ORIGINAL ARTICLE

Event-related potentials, CNV, readiness potential,and movement accompanying potential recorded fromposterior thalamus in human subjects. A SEEG study

I. Rektor1*, P. Kanovsky1, M. Bares1, J. Louvel2, M. Lamarche21 1st Department of neurology, Saint-Anne’s Hospital, Masaryk University, Pekarska 53, 60200 Brno, Czech Republic ;2 groupe de neurophysiologie clinique, hôpital Sainte-Anne, Paris, France

(Received 1 January 2001; accepted 21 June 2001)

Summary – Intracranial recordings were obtained from three patients with intractable chronic pain who underwentanalgesic electrical stimulation of the contralateral thalamus. Multilead electrode made it possible to record from severalthalamic nuclei. The electrode was targeted into the ventroposterolateral (VPL) nucleus of the thalamus. During separaterecording sessions, the following tests were performed: somatosensory evoked potentials (SEP) of the median orposterior tibial nerve, event-related cognitive potentials (auditory oddball P3 wave), readiness potential (RP) andcontingent negative variation (CNV) using auditory warning (S1) and visual imperative (S2) stimuli. The movementaccompanying potential (MAP), which was present in the VPL in all but one of the recordings, behaved as a far-fieldpotential. Recordings obtained from the VPL confirmed its established role as a relay nucleus, processing somatosen-sory information to the primary somatosensory cortex. The VPL generated the ‘thalamic’ SEP, which was the onlypotential regularly recorded in this nucleus. In the recordings from one patient (No. 3), auditory and visual evokedpotentials of the CNV protocol, peaking at approximately 300 ms, were obtained from the VPL and appeared to begenerated in situ. Neither RP, CNV nor ’oddball’ ERPs appeared in the VPL. From the pulvinar, only a visually evokedpotential was recorded. Oddball P3, RP, CNV, and middle and long latency auditory and visual potentials (evoked in theCNV paradigm) appeared to be generated ‘dorsally’ to the VPL, probably in the nucleus posterolateralis (PL). Thisstructure may therefore be involved in both the processing of afferent information and in cognitive operations. © 2001Éditions scientifiques et médicales Elsevier SAS

CNV / event-related potential / evoked potential / intracranial recordings / P300 / posterior thalamus / readinesspotential

Résumé – Potentiels liés à un évènement, variation négative de contingence, potentiel de préparation, et potentielde mouvements accompagnateurs enregistrés dans le thalamus postérieur chez des patients. Étude d’élec-troencéphalogramme sous-cortical. Des enregistrements intracrâniens ont été obtenus chez trois patients présentantune douleur irréductible chronique ayant subi une stimulation électrique analgésique du thalamus controlatéral. Uneélectrode multitêtes a permis d’enregistrer l’activité de plusieurs noyaux thalamiques. Les électrodes ont été dirigéesdans le noyau ventro-postéro-latéral (VPL) du thalamus. Lors de sessions différentes, les enregistrements tests

*Correspondence and reprints. Tel.: + 420-5-4318-2623; fax: +420-5-4318-2624.E-mail address: irektor@med.muni.cz (I. Rektor).

Neurophysiol Clin 2001 ; 31 : 253-61© 2001 Éditions scientifiques et médicales Elsevier SAS. All rights reservedS0987705301002623/FLAwww.elsevier.fr/direct/nc-cn

suivants ont été réalisés: les potentiels évoqués somesthésiques (PES) du nerf médian ou tibialis postérieur, lespotentiels cognitifs liés à un évènement (PE) (l’onde auditive oddball P3), le potentiel de préparation (PP) et la variationnégative de contingence (VNC), en utilisant un stimulus d’avertissement auditif (S1) et un stimulus visuel impératif (S2).Le potentiel de mouvements accompagnateurs (PMA), détecté dans tous sauf un des enregistrements dans le VPL,s’est comporté comme un potentiel de champ lointain. Les enregistrements obtenus dans le VPL confirment son rôlede noyau relais, traitant les informations somesthésiques pour les aires corticales somesthésiques. Le VPL a généré lePES thalamique qui a été le seul potentiel enregistré régulièrement dans ce noyau. Sur les enregistrements de l’un despatients (n° 3), les potentiels évoqués auditifs et visuels du protocole de VNC, dont les pics maximums ont étéenregistrés à environ 300 ms, ont été obtenus à partir du VPL et semblaient être générés in situ. Aucun PP, VNC ou PEoddball ne sont apparus dans le VPL. Seul un potentiel évoqué visuel a été enregistré dans le pulvinar. La oddball P3,le PP, la VNC, et les potentiels visuels et auditifs de latence moyenne et longue (évoqué dans le paradigme du VNC)semblaient être générés en arrière du VPL, probablement dans le noyau postéro-latéral (PL). Cette structure pourraitdonc être impliquée à la fois dans le traitement d’informations afférentes et dans des opérations cognitives. © 2001Éditions scientifiques et médicales Elsevier SAS

VNC / potentiel lié à un évènement / potentiel évoqué / enregistrements intracrâniens / P300 / thalamuspostérieur / potentiel de préparation

Interest in brain function localization is continuallyincreasing. One quite new advance is the recentprogress in the detection of brain functions throughmetabolic studies using positron emission tomographyand functional magnetic resonance imaging. Despitetheir spectacular results, these techniques should bepractised and interpreted in conjunction with electro-physiological methods, since electrophysiology revealsphysiological mechanisms providing informationcomplementary to metabolic studies. This is principallydue to the better time resolution of the electrophysi-ological techniques. Electrophysiological research inman has been primarily targeted towards the study ofthe cortex; studies of subcortical functions have beenrarely performed, and most often through interpreta-tion from distant scalp or cortical recordings. However,recording through electrodes placed on the scalp or thebrain surface can bring only an indirect indication ofthe function of the subcortical structure. Although theuse of mathematical models can provide hypotheses onthe location of subcortical generators, such mathemati-cal modelling can hardly provide the proof of suchpostulates, which often remain somewhat ambiguouson anatomical grounds. For example, Milliken et al.[17] recently postulated a source for the readiness po-tential (RP) in the thalamus, but did not suggest a moreprecise location. Intracerebral recordings may bringinformation relevant to the localisation of physiologicalsources not readily available through other means.There are a number of medical reasons to have elec-trode contacts placed subcortically in human subjects,

including stereotaxic surgery for pain relief and controlof abnormal motor activity: the implanted depth elec-trodes may in turn provide direct information on thefunction of subcortical structures. In human subjects,thalamic recordings have most frequently brought in-formation from motor thalamic nuclei, because tha-lamic electrodes have been most often used in tremorsurgery. Information from the posterior thalamus,other than that concerning somatosensory potentials[4, 22] remains very limited [7, 13]. This paper dealswith recordings of potentials from the posterior tha-lamic structures evoked by several modalities of sensorystimuli, as well as of potentials related to motor regula-tion and to some cognitive functions. Thus, duringseparate recording sessions, we were able to recordsomatosensory evoked potentials (SEPs) following me-dian or tibial nerve stimulation, the auditory oddballP3, the readiness potential (RP), the contingent nega-tive variation (CNV) using auditory warning (S1) andvisual imperative (S2) stimuli, and the movement ac-companying potential (MAP).

We obtained recordings from various thalamic struc-tures in three patients. To clarify the role of thalamicstructures, more recordings from more patients areobviously required. In our hospital, we no longer usethe stimulation of the posterior thalamus for therapeu-tic purposes. Therefore, we have decided to publish ourresults with the hope that other laboratories using thesame or similar techniques will be able to use them toinvestigate the complex functions of posterior thalamicstructures.

254 I. Rektor et al.

PATIENTS AND METHODS

Subcortical EEG was recorded in three patients, each ofwhom had an intracerebral electrode implanted in theventroposterolateral (VPL) nucleus of the thalamus.Because multicontact electrodes were used, it was alsopossible to record from neighbouring thalamic nuclei.In all three patients, the electrode was implanted tocounteract chronic pain through contralateral thalamicstimulation [8]. All three patients were right-handedmen, aged 64, 67 and 76 years old. Patient 1 sufferedfrom residual right-sided spastic hemiparesis, with tha-lamic pain on the right side of the body after anischemic stroke in the left thalamic region in 1994. Hewas implanted twice, as the first stimulation had to beinterrupted due to the development of a depressivestate. Patient 2 had chronic pain following traumaticavulsion of the left plexus brachialis in 1991 and am-putation of the left arm in 1993. Spinothalamic tracto-tomy was performed in 1993 and was inefficacious torelieve neurogenic and phantom pain. Patient 3 had aputaminal haemorrhage in the left basal ganglia regionin 1984, with residual spastic right-sided hemiparesisand ‘thalamic’ pain.

Five-lead stimulating and recording electrodes (Mi-crodeept, Dixi Electronics, France) were used in allpatients. Electrodes diameter was 0.8 mm, each contactwas 2 mm long, and the distance between the contactswas 1.3 mm. Each electrode was implanted using aCT-aided stereotactic procedure with the Reichert-Mundinger stereotactic frame, and the co-ordinateswere chosen to reach the VPL nucleus of the thalamus.The position of the electrode was monitored by CTscan and by the pain relief induced by the thalamicstimulation. The electrode was implanted into the leftVPL nucleus of the thalamus in patients 1 and 3 andinto the right VPL nucleus of the thalamus in patient 2.In patient 1, where the pain was a result of a previousstroke probably involving the thalamic area, the elec-trode was implanted after the CT scan indicated noresidual post-stroke lesion on the thalamus.

The sensory effects of the stimulation, including de-crease or disappearance of pain, indicated the success ofthe electrode implantation. In this way, the preciselocation of a given contact within the VPL was deter-mined. The localization of contacts outside the VPLwas based on the superposition of the patient’s CT scanon a stereotactic atlas [23].

Patients were stimulated several times daily for 10 to15 min. The frequency of stimulation was 50 Hz, and

the intensity of the stimulation current was maintainedat the level at which the pain substantially decreased ordisappeared. The duration of this procedure was 3weeks.

The patients were adequately informed about theneurophysiological testing, and consent was obtainedfrom each patient. The testing was approved by thelocal ethics committee. A Nihon-Kohden Neuropack 8device (Nihon – Kohden Systems, Osaka, Japan) wasused for all recordings. Recording sessions usually be-gan on the third day after implantation. During sepa-rate sessions, the following recordings were performed:somatosensory evoked potentials (SEP) of the medianor tibial nerve, event-related cognitive potentials(mainly P3) during an auditory oddball paradigm, Be-reitschaftpotential, or readiness potential (RP), andcontingent negative variation (CNV). Only one type ofrecording was obtained during each recording session,with at least 2-h intervals between recording sessions.The movement accompanying potential (MAP), re-ferred to as ‘movement accompanying slow potential’in an earlier work [19] was defined as an averaged EEGpotential recorded during the motor tasks, increasing ordecreasing from the baseline with a slopeabove ± 10 µV/s in both the RP paradigm and in theCNV paradigm. The MAP potential started before orduring the movement-related EMG activity, and endedduring or after the EMG activity.

During recording sessions, the subject reclined on acomfortable chair in a calm, semi-darkened room. Foreach protocol, recordings were obtained from all leadsof the implanted electrode. Mutually connected ear-lobes were used as a reference. Resistance was keptbelow 5 kX, and was repeatedly checked during eachrecording session. Sweeps contaminated with artefactswere rejected both automatically and manually. Eachset of recordings during a given paradigm was repeatedat least twice.

Somatosensory evoked potentials (SEP), elicited bythe stimulation of the contralateral median or tibialnerves were obtained in all three patients. The mediannerve contralateral to the implanted electrode wasstimulated above its course on the wrist, and the tibialnerve was stimulated above its course behind the medialankle. In patient 2, the somatosensory evoked poten-tials were recorded from the stimulation of the medianand tibial nerves ipsilateral to the implanted electrode(the contralateral upper limb had been amputated inthis patient). In all examinations, 5 Hz, 0.1 ms square-wave pulses at an intensity 1.5 times higher than the

SEP, ERP, RP and CNV recordings in the human thalamus 255

motor threshold which evoked thumb twitching wereused. Somatosensory evoked potentials were recordedfrom all leads of the implanted electrode, with signalsfiltered between 10–3000 Hz and within a time win-dow encompassing 30 to 50 ms from stimulus onset.There was an average of two runs of 2000 artefact-freesweeps in each recording session. Each run was storedseparately in the computer memory and the runs weresuperimposed when analysed. The peaks were labelledin accordance with the nomenclature of Donchin et al.[6] and Celesia et al. [4]. The latencies of each peakwere measured from the beginning of the stimulus tothe highest level of a negative or a positive deflection.

The RP and CNV were recorded using slightly modi-fied classical protocols [14, 15, 18, 24]. Patients re-clined comfortably on a chair and were instructed tofocus on a given point in their visual field. In the RPparadigm, patients were instructed to perform a move-ment at will (self-paced). The movement was a palmarflexion of the hand or a plantar flexion of the foot.Given an analysis window of 5 s, the patients wereasked to maintain intervals of at least 10 s betweensuccessive movements, without counting the time. Thesignal was filtered with a 0.01–500 Hz bandpass. Sur-face EMG was recorded with a pair of cup electrodesplaced on the skin overlying the flexor digitorum com-munis or the gastrocnemius muscles. The recordingwas triggered by the EMG, and covered a period from3.5 s before the movement to 1.5 s after it. There wereforty to sixty trials, not including the records contain-ing artefacts.

In the CNV auditory-visual paradigm, two successivestimuli were delivered through earphones and throughLED goggles. The first stimulus was a tone (warningstimulus, S1, intensity 70 dB, frequency 1 000 Hz,duration 0.1 s) followed 3 s later by the second stimu-lus (imperative stimulus, S2) which was a sequence offlashes delivered through LED goggles (repetition rate5 Hz) and displayed to both eyes. Patients were in-structed to perform a unilateral movement of the con-tralateral hand or foot (pressing a response switch) assoon as possible after S2 in order to interrupt the S2sequence. Successive trials of the S1-S2 sequences weredelivered at random intervals of approximately 10 s,with a maximum interval of 12 s. The signal was fil-tered with a bandpass of 0.01–500 Hz, during a 5 sanalysis time. Recordings started 500 ms before the S1stimulus. For each recording, there was an average ofthirty trials, not including records containing artefacts.

The event-related P3 cognitive potential was elicitedusing the standard auditory oddball paradigm [4]. Thetarget (rare) and non-target (frequent) stimuli were75 dB binaural tones delivered through earphones andoccurring every 1 600 ms, with a frequency of 2 000and 1 000 Hz, respectively. The occurrence ratio was1:4, presented in random order with the some con-straint that at least two non-target stimuli should occurafter each target stimuli. Patients were instructed toignore the non-target frequent tones, and to detect andcount the target infrequent ones. Signals were recordedwithin a 0.1–70 Hz bandpass. Up to thirty responseselicited by the target tones were collected during onerecording trial. The analysis was made at the end ofrecording, and the peaks were labelled in accordancewith the published nomenclature [4].

We considered that the electrode was located close toa waveform generator in all the tests in which a gradualdecrease in potential amplitude was recorded. Simi-larly, a clear, high amplitude potential recorded at onlyone contact was considered to have been generated inthe vicinity of the electrode. Conversely, when poten-tials with unchanging amplitude were recorded on sev-eral consecutive leads, it was not possible to localize thegenerator. These latter recordings were considered tocorrespond to far-field potentials.

An artefact was defined as inappropriate, high ampli-tude signal at any given point in time which was notrepeated in neighbouring sweeps. If artefacts occurred,they were rejected both automatically and manuallyfrom all recorded traces.

RESULTS

Patient 1

First, an electrode was implanted on the left side dorso-ventrally passing through the nucleus posterolateralisand reaching the ventroposterolateral (VPL) nucleuswith its most deeply inserted lead (figure 1, T1). Thelocation of T1 in the VPL was confirmed both by theanalgesic effect of the stimulation, and by the exclusiveappearance of the thalamic median SEP componentP17/N22 (figure 1). The stimulation had to be discon-tinued on the 8th day of stimulation due to symptomsof mental depression.

A readiness potential (RP) contralateral to the move-ment was recorded at all contacts, with decreasingamplitude from the most superficial (T4) to the deepest(T1) contact (figure 1). The typical two-slope morphol-

256 I. Rektor et al.

ogy (BP I and BP II), like that of RPs on the scalp [13,19] was recorded on T4, and (less clearly) on T3. Theamplitude gradient from T4 to T2 showed that themost dorsal contact, located in the posterolateralnucleus (PL), was closer to the generator than theventral ones. The RP preceding ipsilateral movementswas of lower amplitude and showed only the second(BP II) slope. In contrast to the positive polarity of theRP, the polarity of the movement accompanying po-tential (MAP) was negative with the ipsilateral move-ment. MAP was also recorded in the VPL (T1), whereno RP to ipsilateral movements appeared.

The distribution of the CNV was identical to that ofthe RP. The CNV was recorded with decreasing posi-tivity from T4 to T2. The CNV occurred irregularly oncontact T1 in the VPL (figure 1), and was considered tobe generated outside this nucleus, most probably in thePL. Potentials evoked by both S1 (auditory) and S2

(visual) stimuli in the CNV paradigm occurred in allcontacts, including the VPL, again with the lowestamplitude in the VPL. In the ERP oddball paradigm, asimilar amplitude gradient of the N1/P2/N2 complexand of the P3 wave was displayed. A P3, probably of theP3a type, with a latency of 251 ms, was recorded on T4and T3; it behaved as a ‘far-field’ potential in T1 (theVPL) (figure 1).

Patient 2

The five-lead electrode was inserted in the right hemi-sphere, passing through the VPL into the pulvinar. Thestimulation was performed because of phantom pain inthe left upper arm, which had been amputated follow-ing an injury 7 years before. The effects of the stimula-tion confirmed the position of the most superficial lead(T5) and possibly of T4 in the hand area of the VPL.

Figure 1. Patient 1. Left: position of the electrode in the thalamus in an overprojection of the CT scan on the stereotactic atlas (Talairach andTournoux, 1988). An arrow indicates the position of the electrode. Va: nucleus ventralis anterior, Vl: nucleus ventrolateralis, Vpl: nucleusventroposterolateralis, Vpm: nucleus ventroposteromedialis, dm: nucleus dorsomedialis, P: pulvinar, Pu: putamen, Gp: globus pallidus. Right:recordings in patient 1. Electrode contacts: T1 - nucleus ventroposterolateralis (VPL); T3, 4 - nucleus posterolateralis (PL). A: Somatosensoryevoked potentials, stimulation of the contralateral median nerve. Sensitivity: 2 µV/division, time base: 3 ms/division. B: Contingent negativevariation, auditory-visual paradigm. Arrows indicate S1 and S2 stimuli. Sensitivity: 20 µV/division, time base: 0.5 s/division. C: ERP P3 wave,auditory oddball paradigm. Sensitivity: 10 µV/division, time base: 50 ms/division. D: Readiness potential paradigm, contralateral to theself-paced movement. Sensitivity: 20 µV/division, time base: 0.5 s/division. P: positivity, N: negativity.

SEP, ERP, RP and CNV recordings in the human thalamus 257

The SEPs evoked by contralateral tibial stimulationwere present in all five leads, and their amplitude gra-dient from T5 to T1 strongly supported the location ofT5 in the VPL. The tests, which included movement,were performed with the ipsilateral upper and lowerlimbs and on the contralateral lower limb. No RP wasobserved preceding the movements of the ipsilateralhand and foot, however, a far-field potential MAP wasrecorded. The RP appeared preceding the movementsof the contralateral foot, as a positive two-slope phe-nomenon lasting 1100 ms (600 and 500 ms). The lackof amplitude gradient indicated that the RP was notgenerated in the region where the recording contactswere located. This is in accordance with the fact that thecontacts were placed in the region of the absent upperlimb, while RP was elicited by lower limb activity. Thisfar-field RP may have been generated in the nearbyVIM. MAP occurrence was similar to that observed foripsilateral limbs. The SEP of the stimulation of thecontralateral lower limb showed the shape typicallyseen in the VPL nucleus recordings. The CNV waspresent as a triphasic complex. In the first phase, apositive wave peaking at about 500 ms appeared afterthe S1 stimulus, which was considered as the longlatency evoked potential. It was followed by a long-lasting negative wave with irregular peaks superim-posed on it, that lasted between 900 and 1 065 ms afterS1. The following positive wave, lasting 1500 ms, wasflat and less pronounced. This CNV was considered asa far-field potential because of the absence of clearamplitude gradient. In the CNV protocol, the evokedpotential to the S1 (auditory) stimulus was recorded asa far-field potential, while the response to imperative S2(visual) stimulus showed decreasing amplitude in con-secutive plots, suggesting a near-field generation. Thiswas tentatively explained by the proximity of the opticarea in the pulvinar.

In the ERP standard auditory oddball paradigm, aP3a-type wave [2, 3, 8-10] with a latency of 267 ms wasoccasionally recorded from the electrode contact in theVPL nucleus. However, this finding could not be fur-ther reproduced. Conversely, clearly distinguishableN1, P2 and N2 ERP components were present at allleads during all recordings.

Patient 3

The electrode of this patient reached the thalamusthrough contact T2 and perhaps T1. Contacts T3–T5were most probably located in the internal capsule

(figure 2). Stimulation was performed because ofchronic pain following a putaminal stroke. Contralat-eral median nerve stimulation revealed biphasicpositive-negative SEPs peaking at 14 ms on T2, and anegative 17 ms wave with the highest amplitude beingrecorded on T1. Far-field SEPs, elicited by the stimu-lation of the ipsilateral median nerve, were recorded atall contacts. The RP and CNV could be tested onlywhen moving the ipsilateral side (because of contralat-eral hemiplegia). On the VPL-located T2 contact, nei-ther RP nor MAP appeared. The CNV was not clearlyreproducible in this patient. Potentials evoked by S1and S2 stimulation respectively appeared with oppositepolarity. Given the time-scale constraints of the CNVprotocol, components occurring earlier than 100 mscould not be recognized. The amplitude gradients ofboth the auditory N100 and the visual N100 wererecorded with highest amplitude ‘outside’ the VPL, onthe T3 lead. In the neighbouring VPL, however, bothevoked potentials showed very different shapes. Onlysmall early waves (peaking at about 170 ms) were re-corded on T1 and T2, contrary to the large slow wavespeaking at approximately 300 ms, which were highestin the VPL and decreased from T2 to T5 (figure 2). NoP3 potentials were recorded in the standard auditoryoddball paradigm.

DISCUSSION

Cortical studies of various electrical potentials havebeen performed in several laboratories. We have alreadypublished the results of cortical recordings of RP, CNVand MAP through depth electrodes [15, 18-21]. Cor-tical recordings have also been performed through sub-dural electrodes [11]. However, little is known on theelectrical potentials in the thalamus, particularly in theposterior thalamus, of human subjects. There havebeen some reports on SEP [22] and P3 recordings [13]from posterior thalamic structures. The RP was testedby Fève [7] with a negative result in the VPL of onepatient. However, we are not aware of other informa-tion concerning recordings from the posterior thala-mus.

We performed electrophysiologal studies in severalregions in the posterior thalamus, namely the VPL, thePL, and the pulvinar nuclei, as well as from the internalcapsula in the vicinity of the ventral anterior (VA)nucleus. The different paradigms used enabled us tostudy the simple low-level information processing (SEPand middle latency evoked potentials in the CNV

258 I. Rektor et al.

protocol) as well as some functions linked with motorregulation and cognitive functions (RP, CNV, MAP,P3). The intracerebral potentials occurred with bothpositive and negative polarities, probably due to thevariable position of the electrode contact relative to thedipole generator. It is known that intracerebrally both‘positive’ and ‘negative’ locations exist for many poten-tials; for example, the contingent ‘negative’ variationcan be positive with depth electrode recording. Weconcentrated our attention on potentials which re-vealed a steep amplitude gradient on consecutive con-tacts of an electrode. Such a gradient indicated that therecording potential was generated in the vicinity of theelectrode.

The results obtained from VPL recordings were inconcurrence with its established role as a relay nucleus,processing somatosensory information to the primarysomatosensory cortex. The neurons in the VPL havestrong functional connections with SI neurons in theIIIb and IV layers [12]. In our testing, the VPL gener-ated the ‘thalamic’ SEP, which was the only potentialregularly recorded in this nucleus. The VPL is an estab-lished source of SEPs [1, 4], even if thalamic responsesmay not have clear scalp counterparts [5, 16]. The

MAP, which was present in the VPL in all but one ofthe recordings, clearly behaved as a far-field potential,without amplitude gradient among recording plots. Itwas probably generated in some neighbouring struc-ture, due to its relatively high amplitude. The absenceof the RP and the presence of the MAP in the VPL werealso observed in one recorded patient by Fève [7]. Noevidence of VPL involvement in cognitive activitieslinked with the RP, the CNV or the ERP protocols wasobserved in our recordings. In the recordings frompatient 3, both auditory and visual evoked potentialsobtained in the CNV protocol, peaking at approxi-mately 300 ms, were most probably generated in theVPL, while the earlier components were generated else-where. This could mean that the role of VPL is morecomplex than that of a simple relay in the somatosen-sory system, at least in some protocols. However, thisisolated observation should obviously be replicated inthe future.

Pulvinar recordings were obtained from one patientonly. A visual evoked potential generator is not surpris-ing in this structure strongly involved in the visualfunction. No other generator of the waves studied wasfound in the pulvinar.

Figure 2. Patient 3. Left: position of the electrode in the thalamus in an overprojection on the CT scan on the stereotactic atlas (Talairach andTournoux, 1988). An arrow indicates the position of the electrode. Va: nucleus ventralis anterior, Vl: nucleus ventrolateralis, Vpl: nucleusventroposterolateralis, Vpm: nucleus ventroposteromedialis, dm: nucleus dorsomedialis, P: pulvinar, Pu: putamen, Gp: globus pallidus. Right:Contingent negative variation, auditory-visual paradigm. Electrode contacts T2 (T1): nucleus ventroposterolateralis (VPL), T3, 4, 5: internalcapsule. The steep voltage gradient of the auditory evoked potential component and of the visual evoked potential component (the highestamplitude on the electrode contact T3 is indicated by thick arrows) is visible. Thin arrows indicate S1 and S2 stimuli. Sensitivity: 50 µV/division,time base: 0.5 s/division. P: positivity, N: negativity.

SEP, ERP, RP and CNV recordings in the human thalamus 259

Interesting results were obtained from the nucleusposterolateralis (PL), a region located just dorsally tothe VPL. The PL, which is connected with parietalassociative cortex areas 5 and 7 [23], generated severalelectrophysiological potentials in tests involving cogni-tive operations. Thus, oddball P3-like, CNV, and RP-type potentials appeared to be generated here. It seemsfrom our recordings that there is a generator of RPs inthe PL that is more active with contralateral movementthan with ipsilateral movement (the amplitude of theipsilateral RP was lower and it consisted only of thesecond slope). The RP generator is not identical to theMAP generator, and the MAP recorded here was prob-ably a far-field potential. The local recording of near-field middle and long latency potentials evoked byauditory and visual stimuli suggests that the PL alsoprocesses various modalities of afferent information.This structure, involved in the processing of afferentinformation as well as of cognitive operations, couldtherefore play a role in the integrative operations ofbrain networks, and should be a candidate for furtherexplorations.

It is interesting to note that the posterior thalamusappeared to contribute activity relevant to brain poten-tials related to a number of cognitive functions. Thesources of these potentials are attributed mostly to thecortex; for example, ‘movement related ‘cortical’ poten-tial’ is sometimes used as a synonym for readinesspotential. Nevertheless, it is evident from our record-ings that these potentials are not exclusively a corticalphenomenon. The origins of RPs, CNV and P3 arecortical as well as subcortical. Although more data areneeded to highlight the role of posterior thalamic struc-tures in functions other than somatosensory processing,our results indicate that there are structures in theposterior thalamus which are involved in the handlingof multisensory information, and which may play a rolein motor regulation and in cognitive operations.

ACKNOWLEDGEMENTS

We wish to express our thanks to Ing. Pavel Daniel forphotographic assistance and Anne Johnson for gram-matical assistance. This study was supported by MSCR112801 Research Programme and Yamanouchi Euro-pean Foundation.

REFERENCES

1 Arezzo J, Legatt AD, Vaughan Jr HG. Topography and intrac-

ranial sources of somatosensory evoked potentials in the mon-key I. Early component. Electroencephalogr Clin Neuro-physiol 1979 ; 46 : 155-72.

2 Baudena P, Halgren E, Heit G, Clarke JM. Intracerebralpotentials to rare target and distractor auditory and visualstimuli: III. Frontal cortex. Electroencephalogr Clin Neuro-physiol 1995 ; 94 : 251-4.

3 Brázdil M, Rektor I, Dufek M, Daniel P, Jurák P, Kuba R. Therole of frontal and temporal lobes in visual discrimination task– depth ERP studies. Neurophysiol Clin 1999 ; 29 : 339-50.

4 Celesia GG. Somatosensory evoked potentials recorded di-rectly from human thalamus and SM I cortical area. ArchNeurol 1979 ; 36 : 399-405.

5 Desmedt JE, Cheron G. Non-cephalic reference recording ofearly somatosensory potentials to finger stimulation in adult oraging normal man: differentiation of widespread N18 andcontralateral N22 from the prerolandic P22 and N30 compo-nents. Electroencephalogr Clin Neurophysiol 1982 ; 52 :553-70.

6 Donchin E, Callaway E, Cooper R, Goff RW, Desmedt JE,Hillyard SA, et al. Publication criteria for studies of evokedpotentials in man. In: Desmedt JE, Ed. Attention, voluntarycontraction and event- related potentials. Progr Clin Neuro-physiol. Karger: Basel; 1977. p. 1-11.

7 Fève A. Origine sous-corticale des potentiels pré-moteurs(movement-related potentials) chez l’homme [thèse de doc-torat]. Paris: Université Paris VI; 1993.

8 Gybels J, Kupers R, Nuttin B. Deep brain stimulation in thetreatment of chronic pain in man: where and why? Neuro-physiol Clin 1990 ; 20 : 389-98.

9 Halgren E, Baudena P, Clarke JM, et al. Intracerebral poten-tials to rare target and distractor auditory and visual stimuli: I.Superior temporal plane and parietal lobe. ElectroencephalogrClin Neurophysiol 1995 ; 94 : 191-220.

10 Halgren E, Baudena P, Clarke JM, et al. Intracerebral poten-tials to rare target and distractor auditory and visual stimuli. II.Medial, lateral and posterior temporal lobe. Electroencepha-logr Clin Neurophysiol 1995 ; 94 : 229-50.

11 Ikeda A, Luders HO, Burgess RC, Shibasaki H. Movement-related potentials recorded from supplementary motor areaand primary motor cortex. Brain 1992 ; 115 : 1017-43.

12 Johnson MJ, Alloway KD. Cross-correlation analysis revealslaminar differences in thalamocortical interactions in the so-matosensory system. J Neurophysiol 1996 ; 75 : 1444-57.

13 Kropotov JD, Ponomarev VA. Subcortical neuronal correlatesof component P300 in man. Electroencephalogr Clin Neuro-physiol 1991 ; 78 : 40-9.

14 Kornhuber HH, Deecke L. Hirnpotentialänderung beimMenschen vor und nach Willkurbewegungen, dargestellt mitMagnetbandspeicherung und Rückwärtsanalyse. PflugersArch 1965 ; 281 : 52.

15 Lamarche M, Louvel J, Buser P, Rektor I. Intracerebral record-ings of slow potentials in a contingent negative variationparadigm: an exploration in epileptic patients. Electroen-cephalogr Clin Neurophysiol 1995 ; 95 : 268-76.

16 Mauguiére F, Desmedt JE, Courjon J. Astereognosis and dis-sociated loss of frontal or parietal components of somatosen-sory evoked potentials in hemispheric lesions. Detailed corre-lations with clinical signs and computerized tomographyscanning. Brain 1983 ; 106 : 271-311.

17 Milliken GW, Stokic DS, Tarkka IM. Sources of movement-related cortical potentials derived from foot, finger, and mouthmovements. J Clin Neurophysiol 1999 ; 16 : 361-72.

18 Rektor I, Fève A, Buser P, Bathien N, Lamarche M. Intracere-bral recording of movement related readiness potentials: an

260 I. Rektor et al.

exploration in epileptic patients. Electroencephalogr ClinNeurophysiol 1994 ; 90 : 273-83.

19 Rektor I, Louvel J, Lamarche M. Intracerebral recording ofpotentials accompanying simple limb movements. A SEEGstudy in epileptic patients. Electroencephalogr Clin Neuro-physiol 1998 ; 107 : 277-86.

20 Rektor I. Parallel information processing in motor systems:Intracerebral recordings of readiness potential and CNV inhuman subjects. Neural Plasticity 2000 ; 7 : 65-72.

21 Shibasaki H, Barret G, Halliday E, Halliday AM. Corticalpotentials associated with voluntary foot movement in man.Electroencephalogr Clin Neurophysiol 1981 ; 52 : 507-16.

22 Saira T, Amano K, Kawamura H, Tanikawa T, Kawabatake H,Notani M, et al. Short latency somatosensory evoked poten-tials – direct recording from the human midbrain and thala-mus. Acta Neurochir 1987 ; 39 : 170-3.

23 Talairach J, Tournoux P. Co-Planar Stereotactic atlas of thehuman brain. Stuttgart: Georg Thieme Verlag; 1988.

24 Walter WG, Cooper R, Aldridge VJ, McCallum C, Cohen J.The contingent negative variation: an electro-cortical sign ofsensorimotor association in man. Electroencephalogr ClinNeurophysiol 1964 ; 17 : 340-1.

SEP, ERP, RP and CNV recordings in the human thalamus 261