previous neuroimaging results of the may reflect neural mechanisms underlying recollection
effects of recognition. A meta-analysis of functional MRI [8]. Note that the present study examined changes in
studies showed that the loci of such effects were theta synchronizations during the recognition of old and
concentrated in the left parietal area [12]. Our results new objects without further distinguishing between famil-
suggest that the recognition memory process involves not iarity and recollection of recognition memory, as it focused on
only the local activation of left parietal neural assemblies the two cases in which participants recognized either an old
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

5. Theta phase synchrony for recognition Kim et al. 5
item in its previous location or that a completely new item. 5 Miltner WH, Braun C, Arnold M, Witte H, Taub E. Coherence of gamma-band
Although we did not examine how theta activity would EEG activity as a basis for associative learning. Nature 1999; 397:
434–436.
change if the participants recognized old items in new 6 Lee SH, Kim DW, Kim EY, Kim S, Im CH. Dysfunctional gamma-band activity
locations, answering this latter question would be helpful for during face structural processing in schizophrenia patients. Schizophr Res
understanding how phase synchronization supports familiar- 2010; 119:191–197.
7 Klimesch W, Doppelmayr M, Yonelinas A, Kroll NE, Lazzara M, Rohm D, et al.
ity and recollection. A further study will address this question Theta synchronization during episodic retrieval: neural correlates of
to determine whether dual recognition memory processes conscious awareness. Brain Res Cogn Brain Res 2001; 12:33–38.
8 ¨
Gruber T, Tsivilis D, Giabbiconi CM, Muller MM. Induced
would induce distinct phase synchronization patterns.
electroencephalogram oscillations during source memory: familiarity
is reflected in the gamma band, recollection in the theta band.
Conclusion J Cogn Neurosci 2008; 20:1043–1053.
9 Burgess AP, Ali L. Functional connectivity of gamma EEG activity
This study documented an increase in theta phase is modulated at low frequency during conscious recollection.
synchronization during recognition memory. The results Int J Psychophysiol 2002; 46:91–100.
suggest that theta phase synchronization may be impor- 10 Summerfield C, Mangels JA. Functional coupling between frontal and
parietal lobes during recognition memory. NeuroReport 2005; 16:117–122.
tant in integrating widespread recognition memory traces. 11 Klimesch W, Freunberger R, Sauseng P, Gruber W. A short review of slow
Strong theta phase synchronizations were found between phase synchronization and memory: evidence for control processes in
the right frontal and the left parietal regions, which may different memory systems? Brain Res 2008; 1235:31–44.
12 Vilberg KL, Rugg MD. Memory retrieval and the parietal cortex: a review
reflect large-scale functional connectivity between these of evidence from a dual-process perspective. Neuropsychologia 2008;
regions during recognition memory processes. 46:1787–1799.
13 Kim YY, Kim HJ, Kim EN, Ko HD, Kim HT. Characteristic changes in the
physiological components of cybersickness. Psychophysiology 2005;
Acknowledgements 42:616–625.
This work was supported by the National Research 14 Hahm J, Lee K, Lim SL, Kim SY, Kim HT, Lee JH. Effects of active navigation
Foundation (NRF) of Korea Grant funded by the Korean on object recognition in virtual environments. Cyberpsychol Behav 2007;
10:305–308.
Government (NRF-2010-32A-B00282). S.P.K., J.H.K., and 15 Makeig S, Bell AJ, Sejnowski TJ. Independent component analysis of
S.H.C. were also supported by World Class University electroencephalographic data. Adv Neural Inf Process Syst 1996; 8:
(WCU) program (R31-10008) and the Basic Science 145–151.
16 Bruns A. Fourier-, Hilbert- and wavelet-based signal analysis: are they really
Research Program (2011-0026502) of the NRF of Korea different approaches? J Neurosci Methods 2004; 137:321–332.
funded by the Ministry of Education, Science and 17 Lachaux JP, Rodriguez E, Martinerie J, Varela FJ. Measuring phase synchrony
Technology. in brain signals. Hum Brain Mapp 1999; 8:194–208.
18 Payton ME, Greenstone MH, Schenker N. Overlapping confidence intervals
or standard error intervals: what do they mean in terms of statistical
Conflicts of interest significance? J Insect Sci 2003; 3:34.
There are no conflicts of interest. 19 Piefke M, Fink GR. Recollections of one’s own past: the effects of aging and
gender on the neural mechanisms of episodic autobiographical memory.
Anat Embryol (Berl) 2005; 210:497–512.
References 20 Burgess N, Maguire EA, Spiers HJ, O’Keefe J. A temporoparietal and
1 Nyhus E, Curran T. Functional role of gamma and theta oscillations in prefrontal network for retrieving the spatial context of lifelike events.
episodic memory. Neurosci Biobehav Rev 2009; 34:1023–1035. NeuroImage 2001; 14:439–453.
2 Womelsdorf T, Schoffelen JM, Oostenveld R, Singer W, Desimone R, 21 ¨
Guderian S, Duzel E. Induced theta oscillations mediate large-scale
Engel AK, et al. Modulation of neuronal interactions through neuronal synchrony with mediotemporal areas during recollection in humans.
synchronization. Science 2007; 316:1609–1612. Hippocampus 2005; 15:901–912.
3 Varela F, Lachaux JP, Rodriguez E, Martinerie J. The brainweb: phase 22 Zion-Golumbic E, Kutas M, Bentin S. Neural dynamics associated with
synchronization and large-scale integration. Nat Rev Neurosci 2001; 2: semantic and episodic memory for faces: evidence from multiple
229–239. frequency bands. J Cog Neurosci 2010; 22:263–277.
4 Fell J, Axmacher N. The role of phase synchronization in memory processes. 23 Curran T, Cleary AM. Using ERPs to dissociate recollection from familiarity
Nat Rev Neurosci 2011; 12:105–118. in picture recognition. Brain Res Cogn Brain Res 2003; 15:191–205.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.