Habituation of the visually evoked potential and its vascular response: implications for...

NeuroImage 17, 1–18 (2002)doi:10.1006/nimg.2002.1177

Habituation of the Visually Evoked Potential and Its VascularResponse: Implications for Neurovascular Coupling in the Healthy Adult

Hellmuth Obrig, Heike Israel, Matthias Kohl-Bareis, Kamil Uludag, Rudiger Wenzel,Bianca Muller, Guy Arnold, and Arno Villringer

Neurologische Klinik, Charite, Humboldt Universitat zu Berlin, 10098 Berlin, Germany

Received November 29, 2001

In the human, visually evoked potentials (VEP)and cerebral oxygenation changes, as measuredby near-infrared spectroscopy (NIRS), are assessedto elucidate the relation between electrophysiologi-cal and vascular responses to a checkerboard stim-ulus reversing at 3 Hz. Habituation of either re-sponse is analysed on two time scales. Within the1-min stimulation period we find a decrease inP100N135-component amplitude, closely coupled toa decrease in the amplitude of the oxygenationparameters (concentration changes in oxygenatedand deoxygenated haemoglobin, [oxy-Hb] and[deoxy-Hb]). The N75P100-component amplitude ex-hibits a different behaviour along the 1-min stimu-lation period. An initial increase is overridden byan overall decrease, the latter not reaching statis-tical significance. The analysis across the 13 suc-cessive stimulation blocks separated by resting pe-riods of equal duration yields a trend for andecrease in the VEP-components’ amplitude, not re-flected in the vascular response. When calculating aratio between the amplitude of the P100N135-component and the concentration changes in thehaemoglobins a “coupling index” of a 0.2 �M de-crease in [deoxy-Hb] and an increase of 0.6 �M in[oxy-Hb] is found per 1 �V increase in VEP-compo-nent amplitude. The ratio is the same irrespectiveof its assessment from the difference stimulation/rest or from the habituation effect, i.e., the differ-ence between the amplitudes at the beginningand towards the end of the stimulation period. Al-though supporting the notion that the coupling be-tween neuronal activation and the vascular re-sponse exhibits linear aspects, our findings cannotbe taken as a proof of such a linearity across allbrain regions and activation types. On the contrary,tentatively calculating a coupling index for thedata assessed in the visual system, we intend to stressthe necessity to assess both neuronal and vascularresponse to allow for a comparison between differentsystems and conditions in whom neurovascular

coupling is expected to be altered (Miller et al., 2001;Mechelli et al., 2001). © 2002 Elsevier Science (USA)

INTRODUCTION

Investigating “the brain at work” can rely on theneuronal and the vascular response to a stimulation(Villringer and Dirnagl, 1995). Scalp recorded poten-tials dominated the noninvasive investigations in thehuman adult until the advent of modern vascular im-aging techniques, today clearly the most widely usedapproaches for studies in functional anatomy of thehuman brain (Rosen et al., 1998). The shift in method-ological preference has likewise shifted the connotationof the term “activation” from a primarily electrophysi-ological understanding to defining cerebral activationby the vascular imprint in cortical areas as detected bypositron emission tomography (PET) or functionalmagnetic resonance imaging (fMRI) (Raichle, 1998).The assumption of a close coupling between the neuro-nal excitation and the changes in blood flow in thesame cortical area may have proven valid in the fun-damental sense that areas identified to play a role byinvasive electrophysiological studies in primate modelscorrespond to those in the human as shown by thefunctional imaging techniques, founded on changes inregional cerebral blood flow (rCBF) (van Essen et al.,2001). Also besides this general functional–anatomicalconcordance and despite the clearly lower temporalresolution, BOLD contrast fMRI has been successful totrack sequential activation of different cortical areas ingood agreement with previously reported ERP findings(Wildgruber et al., 1997). There are, however, a num-ber of unresolved questions, one of which is by whichmodel the translation from the neuronal activation tothe vascular response can be best described (Boynton etal., 1996). In a broader sense this is the question as tothe linearity of the vascular response, of BOLD-con-trast changes and of other measures of rCBF (Mechelliet al., 2001), to the underlying neuronal activity. More

1 1053-8119/02 $35.00
© 2002 Elsevier Science (USA)
All rights reserved.

specifically this also touches the question of habitua-tion, upon which we here focus. Finding the BOLDcontrast to decay during ongoing stimulation with cer-tain paradigms (Hathout et al., 1994; Condon et al.,1997) brought up the issue whether the vascular re-sponse shows a “habituation” independent of the neu-ronal input over time (i.e., responds nonlinearly overtime). The issue has been controversially discussed andyielded a theory suggesting changes from oxidative tononoxidative metabolism during ongoing stimulation,readily explaining the increase in lactate found duringprolonged activation of neuronal tissue (Frahm et al.,1996; Magistretti, 2000). At present the controversyprobably has given way to the notion of both linear andnonlinear aspects of the BOLD-contrast (Kruger et al.,1999), nonetheless a principal weakness of many re-lated studies is that neuronal input is assumed con-stant. This may be surprising since habituation ofevent-related potentials (ERP) has been extensivelystudied in the human since the basic mechanisms havebeen described in Aplysia by the Kandel and coworkers(Pinsker et al., 1970). Proceeding from electrophysio-logical work (Wastell and Kleinman, 1980), elucidatingthe time course of the reduction in ERP amplitudemore recent work has even highlighted deficient habit-uation in pathological conditions, especially migraine(Schoenen et al., 1995). Assuming that habituation is aphysiological process preventing stimulus overload im-aging techniques have been used to identify brainstructures governing habituation (Poellinger et al.,2001; Fischer et al., 2000).

The reason why neuronal input has not been rou-tinely assessed in studies concerning aspects of linear-ity is twofold. On the one side there is no electro-physiological standard by which neuronal input can bedefined. For noninvasive studies in the human, scalprecorded ERPs represent a rather blurry net-effect ofexcitatory and inhibitory processes in the cerebralcortex. For invasive primate models spike rate hasbeen related to the vascular response (Rees et al., 2000;Heeger et al., 2000), but recently it has been shownthat field potentials may be a better measure forneuronal activity eliciting the vascular response (Logo-thetis et al., 2001). On the other side there are meth-odological limitations. The registration of electro-physiological responses in the magnet are feasible(Bonmassar et al., 2001), but signal to noise ratio(SNR) is decreased by the simultaneous registration.Concerning PET the technical limitations may belesser; however, PET’s temporal resolution, rather onthe order of minutes, restrains potential correlationanalyses, besides PET’s limitations as to repetitivemeasurements in an individual (Sadato et al., 1998).

We here present a combined approach using nearinfrared spectroscopy (NIRS) to monitor changes incortical haemoglobin oxygenation and a standard EEGset-up to simultaneously measure VEP amplitudes.

The approach is directed towards a comparatively un-demanding assessment of neurovascular coupling. Themajor problem of either method is that of exactly de-fining the cortical structures generating the VEP andthe volume sampled, to which the changes in oxygen-ation changes can be attributed. Aware of these limi-tations we define a coupling index relating the magni-tude of the VEP amplitude in �V to the changes in thehaemoglobins’ concentrations in �M. This may be con-sidered too confident in the reliability of either mea-surement; however, the rising interest in correlatingelectrophysiological measures to parameters based onthe vascular response (spike rate, field potential/BOLD (Heeger et al., 2000; Logothetis et al., 2001),photic driving/TCD-flow velocity (Diehl et al., 1998))has encouraged us to operationally define such an in-dex. We invoke to control for neuronal input whenissues of neurovascular coupling are investigated,while we aim to further elucidate the versatility of suchan index, and test its applicability also under patho-physiological conditions, as has been reported in a pre-liminary form for migraine patients (Israel et al.,2000).

MATERIALS AND METHODS

Fifteen volunteers were examined (mean age 26years SD 2.1 years; 8 female, 7 male) none of whom hada history of a neurological or other medical disorder.All subjects had normal or corrected to normal vision.Special care was taken to exclude migraine sufferers.The protocol was agreed on by the local ethics commit-tee and volunteers were financially rewarded for theirparticipation.

For the EEG measurements 5 channels of a 16-chan-nel EEG (Schwarzer, Munchen) were used. Data wererecorded from Oz, O1, and O2 and referenced to Fz

(electrode positions according to international 10–20-system). EOG (electrooculogram) was registered butnot used for further artefact correction. Data weresampled at a rate of 1 kHz, applying an online band-pass filter between 0.3 and 70 Hz, and stored for offlineanalysis.

For the optical measurements a frequency domainspectrometer (ISS Inc., U.S.A.) was used allowing forthe detection of intensity and phase changes (modula-tion frequency 110 MHz) at 4 wavelengths (750, 780,810, and 830 nm) in two independent probe positions.The light-emitter to -detector distance was 3.0 cm andthe two probe pairs were centered around O1 and O2

(see Fig. 1). Data were sampled at 5 Hz and stored foroffline analysis.

The stimulus consisted of an annular black andwhite checkerboard (compensating for cortical magni-fication) displayed on a computer monitor (17 inches)at a distance of �1 m (Fig. 1). Reversion rate was �3Hz. It was triggered by an impulse generator, whose

2 OBRIG ET AL.

impulses were co-registered on the EEG and the opti-cal recordings for synchronisation in the offline analy-sis. Trigger frequency slightly varied between subjects(3.1 Hz � 0.05). During the resting periods a blankscreen with the fixation cross was displayed.

Subjects were comfortably seated in an EEG-chair ina quiet room. Room light was dimmed and the opticalprobes were covered by black cloth to avoid ambientlight. Fixation was maintained throughout the exper-iment. The experiment was stopped when the subjectbecame tired or encountered problems to maintain fix-ation. Blocks of 180 reversals (� 1 min) alternated withresting periods (blank screen) of the same duration (seeFig. 2). Subjects completed 12–15 of such stimulation-rest cycles (overall experimental duration �30 min).

ANALYSIS

EEG Data

EEG data were digitally filtered offline with a low-pass Butterworth filter at 30 Hz (filter power 4). Next,the data were averaged across 15 and 180 successivetriggers respectively (VEP15 and VEP180). The restingperiods were analysed in the same fashion to obtainbaseline values. The time courses of the VEP-compo-nent amplitudes thus consist of 12 values for eachstimulation and each resting period (VEP15), or 1 valuewhen 180 reversals were averaged (VEP180). Also thegrand averages across all responses in a single subjectwere assessed (VEPall). Since only subjects with a nor-mal P100 latency were included and latencies are nofocus in this paper the P100 peak was set to 100 ms.The latencies of the two other components were alsowithin the normal range (N75: 70.3 ms � 3.6; N135:139.2 ms � 6.2; mean � SD). Since the focus of thepresent paper is on the amplitude and its change overtime the following values were assessed in each subjectand each epoch:

N75 P100: difference between maximum at 80–120 ms andminimum at 60–90 ms.

P100 N135: difference between maximum at 80–120 ms andminimum at 110–190 ms.

The temporal windows for the individual compo-nents were adopted from a related study (Afra et al.,1998); however, a slight narrowing of the differentwindows (as used by other groups) did not qualitativelychange the results.

Both values were also calculated as percentage of themean amplitude across all trials in the respective sub-ject. Finally, a fitting procedure was applied where theindividual VEPall was linearly fitted into the successiveepoch averages (VEP15/VEP180) in the following tempo-ral windows, yielding a normalised time course of theVEP amplitude in the individual volunteer (comparewith (Guy et al., 1999)):

from 60–120 ms (corresponding to N75 P100),from 80–190 ms (corresponding to P100 N135)and from 40–210 ms (as a measure of all three VEP

components).

These different procedures are illustrated in Fig. 3.In summary, it was our goal to define different mea-sures to test for a habituation of the VEP amplitude ontwo different time-scales (1) within the individual 1min stimulation blocks (VEP15-epochs) and (2) over thesuccessive stimulation blocks (VEP180).

Vascular Parameters

The phase and intensity changes acquired by NIRSwere converted into concentration changes of oxygen-ated and deoxygenated haemoglobin (�[oxy-Hb] and�[deoxy-Hb]) according to a previously described mul-tiple discrete wavelengths algorithm (Uludag et al.,2001). Generally, the NIRS approach in functional ac-tivation studies has been multiply used in previouswork of our and other groups (for review see (Obrig andVillringer, 1996)). Principally, the method relateschanges in light absorption to changes in chromophoreconcentration in the illuminated tissue. Due to therelative transparency of biological tissue to light in thenear-infrared (especially between 600 and 950 nm), thesampling volume reaches down to the cerebral cortexeven in human adults. The physical assumption is aconstant optical path length during the measurement,allowing for the application of a modified Beer–Lam-bert law to calculate and quantify concentrationchanges. The physiological assumption is that the hae-modynamic changes induced by the functional stimu-lation take place in the intracerebral layer of the sam-pling volume and are negligibly confounded by changesin extracerebral haemodynamics. Based on these as-sumptions NIRS has been shown to reliably detect ahaemodynamic response over an activated corticalarea typically consisting of an increase in [oxy-Hb] anda decrease in [deoxy-Hb]. The latter corresponds to theBOLD-contrast increases in response to functional ac-tivation as monitored by T2* weighted fMRI (Klein-schmidt et al., 1996).

In the present study we used a monitor providingdata on changes and in intensity (that is: attenuation,�I) and in phase (a measure of the time of flight of thephotons, ��). The concentration changes in the hae-moglobins were therefore assessed based on either pa-rameter. Theoretically, the ��-derived changes aremore sensitive for changes in the deeper layers; how-ever, the signal to noise ratio is smaller than for the�I-based values (Hemelt and Kang, 1999) (Kohl et al.,2002). We do not focus on the issue in this paper, butreport the concentration changes in [oxy-Hb] and [de-oxy-Hb] based on either procedure and discuss poten-tial implications below (see Discussion). Yet anothermethodological issue beyond the scope of the present

3HABITUATION OF VEP AND ITS VASCULAR RESPONSE

paper needs mentioning. Besides the changes in thehaemoglobins, cytochrome oxidase exhibits a differen-tial spectrum in the near-infrared. Therefore changesin its redox state can also be monitored by the meth-odology (Heekeren et al., 1999a; Springett et al., 2000).Since the change in Cyt-ox redox-state to physiologicalstimulation in the human adult is still under debateand the here used system does not use optimal wave-lengths for the assessment of Cyt-ox, we restrict theanalysis to the haemoglobins, the assessment of whichis not controversial.

The following paragraph describes the differentsteps of the time course and amplitude analysis in�[oxy-Hb] and �[deoxy-Hb]. First, the parameters,sampled at 5 Hz, were digitally resampled to the indi-vidual stimulus frequency (�3 Hz) to allow for better

comparison with the VEP amplitudes assessed. Tominimize artefacts from slow hemodynamic changes,most probably unrelated to the stimulus, the timecourse of the changes within each stimulation/rest cy-cle were linearly detrended. More refined methods ofbaseline correction such as high-pass filtering or high-er-order polynomial fit subtractions were tested, butbear the undesirable danger of introducing a change inresponse magnitude across the different cycles. Next,

FIG. 1. Experimental set-up. Two pairs of optical probes arefixed on the occiput of the subject at a source-detector distance of 3cm. At the midpoint of the probe pairs (O1,O2) and at between the twopairs (O2) electrodes are placed, referenced to another electrode at Fz.The annular checkerboard is presented on a 17-inch computer screenat �1 m distance.

FIG. 2. Stimulation protocol. At least 13 cycles of �1 min stim-ulation blocks, corresponding to 180 checkerboard reversals at �3Hz, separated by equally long resting periods were acquired in eachsubject.

FIG. 3. Assessment of the response amplitude of the VEP-com-ponents (a), and the oxygenation changes (b), exemplified for the[deoxy-Hb] response. (a) The left graph demonstrates the fittingprocedure showing the grand average (thick line), the average across180 reversals (medium line) and the average across 15 reversals(thin line) in a single subject. The three graphs on the right illustratethe different temporal windows, in which the fitting procedure andthe assessment of the absolute component amplitude (differencebetween maximum and minimum) were performed. (b) Fitting pro-cedure to assess the magnitude of the oxygenation response in asingle stimulation block (broken line) relative to the average acrossall 13 stimulation blocks (solid line).

4 OBRIG ET AL.

FIG. 4. Grand average across 12 volunteers. The upper graph shows the vascular response, i.e., the increase in [oxy-Hb] and decreasein [deoxy-Hb] in response to the checkerboard stimulation (grey symbols denote the respective ��-based changes, bars indicate stimulusonset). The lower two graphs show the VEP15 magnitudes as assessed by the difference between maximum and minimum in the respectivewindow (compare with Fig. 2). The fact that during resting period the amplitude of the two components does not equal zero is due to theanalysis procedure and reflects “baseline” noise.


an average across all subjects (Fig. 4) and the averagetime course across all cycles in all subjects (Fig. 5) wereassessed. To determine a potential attenuation of thevascular response across the stimulation-blocks, ineach volunteer the average response was fitted into theresponse in the individual block resulting in a vascularmeasure comparable to the VEP180 (see Fig. 3). Toexamine within block habituation of the vascular re-sponse data were resampled at the interval of 15 re-versals (�4.8 s) resulting in a vascular measure com-parable to the measure obtained for the VEP15-epochs(Fig. 3). To clarify the terminology chosen in thepresent paper: A cycle consists of 1 stimulation block(i.e., VEP180 or mean across �1 min of the NIRS-pa-rameter) and 1 resting period. The stimulation block issplit up into epochs of 15 reversals (i.e., VEP15 corre-sponding to means across �4.8 s of the NIRS-parame-ter).

Coupling Analysis

The design of the present study investigates the ha-bituation effects caused by repetitive checkerboardstimulation on two different time scales (within blockand across blocks). Since electrophysiological and vas-cular responses were measured simultaneously thereis the option to also analyse their relation determinedby the so-termed neurovascular coupling. Though ei-ther method exhibits specific problems concerningquantification and localisation, we consider it justifiedto operationally define a ratio between the two param-eters, the numeric value of which may be rather pre-liminary due to the above mentioned uncertainties con-

cerning absolute magnitudes of vascular andelectrophysiological parameters. The here introducedcoupling index (CI) is defined by the ratio:

CI ��(vascular parameter)

�(VEP�component amplitude)(1)

As the vascular parameter is measured in units of�M and the VEP in �V, the CI is expressed in units of�M/�V. The ratio implies a linear relationship betweenthe respective vascular and electrophysiological pa-rameter. We cannot definetly test this hypothesis bythe present experimental design. However, the habit-uation effect found in the electrophysiological and thevascular response within the 1-min stimulation periodoffers a rough check of the assumption. In case theassumption is correct we expect a decrease of 1% inVEP-component amplitude to result in a 1% decrease/increase in [oxy-Hb] and [deoxy-Hb] respectively. Theanalysis therefore comprised three steps.

(1) The mean amplitudes of the vascular parametersand the N75P100 / P100N135 components were calcu-lated. All parameters were checked for habituationwithin the 1 min stimulation block (see Table 1).

(2) All responses were normalised to their respectivemean amplitude (i.e., the mean amplitude across allstimulation blocks was set 100%, for the vascular re-sponse. For the electrophysiological response the fit-ting procedure results in normalised values, the VEPall

being set to 100%). The habituation expressed in per-centage is given in Table 2.

TABLE 1

Subject

Intensity �[. . .]/�M Phase �[. . .]/�M Maximum–minimum/�V

Oxy � Deoxy � Oxy � Deoxy � N75P100 � P100N135 �

as_74 0.62 �0.14 �0.21 0.05 1.30 �0.49 �0.45 0.17 5.7 �1.7 11.4 �3.3**cb_77 0.70 �0.07 �0.19 �0.01 1.11 �0.22* �0.35 0.07 17.2 3.8 16.9 �1.2dl_73 0.84 �0.02 �0.19 0.06$ 1.30 �0.11 �0.33 0.19 6.5 �0.7 11.9 �4.8**dp_75 1.06 �0.09 �0.45 0.05$ 1.40 �0.24$ �0.63 �0.01 7.9 0.1 8.3 �2.9$ik_78 0.71 �0.12* �0.18 0.04** 1.28 �0.18 �0.44 0.07 12.0 1.8 16.1 �4.5**is_73 1.05 �0.36* �0.49 0.27** 1.39 �0.40** �0.62 0.20* 8.7 1.2 15.9 �4.9*iw_74 0.87 �0.46** �0.27 0.04 1.32 �0.74** �0.46 0.30$ 5.4 0.5 8.4 �0.2kn_76 0.16 �0.07 �0.09 0.12* 0.13 �0.05 �0.07 0.07 6.6 1.9 13.1 �2.0**ob_74 0.59 �0.63* �0.37 0.05 1.36 �0.93** �0.66 0.29$ 5.8 3.8 12.4 �3.1**ss_76 0.38 �0.29$ �0.15 0.08$ 0.35 �0.20 �0.12 0.09 6.9 1.1 9.7 �0.4su_76 0.88 �0.19 �0.46 0.03 1.18 �0.10 �0.59 �0.01 8.3 �0.2 12.4 �4.0**wc_73 1.01 �0.27$ �0.30 0.11** 1.47 �0.41* �0.52 0.11$ 12.8 0.4 19.5 �5.7*

Mean 0.74 �0.23** �0.28 0.07** 1.13 �0.34** �0.44 0.13** 8.7 1.0 n.s. 13.0 �3.1**SD 0.27 0.18 0.13 0.07 0.43 0.27 0.19 0.10 3.6 1.7 3.5 1.8

Note. Absolute changes of the oxygenation parameters in �M (��-based and �I-based assessment) and the VEP-components in �V. Thechanges over the 1 min stimulation block (i.e., the habituation) is given for each parameter (rows marked by �). To check for statisticalsignificance of this habituation-effect t statistics were performed across the 13 blocks for the individual volunteer (paired, 12 df). The last rowgives the mean across subjects. Here the asterisks denote the result of t statistics performed across the habituation in the 12 subjects(one-sample, 11 df). $ P � 0.1; * P � 0.05; ** P � 0.01; n.s., P � 0.1, indicated only in the last row.

6 OBRIG ET AL.

(3) The above defined coupling index can be calcu-lated for any pair of the electrophysiological (N75P100,P100N135, 40–210 ms) and the two oxygenation pa-rameters ([oxy-Hb] & [deoxy-Hb]), which in turn werecalculated based on the ��- and the �I-values. Weassessed the correlation coefficient between all of the12 possible combinations between an electrophysiolog-ical and a vascular parameter, based on the relativeamplitudes of all 312 VEP15 and the corresponding 312means across 4.8 s of [deoxy-Hb]- and [oxy-Hb]-changes (see Table 4). The resulting correlation coeffi-cient gives a general measure of the correlation for theparameter pairs but is also determined by the noiselevel of the respective parameter combination. Thisanalysis was performed to find out, which of the elec-trophysiological parameters would be the best predic-tor of the response in which of the vascular parame-ters.

Statistical Methods

Across all subjects. A General Linear Model (GLM)analysis for repeated measures with a simple contrastagainst the first (for the vascular response the second)epoch was used to check for statistically significantchanges within and between blocks across all subjects(see Table 3). Also the difference between the 1st (forthe vascular parameters 3rd) compared to the last ep-och in a stimulation block was tested by t tests (1st vs12th VEP15 and 3rd vs 12th epoch of the vascularparameters).

Within the individual subject. Paired t tests acrossthe successive 13 stimulation blocks were performed totest for a statistically significant change in the re-sponse from the 1st 15 VEPs compared to the last 15

VEPs (i.e., 1st VEP15 vs last VEP15, 12 df P � 0.05). Inanalogy the vascular response was tested for ‘habitua-tion’ in the individual subject, respecting the vascularlatency (i.e., the 3rd instead of the 1st epoch duringstimulation was tested against the last).

The analyses were performed in Matlab 5.3 (TheMath Work Inc.) in part using prewritten routines. Forthe GLM analyses SPSS 7.5 (SPSS Inc.) was used.

RESULTS

The grand averages across all 12 subjects for thevascular and the electrophysiological data are shownin Fig. 4. The remaining three subjects were excludedfrom the analysis, because either NIRS or EEG mea-surements showed strong artefacts (baseline shifts) orP100 abnormalities (W-form). The results of the oxy-genation measurements are a decrease in [deoxy-Hb]mirrored by an increase in [oxy-Hb] of about threetimes the magnitude (compare with Table 1). This find-ing is in line with the previously reported oxygenationchanges over a functionally activated cortical area(Heekeren et al., 1999a; Obrig et al., 2000). The oxy-genation changes derived from the phase measure-ments show a greater amplitude, while noise level isalso greater especially for the [deoxy-Hb] parameter(grey symbols without SD in the upper two graphs).The two VEP components (N75P100, P100N135) aregiven as averages across 15 reversals (VEP15) in thelower two graphs. Even in this grand average the dif-ferent behaviour within the blocks between the twoparameters can be seen: there is an initial maximumfor P100N135-component, while a slower developmentof the maximum is seen for the N75P100-component.

TABLE 2

Subject

Intensity Phase Fit

�[oxy-Hb] /% �[deoxy-Hb] /% �[oxy-Hb] /% �[deoxy-Hb] /% N75P100 /% P100 N135 /% 40–210ms /%

as_74 �48 0.13 �16 0.34 �32 0.12 �48 0.16 �60 0.16 �70* 0.03 �70* 0.03cb_77 �10 0.23 3 0.60 �20* 0.02 �19 0.10 29 1.00 1 0.53 7 0.76dl_73 �2 0.45 �32$ 0.08 �8 0.37 �59 0.22 �19 0.24 �30* 0.02 �38** 0.01dp_75 �9 0.24 �10$ 0.07 �17$ 0.07 1 0.53 �13 0.29 �58** 0.00 �35$ 0.09ik_78 �41* 0.03 �26** 0.00 �13 0.18 �12 0.37 �6 0.19 �32** 0.00 �30** 0.00is_73 �35* 0.02 �56** 0.00 �29** 0.01 �32* 0.03 �11 0.37 �62** 0.01 �64* 0.01iw_74 �53** 0.01 �15 0.19 �56** 0.01 �66$ 0.06 �69 0.21 �22 0.34 �14 0.40kn_76 �8 0.46 �123** 0.01 �55 0.24 �96 0.20 13 0.84 �19** 0.01 �27** 0.00ob_74 �108* 0.04 �12 0.18 �68** 0.00 �45$ 0.06 23 0.89 �42** 0.00 �52** 0.00ss_76 �53$ 0.09 �51$ 0.07 �25 0.34 �90 0.16 29 0.72 �13 0.41 �11 0.41su_76 �22 0.14 �7 0.28 �8 0.34 2 0.54 �6 0.38 �42* 0.03 �50** 0.01wc_73 �26$ 0.06 �36** 0.01 �28* 0.03 �22$ 0.07 �32$ 0.05 �51** 0.00 �48** 0.00

Mean �35** 0.00 �32** 0.01 �30** 0.00 �40** 0.01 �10 n.s. 0.29 �37** 0.00 �36** 0.00SD 29 34 20 33 32 21 23

Note. Relative changes of oxygenation parameters and VEP components within the 1 min stimulation block in percentage of mean responseas obtained by the fitting procedure. Here the last row gives the P values according to t statistics. Symbols and statistical procedures asdescribed for Table 1.


The baseline values of about 5–6 �V results from theanalysis strategy (maximum minus minimum in thedifferent temporal windows, thus representing thenoise level of the filtered data).

Across Block Changes in Amplitude

The following figure (Fig. 5) demonstrates the am-plitudes’ dynamics along the successive blocks (180reversals, �58 s duration). The values shown in thefigure (% of mean response) are calculated by the fit-ting procedure described above. The offset at the be-ginning is set at 100% for easier comparison. While thetwo VEP-components show some decrease in relativeamplitude over the first 6 blocks, the vascular param-eters show no clearly identifiable trend. For neitherVEP nor vascular parameters GLM statistics revealeda significance. The decrease in amplitude seen in Fig. 5may is but a brittle indication of a slight electrophysi-ological habituation across successive 1 min blocks sep-arated by equally long recovery periods for theP100N135-component.

Within Block Changes in Amplitude

Figure 6 illustrates the oxygenation changes withinthe �1 min stimulation blocks. The response exhibitsdynamic features well known from previous functional

NIRS studies. [oxy-Hb] and [deoxy-Hb] reach their re-spective maximum/minimum some 5–8 s after stimu-lus onset, with [oxy-Hb] showing a slightly steeperinitial slope. Both parameters show some attenuationof the response amplitude during the stimulation pe-riod, following slightly different shapes (initial over-shot phenomenon with following plateau in [oxy-Hb]).

FIG. 5. Response amplitude in percent change relative to the firststimulation block for the VEP180 amplitudes (left) and the oxygenationchanges (right). Note that for the electrophysiological parameters thereis a trend to decrease along the successive blocks. However this trenddid not reach significance in the statistical analysis (P values � 0.05according to the GLM-analysis for repeated measures).

FIG. 6. Average of the oxygenation response across all subjectsand all stimulation blocks. There is an increase in [oxy-Hb] and adecrease in [deoxy-Hb] during the stimulation period. ��-based val-ues for [oxy-Hb] and [deoxy-Hb] changes are indicated by the smallergrey symbols.

FIG. 7. Habituation of the VEP components within the 1-min stimulation block (average across all subjects and cycles given in % of meanresponse). The upper three graphs show the results obtained by the fitting procedure in the three windows (indicated by the inlays). The lowertwo graphs allow for the comparison between the results obtained by the fitting procedure compared to the assessment of the maximum-minimum for N75P100 and P100N135 components.

FIG. 8. Linear regression plots between the absolute changes in the oxygenation parameters ([oxy-Hb], [deoxy-Hb]) in �M, and the twoVEP components in �V. The regression is based on the eleven data points during the 1 min stimulation block (omitting the first value toaccount for the vascular latency, see small graph). The regression line is given by the function �[. . .-Hb] � a*(�VEP-component) b. Thenormalized residual (rnorm) provides an estimate of the goodness of the prediction by the regression line. Note that the coefficient a, i.e., theslope of the regression line, corresponds to the coupling index as derived from the habituation effect. The prediction of the vascular responseby the P100N135 is rather well approximated by the linear regression function. In this case the coupling index (i.e., the slope) has the samemagnitude as the value obtained by dividing the average response magnitudes (compare text and Table 1).

8 OBRIG ET AL.

The under-/overshoot phenomenon after the cessationof the stimulus is more prominent in the [deoxy-Hb]trace. Besides the systematic difference in magnitudethere were no relevant differences between the valuesderived from the intensity when compared to thosederived from phase changes (grey symbols in bothgraphs).

Correspondingly Fig. 7 shows the results from theelectrophysiological measurements. The upper row ofgraphs shows the relative magnitudes for theN75P100- and the P100N135-component as well as thevalues obtained by applying the fitting procedure to thetemporal window from 40–210 ms after pattern rever-sal. The lower two graphs illustrate the general agree-ment between the relative changes in amplitude whencalculated from the fitting procedure as compared tothe maximum–minimum assessment (here, the latteris also given in percentage of mean). The habituationeffect is slightly more prominent in the values obtainedby the fitting procedure. The obvious habituation of theP100N135-component is in agreement with the litera-ture (Afra et al., 1998; Schoenen et al., 1995; Sandor etal., 1999); however, we find a different behavior of theN75P100-component, which shows an increase in am-plitude over some 10 s to habituate only over the last10 s of the stimulation block. The fitting procedurecomprising all three components (40–210 ms, also com-pare the small inlays denoting the temporal windowfitted in the three upper graphs) shows a behaviorsimilar to the P100N135-component fit. We introducedthis measure, since it may best represent global elec-trophysiological activity when comparing it to the vas-cular response which cannot be differentiated tempo-rally with respect to the individual VEP-components.

Statistical analysis generally supported these find-ings. For the absolute response magnitudes we foundin all subjects the [oxy-Hb]- and in 11 subjects the[deoxy-Hb]-response to be attenuated during the stim-ulation period. The mean �I-derived changesamounted to �0.23 �M �0.18 for oxy-Hb and 0.07�M �0.07 for deoxy-Hb, ��-derived values were cal-culated to �0.34 �M �0.27 for [oxy-Hb] and 0.13 �M�0.10 for [deoxy-Hb] (see Table 1). Across all subjectsthese changes were statistically significant at the P �0.01 level (1-sample t test 11 df, last line in Table 1),while within subject statistics became significant in 7subjects in one or two parameters.

Concerning the VEP-components the N75P100-com-ponent showed statistically significant changes in thesingle subject analysis in only 1 subject (cb_77 showingan increase in amplitude). The P100N135-componentand the fit over 40–210ms on the other hand consis-tently decreased in amplitude in 11 subjects, showing astatistically significant change in 9 subjects. Acrosssubject analysis proved a similar difference betweenthe components. A highly significant decrease of 3.1 �V�1.8 was found for the P100N135-component, while

the changes in the N75P100-component did not reachstatistical significance.

The analysis of the normalised vascular changes (Ta-ble 2) and the VEP-component changes obtained from thefitting procedure yielded similar results with a 30–40%decrease in amplitude in the vascular and a decrease of36 and 37% for the P100N135 and 40–210 ms fit, respec-tively. The fit for the N75P100 component did not showstatistically significant changes (see Table 2).

For the across subject analysis of parameters dynamicsduring the �1-min stimulation block, Table 3 summa-rizes the results of the GLM repeated measures analysis.The contrast analysis shows a significant decrease (com-pared to the initial value) for the N75P100-componentafter about half the block length. Interestingly, whencompared to the first value, the initial increase inN75P100 amplitude also reached statistical significance.For the vascular parameters we chose the second epochas a reference for the following 10 epochs, taking thevascular latency into account.1 The decrease in [oxy-Hb]and the increase in [deoxy-Hb] reached statistical signif-icance only at the end of the stimulation.

In summary, either vascular and electrophysiologi-cal parameters exhibit dynamic changes within the �1min stimulation period, compatible with a habituationto the stimulus. The comparison across the successiveblocks yields only a brittle indication of a neuronalhabituation. In the following paragraph we report theresults from the coupling analysis based on the dynam-ics of the oxygenation and the VEP-component changeswithin the �1 min stimulation block.

Correlating the normalised vascular parameterswith the normalised electrophysiological parametersprovides a rough measure of the coupling. Table 4 givesthe correlation coefficients for all comparisons. As ex-pected by the results described above the correlationbetween the N75P100-component and the differentvascular parameters was generally lower than with thetwo other electrophysiological measures. Interestinglythe correlation between the [oxy-Hb] changes with theelectrophysiological components was much higher forthe [oxy-Hb] changes calculated from the phase mea-surements, while for [deoxy-Hb] the opposite held true.As can be seen in table 4 the correlation coefficient wasvariable between subjects. It should be noted, that thecoefficient is also dependent on the relative technicalnoise level of either vascular and electrophysiologicalparameter in the respective comparison. I.e., if thetechnical noise in the individual electrophysiologicalmeasurement was high this yields in a lower coeffi-cient, which does not necessarily represent the physi-

1 When including the first epoch all parameters showed statisti-cally significant changes according to the GLM repeated measuresanalysis. These, however, represent the vascular response latency ofthe oxygenation changes, which are not the issue of the currentpaper.

10 OBRIG ET AL.

ological correlation between the parameters. The effectis augmented in cases in whom both electrophysiolog-ical and vascular noise levels were high due to techni-cal (i.e., uncorrelated) noise.

The numerical values of the here proposed couplingindex must be judged preliminary. Since the compari-son between the vascular changes and the P100N135-component yielded highest values in the correlationanalysis we here report the value for this comparison.

CI deoxy

P100N135�

�[deoxy�Hb]

� P100N135� �0.02 �M * �V�1

(2)

CI oxy

P100N135�

�[oxy�Hb]

� P100N135� 0.06 �M * �V�1

For the vascular changes obtained by the phase mea-surements the values are �0.03 �M * �V�1 for de-

TABLE 3

Block

GLMrepeatedmeasures

Contrast analysisvs 1st block (2nd block for vascular parameters)

2 3 4 5 6 7 8 9 10 11 12

Electrophysiological parametersFitting

40–210 ms �0.000 n.s. n.s. n.s. n.s. n.s. n.s. * * ** ** **N75P100 0.001 * * ** * ** ** ** n.s. n.s. n.s. n.s.P100N135 �0.000 n.s. n.s. n.s. * * * ** ** ** ** **

Max-minN75P100 0.001 ** ** ** ** ** ** ** * * n.s. *P100N135 �0.000 n.s. n.s. n.s. * * * * ** ** ** **

Vascular parametersIntensity

�[oxy-Hb] �0.000 n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. *�[deoxy-Hb] �0.000 $ n.s. n.s. n.s. n.s. n.s. n.s. n.s. $ *

Phase�[oxy-Hb] 0.006 n.s. n.s. n.s. n.s. n.s. n.s. n.s. $ n.s. *�[deoxy-Hb] n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. n.s. *

Note. Results of the general linear model (GLM) analysis for repeated measures. The first column provides the P values obtained by theGLM. The asterisks and colours in columns 2–12 denote the level of statistical significance of the post-hoc testing (simple contrast against1st epoch for the electrophysiological parameters). For the vascular parameters the comparison to the 1st epoch (i.e., the first �4.8 s afterstimulus onset) resulted in statistically significant post hoc results for all successive epochs, due to the vascular response latency (data notshown see footnote in the text). A comparison to the 3rd value (i.e., assuming an even longer vascular response latency) did not change theresults (data not shown) n.s., P � 0.1; $ P � 0.1; * P � 0.05; ** P � 0.01.

TABLE 4

Subject

Intensity Phase

Mean

�[oxy-Hb] �[deoxy-Hb] �[oxy-Hb] [deoxy-Hb]

N75P100 P100N135 40–210ms N75P100 P100N135 40–210ms N75P100 P100N135 40–210ms N75P100 P100N135 40–210ms

as_74 0.27 0.33 0.32 0.28 0.41 0.42 0.36 0.47 0.47 0.28 0.38 0.38 0.36cb_77 0.72 0.74 0.75 0.70 0.69 0.71 0.74 0.77 0.78 0.62 0.65 0.66 0.71dl_73 0.44 0.49 0.49 0.43 0.47 0.47 0.36 0.45 0.46 0.12 0.16 0.17 0.38dp_75 0.57 0.50 0.63 0.60 0.57 0.68 0.57 0.54 0.62 0.49 0.49 0.54 0.57ik_78 0.60 0.60 0.61 0.76 0.79 0.79 0.66 0.69 0.69 0.31 0.34 0.33 0.60is_73 0.39 0.46 0.46 0.44 0.51 0.51 0.48 0.58 0.57 0.43 0.53 0.53 0.49iw_74 0.13 0.25 0.26 0.16 0.31 0.31 0.19 0.30 0.30 0.14 0.20 0.20 0.23kn_76 0.19 0.20 0.20 0.35 0.32 0.31 0.12 0.18 0.19 0.09 0.16 0.17 0.21ob_74 0.35 0.37 0.37 0.55 0.62 0.62 0.55 0.64 0.64 0.43 0.51 0.50 0.51ss_76 0.17 0.27 0.29 0.21 0.25 0.28 0.08 0.16 0.17 �0.02 0.06 0.07 0.17su_76 0.61 0.50 0.56 0.61 0.49 0.55 0.55 0.48 0.52 0.48 0.44 0.47 0.52wc_73 0.55 0.62 0.62 0.60 0.67 0.67 0.61 0.68 0.68 0.59 0.67 0.67 0.64

Mean 0.42 0.44 0.46 0.47 0.51 0.53 0.44 0.50 0.51 0.33 0.38 0.39

Note. Correlation coefficients for all possible combinations between the vascular and electrophysiological parameters based on all epochsin the individual volunteers. The measure does not reflect the true correlation between the parameters but serves as an indicator for thesignal-to-noise ratio for the respective comparison and subject.


oxy-Hb and 0.09 �M * �V�1 for oxy-Hb. Calculatingthe index only for those subjects with a mean correla-tion coefficient of �0.5 (see last column in Table 4),only slightly changes the result.

To supply another check of the assumed linearity weplotted the grand average of the VEP components’ am-plitude against the changes in the vascular parametersduring the stimulation period (omitting the first valueto compensate for the lag in vascular response). As canbe seen in Fig. 8, the 1st order regression line ratherwell explains the relationship between the habituationin P100N135 amplitude and the habituation in thevascular parameters. The slope of the regression cor-responds well to the coupling index assessed by thedifference between stimulation and rest. This meansthat for the P100N135 component a close (most prob-ably linear) coupling can be postulated also for thehabituation effect. For the N75P100-component such arelation cannot be assumed.

DISCUSSION

The present study examines the habituation of thevascular and the electrophysiological response toblocks of 1 min checkerboard stimulation. The resultsshow a clear habituation effect for the P100N135 VEP-component and the vascular parameters. Since datawere assessed simultaneously, a preliminary couplingindex can be given with an increase of 0.06 �M in[oxy-Hb] and a decrease of 0.02 �M in [deoxy-Hb] elic-ited per �V increase in P100N135 amplitude.2 Thisindex is the same when calculated by the averageacross the stimulation versus resting periods and whenassessed for the habituation effect within the 1 minstimulation block. This suggests a close coupling be-tween the N100P135-component and the vascular re-sponse.

The N75P100-component did not show a habituationeffect within the stimulation block. Also the statisti-cally non-significant trend of an electrophysiologicalhabituation across the 13 successive stimulationblocks, separated by equally long resting periods wasnot reflected in a similar trend in the vascular param-eters (Fig. 5).

It should be noted here that the present report of acoupling index is not meant to imply that coupling is ahomogeneous phenomenon across different brain re-gions. There are nonlinear aspects of neurovascularcoupling (Miller et al., 2001; Mechelli et al., 2001),

which, however, stress the necessity to not simply takea vascular response as the sole marker of cerebralactivity. We here propose an approach of co-registeringneuronal and vascular response to highlight that alinear relationship must be tested rather than as-sumed. In many complex paradigms a very obviousnonlinearity is the difference between excitatory andinhibitory neuronal processes for which network anal-ysis has been proposed to define a more realistic imageof brain activation (Nyberg et al., 1996).

The discussion of the principal findings is structuredas follows: we first discuss some issues of the localisa-tion with respect to either VEP and haemodynamicresponse. This fairly general issue is relevant sincehabituation phenomena in either response can only becompared, if there is a rough spatial correspondence ofthe neuronal structures generating the VEP-compo-nents with the sampling volume of the NIRS measure-ments. Next, the findings are discussed with respect tothe VEP habituation and the dynamics of the vascularresponse. Finally, we discuss some of the recent reportson the assessment of neurovascular coupling in thelight of the present findings.

Localization of the Electrophysiological andHemodynamic Response

Despite the wide use of visually evoked potentials inclinical routine and research, the generators of theindividual VEP-components are not clearly identified.The gold-standard to localise neuronal generators ofscalp potentials, single cell recording, is not feasible inthe human. In the cat it was shown that 60% of thecells in area 17 are activated by a grating stimulus(Spileers et al., 1994). At the same time the wide vari-ety of responses for different cell populations is dem-onstrated using contrast sweeping to define responsethresholds. Based on the recording of intracortical andsurface potentials in the monkey Kraut and coworkers(Kraut et al., 1985) tentatively ascribe the N75-compo-nent to an initial excitation in layer IV of the striatecortex. P100-component is thought to reflect inhibitionin thalamorecipient cells while the N135-component isascribed to a stellate input into pyramidal cells (Krautet al., 1990). This finding of a mainly striate origin ofthe three components is supported by subdural elec-trode studies in the monkey (Dagnelie et al., 1989) andthe human (Arroyo et al., 1997). In the human, a largenumber of studies have been dedicated to the investi-gation in VEP-generators by use of scalp recordedpotentials. Even magno- and parvocellular input (Klis-torner et al., 1997) and retinotopic representation(Slotnick et al., 1999) have been differentiated. Suchstudies analyse temporal characteristics or apply di-pole localisation techniques to isolate different inputsand generators of the scalp recorded potentials. How-ever, due to the inverse problem assumptions must be

2 As stated above the fit across all VEP components may betterrepresent the neuronal response elicited by the reversal. However,the fitting procedure yields no absolute response magnitude in �V.The calculation of the Coupling Index based on the P100N135 com-ponent may still be justified, considering that the fit from 80–190 ms(corresponding to the P100N135 component) yields essentially thesame results as the fit 40–210ms.

12 OBRIG ET AL.

made, in order to approximate the localisation of thesources corresponding to the neuronal generators ofthe scalp recorded potentials. In the present study aprimarily cortical origin of the N75-P100-N135 swingis assumed. Since relative VEP-magnitude at Oz wasanalyzed we consider subcortical and extrastriate con-tribution small.

Concerning the NIRS measurements the issue ofsampling volume is dominated by the question of depthpenetration and extracerebral contribution. Extensivework has been dedicated to the question in how far theoxygenation changes monitored are truly cortical innature (for review see (Obrig et al., 2000)) and thuscorrespond to the haemodynamic responses demon-strated by imaging techniques most notably BOLD-contrast fMRI (Kleinschmidt et al., 1996). Qualita-tively the present results are in agreement with ourprevious studies using visual stimuli (Wenzel et al.,1996) (Heekeren et al., 1999b) and the probes are per-fectly colocated with the electrodes due to the experi-mental set-up (Fig. 1). Since the area sampled by thetwo probe pairs is rather large it may include areasbeyond V1 (V2–V4). However, potential activation ofthese areas will also contribute to the VEP and itscomponents.

A further issue concerning sampling volume needsmentioning. Since the measurement was performedwith a frequency-domain monitor, we here report con-centration changes calculated based on both, phase(��) and intensity (attenuation) changes (�I). The re-sults are qualitatively similar with a generally largeramplitude for the ��-based values (�[. .-Hb]�� 1.5 *�[. .Hb]�I). This can be explained by the higher sensi-tivity of ��-measurements to deeper layers (Hemeltand Kang, 1999; Kohl et al., 2002) reducing the partialvolume effect. Interestingly the ��-based [oxy-Hb]changes correlated better with electrophysiologicalchanges than the �I-derived [oxy-Hb] changes. Theopposite was true for [deoxy-Hb]. This finding may beexplained by different contribution of physiological andtechnical noise to the respective parameter. While thetechnically lower SNR of the ��-derived parametersexplains the finding for [deoxy-Hb] changes, the phys-iological noise (heart beat, respiration, vasomotion) ismore prominent in the [oxy-Hb] changes measured. Byreducing the partial volume effect in the ��-derived[oxy-Hb] changes, this overrides the technically in-duced loss in SNR and yields the better correlationwith the electrophysiological findings. The finding that[oxy-Hb] changes are more prone to physiological noise(heart-beat and slow spontaneous oscillations) hasbeen reported by our group (Obrig et al., 2000), alsovisual inspection of the individual non-averaged time-courses supports this explanation, showing less uncor-related [oxy-Hb]-changes in the ��-based traces.

To sum up, although neither electrophysiological norhaemodynamic measurements can be exactly localised

due to the respective methodological limitations, itseems a reasonable first approximation that the am-plitude of the VEP-components measured by EEG andthe oxygenation response as measured by NIRS repre-sent a net effect of the respective neuronal and vascu-lar events. The finding of a similar response behaviourfor one of the VEP components and the [deoxy-Hb]changes may therefore result from linear and nonlin-ear aspects of neuovascular interaction in the visualsystem, whose primary and secondary areas could notbe differentiated in the present approach. In furtherstudies, we will challenge this simplification by extend-ing the approach to a multisite EEG recording com-bined with a NIRS imaging device.

Habituation of the VEP Components

Habituation of event-related potentials has been re-ported for the auditory (Wang and Schoenen, 1998) andvisual systems (Wastell and Kleinman, 1980) and cog-nitive functions (Wintink et al., 2001; Kropp et al.,2000; Kropp and Gerber, 1998). Concerning VEP ha-bituation a number of studies is concerned with thelack of habituation in interictal migraine patients.These papers all agree on the habituation in normalvolunteers. We also find a 37% decrease of theP100N135-component amplitude over the 1-min stim-ulation block; however, the 10% decrease in N75P100amplitude is not statistically significant. This is evenmore remarkable when the results of the GLM analysisare considered, showing an initial increase of theN75P100-component’s amplitude. A closer look at thecontrol groups of the above mentioned studies explainssome of this discrepancies. Extending on a originalstudy with 2 min repetitive checkerboard stimulation(Schoenen et al., 1995) Afra and coworkers find a ha-bituation of 6.8% for the 3rd block of 100 reversals (i.e.,after �66 s). This decrease in N75P100 amplitude be-comes statistically significant only after 4 min,whereas the 10–11% decrease in P100N135-compo-nent amplitude is significant already in the 2nd block(compare Table 2 in (Afra et al., 1998)). The study bySandor et al. (1999), using the same paradigm, shows anonsignificant decrease of 13.7% between the 1st and5th block of 50 reversals (�80 s) in migraine patientsand a recent study by the same group (Judit et al.,2000) reports normalised VEP habituation just prior tothe migraine attack. Here a decrease in N75P100 am-plitude of 13.6% is reported, which is more pronouncedthan the 6.8% decrease in the control group. The sta-tistical comparisons are performed between groups,thus not allowing to directly compare the findingsand the rather large standard deviation (20.5%)does not allow to infer statistical significance of thereported N75P100 habituation within each group (theP100N135 component is not reported). To explain thedifferences between the different studies it is of further


relevance that stimulus parameters strongly influencehabituation. Check size is of major importance to therelative activation of the parvo- and magnocellularpathway, whose relative input is thought to be re-flected in the different components (Kubova et al.,1995). The study of Oelkers et al. (1999) shows thatonly small check size elicits habituation dominated bythe magnocellular input presumably reflected in theN135 component. Large checks (primarily activatingthe parvocellular pathway, allegedly dominating P100amplitude) do not show this behaviour. The stimulusused in the present experiment consists of an annularcheckerboard with different check sizes. To our knowl-edge the relative input in the two visual pathwayscompared to an unevenly sized checkerboard has notbeen studied. We cannot exclude the possibility thatour finding of a differential behaviour of the two com-ponents is partly explained by this modification of thestimulus.

“Habituation” of the Vascular Response

The issue of a potential habituation of the vascularresponse is essentially the question in how far thecoupling between neuronal activity and vascular re-sponse behaves like a linear system when the timecourse over the lengths of the stimulation period isanalysed. The issue of linearity has been discussedalong different lines of evidence:

Analysis of time course of the vascular response. Ac-tivation induced focal increases in blood flow overridethe focal oxygen demand, leading to a focal hyperoxy-genation. Evidence has been put forward that this focaluncoupling (Fox and Raichle, 1986) is caused by achange to nonoxidative metabolism (Magistretti, 2000;Magistretti et al., 1999) eliciting a lactate increase inthe activated area during prolonged stimulation(Sappey-Marinier et al., 1992). Since some authors re-ported the BOLD-contrast to decrease over extendedperiods of stimulation (Hathout et al., 1994) with asimilar temporal dynamic as the rise in lactate thefinding spurred a theory that during prolonged stimu-lation the focal uncoupling is reversed within 3–4 minresulting in a focal recoupling (Frahm et al., 1996;Kruger et al., 1996). In a previous NIRS study by ourgroup, we also found the decrease in [deoxy-Hb] (whichis inversely related to BOLD contrast) to partly returnto baseline during sustained stimulation (Heekeren etal., 1997). These considerations suggest a truly vascu-lar habituation independent of the neuronal input.However, the theory must be modified in the light of acurrent view of the literature. Two other groups reporta stability of the BOLD contrast over extended stimu-lation periods (Bandettini et al., 1997; Howseman etal., 1998). Also the lactate shuttle between astrocytesand neurones as described by recent studies does notimply that substantial non-oxidative metabolism is in-

duced by physiological activation of the cerebral cortex.Recently both linear and nonlinear aspects of the neu-rovascular coupling have been described. Kruger andcoworkers reported BOLD-contrast- and perfusion-sensitized-fMRI yielding almost identical dynamicsduring ongoing stimulation (Kruger et al., 1999). In amore recent study, however, the model of a linear cou-pling between neuronal activation and rCBF but anonlinear translation of the latter into BOLD-contrastchanges has been postulated based on the simulta-neous assessment of BOLD-contrast and arterial spinlabelling sequences (Miller et al., 2001). Differencesbetween the motor and the visual systems are re-ported. Also the rat cerebellar coupling model investi-gated by Mathiesen et al. (2000) stresses the impor-tance of comparing identical stimulus conditions, sincedifferent stimulation frequencies led to different re-sults concerning neurovascular coupling.

Hence, when BOLD-contrast changes in response tofunctional stimulation are to be modelled, neuronalinput, the transform of this input into a vascular re-sponse and the ensuing changes in the parameters offlow velocity, flow volume and focal deoxy-Hb concen-tration must be respected. Such a model to describelinear and non-linear behaviour, which does not relyon the—questionable—assumption of a substantialchange in the metabolic strategy to meet the energydemand induced by functional activation, has been de-veloped. The Balloon and Windkessel model (Buxton etal., 1998; Mandeville et al., 1999) proceeds from themore simple assumptions of an elasticity of the venouscompartment and changes in oxygen diffusion capacityinduced by changes in vessel diameter and flow veloc-ity. The model has been shown to readily explainchanges in BOLD-contrast in response to functionalstimulation and it has been refined and generalised toallow to explain both linear an nonlinear aspects of theBOLD-contrast response for almost any given stimula-tion paradigm (Friston et al., 2000).

However, most studies investigating the relation be-tween neuronal activation and vascular response sharethe uncertainty as to the stability of the neuronal in-put. Surrogate markers such as flow-sensitized fMRIsequences are taken as evidence for a lack in habitua-tion of the neuronal response or decrease in attention.To our knowledge the present study is the first toaddress the issue by a simultaneous assessment of theneuronal and the vascular response. The results are inagreement rather with the view that the decrease inBOLD-contrast (corresponding to the increase in [de-oxy-Hb]) reflects neuronal habituation. We do not findevidence of a decrease in the vascular response inde-pendent of a concomitant neuronal habituation, on thecontrary the attenuation of both the P100N135-compo-nent amplitude and the [deoxy-Hb] response duringthe 1 min stimulation block are statistically signifi-cant. This finding of a habituation is not in full accor-

14 OBRIG ET AL.

dance with those reported in those fMRI studies, whichreport a stable response for prolonged stimulation(Bandettini et al., 1997; Howseman et al., 1998). It maybe argued that the stimulation period of 1 min is tooshort and the early partial habituation effect docu-mented here may be missed when much longer stimu-lation periods are analysed (there is in fact an initialovershoot of the BOLD response seen in Fig. 4; p. 5 ofHowseman et al. (1998) and Fig. 10, p. 102, of Bandet-tini et al. (1997)). However, the neuronal habituationhas been found to extend over a much longer periodand the control group in the above mentioned study byAfra and coworkers shows a neuronal habituation ofabout 20% for both components maximal at the end ofthe 15-min stimulation period. If the vascular responsewere to perfectly match the neuronal activation a de-crease in BOLD contrast and flow-sensitised fMRI sig-nal would be expected. This may highlight, that thepresent study supplies evidence for just one aspect oflinear behaviour between VEP-component amplitudeand oxygenation parameters as monitored by NIRS.The question of how the different vascular parametersare related is not addressed by the study. A recentstudy by Janz and coworkers has focussed the nonlin-ear interplay between neuronal activation and BOLDcontrast (Janz et al., 2001). Though they find thatsuperposition of single-event responses reasonablywell explains the response to longer stimulation peri-ods, the convolution of the stimulus input function andthe decay-function of the VEP, measured prior to thefMRI experiment, still overestimates the transientchanges at the beginning and the end of the stimula-tion period. The analysis of the time course of thevascular response may hence be complicated by thefact that at the beginning and at the end of a stimula-tion periods over- and undershoot phenomena appear,probably related to different latencies of the changes inblood volume and flow velocity (Mandeville et al.,1999). The question concerning linearity of neurovascu-lar coupling has therefore also been addressed by a num-ber of recent studies applying graded stimulus inten-sities. Generally these studies investigate whethergraded stimuli, of whom the neuronal response behav-iour is known, elicit a correspondingly graded vascularresponse.

Response to graded stimulation. Using graded vi-sual stimulation and balancing the flow velocity(CBFv) change induced by visual stimulation by hyper-capnia induced CBFv increases, Hoge and coworkers(Hoge et al., 1999) have recently shown that theCMRO2 does not reach a plateau but continuously in-creases with stimulus intensity, thus supporting thediffusion-based theory for functionally induced focalhyperoxygenation. A number of studies relate eventrelated potential amplitudes to the vascular response.A linear relationship between N20P22 amplitude of the

somatosensory evoked potential and BOLD-contrast inresponse to graded stimulus intensity has been re-ported in a study by Arthurs and co-workers (Arthurset al., 2000). In the rat, Ngai and coworkers found alinear coupling between SEP amplitude and vascularresponse as measured by laser doppler flowmetry inresponse to different stimulation frequencies (Ngai etal., 1999). Interestingly this linearity was observedonly if the integral instead of the mean across all SEPswas correlated to the vascular response. This findingfosters the view that the vascular response mirrors thesum of electrophysiological activity. We similarly findthe fit across the whole N75-P100-N135 swing to cor-relate well with the oxygenation changes, while theN75P100 component did not correlate with the vascu-lar changes during the stimulation period.

Besides the different aspects of linearity with respectto time course and graded stimulus intensity an essen-tial question is which electrophysiological parameter isbest able to represent the activity eliciting the vascularresponse. Comparing coherence of a motion stimulus,spike activity in V5 assessed by single cell recording inthe macaque and BOLD-contrast changes assessed inthe human, Rees and co-workers conclude, that “a verysimple mathematical model links changes in the stim-ulus coherence, [. . .] firing rates [. . .] and modulationof BOLD contrast” (Rees et al., 2000). The ratio theyreport is an increase in firing rate of 9 spikes/s per 1%increase in BOLD contrast. This finding encouragesHeeger and coworkers to apply this trans-species ap-proach to their data from the primary visual cortex.From the analysis they derive an increase in 0.4 spike/second per neurone to elicit a 1% BOLD contrast in-crease (Heeger et al., 2000). The gross difference be-tween these results remains unexplained.

Also, many stimuli will lead to a broad variety of spike-response patterns in different neuronal populations,which holds definitely true for the checkerboard stimulusapplied in the present study (see above localization of theelectrophysiological response). Also synaptic activity ismetabolically demanding for both excitatory and inhibi-tory processes in the cortex (Jueptner and Weiller, 1995).For a model of excitatory climbing fibres in the rat cere-bellum blood flow correlated with evoked field potentials,which probably better represent neuronal activity(Mathiesen et al., 1998, 2000). Simultaneously assessingdifferent measures of the electrophysiological activityand BOLD-contrast changes in the monkey, Logothetisand coworkers also reported a better correlation betweenthe BOLD response and the local field potentials as op-posed to multiunit responses (Logothetis et al., 2001).These findings may indicate that scalp recorded poten-tials as assessed in the present study are a justifiedapproximation to neuronal activity. A study by Guy andcoworkers shows a good accordance between BOLD con-trast and the VPEG, a measure of VEP amplitudechanges similar to the here reported changes the value


obtained by 40–210 ms fitting-procedure (Guy et al.,1999).

In the present study we provide evidence for just oneaspect of the assumption that the translation of theneuronal to the vascular response behaves like a linearsystem, also we cannot ascertain that the volumessampled are precisely the same for our electrophysio-logical recordings and the oxygenation changes mea-sured by NIRS. However, based on the above discussedliterature we consider it justified to calculate a ratiobetween simultaneously assessed changes in VEPmagnitude and oxygenation changes. We find that theratio between the oxygenation parameters and theVEP response is the same when assessed by the aver-age across the 1-min stimulation block and when as-sessed by the difference between the beginning and theend of the stimulation (i.e., the habituation). Also theregression between the P100N135 amplitude and theoxygenation parameters is sufficiently approximatedby a linear equation (see Fig. 8). These findings supportthe usefulness of the here presented approach to assessneurovascular coupling noninvasively in the human.

ACKNOWLEDGMENTS

This work was supported by the Deutsche Forschungs Gemein-schaft (Klinische Forschergruppe Funktionsdiagnostik des ZNS,Charite).

REFERENCES

Afra, J., Cecchini, A. P., De, P. V., Albert, A., and Schoenen, J. 1998.Visual evoked potentials during long periods of pattern-reversalstimulation in migraine. Brain 121: 233–241.

Arroyo, S., Lesser, R. P., Poon, W. T., Webber, W. R., and Gordon, B.1997. Neuronal generators of visual evoked potentials in humans:Visual processing in the human cortex. Epilepsia 38: 600–610.

Arthurs, O. J., Williams, E. J., Carpenter, T. A., Pickard, J. D., andBoniface, S. J. 2000. Linear coupling between functional magneticresonance imaging and evoked potential amplitude in human so-matosensory cortex. Neuroscience 101: 803–806.

Bandettini, P. A., Kwong, K. K., Davis, T. L., Tootell, R. B., Wong,E. C., Fox, P. T., Belliveau, J. W., Weisskoff, R. M., and Rosen,B. R. 1997. Characterization of cerebral blood oxygenation andflow changes during prolonged brain activation. Hum. BrainMapp. 5: 93–109.

Bonmassar, G., Schwartz, D. P., Liu, A. K., Kwong, K. K., Dale,A. M., and Belliveau, J. W. 2001. Spatiotemporal brain imaging ofvisual-evoked activity using interleaved EEG and fMRI record-ings. NeuroImage 13: 1035–1043.

Boynton, G. M., Engel, S. A., Glover, G. H., and Heeger, D. J. 1996.Linear systems analysis of functional magnetic resonance imagingin human V1. J. Neurosci. 16: 4207–4221.

Buxton, R. B., Wong, E. C., and Frank, L. R. 1998. Dynamics of bloodflow and oxygenation changes during brain activation: The balloonmodel. Magn. Reson. Med. 39: 855–864.

Condon, B., McFadzean, R., Hadley, D. M., Bradnam, M. S., andShahani, U. 1997. Habituation-like effects cause a significant de-crease in response in MRI neuroactivation during visual stimula-tion. Vision Res. 37: 1243–1247.

Dagnelie, G., Spekreijse, H., and van, D. B. 1989. Topography andhomogeneity of monkey V1 studied through subdurally recordedpattern-evoked potentials. Vis. Neurosci. 3: 509–525.

Diehl, B., Stodieck, S. R., Diehl, R. R., and Ringelstein, E. B. 1998.The photic driving EEG response and photoreactive cerebral bloodflow in the posterior cerebral artery in controls and in patientswith epilepsy. Electroencephalogr. Clin. Neurophysiol. 107: 8–12.

Fischer, H., Furmark, T., Wik, G., and Fredrikson, M. 2000. Brainrepresentation of habituation to repeated complex visual stimula-tion studied with PET. Neuroreport 11: 123–126.

Fox, P. T., and Raichle, M. E. 1986. Focal physiological uncoupling ofcerebral blood flow and oxidative metabolism during somatosen-sory stimulation in human subjects. Proc. Natl. Acad. Sci. USA 83:1140–1144.

Frahm, J., Kruger, G., Merboldt, K. D., and Kleinschmidt, A. 1996.Dynamic uncoupling and recoupling of perfusion and oxidationmetabolism during focal brain activation in man. Magn. Reson.Med. 35: 143–148.

Friston, K. J., Mechelli, A., Turner, R., and Price, C. J. 2000. Non-linear responses in fMRI: The Balloon model, Volterra kernels,and other hemodynamics. NeuroImage 12: 466–477.

Guy, C. N., ffytche, D. H., Brovelli, A., and Chumillas, J. 1999. fMRIand EEG responses to periodic visual stimulation. NeuroImage 10:125–148.

Hathout, G. M., Kirlew, K. A., So, G. J., Hamilton, D. R., Zhang, J. X.,Sinha, U., Sinha, S., Sayre, J., Gozal, D., and Harper, R. M. 1994.MR imaging signal response to sustained stimulation in humanvisual cortex. J. Magn. Reson. Imag. 4: 537–543.

Heeger, D. J., Huk, A. C., Geisler, W. S., and Albrecht, D. G. 2000.Spikes versus BOLD: What does neuroimaging tell us about neu-ronal activity? Nat. Neurosci. 3: 631–633.

Heekeren, H. R., Kohl, M., Obrig, H., Wenzel, R., von, P. W.,Matcher, S. J., Dirnagl, U., Cooper, C. E., and Villringer, A. 1999a.Noninvasive assessment of changes in cytochrome-c oxidase oxi-dation in human subjects during visual stimulation [In ProcessCitation]. J. Cereb. Blood Flow Metab. 19: 592–603.

Heekeren, H. R., Kohl, M., Obrig, H., Wenzel, R., von, P. W.,Matcher, S. J., Dirnagl, U., Cooper, C. E., and Villringer, A. 1999b.Noninvasive assessment of changes in cytochrome-c oxidase oxi-dation in human subjects during visual stimulation. J. Cereb.Blood Flow Metab. 19: 592–603.

Heekeren, H. R., Obrig, H., Wenzel, R., Eberle, K., Ruben, J.,Villringer, K., Kurth, R., and Villringer, A. 1997. Cerebral haemo-globin oxygenation during sustained visual stimulation—A near-infrared spectroscopy study. Philos. Trans. R. Soc. Lond. B. Biol.Sci. 352: 743–750.

Hemelt, M. W., and Kang, K. A. 1999. Determination of a biologicalabsorber depth utilizing multiple source-detector separations andmultiple frequency values of near-infrared time-resolved spectros-copy. Biotechnol. Prog. 15: 622–629.

Hoge, R. D., Atkinson, J., Gill, B., Crelier, G. R., Marrett, S., andPike, G. B. 1999. Linear coupling between cerebral blood flow andoxygen consumption in activated human cortex. Proc. Natl. Acad.Sci. USA 96: 9403–9408.

Howseman, A. M., Porter, D. A., Hutton, C., Josephs, O., and Turner,R. 1998. Blood oxygenation level dependent signal time coursesduring prolonged visual stimulation. Magn. Reson. Imag. 16: 1–11.

Israel, H., Obrig, H., Kohl, M., Uludag, K., Muller, B., Wenzel, R.,Buckow, C., Arnold, G., and Villringer, A. 2000. Is neuro-vascularcoupling altered in interictal migraineurs? A combined visuallyevoked potential (VEP) and Near-Infrared-Spectroscopy (NIRS)Approach. Conference Proceedings Human Brain Mapping, Brigh-ton Abstr. 10434.

Janz, C., Heinrich, S. P., Kornmayer, J., Bach, M. and Hennig, J.2001. Coupling of neural activity and BOLD fMRI response: New

16 OBRIG ET AL.

insights by combination of fMRI and VEP experiments in transi-tion from single events to continuous stimulation. Magn. Reson.Med. 46: 482–486.

Judit, A., Sandor, P. S., and Schoenen, J. 2000. Habituation of visualand intensity dependence of auditory evoked cortical potentialstends to normalize just before and during the migraine attack.Cephalalgia 20: 714–719.

Jueptner, M., and Weiller, C. 1995. Review: does measurement ofregional cerebral blood flow reflect synaptic activity? Implicationsfor PET and fMRI. NeuroImage 2: 148–156.

Kleinschmidt, A., Obrig, H., Requardt, M., Merboldt, K. D., Dirnagl,U., Villringer, A., and Frahm, J. 1996. Simultaneous recording ofcerebral blood oxygenation changes during human brain activa-tion by magnetic resonance imaging and near-infrared spectros-copy. J. Cereb. Blood Flow Metab. 16: 817–826.

Klistorner, A., Crewther, D. P., and Crewther, S. G. 1997. Separatemagnocellular and parvocellular contributions from temporalanalysis of the multifocal VEP. Vision Res. 37: 2161–2169.

Kohl, M., Obrig, H., Steinbrink, J., Malak, J., Uludag, K., andVillringer, A. 2002. Non-invasive monitoring of cerebral blood flowby a dye bolus method: Separation of brain from skin and skullsignals. J. Biomed. Opt., in press.

Kraut, M. A., Arezzo, J. C., and Vaughan, H. G. J. 1985. Intracorticalgenerators of the flash VEP in monkeys. Electroencephalogr. Clin.Neurophysiol. 62: 300–312.

Kraut, M. A., Arezzo, J. C., and Vaughan, H. G. J. 1990. Inhibitoryprocesses in the flash evoked potential of the monkey. Electroen-cephalogr. Clin. Neurophysiol. 76: 440–452.

Kropp, P., and Gerber, W. D. 1998. Prediction of migraine attacksusing a slow cortical potential, the contingent negative variation.Neurosci. Lett. 257: 73–76.

Kropp, P., Siniatchkin, M., and Gerber, W. D. 2000. Contingentnegative variation as indicator of duration of migraine disease.Funct. Neurol. 15(Suppl. 3): 78–81.

Kruger, G., Kastrup, A., Takahashi, A., and Glover, G. H. 1999.Simultaneous monitoring of dynamic changes in cerebral bloodflow and oxygenation during sustained activation of the humanvisual cortex. Neuroreport 10: 2939–2943.

Kruger, G., Kleinschmidt, A., and Frahm, J. 1996. Dynamic MRI sen-sitized to cerebral blood oxygenation and flow during sustained acti-vation of human visual cortex. Magn. Reson. Med. 35: 797–800.

Kubova, Z., Kuba, M., Spekreijse, H., and Blakemore, C. 1995. Con-trast dependence of motion-onset and pattern-reversal evoked po-tentials. Vision Res. 35: 197–205.

Logothetis, N. K., Pauls, J., Augath, M., Trinath, T., and Oelter-mann, A. 2001. Neurophysiological investigation of the basis of thefMRI signal. Nature 412: 150–157.

Magistretti, P. J. 2000. Cellular bases of functional brain imaging:insights from neuron-glia metabolic coupling. Brain Res. 886:108–112.

Magistretti, P. J., Pellerin, L., Rothman, D. L., and Shulman, R. G.1999. Energy on demand. Science 283: 496–497.

Mandeville, J. B., Marota, J. J., Ayata, C., Zaharchuk, G., Moskow-itz, M. A., Rosen, B. R., and Weisskoff, R. M. 1999. Evidence of acerebrovascular postarteriole windkessel with delayed compliance.J. Cereb. Blood Flow Metab. 19: 679–689.

Mathiesen, C., Caesar, K., Akgoren, N., and Lauritzen, M. 1998.Modification of activity-dependent increases of cerebral blood flowby excitatory synaptic activity and spikes in rat cerebellar cortex.J. Physiol. 512: 555–566.

Mathiesen, C., Caesar, K., and Lauritzen, M. 2000. Temporal cou-pling between neuronal activity and blood flow in rat cerebellarcortex as indicated by field potential analysis. J. Physiol. 523 Pt 1:235–246.

Mechelli, A., Price, C. J., and Friston, K. J. 2001. Nonlinear Couplingbetween Evoked rCBF and BOLD Signals: A Simulation Study ofHemodynamic Responses. Neuroimage 14: 862–872.

Miller, K. L., Luh, W. M., Liu, T. T., Martinez, A., Obata, T., Wong,E. C., Frank, L. R., and Buxton, R. B. 2001. Nonlinear temporaldynamics of the cerebral blood flow response. Hum. Brain Mapp.13: 1–12.

Ngai, A. C., Jolley, M. A., D’Ambrosio, R., Meno, J. R., and Winn,H. R. 1999. Frequency-dependent changes in cerebral blood flowand evoked potentials during somatosensory stimulation in therat. Brain Res. 837: 221–228.

Nyberg, L., McIntosh, A. R., Cabeza, R., Nilsson, L. G., Houle, S.,Habib, R., and Tulving, E. 1996. Network analysis of positronemission tomography regional cerebral blood flow data: ensembleinhibition during episodic memory retrieval. J. Neurosci. 16:3753–3759.

Obrig, H., Neufang, M., Wenzel, R., Kohl, M., Steinbrink, J., Ein-haupl, K., and Villringer, A. 2000. Spontaneous low frequencyoscillations of cerebral hemodynamics and metabolism in humanadults. NeuroImage 12: 623–639.

Obrig, H., and Villringer, A. 1996. Near-infrared spectroscopy infunctional activation studies. Can NIRS demonstrate cortical ac-tivation? Adv. Exp. Med. Biol. 413: 113–127.

Obrig, H., Wenzel, R., Kohl, M., Horst, S., Wobst, P., Steinbrink, J.,Thomas, F., and Villringer, A. 2000. Near-infrared spectroscopy:does it function in functional activation studies of the adult brain?Int. J. Psychophysiol. 35: 125–142.

Oelkers, R., Grosser, K., Lang, E., Geisslinger, G., Kobal, G., Brune,K., and Lotsch, J. 1999. Visual evoked potentials in migrainepatients: Alterations depend on pattern spatial frequency. Brain122: 1147–1155.

Pinsker, H., Kandel, E. R., Castellucci, V., and Kupfermann, I. 1970.An analysis of habituation and dishabituation in Aplysia. Adv.Biochem. Psychopharmacol. 2: 351–373.

Poellinger, A., Thomas, R., Lio, P., Lee, A., Makris, N., Rosen, B. R.,and Kwong, K. K. 2001. Activation and habituation in olfac-tion—An fMRI study. NeuroImage 13: 547–560.

Raichle, M. E. 1998. Behind the scenes of functional brain imaging:A historical and physiological perspective. Proc. Natl. Acad. Sci.USA 95: 765–772.

Rees, G., Friston, K., and Koch, C. 2000. A direct quantitative rela-tionship between the functional properties of human and macaqueV5. Nat. Neurosci. 3: 716–723.

Rosen, B. R., Buckner, R. L., and Dale, A. M. 1998. Event-relatedfunctional MRI: past, present, and future. Proc. Natl. Acad. Sci.USA 95: 773–780.

Sadato, N., Nakamura, S., Oohashi, T., Nishina, E., Fuwamoto, Y.,Waki, A., and Yonekura, Y. 1998. Neural networks for generationand suppression of alpha rhythm: A PET study. Neuroreport 9:893–897.

Sandor, P. S., Afra, J., Proietti-Cecchini, A., Albert, A., andSchoenen, J. 1999. Familial influences on cortical evoked poten-tials in migraine. Neuroreport 10: 1235–1238.

Sappey-Marinier, D., Calabrese, G., Fein, G., Hugg, J. W., Big-gins, C., and Weiner, M. W. 1992. Effect of photic stimulation onhuman visual cortex lactate and phosphates 1H and 31P mag-netic resonance spectroscopy. J. Cereb. Blood Flow Metab. 12:584 –592.

Schoenen, J., Wang, W., Albert, A., and Delwaide, P. J. 1995. Poten-tiation instead of habituation characterizes visual evoked poten-tials in migraine patients between attacks. Eur. J. Neurol. 2:115–122.

Slotnick, S. D., Klein, S. A., Carney, T., Sutter, E., and Dastmalchi,S. 1999. Using multi-stimulus VEP source localization to obtain a


retinotopic map of human primary visual cortex. Clin. Neuro-physiol. 110: 1793–1800.

Spileers, W., Maes, H., Lagae, L., and Orban, G. A. 1994. Contrastmodulated steady-state visual evoked potentials (CMSS VEPs):Recording evoked potentials and related single cell responses inarea 17 of the cat. Electroencephalogr. Clin. Neurophysiol. 92:64–77.

Springett, R., Newman, J., Cope, M., and Delpy, D. T. 2000. Oxygendependency and precision of cytochrome oxidase signal from fullspectral NIRS of the piglet brain. Am. J. Physiol. Heart Circ.Physiol. 279: H2202–H2209.

Uludag, K., Kohl, M., Steinbrink, J., Obrig, H., and Villringer, A.2001. Crosstalk in the Lambert–Beer calculation for near-infrared wavelengths estimated by Monte simulations.J. Biomed. Opt.

van Essen, D. C., Lewis, J. W., Drury, H. A., Hadjikhani, N., Tootell,R. B., Bakircioglu, M., and Miller, M. I. 2001. Mapping visualcortex in monkeys and humans using surface-based atlases. VisionRes. 41: 1359–1378.

Villringer, A., and Dirnagl, U. 1995. Coupling of brain activity andcerebral blood flow: Basis of functional neuroimaging. Cerebrovasc.Brain Metab. Rev. 7: 240–276.

Wang, W., and Schoenen, J. 1998. Interictal potentiation of passive“oddball” auditory event-related potentials in migraine. Cephalal-gia 18: 261–265.

Wastell, D. G., and Kleinman, D. 1980. Potentiation of the habitua-tion of human brain potentials. Biol. Psychol. 10: 21–29.

Wenzel, R., Obrig, H., Ruben, J., Villringer, K., Thiel, A., Bernard-ing, J., Dirnagl, U., and Villringer, A. 1996. Cerbral blood oxygen-ation changes induced by visual stimulation in humans. J. Cereb.Blood Flow Metab. 1: 399–404.

Wildgruber, D., Erb, M., Klose, U., and Grodd, W. 1997. Sequentialactivation of supplementary motor area and primary motor cortexduring self-paced finger movement in human evaluated by func-tional MRI. Neurosci. Lett. 227: 161–164.

Wintink, A. J., Segalowitz, S. J., and Cudmore, L. J. 2001. Taskcomplexity and habituation effects on frontal P300 topography.Brain Cogn. 46: 307–311.

18 OBRIG ET AL.

Habituation of the visually evoked potential and its vascular response: implications for...

Documents

Transcript of Habituation of the visually evoked potential and its vascular response: implications for...