NeuroImage 138 (2016) 100–108

Contents lists available at ScienceDirect

NeuroImage

j ourna l homepage: www.e lsev ie r .com/ locate /yn img

Pre-stimulus thalamic theta power predicts human memory formation

Catherine M. Sweeney-Reed a,⁎, Tino Zaehle a, Jürgen Voges a,b, Friedhelm C. Schmitt a, Lars Buentjen a,Klaus Kopitzki a,b, Alan Richardson-Klavehn a, Hermann Hinrichs a,b,c, Hans-Jochen Heinze a,b,c,Robert T. Knight d, Michael D. Rugg e

a Departments of Neurology and Stereotactic Neurosurgery, Otto von Guericke University, Leipziger Strasse 44, 39120 Magdeburg, Germanyb Department of Behavioral Neurology, Leibniz Institute for Neurobiology, Otto von Guericke University, Leipziger Strasse 44, 39120 Magdeburg, Germanyc German Centre for Neurodegenerative Diseases (DZNE), Otto von Guericke University, Leipziger Strasse 44, 39120 Magdeburg, Germanyd Helen Wills Neuroscience Institute and Department of Psychology, University of California, Tolman Hall, MC 3192, Berkeley, CA 94720, USAe Center for Vital Longevity and School of Behavioral and Brain Sciences, University of Texas, Dallas, TX 75235, USA

⁎ Corresponding author.E-mail addresses: catherine.sweeney-reed@med.ovgu

tino.zaehle@ovgu.de (T. Zaehle), juergen.voges@med.ovgfc.schmitt@med.ovgu.de (F.C. Schmitt), lars.buentjen@meklaus.kopitzki@med.ovgu.de (K. Kopitzki), alan.richardson(A. Richardson-Klavehn), hermann.hinrichs@med.ovgu.dehans-jochen.heinze@med.ovgu.de (H.-J. Heinze), rtknightmdr104020@utdallas.edu (M.D. Rugg).

http://dx.doi.org/10.1016/j.neuroimage.2016.05.0421053-8119/© 2016 Elsevier Inc. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 January 2016Revised 13 April 2016Accepted 16 May 2016Available online 19 May 2016

Pre-stimulus theta (4–8 Hz) power in the hippocampus and neocortex predicts whether a memory for asubsequent event will be formed. Anatomical studies reveal thalamus-hippocampal connectivity, and lesion,neuroimaging, and electrophysiological studies show that memory processing involves the dorsomedial(DMTN) and anterior thalamic nuclei (ATN). The small size and deep location of these nuclei have limitedreal-time study of their activity, however, and it is unknown whether pre-stimulus theta power predictive ofsuccessfulmemory formation is also found in these subcortical structures.We recorded human electrophysiolog-ical data from theDMTNand ATN of 7 patients receiving deep brain stimulation for refractory epilepsy.We foundthat greater pre-stimulus theta power in the right DMTN was associated with successful memory encoding,predicting both behavioral outcome and post-stimulus correlates of successful memory formation. In particular,significant correlations were observed between right DMTN theta power and both frontal theta and right ATNgamma (32–50 Hz) phase alignment, and frontal-ATN theta-gamma cross-frequency coupling. We draw thefollowing primary conclusions. Our results provide direct electrophysiological evidence in humans of a role forthe DMTNaswell as theATN inmemory formation. Furthermore, prediction of subsequentmemoryperformanceby pre-stimulus thalamic oscillations provides evidence that post-stimulus differences in thalamic activity thatindex successful and unsuccessful encoding reflect brain processes specifically underpinningmemory formation.Finally, the findings broaden the understanding of brain states that facilitate memory encoding to includesubcortical as well as cortical structures.

© 2016 Elsevier Inc. All rights reserved.

Keywords:Pre-stimulus thetaThalamusMemoryEncodingDorsomedial thalamic nucleusAnterior thalamic nucleus

Introduction

Memory formation arises from interactions between environmentalevents and a continually varying brain state (Fox et al., 2007).Pre-stimulus activity, reflecting the brain state preceding memoryformation, has recently been shown to predict whether a memory willbe formed (e.g., Cohen et al., 2015; Guderian et al., 2009; Otten et al.,2006, 2010; Park and Rugg, 2011). Studies have focused on the medialtemporal lobe (MTL) (Fell et al., 2011; Guderian et al., 2009) and frontalcortex (Otten et al., 2006) due to the well-known roles of these regions

.de (C.M. Sweeney-Reed),u.de (J. Voges),d.ovgu.de (L. Buentjen),-klavehn@med.ovgu.de(H. Hinrichs),

@berkeley.edu (R.T. Knight),

in memory formation. For instance, hippocampal and rhinal corticaltheta (4–8 Hz) oscillations predict encoding success (Fell et al., 2011).However, whether pre-stimulus activity in subcortical structures orpre-stimulus subcortical–cortical interactions play a role in determiningencoding success remains unknown. Recent evidence highlights impor-tant roles for the anterior and dorsomedial thalamic nuclei (ATNand DMTN respectively) in memory processing (Aggleton, 2012;Staudigl et al., 2012; Sweeney-Reed et al., 2014, 2015). Based on thesefindings and established thalamo-hippocampal connectional anatomy(Aggleton, 2012; Aggleton et al., 2010; Vertes et al., 2001), we hypoth-esized that pre-stimulus thalamic theta power preceding presentationwould predict successful memory performance.

To test this hypothesis, we examined electrophysiological activityrecorded directly from the DMTN and ATN during memory encodingin human participants who had received electrodes implanted fordeep brain stimulation therapy for refractory epilepsy. Our primaryobjective was to investigate whether pre-stimulus theta oscillationsin thalamic nuclei predict successful memory formation. To assess

101C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

whether thalamic theta activity before an event has a direct relationshipto event memory, we also examined whether individual differences inmemory performance were predicted by pre-stimulus theta power.Additionally, we assessed the correlation between pre-stimulusthalamic theta power and recently identified post-stimulus neural corre-lates of successful memory formation. Pre- and post-stimulus corticalactivity predicting successful memory formation are partially correlated(Otten et al., 2006), and we hypothesized that thalamic pre-stimulustheta power would also predict some of the post-stimulus electro-physiological measures of successful encoding. These included post-stimulus frontal-ATN theta-gamma cross-frequency coupling (CFC) andtheta phase synchrony, as well as ATN theta phase alignment, all predic-tive of successful memory formation (Sweeney-Reed et al., 2014, 2015).We discuss our findings in the context of the hypothesis that the DMTNand ATN are differentially involved in memory systems supportingencoding resulting in enhanced familiarity (perirhinal-medial dorsalthalamic system) or in enhanced recollection (hippocampal-anteriorthalamic: ‘extended hippocampus system’) (Aggleton and Brown,1999). We also consider the findings from the perspective ofanatomical connectivity between the DMTN and the amygdala and inthe context of a role in oculomotor control.

Materials and methods

We analyzed intracranial electroencephalogram (EEG) recordeddirectly from 8 bilateral contacts (4 on each electrode probe) in the

Fig. 1. Electrode localization. (A–D) Illustration of electrode localization in Participant 2. (A) Theon the pre-operative structural MRI scan (coronal view). Inset: sagittal view. (B) Intrathalamistructural MRI scan was co-registered with the post-operative CT scan, and the electrode locatthe intraoperative x-ray coordinates. (D) Atlas illustration of thalamic nuclei (line drawing froBasal Ganglia’ by A. Morel. Reproduced by permission of Taylor and Francis Group, LLC, a divisthe angulation of the stereotactic electrode trajectory means that the two deepest contactslocations calculated by transformation into Montreal Neurological Institute (MNI) space, corepresentation of the DMTN (purple).

ATN and DMTN in 7 human volunteers receiving electrodes implantedfor stimulation treatment of pharmacoresistant focal epilepsy (Fig. 1),as well as from a single frontal scalp EEG contact (Table 1). The frontalelectrode location (Fz, AFz, or Fpz) differed slightly across participants,because differing bandage placements dictated the electrode location.We note that post-stimulus aspects of this dataset have been describedin prior reports (Sweeney-Reed et al., 2014, 2015). The analyses andresults presented here focus on the pre-stimulus interval, and thedependency between pre-stimulus activity and previously identifiedpost-stimulus correlates of encoding.

We begin by describing how intrathalamic electrode positioning wasperformed and providing clinical information regarding the patients.Details of the memory encoding paradigm employed are then given,followed by details of the data analysis. The EEG data were analyzedtaking two complementary approaches. Prediction of subsequentmemory formation from pre-stimulus thalamic data was evaluated,then correlations between pre-stimulus DMTN theta power and post-stimulus neural correlates of successful memory encoding were calcu-lated. We describe the steps taken for each analysis, including assess-ment for statistical significance taking account ofmultiple comparisons.

Electrode localization

A stereotactic neurosurgeon performed trajectory planning forthe electrode positioning using ATN atlas coordinates, adjusting theentry point to bypass the ventricular vessels safely. The DMTN is a

twomost dorsal right-sided contacts in the anterior thalamic nucleus (ATN), superimposedc electrodes visible on the intraoperative stereotactic x-ray image. (C) The pre-operativeion (second most dorsal contact shown here) in the ATN (blue) was confirmed based onm Morel, 2007: Copyright © 2007. From ‘Stereotactic Atlas of the Human Thalamus andion of Informa plc): the target for the two most dorsal contacts was the ATN (green), andlie in the dorsomedial thalamic nucleus (DMTN) (purple). (E) The right-sided contactlor-coded for each participant, and projected onto the Pick Atlas (Maldjian et al., 2003)

Table 1Electrode locations. Thalamic contacts used in bipolar montage and frontal contacts. * = anelectrophysiological signal was obtained with similar amplitude to that from the ATNcontact beneath, but imaging indicated a contact location in cerebrospinal fluid just abovethe ATN. ATN = anterior thalamic nucleus; DMTN= dorsomedial thalamic nucleus.

Pt Left bipolar channels Right bipolar channels Frontal electrode

1 ATN – ATN edge ATN – ATN edge FzDMTN – DMTN DMTN – DMTN

2 ATN edge – ATN ATN – ATN FpzLamina/DMTN – DMTN DMTN – DMTN

3 ATN – ATN/lamina ATN – ATN/lamina FpzDMTN – DMTN DMTN – DMTN

4 ATN edge – ATN ATN – lamina/DMTN Fpzlamina/DMTN – DMTN DMTN – DMTN

5 border* ATN – ATN border* ATN – ATN AFzDMTN edge – DMTN DMTN – DMTN

6 ATN edge – ATN ATN – ATN FpzDMTN – DMTN DMTN – DMTN

7 ATN edge – ATN ATN – ATN FpzDMTN edge – DMTN DMTN – DMTN

Fig. 2. Memory encoding paradigm. A fixation cross was shown during the pre-stimulusinterval investigated, in order to minimize eye movements.

102 C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

comparatively large thalamic structure, positioned beneath the ATN,and the angulation involved in the standard stereotactic approach totarget the ATN necessarily guides the trajectory into the DMTN.

Electrode localization was carried out in several steps (Fig. 1). First,intraoperative stereotactic x-ray images confirmed the alignment ofimplanted hardware and the MRI- and CT-based trajectory. Next,a post-operative CT scan was co-registered with the pre-operativestructural MRI. Electrode localization was then carried out by anindependent stereotactic neurosurgeon, using information from thelatter images and the intraoperative stereotactic x-ray. Because theATN is located directly beneath the ventricular wall, the most dorsalelectrode was judged to be in the ATN when a neuronal signal wasdetected and electrode resistance indicated tissue contact instead ofcerebrospinal fluid. Further electrode positions were determinedthrough comparison of atlas data with individual anatomy. Stereotacticmeasurements were plotted into contours of the DMTN as depictedin the Morel stereotactic atlas (Morel, 2007). The locations of theright-sided contacts for all 7 participants, converted into MontrealNeurological Institute (MNI) space and projected onto the Pick Atlas(Maldjian et al., 2003), are shown in Fig. 1(E).

Participants

Mean participant age was 37.1 years (standard deviation, SD8.8 years), and four participants were female. Pre-operative neuropsy-chological testing is described in Sweeney-Reed et al. (submitted forpublication), and clinical andmemory performance details are providedin Table 1 of Sweeney-Reed et al. (submitted for publication). Thecurrent data are from the 7 participants who had contacts in both theDMTN and the ATN, out of the 8 participants whose post-stimuluscorrelates of memory formation in the ATN were previously reported(Sweeney-Reed et al., 2014, 2015).

Memory encoding paradigm

The recordings weremade during the incidental encoding of a seriesof 200 photographic scenes (Fig. 2), which were judged by participantsas depicting an indoors or an outdoors setting (see Sweeney-Reed et al.,2014). A fixation cross, on which participants were instructed to focustheir gaze,was present during thepre-stimulus timeperiod tominimizeeye movement. During a subsequent memory test, the scenes were re-presented in a new random order, interspersed with 100 similar newscenes, and judged by participants as ‘old’ or ‘new'’. The electrophysio-logical data were divided into epochs that onset 1 s pre-stimulusand continued to 2 s post-stimulus, and were sorted according tosubsequent memory performance. The experiment was performedbefore intrathalamic stimulation therapy commenced. Approval for

these recordings was granted by the Institutional Review Board of theUniversity of Magdeburg Medical Faculty, and all participants gaveinformed, written consent.

Intracranial EEG recording and data pre-processing

Intrathalamic and scalp EEG data were recorded with respect to anose referencewith aWalter Graphtek amplifier, at a sampling frequen-cy of 512 Hz, using eight 1.5 mm platinum intrathalamic contactsspaced at 1.5 mm intervals (4 on the left electrode probe and 4 on theright), and scalp electrodes. Electrodes were positioned stereotacticallyon the basis of pre-operative MRI scans, and localization was confirmedusing intra-operative x-ray and post-operative CT images, as perElectrode localization section. Pre-processing included artifact removalusing temporal decorrelation separation independent componentanalysis, as introduced by (Nicolaou and Nasuto, 2007), following theprocedure described in Sweeney-Reed et al. (2012). Each thalamiccontact was re-referenced to its ipsilateral neighbor to produce abipolar montage that increased spatial resolution (Kühn et al., 2006;Oswal et al., 2013; Rutishauser et al., 2010; Staudigl et al., 2012;Sweeney-Reed et al., 2014).

EEG analysis

We briefly describe the calculation of phase synchrony, CFC,and phase alignment metrics that were previously reported for post-stimulus processing (Sweeney-Reed et al., 2014, 2015). The novelfindings presented here pertain to pre-stimulus oscillatory power andits relationship with behavioral and post-stimulus electrophysiologicalfindings.

EEG pre- and post-stimulus featuresThe data from each trial were wavelet-transformed using a 6-cycle

Morlet wavelet to yield amplitude and phase values, from whichpower and frequency measures were derived (Lachaux et al., 1999),and the resultswere averaged across trials and participants. Phase align-ment with respect to stimulus onset was estimated by calculating thestimulus-locked average of the phase distribution across trials at agiven time-frequency, yielding values between 0 (no phase alignment)and 1 (complete phase alignment) (for details see: Tallon-Baudry et al.,1996; Düzel et al., 2005; Sweeney-Reed et al., 2015). The wavelet-derived phase series were also used to calculate phase synchrony,defined as a constant value of the phase difference between frontaland thalamic channels for each epoch, providing phase-locking valuesbetween 0 (no phase synchrony) and 1 (complete phase synchrony)

103C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

(Daly et al., 2012; Sweeney-Reed et al., 2014; Varela et al., 2001). CFCwas assessed between the phase of frontal cortical theta oscillationsand right ATN (RATN) gamma amplitude using the methods describedby Axmacher et al. (2010), Canolty et al. (2006), Sweeney-Reed et al.(2014). Trial numbers were balanced for phase alignment, phasesynchrony, and CFC analyses, because a valid statistical comparison be-tween these measures during successful compared with unsuccessfulencoding requires an equal number of epochs in each category. Toachieve this, epochs were randomly selected from the larger categoryto match the size of the smaller one (Sweeney-Reed et al., 2014).Corrections for multiple time-frequency comparisons were effected byusing a cluster-size permutation test (CSPT) (Maris and Oostenveld,2007; Sweeney-Reed et al., 2014), where a cluster was defined as con-tiguous time-frequency points at which the power differed significantly(criterion p = 0.05) according to T-tests. Further details regarding cal-culation of post-stimulus measures can be found in Sweeney-Reedet al. (2014, 2015). We applied a false discovery rate (FDR) correctionto account for testing four thalamic channels (left and right DMTNand ATN) and a frontal channel, resulting in a threshold of p = 0.02(criterion, q = 0.05). All significance tests had 6 degrees of freedom,based on means over trials calculated for 7 participants.

Late post-stimulus effects and attentionWe also evaluated the possible influences of late post-stimulus

effects and of attention on our pre-stimulus findings. We assessedthe former by investigating whether pre-stimulus theta differencesbetween successful and unsuccessful encoding identified in the rightDMTN (RDMTN) correlated with differences in late post-stimulus pro-cessing. To accomplish this, the 1 s of data preceding each stimulus(−1–0 s) was relabeled as late post-stimulus (2.8–3.8 s) data followingeither successful or unsuccessful encoding of the preceding stimulus.The mean power across these new categories was calculated followingthe same procedure as described above for the analysis of the pre-stimulus data, and the same statistical comparison was performed,applying T-tests to each time-frequency point. We also assessed possi-ble attentional effects by examining whether there was a dependencyin the subsequent memory performance for sequentially presentedstudy items. The probability of consecutively presented items bothbeing correctly recognized did not differ from the probability that amissed item would be followed by a recognized item, suggesting thatslow fluctuations in attentional state were not responsible for ourfindings (Sweeney-Reed et al., 2014).

Prediction of post-stimulus electrophysiological correlates of memoryformation

If pre-stimulus RDMTN theta power primes successful memoryformation, it should predict post-stimulus neural correlates of memoryformation. We previously described post-stimulus differences infrontal-RATN theta phase synchrony, frontal-RATN CFC, and RATNtheta phase alignment, during successful compared with unsuccessfulencoding (Sweeney-Reed et al., 2014, 2015). We assessed whether thepre-stimulus DMTN theta power (4–8Hz, 0.9–0 s), predicting successfulmemory formation, was correlated with these measures in the RATN orfrontal cortex, or with gamma (32–50 Hz) phase alignment 1 s post-stimulus in those structures, given that RATN gamma oscillations werecoupled with theta timing (Sweeney-Reed et al., 2015). We usedmean values over time and frequency to limit the number of compari-sons (see Addante et al., 2011). The frequency and temporal rangeswere chosen on the basis of our previous findings and the literature.The correlation significance threshold was adjusted by FDR correctionto account for the total number of computed correlations, resulting ina threshold of p = 0.039 (criterion, q = 0.05, min r = 0.78).

Because early (0–0.5 s post-stimulus) fronto-thalamic thetaphase synchrony was greater during successful than unsuccessfulencoding (Sweeney-Reed et al., 2014), and early frontal theta phasealignment predicted subsequent CFC during successful encoding

(Sweeney-Reed et al., 2015), we also investigated whether RDMTNpre-stimulus theta power predicted early theta phase alignment.

Results

Webeginwith the behavioral analysis, in which thememory perfor-mance during the experimental paradigm was assessed, before provid-ing the results of the electrophysiological data analysis. The EEG analysisis presented in three stages. First, the difference in pre-stimulus powerpreceding successful and unsuccessful memory formation is evaluated,then assessment is made of prediction of behavioral and of post-stimulus neurophysiological correlates of successful memory formationby DMTN pre-stimulus theta power.

Behavioral analysis

Ameanacross participants of 54.6% (SD 18.6) of the sceneswere cor-rectly identified as having been seen during the encoding phase (hits),whereas 27% (SD 19.1) of new items were falsely judged as old (falsealarms). The hit minus false alarm rate of 35.4% (SD 9.6), resulting in ad′ of 0.73 (Snodgrass and Corwin, 1988), differed significantly fromthe chance rate of 0 (T-test: T = 6.7, p = 0.0005), indicating reliablediscrimination between studied and unstudied test items acrossparticipants. The behavioral results from the simple paradigmemployed during the intracranial recording indicate that all partici-pants had adequate memory capabilities to perform the task.Notably, despite having the lowest recognition score during thepre-operative neuropsychological assessment, Participant 1 had thehighest hit minus false alarm rate of all participants, at 45%.

Pre-stimulus theta analysis

In all 4 thalamic nuclei, pre-stimulus (0.9–0 s) theta (4–8Hz) power,averaged across epochs for each participant, was enhanced precedingsuccessful compared with unsuccessful encoding (Fig. 3). Aftercorrection for multiple comparisons the difference (the pre-stimulussubsequent memory effect) was statistically significant only in theRDMTN (cluster-size permutation test, CSPT: p = 0.012; FDR correctedp-value = 0.02 with a maximum FDR level of q = 0.05). There was atrend for greater pre-stimulus theta power in the LATN (p = 0.021)and lower power in frontal EEG (p = 0.037), preceding successfulcomparedwith unsuccessful encoding; neither of these effects survivedFDR correction. Remarkably, the pre-stimulus subsequent memoryeffects were evident in the RDMTN in all 7 participants (Fig. 4(A); seealso Fig. 1 in Sweeney-Reed et al., submitted for publication). Wetherefore assessed the significance of the differences between success-ful and unsuccessful encoding on an individual level, shuffling thecategories 1000 times and calculating differences between two artifi-cially constructed groups to produce a distribution against which tocompare the individual differences. The individual difference was sig-nificant (criterion p = 0.05) for 4 of the 7 participants (p = 0.040;p=0.11; p=0.015; p N 0.3; p=0.21; p=0.042; p=0.012). Addition-al analysis excluded the possibility that these effects were late post-stimulus effects from the preceding trial. When each pre-stimulusepoch was re-categorized as the late (2.8–3.8 s) post-stimulusperiod following successful or unsuccessful encoding of the precedingstimulus (see Late post-stimulus effects and attention section), thetime-frequency points at which theta power differed significantly(according to uncorrected T-tests at each point) between categoriesdid not coincide in time-frequency with the significant pre-stimulustheta power difference (Inline Supplementary Fig. S1). The variance ofpre-stimulus DMTN theta power was stable across memory conditions.The variance of the power preceding items subsequently successfullyremembered did not differ from that preceding items subsequentlyforgotten (F = 4.04; p = 0.11), and it did not differ following correctly

Fig. 3. Pre-stimulus theta power. Mean group difference between unbaselined pre-stimulus theta power preceding successful versus unsuccessful encoding. (A) Left anterior thalamicnucleus (ATN). (B) Right ATN. (C) Left dorsomedial thalamic nucleus (DMTN). (D) Right DMTN. The difference was significant after correction for multiple comparisons (region ofsignificance indicated by the contour). T-tests were used to assess the significance of the differences shown in (A–D), to which cluster-size permutation tests for multipletime-frequency comparisons were then applied. False discovery rate correction for multiple locations with criterion q = 0.05 yielded a significance threshold of p = 0.02, so that onlytheta power in the right DMTN (D) was significantly greater preceding successful than unsuccessful encoding (p = 0.012).

104 C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

compared with following incorrectly encoded stimuli (F = 0.68;p = 0.68).

Inline Supplementary Fig. S1 can be found online at doi:http://dx.doi.org/10.1016/j.neuroimage.2016.05.042.

Additional support for the proposal that pre-stimulus RDMTN thetapower predicts successful encoding was provided by assessment ofwhether a task independent measure would independently predictmemory performance. Data were additionally available for five partici-pants during rest with eyes closed. The resting data were pre-processed identically to the encoding data. Resting RDMTN thetapower correlated with memory performance, as measured by the hitminus false alarm rate (r = 0.89, p = 0.042). Neither RDMTN alpha(8–12 Hz) nor beta (12–20 Hz) power correlated significantly withthe hit minus false alarm rate (r = 0.49, p = 0.40; r = 0.50, p = 0.39,respectively). Mean RDMTN pre-stimulus theta power preceding suc-cessful encoding was greater than resting RDMTN theta power (pairedT-test: T = 6.6, p = 0.0027). Theta power preceding unsuccessfulencoding was also greater than the power at rest (T = 3.8, p = 0.02),

but the difference was smaller. Resting RDMTN theta power was notcorrelated with RDMTN theta power preceding successful or precedingunsuccessful encoding (r = 0.60, p = 0.28; r = −0.52, p = 0.37,respectively).

Pre-stimulus theta and memory performanceThe correspondence between pre-stimulus theta power and sub-

sequent memory performance was also examined on an epoch-by-epoch basis. We focused on the RDMTN and the theta frequencyrange, because the difference between power preceding successfuland unsuccessful encoding was only significant at the RDMTN loca-tion at the theta frequency range. Mean pre-stimulus theta powerwas calculated for each epoch for each participant. Based on perfor-mance on the subsequent memory test, the epochs were labeled asleading to successful or unsuccessful encoding, and were then sortedin order of pre-stimulus theta power. The ordered epochs wereplaced into 7 equally sized bins, such that bin 1 contained the lowesttheta values and bin 7 the highest, referred to as the theta rank

Fig. 4.Relationship between pre-stimulus theta power and behavioral measures. (A)Mean right dorsomedial thalamic nucleus (RDMTN) pre-stimulus theta powerwas greater precedingsuccessful than unsuccessful encoding in each of the 7 individual participants with RDMTN electrodes in the time-frequency window where mean group pre-stimulus theta power wassignificantly greater preceding successful than unsuccessful encoding. (B) Ascendingmean pre-stimulus RDMTN theta power values for each participantwere divided into 7 equally-sizedbins. The theta rank is shown as a function of recognition performance (hits minus false alarms). The recognition performance in each bin was normalized by subtracting the overallperformance for each participant. A positive correlation was found between pre-stimulus theta amplitude and recognition performance (r = 0.76, p = 0.049). (The thick line showsthe mean across participants, and the thin lines show the recognition rate for each participant individually.)

105C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

(Guderian et al., 2009). The recognition rate (pHit-pFalse Alarm)was calculated for each rank, and then normalized by subtractingthe overall recognition rate for each participant. Across participants,

Fig. 5. Correlation between pre- and post-stimulus electrophysiological correlates of successfunsuccessful encoding (UE). Third and fourth columns: significance of correlations in firspermutation test (p b 0.001). Final column: illustrative scatter plots, calculating the means ovof pre-stimulus (0.9–0 s) right dorsomedial thalamic nucleus (RDMTN) theta power: UpperCorrelation in scatter plot (pre-stimulus theta power: mean over 4–8 Hz; post-stimulus thet(0.9–1.1 s) post-stimulus right anterior thalamic nucleus gamma phase alignment. Correlatiphase alignment: mean over 32–40 Hz): r = −0.76, p = 0.049. (B) Correlation of post-stipre-stimulus (0.9–0 s) RDMTN theta (4–8 Hz) power. Correlation in scatter plot (pre-stimugamma amplitude): r = 0.81, p = 0.0026. Lower panels: with poststimulus (0.9–1.1 s) gammscatter plot above): r = −0.78, p = 0.0036.

the epoch-by-epoch mean pre-stimulus RDMTN power levelcorrelated with recognition performance as indexed by pHit-pFalseAlarm (r = 0.76, p = 0.049) (Fig. 4(B)).

ul encoding (SE). First column: correlation during SE. Second column: correlation duringt two columns. The correlations shown were all significant during SE on cluster-sizeer the frequency ranges of the significant clusters, with linear regression. (A) Correlationpanels — with early (0–0.5 s) post-stimulus theta phase alignment in frontal neocortex.a phase alignment: mean over 4–5 Hz): r = 0.94, p = 0.0015. Lower panels — with lateon in scatter plot (pre-stimulus theta power: mean over 4–8 Hz; post-stimulus gammamulus (0.5–1.5 s) frontal-RATN cross-frequency coupling (CFC): Upper panels — withlus theta power: mean over 4–8 Hz; CFC: mean over 5–6 Hz theta phase and 32–50 Hza (31–66 Hz) phase alignment and CFC. Correlation in scatter plot (CFC parameters as for

106 C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

Pre-stimulus theta and post-stimulus correlates of memory formationCorrelation between pre-stimulus RDMTN theta power and early

(0–0.5 s) frontal theta phase alignment was significant duringsuccessful encoding (r = 0.80, p = 0.032; CSPT: p b 0.001) but notduring unsuccessful encoding (r = −0.28, p = 0.54) (Fig. 5(A)).These correlation coefficients differed significantly (p b 0.05 onFisher's z-transformed values). The correlation between pre-stimulusRDMTN theta power and early RATN theta phase alignment duringsuccessful encoding was not significant (r = −0.31, p = 0.50), nordid pre-stimulus RDMTN theta power correlate with late (0.9–1.1 s)post-stimulus theta phase alignment, either frontally (r = −0.19,p = 0.69) or in the RATN (r = 0.19, p = 0.69). However, pre-stimulustheta RDMTN power negatively correlated with post-stimulus RATNgamma phase alignment during successful encoding (r = −0.78, p =0.039; CSPT: p b 0.001) (Fig. 5(B)). When the correlation betweenpre-stimulus RDMTN theta power and post-stimulus RATN gammaphase alignment was directly compared between later rememberedand later forgotten items, the correlation was greater for later remem-bered items (p b 0.05). Pre-stimulus RDMTN theta power also correlat-ed with frontal-RATN CFC (5–6 Hz theta phase coupled with 32–50 Hzgammaamplitude: r=0.72, p=0.069; CSPT: p b 0.001) (Fig. 5(C)). Ad-ditionally, the gamma RATN phase alignment around 1 s post-stimulus,which itself was predicted by the pre-stimulus RDMTN theta power,was correlated with frontal-RATN CFC (r = 0.82, p = 0.023; CSPT:p b 0.001) (Fig. 5(D)) at that time. Illustrative scatter plots of the signif-icant correlations are provided in the final column of Fig. 5. The panelsin (A) show pre-stimulus RDMTN theta power against frontal cortexand RATN phase alignment for each participant, and the panels in(B) show correlations between CFC and pre-stimulus RDMTN powerand post-stimulus gamma phase alignment. The frequencies of thephase alignment and CFC were those at which correlation was signifi-cant on cluster-size testing. Linear regression is shown for each.

Discussion

We observed that pre-stimulus theta power in the RDMTN predictsencoding success. Enhanced theta power predicted both better perfor-mance on a later test of recognition memory as well as frontal andRATN post-stimulus neural correlates of successful memory formation.Post-stimulus RATN theta synchrony correlateswith successfulmemoryencoding and is a core component of an ‘extended hippocampal system’proposed to support episodic memory formation (Aggleton and Brown,1999; Sweeney-Reed et al., 2014, 2015). Here, we present evidence thatthe RDMTN also plays a key role in memory encoding, and provide fur-ther evidence that a DMTN-frontal subsystem modulates ATN activity.These results were robust and prediction of encoding success at thegroup level was observed in all seven participants, each of whom hadgreater mean pre-stimulus theta power in epochs preceding later re-membered than later forgotten stimuli. Our findings extend evidencefor involvement of the DMTN in memory processing (Aggleton andBrown, 1999), suggesting it plays a critical role even before the eventto be remembered occurs.

We discuss our findings in the context of what is known about pre-stimulus prediction of successfulmemory formation and the anatomicaland functional connectivity of the DMTN and ATN. We consider threepossible mechanisms by which DMTN pre-stimulus oscillatory activitycould modulate post-stimulus ATN activity: via a DMTN-frontal-ATNroute in the context of separate recollection and familiarity subsystems,via a DMTN-frontal-ATN route in the context of an early oculomotorcontrol system affecting readiness for stimulus perception, and via aDMTN-amygdala-frontal-ATN or DMTN-amygdala-hippocampus-ATNroute in which emotional context affects memory formation. We nextdiscuss the current contribution to growing evidence that the ATN per-forms an integrative role in memory processing, expanding further onits possible role in recollection and familiarity. We then consider thetiming of the greater DMTN pre-stimulus theta power in the context

of pre-stimulus theta in other brain structures associated with subse-quent memory formation. Finally, we consider the influence of pre-stimulus DMTN theta power on ATN gamma oscillations in memoryformation.

Theta activity in the DMTN pre-stimulation and in the ATN post-stimulation predicts memory formation. There are no direct connec-tions between the DMTN and the ATN, and other evidence indicatesthat these nuclei belong to separate memory systems supportingfamiliarity and recollection respectively (Aggleton, 2012). These factorssuggest that the RDMTN and RATN contribute to successful memoryencoding via distinct, albeit interacting mechanisms. Indeed, pre-stimulus RDMTN theta power predicted gamma RATN phase alignmentand frontal-RATN theta-gamma CFC around 1 s post-stimulus, but pre-dicted neither RATN theta phase alignment nor frontal-RATN thetaphase synchrony during this period, suggesting that these measures re-flect different aspects of encoding. Unlike the strong ATN connections,the DMTN is not directly connected with the hippocampus (Aggleton,2012). The existence of strong and direct connections between theRDMTN and prefrontal cortex (Aggleton, 2012; Klein et al., 2010)supports the proposal that RDMTN theta power influences encodingprocesses in the ‘extended hippocampal system’ via the frontal cortex,likely involving top-down prefrontal pathways through the entorhinalcortex and hippocampus to the ATN (Aggleton and Brown, 1999). In ac-cord with these ideas, we observed a correlation between pre-stimulusRDMTN theta power and early post-stimulus frontal theta phase align-ment, both of which correlated with post-stimulus frontal-RATN CFC(Sweeney-Reed et al., 2015). These findings are consistent with thenotion that both of these thalamic nuclei play a modulatory role inmemory encoding (Aggleton and Brown, 1999).

There is also evidence that the DMTN plays a role in oculomotorcontrol (Huerta and Kaas, 1990; Rafal et al., 2004), which is compatiblewith DMTN pre-stimulus theta power facilitating subsequent memoryencoding via a mechanism in which visual attention for the subsequentstimulus is heightened. The current findings are in line with lesionevidence suggesting that the DMTN is involved in the learning of condi-tioned responses involving attention (reviewed by (Bradfield et al.,2013)). The DMTN is activated earlier than the pre-frontal cortex in pro-spective memory for saccadic direction in oculomotor-based workingmemory (Watanabe and Funahashi, 2012), which could be interpretedas a form of anticipatory processing and also fits with our suggestionthat pre-stimulus DMTN activity influences post-stimulus ATN activityvia the prefrontal cortex. Our finding that DMTN theta power is greaterduring the encoding task than during rest is consistent with the oculo-motor control system maintaining heightened perceptual attentionalreadiness throughout the task, in anticipation of the next stimulus.Moreover, we note that greater resting RDMTN theta power is predic-tive of better overall memory performance, which provides additionalsupport for the proposal that DMTN theta activity facilitates memoryformation. Resting RDMTN theta power not being correlated withRDMTN theta power preceding successful or preceding unsuccessfulencoding suggests that high theta power in the RDMTN is associatedwith trial by trial memory formation independently of a participant'sbackground or resting level of RDMTN theta power. These findings fitwith the proposal that an association between greater RDMTN thetapower and better memory formation not only reflects an overall predis-position to successful memory formation, but fluctuations in this powerover time in an individual also influence memory formation, possiblydue to attentional processes. If attentional fluctuations are involved,the time scale is short, because no dependency was found in theprobability of successful encoding of sequential items (Sweeney-Reedet al., 2014).

The correlation between pre-stimulus DMTN and post-stimulusATN activity could also be mediated via pathways through theamygdala and hippocampus, given DMTN-amygdala (Gaffan andMurray, 1990; Mitchell and Chakraborty, 2013), amygdala-hippocampus (Aggleton, 2012), hippocampus-ATN (Aggleton and

107C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

Brown, 1999), and amygdala-frontal (Ishikawa and Nakamura,2003) anatomical connectivity. Moreover, functional connectivityin the theta range has been observed between the hippocampusand amygdala (Benchenane et al., 2011). Indeed, DMTN lesionshave produced similar associative learning deficits to amygdalalesions (Gaffan and Murray, 1990). The current paradigm did notinvolve an assessment by participants of their emotional reactionto stimuli, so we are unable to determine whether stimuli elicitingan individual emotional response were more readily successfullyencoded.

The diverse inputs received by the ATN, including frontal,retrosplenial and hippocampal inputs (Aggleton, 2012), and theinvolvement of theta as well as theta-gamma long-range synchronybetween the frontal cortex and the ATN during memory encoding(Sweeney-Reed et al., 2014), indicate that the ATN has an integrativerole in memory formation (Sweeney-Reed et al., 2015). The correlationof ATN post-stimulus activity with pre-stimulus activity in the RDMTN,a structure thought to be involved in a distinct memory system to thatinvolving the ATN, further supports the hypothesis that the ATN inte-grates information from diverse sources during encoding. Specifically,the correlation between DMTN and ATN activity raises the possibilitythat ATN activity is modulated by the familiarity system of which theDMTN is a member (Aggleton, 2012). Although it has been proposedthat familiarity and recollection reflect points on a unidimensional con-tinuum of memory strength (Squire et al., 2007), substantial evidencesuggests that these two forms of memory are functionally distinct anddepend in part on different brain networks (Aggleton and Brown,2006; Sauvage et al., 2008; Yonelinas et al., 2005). Although the presentparadigm does not permit separate estimation of the contributions ofrecollection and familiarity to memory performance, our findings areconsistent with the hypothesis of Aggleton and colleagues.

The strong anatomical connectivity between the ATN andhippocampus supports a role for the ATN in encoding that involvesthe hippocampal system and thus, perhaps, a critical role for thisstructure in recollection. The current findings, which implicate theDMTN in encoding, suggest that performance on our recognitiontask also depended on familiarity. Interestingly, pre-stimulus hippo-campal theta power has recently been reported to predict successfulrecognition memory, but not recall (Merkow et al., 2014). One inter-pretation of this finding is that, as we have proposed for the DMTN,pre-stimulus theta in the hippocampus selectively supports laterfamiliarity-based memory judgments. The finding of Merkow et al.could, however, also be explained by the idea that theta is more im-portant for encoding of items rather than associations (Merkowet al., 2014). Although selective hippocampal lesions generallyleave familiarity-driven recognition memory intact (Eichenbaumet al., 2007), impairment of familiarity is sometimes reported follow-ing hippocampal damage (Song et al., 2011). Thus, we cannot dis-count the possibility that the effects we observed in the DMTNreflect processes that enhance encoding operations supportingfamiliarity-based memory within the ‘extended hippocampal sys-tem’, to which the ATN belong. Given that prefrontal cortex hasbeen implicated in both recollection and familiarity (Duarte et al.,2005), an interaction between the two memory systems via frontalcortex is plausible. Based on our finding that the DMTN activity pre-dicts activity in both the RATN and frontal cortex, we postulate that aDMTN-frontal familiarity systemmodulates activity in the ‘extendedhippocampal system’ (involving the ATN). Such a modulatory role isconsistent with a role for the ATN in integrating information fromdifferent anatomical subsystems underpinning memory formation.

The peak of the present RDMTN pre-stimulus theta power effectcoincided in time with the analogous theta power increase previouslyreported in direct hippocampal recordings (Fell et al., 2011). Bycontrast, the RDMTNpeak preceded the fronto-temporal and right fron-tal pre-stimulus theta peak reported previously in a free recall task(Guderian et al., 2009), a task dependent on recollection. Together,

these findings suggest that early enhancement of pre-stimulus thetain the hippocampus and RDMTN provide a neural context conduciveto successful memory encoding. In contrast, whether encodingsupporting later recollection or familiarity subsequently ensues isdetermined by other pre-stimulus factors operating closer in timeto stimulus onset, as well as post-stimulus factors.

Higher pre-stimulus RDMTN theta powerwas associatedwith lowergamma phase alignment and with greater frontal-RATN CFC. Thecorrelation between pre-stimulus DMTN theta power and ATN gammaphase alignment must be mediated indirectly, given the absence ofdirect anatomical connectivity between these structures and otherproposals that gamma oscillations do not facilitate long-range inter-regional communication in memory (Kahana et al., 2001; Ray andMaunsell, 2015; Stein and Sarnthein, 2000). During successful encoding,pre-stimulus DMTN theta power was positively correlated with frontal-ATN CFC. We hypothesize that the pre-stimulus state facilitated anactive selection of particular neural assemblies related to the encodedevent, such that CFC continued only at the frequency of the neuralassemblies relevant to the encoding of that specific event. Indeed, aclose relationship has been observed between the theta rhythm andMTL local spike timing during successful encoding, highlighting theimportance of the coordination of theta and high frequency activity inmemory processing (Rutishauser et al., 2010).

It should be noted that the correlation between memory perfor-mance and pre-stimulus RDMTN theta power was determined acrossparticipants. The DMTN is a part of one of many systems supportingmemory formation. Greater theta power preceding a given stimulusis associated with a greater probability that the stimulus is latersuccessfully remembered but is only one of a myriad of factors thatcan contribute tomemory encoding and is not necessarily a prerequisiteto memory formation but rather a potential facilitator.

Conclusions

Wepresent direct electrophysiological evidence in humans for a keyrole of the DMTN in memory processing, These findings are consistentwith reports of the involvement of the DMTN in encoding in bothanimal lesion and human fMRI studies (Aggleton and Brown, 1999;Mitchell and Gaffan, 2008; Pergola et al., 2013). Notably, intrinsicpre-stimulus DMTN theta activity predicted both subsequent memoryand its post-stimulus neural correlates in the ATN. Our finding of pre-stimulus thalamic effects on subsequentmemory performance providesevidence that the post-stimulus differences in thalamic activity that arepredictive of successful and unsuccessful encoding are in part driven bypre-stimulus oscillatory signals in line with similar arguments based onpre-stimulus findings in scalp EEG (Otten et al., 2006). Furthermore,evidence that pre-stimulus activity that facilitates encoding can beunder voluntary control (Gruber and Otten, 2010) opens the way fornew approaches to treatment of memory disorders (Gruber et al.,2013). The current work suggests that activity in subcortical as well ascortical structures could be relevant targets for such treatments.

References

Addante, R.J.,Watrous, A.J., Yonelinas, A.P., Ekstrom, A.D., Ranganath, C., 2011. Prestimulustheta activity predicts correct source memory retrieval. PNAS 108, 10702–10707.http://dx.doi.org/10.1073/pnas.1014528108/-/DCSupplemental.www.pnas.org/cgi/doi/10.1073/pnas.1014528108.

Aggleton, J.P., 2012. Multiple anatomical systems embedded within the primate medialtemporal lobe: implications for hippocampal function. Neurosci. Biobehav. Rev. 36,1579–1596. http://dx.doi.org/10.1016/j.neubiorev.2011.09.005.

Aggleton, J.P., Brown, M.W., 1999. Episodic memory, amnesia, and the hippocampal-anterior thalamic axis. Behav. Brain Sci. 22, 425–444.

Aggleton, J.P., Brown, M.W., 2006. Interleaving brain systems for episodic and recognitionmemory. Trends Cogn. Sci. 10, 455–463. http://dx.doi.org/10.1016/j.tics.2006.08.003.

Aggleton, J.P., O'Mara, S.M., Vann, S.D., Wright, N.F., Tsanov, M., Erichsen, J.T., 2010.Hippocampal-anterior thalamic pathways for memory: uncovering a network ofdirect and indirect actions. Eur. J. Neurosci. 31, 2292–2307. http://dx.doi.org/10.1111/j.1460-9568.2010.07251.x.

108 C.M. Sweeney-Reed et al. / NeuroImage 138 (2016) 100–108

Axmacher, N., Henseler, M.M., Jensen, O., Weinreich, I., Elger, C.E., Fell, J., 2010. Cross-frequency coupling supports multi-item working memory in the human hippocam-pus. PNAS 107, 3228–3233. http://dx.doi.org/10.1073/pnas.0911531107.

Benchenane, K., Tiesinga, P.H., Battaglia, F.P., 2011. Oscillations in the prefrontal cortex:a gateway to memory and attention. Curr. Opin. Neurobiol. 21, 475–485. http://dx.doi.org/10.1016/j.conb.2011.01.004.

Bradfield, L.A., Hart, G., Balleine, B.W., 2013. The role of the anterior, mediodorsal, andparafascicular thalamus in instrumental conditioning. Front. Syst. Neurosci. 7, 51.http://dx.doi.org/10.3389/fnsys.2013.00051.

Canolty, R.T., Edwards, E., Dalal, S.S., Soltani, M., Nagarajan, S.S., Kirsch, M., Berger, M.S.,Barbaro, N.M., Knight, R.T., 2006. High gamma power is phase-locked to thetaoscillations in human neocortex. Science 313, 1626–1628. http://dx.doi.org/10.1126/science.1128115.High.

Cohen, N., Pell, L., Edelson, M.G., Ben-Yakov, A., Pine, A., Dudai, Y., 2015. Peri-encodingpredictors of memory encoding and consolidation. Neurosci. Biobehav. Rev. 50,128–142. http://dx.doi.org/10.1016/j.neubiorev.2014.11.002.

Daly, I., Sweeney-Reed, C.M., Nasuto, S.J., 2012. Testing for Significance of PhaseSynchronisation Dynamics in the EEG. http://dx.doi.org/10.1007/s10827-012-0428-2.

Duarte, A., Ranganath, C., Knight, R.T., 2005. Effects of unilateral prefrontal lesions onfamiliarity, recollection, and source memory. J. Neurosci. 25, 8333–8337. http://dx.doi.org/10.1523/JNEUROSCI.1392-05.2005.

Düzel, E., Richardson-Klavehn, A., Neufang, M., Schott, B.H., Scholz, M., Heinze, H.-J., 2005.Early, partly anticipatory, neural oscillations during identification set the stagefor priming. NeuroImage 25, 690–700. http://dx.doi.org/10.1016/j.neuroimage.2004.11.038.

Eichenbaum, H., Yonelinas, A., Ranganath, C., 2007. The medial temporal lobe andrecognition memory. Annu. Rev. Neurosci. 30, 123–152. http://dx.doi.org/10.1016/j.biotechadv.2011.08.021.Secreted.

Fell, J., Ludowig, E., Staresina, B.P., Wagner, T., Kranz, T., Elger, C.E., Axmacher, N., 2011.Medial temporal theta/alpha power enhancement precedes successful memoryencoding: evidence based on intracranial EEG. J. Neurosci. 31, 5392–5397. http://dx.doi.org/10.1523/JNEUROSCI.3668-10.2011.

Fox, M.D., Snyder, A.Z., Vincent, J.L., Raichle, M.E., 2007. Intrinsic fluctuations withincortical systems account for intertrial variability in human behavior. Neuron 56,171–184. http://dx.doi.org/10.1016/j.neuron.2007.08.023.

Gaffan, D., Murray, E.A., 1990. Amygdalar interaction with the mediodorsal nucleus of thethalamus and the ventromedial prefrontal cortex in stimulus-reward associativelearning in the monkey. J. Neurosci. 10, 3479–3493.

Gruber, M.J., Otten, L.J., 2010. Voluntary control over prestimulus activity related toencoding. J. Neurosci. 30, 9793–9800. http://dx.doi.org/10.1523/JNEUROSCI.0915-10.2010.

Gruber, M.J., Watrous, A.J., Ekstrom, A.D., Ranganath, C., Otten, L.J., 2013. Expected rewardmodulates encoding-related theta activity before an event. NeuroImage 64, 68–74.http://dx.doi.org/10.1016/j.neuroimage.2012.07.064.

Guderian, S., Schott, B., Richardson-Klavehn, A., Düzel, E., 2009. Medial temporal thetastate before an event predicts episodic encoding success in humans. PNAS 106,5365–5370. http://dx.doi.org/10.1073/pnas.0900289106.

Huerta, M., Kaas, J., 1990. Supplementary eye field as defined by intracorticalmicrostimulation: connections in macaques. J. Comp. Neurol. 293, 299–330.

Ishikawa, A., Nakamura, S., 2003. Convergence and interaction of hippocampal andamygdalar projections within the prefrontal cortex in the rat. J. Neurosci. 23,9987–9995 (doi:23/31/9987 [pii]).

Kahana, M.J., Seelig, D., Madsen, J.R., 2001. Theta returns. Curr. Opin. Neurobiol. 11,739–744.

Klein, J.C., Rushworth, M.F.S., Behrens, T.E.J., Mackay, C.E., de Crespigny, A.J., D'Arceuil, H.,Johansen-Berg, H., 2010. Topography of connections between human prefrontalcortex and mediodorsal thalamus studied with diffusion tractography. NeuroImage51, 555–564. http://dx.doi.org/10.1016/j.neuroimage.2010.02.062.

Kühn, A.A., Doyle, L., Pogosyan, A., Yarrow, K., Kupsch, A., Schneider, G., Hariz, M.I.,Trottenberg, T., Brown, P., 2006. Modulation of beta oscillations in the subthalamicarea during motor imagery in Parkinson's disease. Brain 129, 695–706. http://dx.doi.org/10.1093/brain/awh715.

Lachaux, J.-P., Rodriguez, E., Martinerie, J., Varela, F.J., 1999. Measuring phase synchronyin brain signals. Hum. Brain Mapp. 8, 194–208. http://dx.doi.org/10.1002/(SICI)1097-0193.

Maldjian, J.A., Laurienti, P.J., Kraft, R.A., Burdette, J.H., 2003. An automated method forneuroanatomic and cytoarchitectonic atlas-based interrogation of fMRI data sets.NeuroImage 19, 1233–1239. http://dx.doi.org/10.1016/S1053-8119(03)00169-1.

Maris, E., Oostenveld, R., 2007. Nonparametric statistical testing of EEG- and MEG-data. J. Neurosci. Methods 164, 177–190. http://dx.doi.org/10.1016/j.jneumeth.2007.03.024.

Merkow, M.B., Burke, J.F., Stein, J.M., Kahana, M.J., 2014. Prestimulus theta in the humanhippocampus predicts subsequent recognition but not recall. Hippocampus 1569,1–23. http://dx.doi.org/10.1002/hipo.22335.

Mitchell, A.S., Chakraborty, S., 2013. What does themediodorsal thalamus do? Front. Syst.Neurosci. 7, 37. http://dx.doi.org/10.3389/fnsys.2013.00037.

Mitchell, A.S., Gaffan, D., 2008. The magnocellular mediodorsal thalamus is necessary formemory acquisition, but not retrieval. J. Neurosci. 28, 258–263. http://dx.doi.org/10.1523/JNEUROSCI.4922-07.2008.

Morel, A., 2007. Stereotactic Atlas of the Human Thalamus and Basal Ganglia. CRC Press.Nicolaou, N., Nasuto, S., 2007. Automatic artefact removal from event-related potentials

via clustering. J. VLSI Sig. Proc. (Invited paper).Oswal, A., Litvak, V., Bru, C., Huebl, J., Schneider, G., Ku, A.A., 2013. Cognitive factors

modulate activity within the human subthalamic nucleus during voluntarymovement in Parkinson's disease. J. Neurosci. 33, 15815–15826. http://dx.doi.org/10.1523/JNEUROSCI.1790-13.2013.

Otten, L.J., Quayle, A.H., Akram, S., Ditewig, T.A., Rugg, M.D., 2006. Brain activity before anevent predicts later recollection. Nat. Neurosci. 9, 489–491. http://dx.doi.org/10.1038/nn1663.

Otten, L.J., Quayle, A.H., Puvaneswaran, B., 2010. Prestimulus subsequent memory effectsfor auditory and visual events. J. Cogn. Neurosci. 22, 1–20. http://dx.doi.org/10.1162/jocn.2009.21298.Prestimulus.

Park, H., Rugg, M.D., 2011. Pre-stimulus hippocampal activity predicts later recollection.Hippocampus 20, 24–28. http://dx.doi.org/10.1002/hipo.20663.Pre-stimulus.

Pergola, G., Ranft, A., Mathias, K., Suchan, B., 2013. The role of the thalamic nuclei inrecognition memory accompanied by recall during encoding and retrieval: anfMRI study. NeuroImage 74, 195–208. http://dx.doi.org/10.1016/j.neuroimage.2013.02.017.

Rafal, R., McGrath, M., Machado, L., Hindle, J., 2004. Effects of lesions of the humanposterior thalamus on ocular fixation during voluntary and visually triggeredsaccades. J. Neurol. Neurosurg. Psychiatry 75, 1602–1606. http://dx.doi.org/10.1136/jnnp.2003.017038.

Ray, S., Maunsell, J.H.R., 2015. Do gamma oscillations play a role in cerebral cortex?Trends Cogn. Sci. 19, 78–85. http://dx.doi.org/10.1016/j.tics.2014.12.002.

Rutishauser, U., Ross, I.B., Mamelak, A.N., Schuman, E.M., 2010. Human memory strengthis predicted by theta-frequency phase-locking of single neurons. Nature 464,903–907. http://dx.doi.org/10.1038/nature08860.

Sauvage, M.M., Fortin, N.J., Owens, C.B., Yonelinas, A.P., Eichenbaum, H., 2008. Recognitionmemory: opposite effects of hippocampal damage on recollection and familiarity.Nat. Neurosci. 11, 16–18. http://dx.doi.org/10.1038/nn2016.

Snodgrass, J.G., Corwin, J., 1988. Pragmatics of measuring recognition memory:applications to dementia and amnesia. J. Exp. Psychol. Gen. 117, 34–50. http://dx.doi.org/10.1037/0096-3445.117.1.34.

Song, Z., Wixted, J.T., Hopkins, R.O., Squire, L.R., 2011. Impaired capacity for familiarityafter hippocampal damage. PNAS 108, 9655–9660. http://dx.doi.org/10.1073/pnas.1107247108.

Squire, L.R., Wixted, J.T., Clark, R.E., 2007. Recognition memory and the medial temporallobe: a new perspective. Nat. Rev. Neurosci. 8, 872–883. http://dx.doi.org/10.1038/nrn2154.

Staudigl, T., Zaehle, T., Voges, J., Hanslmayr, S., Esslinger, C., Hinrichs, H., Schmitt, F.C.,Heinze, H.-J., Richardson-Klavehn, A., 2012. Memory signals from the thalamus:early thalamocortical phase synchronization entrains gamma oscillations duringlong-term memory retrieval. Neuropsychologia 50, 3519–3527. http://dx.doi.org/10.1016/j.neuropsychologia.2012.08.023.

Stein, A.V., Sarnthein, J., 2000. Different frequencies for different scales of corticalintegration: from local gamma to long range alpha/theta synchronization. Int.J. Psychophysiol. 38, 301–313.

Sweeney-Reed, C.M., Riddell, P.M., Ellis, J.A., Freeman, J.E., Nasuto, S.J., 2012. Neuralcorrelates of true and false memory in mild cognitive impairment. PLoS One 7,e48357. http://dx.doi.org/10.1371/journal.pone.0048357.

Sweeney-Reed, C.M., Zaehle, T., Voges, J., Schmitt, F.C., Buentjen, L., Kopitzki, K.,Esslinger, C., Hinrichs, H., Heinze, H.-J., Knight, R.T., Richardson-Klavehn, A., 2014.Corticothalamic phase synchrony and cross-frequency coupling predict humanmemory formation. eLife 3, e05352. http://dx.doi.org/10.7554/eLife.05352.

Sweeney-Reed, C.M., Zaehle, T., Voges, J., Schmitt, F.C., Buentjen, L., Kopitzki, K., Hinrichs,H., Heinze, H.-J., Rugg, M.D., Knight, R.T., Richardson-klavehn, A., 2015. Thalamic thetaphase alignment predicts human memory formation and anterior thalamic cross-frequency coupling. eLife http://dx.doi.org/10.7554/eLife.07578.

Sweeney-Reed, C., Zaehle, T., Voges, J., Schmitt, F., Buentjen, L., Kopitzki, K., Richardson-Klavehn, A., Hinrichs, H., Heinze, H.-J., Knight, R., Rugg, M., 2016. Neuropsychologicalprofile and pre-stimulus dorsomedial thalamic nucleus theta power associated withsuccessful memory formation in deep brain stimulation patients. Data Br (submittedfor publication).

Tallon-Baudry, C., Bertrand, O., Delpuech, C., Pernier, J., 1996. Stimulus specificity ofphase-locked and non-phase-locked 40 Hz visual responses in human. J. Neurosci.16, 4240–4249.

Varela, F., Lachaux, J., Rodriguez, E., Martinerie, J., 2001. The brainweb: phase synchroni-zation and large-scale integration. Nat. Rev. Neurosci. 2, 229–239. http://dx.doi.org/10.1038/35067550.

Vertes, R.P., Albo, Z., Viana di Prisco, G., 2001. Theta-rhythmically firing neurons in the an-terior thalamus: implications for mnemonic functions of Papez's circuit. Neuroscience104, 619–625. http://dx.doi.org/10.1016/S0306-4522(01)00131-2.

Watanabe, Y., Funahashi, S., 2012. Thalamic mediodorsal nucleus and workingmemory. Neurosci. Biobehav. Rev. 36, 134–142. http://dx.doi.org/10.1016/j.neubiorev.2011.05.003.

Yonelinas, A.P., Otten, L.J., Shaw, K.N., Rugg, M.D., 2005. Separating the brain regionsinvolved in recollection and familiarity in recognition memory. J. Neurosci. 25,3002–3008. http://dx.doi.org/10.1523/JNEUROSCI.5295-04.2005.