Intensity dependence of auditory evoked dipole source activity

INTERNATIONAL JOURNALOF PSYCHOPHYSIOLOGY

ELSEVIER

Intensity

International

dependence Ulrich Hegerl

Journal

of Psychophysiology

of auditory

17 (1994) 1-13

evoked dipole source activity

a,*, Jiirgen Gallinat

a, Dieter

Mrowinski

b

a Department of Psychiatry, Laboratory of Clinical Psychophysiology, Freie Unicersitiit Berlin, Eschenallee 3, 14050 Berlin, Germany h ENT Clinic, Freie UnirsersittitBerlin, Uetzrr Steig 24, 14089 Berlin, Germany (Accepted

16 November

1993)

Abstract The spatio-temporal approach of dipole source analysis represents a crucial methodological progress in research on individual differences in the dependence of auditory evoked potential amplitude on stimulus intensity (augmenting/reducing) because overlapping subcomponents of the Nl/P2-component can be separated and can be related to their generating cortical structures. Basic aspects of the intensity dependence of auditory evoked dipole source activity were analysed in 40 healthy subjects. The evoked responses to binaural lOOO-Hz tones at five levels of intensity (60, 70, 80, 90, 100 dB sound pressure level) were recorded at 33 sites across the scalp. The dipole source analysis of the grand average data confirms the reports in the literature that the Nl/P2 potentials at the scalp can be explained by two dipoles per hemisphere: a tangential dipole, representing activity of the superior temporal cortex (including primary auditory cortex), and a radial dipole, representing activity of the lateral temporal cortex (secondary auditory areas). The intensity dependence of the tangential dipole activity was significantly more pronounced than that of the radial dipoles, supporting the assumption that radial and tangential dipoles represent different physiological processes. A high reliability of the intensity dependence of the tangential dipole (Pearson correlation: r = 0.88) was found when retesting the subjects after three weeks. Age was negatively correlated with the intensity dependence of the tangential dipole. Dipole source analysis proved to be a reliable method which allows, at least in part, to study separately the intensity dependence of the evoked responses from primary and secondary auditory cortices. This is of importance with regard to the hypothesis that the central serotonergic system modulates the intensity dependence of the evoked Nl/P2-response of primary auditory cortex. Key words: Augmenting/reducing;

Auditory

evoked potential;

1. Introduction

The dependence of event related potential (ERP) amplitude on the stimulus intensity shows a pronounced interindividual variability. Origi-

* Corresponding

author.

Fax: + 49 30 3003393.

0167-8760/94/$07.00 0 1994 Elsevier SSDI 0167-8760(93)E0054-4

Science

Dipole source; Auditory

cortex; Serotonin;

Age effect

nally, within the augmenting/reducing concept for evoked potentials proposed by Buchsbaum and Silverman [ll, a flat or negative slope of the amplitude/ intensity function (reducing) was seen as reflecting a central mechanism regulating the sensitivity in all sensory modalities and protecting the organism from sensory overstimulation. Within the augmenting/reducing paradigm, rela-

B.V. All rights reserved

2

U. Hegerl et al. /Itttrmationul

Journal

tionships between the intensity dependence of sensory evoked potentials on the one hand and personality, diagnostic categories and biochemical variables on the other hand have been investigated [l-51 (see also the review by Carrillo-dela-Pena [6]). In our own work we focused on the stimulus intensity dependence of the auditory evoked Nl/P2-component [7], since this parameter was found to be of potential clinical value as a predictor of clinical response to lithium prophylaxis in affective psychoses [8,9]. Furthermore, the hypothesis has been proposed, that the intensity dependence of the evoked response of the primary auditory cortex is modulated by the central serotonergic system and may therefore be useful as an indicator of serotonergic dysfunction in psychiatric patients [lo]. However, research on augmenting/reducing has come in for a lot of criticism because the lack of standardization concerning, e.g., the recording sites, the stimulus intensity range or the parametrization precludes generalization between experiments [ 111). Additional confusion resulted from the fact that the augmenting/reducing concept was originally proposed by Petrie [12] for the classification of subjects according to kinaesthetic figural aftereffects. Augmenters in this procedure, however, were not also augmenters in the evoked potential procedure [131. Therefore, we prefer the more neutral term ‘intensity dependence’ to augmenting/ reducing. Even more troublesome for the early augmenting-reducing concept has been the insufficient knowledge about anatomical structures generating the observed evoked potentials. It has been proposed that augmenting/reducing is a nonspecific cortical phenomenon best recorded at vertex [ 141. However, as supported by recent studies [15,16], auditory evoked potentials (AEP) measured from vertex to mastoids with latencies around 100 ms reflect activity from the superior temporal plane including primary auditory cortex, whereas for visual evoked potentials (VEP), the same electrodes may measure mainly activity of non-specific centro-parietal cortex. With this in mind, it is not surprising that no cross-modal correlations between the intensity dependence of

of Psychophysiology

17 (1994) l-13

VEP and AEP have been found in studies recording from vertex to mastoids for both modalities [17-191 because differences in the intensity dependence between the evoked responses from primary and secondary sensory cortices have to be expected 1201. Such differences have been found in epidural recordings in monkeys [21] and are in line with the low levels of agreement between amplitude/ intensity response patterns found in recordings using vertex versus temporal leads for AEP [22] and central versus occipital leads for VEP [2,14,23]. This lack of cross-modal correlation has been a further point of criticism because augmenting/ reducing was originally proposed as a modality-independent concept. The situation is further complicated by the fact that a certain scalp recorded potential can be composed of overlapping subcomponents reflecting activity in different cortical structures (e.g., Refs. 15,24). This component overlap impairs not only the physiological validity of the amplitude measures but also the reliability because peak identification becomes problematic. Double peaks, e.g. of the auditory Nl/P2-complex, are a frequent and intraindividually stable finding. These subcomponents can differ in their intensity dependence. The moderate test-retest stability of the intensity dependence of AEP, as reported in the literature [25], may result in part from such problems. Considering these points of criticism, dipole source analysis represents a crucial methodological progress in augmenting/ reducing research. The spatio-temporal approach of dipole source analysis has been proposed as a useful method for separating the overlapping subcomponents of the Nl/P2-component and relating them to their generating cortical structures [ 15,261. The potential distribution at the scalp in the time range of the Nl/P2-component can be explained by the activity of two dipoles per hemisphere: the activity of a tangential dipole located in the superior temporal plane (including primary auditory cortex) and the somewhat later activity of a radial dipole located near the lateral temporal cortex (secondary auditory areas). This finding is in line with converging evidence from magnetencephalographic studies [27-291, intracranial recordings

U. Hegerl et al. /International

Journal of Psychophysiology 17 (1994) l-13

[30], and lesion studies [31,32] supporting the view that the superior temporal plane and the gyri temporales laterales are the main generators of this component (for reviews see Refs. 24,33,34). By separating the overlapping subcomponents and investigating the intensity dependence of the radial and tangential dipoles independently, it should be possible to enhance the reliability as well as the validity of the intensity dependence measures. Our aim was therefore to study whether or not dipole source analysis can be a solid methodological basis for restarting investigations on individual differences in the intensity dependence of AEP. The intensity dependence of the auditory evoked dipole activity in the time range of the Nl/PZcomponent was investigated in healthy subjects using Brain Electric Source Analysis (BESA, [15,261X The following questions were addressed: (i) Can the scalp distribution of the auditory evoked Nl/PZcomponent be explained by the activity of one radial and one tangential dipole per hemisphere? (ii> Do radial and tangential dipoles differ in their intensity dependence? (iii) How reliable is the intensity dependence of tangential and radial dipole activities? (iv) What are the influences of age or gender on the intensity dependence of the dipole activities?

2. Methods

and materials

2.1. Subjects 40 healthy drug-free volunteers (mean age f S.D.: 39.1 + 10.9 y, 19 male, 21 female) were investigated. The subjects were well-known to the investigators: six were recruited from the hospital personnel, the others from the local community. For all subjects auditory threshold was below 10 dB measured by the BCktsy method using a PCaudiometer.

2.2. Stimulation, recording and averaging The evoked potentials were recorded twice for each subject with a time interval of 3 weeks (run 1 and run 2). Both runs were performed within the same hour of day. Recordings took place in a sound-attenuated and electrically shielded room adjacent to the recording apparatus. The subjects were seated with eyes open in a slightly reclining chair with a head rest and were asked to look at the wall 3 m in front of them. No strict fixation was demanded. Evoked responses were recorded using an electrode cap with 26 electrodes. 7 additional periocular and cerebellar electrodes were placed. Cz was used as reference. An electrode 2 cm in front of Fz was used as ground. Sinus tones (1000 Hz, 30 ms duration with 10 ms rise and fall time, IS1 randomized between 1600 and 2100 ms) of five intensities (60, 70, 80, 90, 100 dB sound pressure level) were presented binaurally in pseudo-randomized form by headphones. Data were collected with a sampling rate of 500 Hz and an analog low-pass filter of 150 Hz (six-pole Butterworth-filter, 48 dB/ octave). Digital bandpass filters were set to 0.5 and 30 Hz (six-pole IIR low-pass filter with 48 dB/octave roll-off, six-pole Butterworth high-pass filter with 24 dB/octave roll-off). The sampling period reached from 200 ms prestimulus to 500 ms poststimulus. 204 sweeps were recorded for each intensity. Before averaging, the first five responses to each intensity were excluded in order to reduce short-term habituation effects. Furthermore, for artifact suppression, all trials were automatically excluded from averaging, when the voltage exceeded +50 PV in any one of the 32 channels at any point during the averaging period. The mean of the rejected sweeps per person was 104 (range: 0 -497) corresponding to 10.2% of the 1020 sweeps of each person. The artifact rate did not differ between the five intensity levels (60 dB: 10.5%; 70 dB: 10%; SO dB: 10.5%; 90 dB: 10%; 100 dB: 10%). Most artifacts occured at the periocular leads.

4

U. Hegerl c’t al. / Intemutional

Journal uf f’sychophysiology

I7 (1994) I-1.7

For each subject the remaining sweeps were averaged separately for the five intensity Icvels. 2.3. Dipole source analysis Dipole source analysis was performed with Brain Electrical Source Analysis (BESA, [35]). The dipoles are thought to represent activity of circumscribed cortical areas. Therefore they change dipole strength over time, but not location or orientation. BESA decomposed the scalp potentials into four dipole source activities. The optimal location and orientation of the dipoles are found by an iterative process (simplex algorithm) optimizing the residual variance (variance of the measured scalp data unexplained by the model), whereas the dipole source potentials are determined by the direct linear approach, as described by Scherg and Picton [351. An approximation of a three-shell head model is used by BESA. Since the Nl/P2-component seems to be mainly generated by the primary and secondary auditory cortices [15], we assumed two dipoles per hemisphere to be simultaneously active in the time range of the Nl/P2-component. The averaged curves of each subject were entered into BESA, where data reduction (from 350 to 90 data points), baseline correction, digital filtering (low-pass 20 Hz, high-pass 1 Hz, 24 dB/ octave roll-off)), and transformation to average reference data (to give equal weight to each location) took place. In order to get a ‘basic dipole model’, a dipole fit was performed on the grand mean curves of all subjects for the period of the Nl/P2-component (63.5-207 ms). The periocular and frontopolar electrodes (Lol, Lo2, nasion, Fpl, Fp2) were not considered in the fit procedure in order to reduce effects of ocular artifacts. All dipole fit procedures were performed using interhemispherical symmetry constraints for location and orientation in order to reduce the number of independent parameters to be determined. Fitting was done by strictly adhering to the following steps: (i) the iterative fit procedure was started with bilateral regional sources (regional source: three orthogonal dipoles with a common location, see Ref. 15). Locations were fitted for

a 1

l 3

-

288

RS

2 JJUcff

\

n

4 f

Fig. 1. Brain Electric Source Analysis (BESA) of grand mean auditory evoked potentials from 32 healthy subjects (run 1). With two equivalent dipoles per hemisphere more than 98% of the variance of the scalp potentials in the time range of the Nl /P2-complex can be explained (RV = residual variance). The dipole source potentials of the radial dipoles (3 and 4). reflecting activity of secondary auditory areas, occur about 40 ms later than those of the tangential dipoles (1 and 2) which are supposed to reflect mainly activity from primary auditory cortex.

these two regional sources in order to find the centers of activity in both hemispheres. (ii) in each hemisphere the dipole with a tangential/ horizontal orientation was switched off. Then orientation and location were fitted, first for the two tangential/vertical and then for the two radial dipoles. This fit procedure was performed independently on the grand mean curves of run 1 and run 2. The resulting ‘basic dipole models’ are shown in Figs. 1 and 2. These basic dipole models are very similar for runs 1 and 2. Only 1.45 and 1.42% of the variance of the scalp data remains unexplained in the time range of the Nl/P2-component. This dipole configuration is a very stable solution which is also found when a different sequence of fitting steps is used. The tangential/ vertical dipoles, named ‘tangential dipoles’ in the following, contributed most to the explained variance.

U. Hegeri et al. /International

0

Journal of Psychophysiology 17 (1994) l-13

6-2 2

I L

1

-3

R

4

-

288 ns

2 puerr

\

Fig. 2. BESA of grand mean auditory the same 32 healthy subjects, retested nearly identical dipole configuration found.

R

f

evoked potentials from after 3 weeks (run 2). A and dipole activity is

The ‘individual dipole model’ was found by starting with the ‘basic dipole model’ and adjusting individually the orientation and location of the tangential dipoles: for each subject the average curves for the five intensities were collapsed to individual mean curves. Using these individual mean curves and starting with the basic dipole model the orientation and location of the tangential dipoles were iteratively adjusted for each subject, resulting in the ‘individual dipole model’. The radial dipoles were not fitted concerning location and orientation because they explained only a small amount of variance and correspondingly showed higher variability. the individual curves were entered into the ‘individual dipole model’ separately for the five intensities. No additional fitting of location or orientation was performed for the different intensities because changes in location or orientation would be confounded with changes in dipole amplitude. The magnitude of the dipole source activity is directly proportional to the recorded voltage on

5

the scalp. The magnitude of the Nl/P2-dipole activity can be measured in effective-amplitude CpV,,, for a definition see Ref. 16) as peak-topeak amplitude (Nl/P2-peak-amplitude) or as the root mean squared effective amplitude over the epoch of the Nl/P2-component (Nl/P2epoch-amplitude 1161). Nl /PZpeak-amplitude is preferred as parameter because the Nl/P2-epoch-amplitude is sensitive to latency shifts. Such latency shifts of the Nl/P2-response have to be expected in different stimulus intensities and can considerably influence epoch-based amplitude as shown by Connolly and Gruzelier [23]. However, the peaks of the radial dipole activities are sometimes not clearly defined making peak identification difficult. Therefore, the N 1/ P2-epoch-amplitude was used for studying the radial dipole activity. The Nl-component of the individual dipole source potentials was measured as the most negative peak within 60 and 125 ms, and the P2-component as the most positive peak within 110 and 210 ms. The Nl /PZepoch-amplitude was determined for the 63.5-207 ms epoch. Amplitudes were determined independently for both hemispheres, as well as for tangential and radial dipoles. Additionally, the peak-amplitudes of the Nland PZcomponents were also considered independently. 2.4. Intensity dependence of the dipoles The intensity dependence &V/l0 dB) of the dipoles was measured using the median-slope. The median-slope was calculated from the slopes of all possible connections (n = 10) between the five amplitude values to the five intensities. In the literature the intensity dependence of evoked potentials is frequently measured as the slope of a straight line fitted to the amplitudes at the different intensity levels (slope of the amplitude/ stimulus intensity function, ASF-slope). The use of this least square slope method has been critisized because of low coefficients of determination in individual data [23]. Therefore, the

U. Hegerl et al. /International

6

Journal of Psychophysiology 17 (1994) I - 13

median-slope is preferred in this paper and the ASF-slopes are only given for comparison.

3.1. Basic dipole models

2.5. Signal-to-noise ratio (SNR) SNR for the averaged potentials (excluding prestimulus epochs) was calculated according to Mijcks et al. [36] for each channel. The mean SNR of the 27 channels included in the dipole analysis was taken as the individual SNR. The mean SNR of the 40 subjects at the five intensity levels are presented in Table 1. High SNR are found, especially at higher stimulus intensities.

2.6. Statistics Because of the exploratory per, no alpha-error correction

nature of the pawas performed.

of the grand

Fig. 3 presents the grand average potentials (run 1, common average reference), used for calculating the basic dipole model. The basic dipole models for run 1 and run 2 (retest after 3 weeks) are presented in Figs. 1 and 2. The two dipoles per hemisphere explain more than 98.5% of the variance of the scalp data in the epoch of the Nl /P2-component. One of the dipoles is located in the superior temporal region and has a tangential and vertical orientation (tangential dipoles 1 and 2). This dipole explains most of the variance of the scalp data. The other dipole is located in the temporal

auer .ref

488 ms Fig. 3. Distribution

3. Results

mean auditory

evoked potentials

. (run 1, common

average

reference.

negativity

down)

7

U. Hegerl et al. /International Journal of Psychophysiology I7 (1994) l-13

3.2. Intensity dependence radial dipole activities

Table 1 Mean signal-to-noise ratios (SNR) of the averaged poststimulus potentials in 40 subjects (run 1). The SNR of the individual subject was calculated as the mean value of the 27 channels included in the dipole analysis Stimulus

60 70 80 90 100

dB dB dB dB dB

intensity

SPL SPL SPL SPL SPL

Signal-to-noise (mean + S.D.)

of the tangential and

Table 2 presents the median-slopes and for comparison the ASF-slopes of Nl/P2-activities of the tangential dipoles for both hemispheres and both runs. No significant interhemispherical differences were found. As can be seen in Table 2 the median-slopes and ASF-slopes have nearly identical values. For the radial dipoles the median-slopes were calculated only for Nl/P2-epoch-amplitudes. Steeper slopes were found in the left than in the right hemisphere in the first but not in the second run (run 1: 0.06 + 0.06 versus 0.08 + 0.06, p < 0.05; run 2: 0.07 f 0.07 versus 0.08 & 0.06; paired t-test). Only the median-slopes will be considered in further analysis because this parameter is less affected by a nonlinear relationship between amplitude and intensity. Fig. 4 presents the intensity dependence of the Nl/P2-activities of the tangential and radial dipoles (run 1). The intensity dependence is significantly more pronounced for the tangential than for the radial dipoles. This difference is clearly significant both for the right and for the left dipoles in both runs (Table 3). This difference is still significant if the individual amplitude increase with increasing stimulus intensity is not

ratio

t3.9* 6.5 11.9* 7.4 16.9 k 10.5 26.3 k 17.3 41.8k33.6

lobe with a radial orientation (radial dipoles 3 and 4). The source activities of the dipoles are also presented in Figs. 1 and 2. A delay of about 40 ms can be observed for the Nl/P2-activity of the radial dipoles as compared to the tangential dipoles (Nl-tang.: right 91.2 ms, left 92.5 ms, Pa-tang.: right 148.6 ms, left 148.5 ms, Nl-rad.: right 136.1 ms, left 137.2 ms, P2-rad.: right 188.2 ms, left 183.3 ms). This corresponds to double peaks separated by about 40 ms, which can be found with scalp recordings in some subjects. Nearly identical dipole activities and dipole configurations are found when the same subjects are retested after 3 weeks supporting the stability of the dipole solution.

Table 2 The intensity dependence (median-slopes and ASF-slopes) of the tangential dipoles (right and left hemisphere) are presented (run 1: n = 40, run 2: n = 36). The slopes are calculated for both the Nl/P2-peak-amplitude and the Nl/P2-epoch-amplitude. Slopes are given in ~V,,/10 dB Tangential right Run 1 Median-slope ASF-slope

Run 2 Median-slope ASF-slope

dipoles left

paired t-test

Nl/P2-peak-ampl. Nl/PZ-epoch-ampl. Nl/P2-peak-ampl. Nl/P2-epoch-ampl,

0.52 0.18 0.53 0.18

* f * *

0.33 0.13 0.33 0.13

0.52 0.18 0.52 0.18

k * + +

0.29 0.10 0.29 0.10

n.s. n.s. n.s. n.s.

Nl/PZ-peak-ampl. Nl/PZ-epoch-ampl. Nl/P2-peak-ampl. Nl/P2-epoch-ampl.

0.53 0.19 0.53 0.19

k 0.24 * 0.09 & 0.24 * 0.09

0.53 0.18 0.53 0.18

? f k f

0.25 0.10 0.24 0.09

n.s. ns. n.s. n.s.

U. Hqerl et al. /Internutionul

1;

Journal of P.sychophysiology 17 (1994) I-13

Test-retest of

the

stability (Pearson

intensity

correlation

dependence

of

the

coefficients, tangential

II = 36)

and

radial

dipoles Median-slope

Tangential right

dipoles

Radial dipoles

left

mean

right

left

mean

-

-

N I /P2-peak-ampl.

0.82

0.81

0.88

-

N I /P?-epoch-ampl.

0.85

0.84

0.91

0.5

I

0.Ih 0.3X

peak-amplitudes measured to prestimulus baseline) of P2 is significantly more pronounced in both runs than that of Nl (run 1:0.33 k 0.22 versus 0.20 _t 0.10, p < 0.001; run 2:0.31 t_ 0.18 versus 0.22 i 0.10, p < 0.01; paired t-test). 3.3. Test-retest

,

90

100 dE

Fig. 4. Mean healthy

dipole

subjects

activity (Nl/P2-epoch-amplitude)

for

tangential

stimulus intensities: (hemisphere.

The

SPL

-

-_)

intensity

dipoles is more pronounced

and

radial

left hemisphere; dependence

of

of 40

dipoles

five

) right

(-the

to

tangential

than that of the radial dipoles.

measured in PV but in percentage of mean amplitude ( p < 0.001). The intensity dependence was also calculated separately for the Nl-and for the P2-component of the tangential dipole (mean between right and left). The intensity dependence (median-slope of

Table

3

The intensity dependence plitudes.

Run I (II right

pV,,, /IO

dB) of the tangential

Paired

dipole

dipole

t-test

0.18~0.13 0. ICI+ 0.10

IO& 0.0’)

right

0.

left

o.Ixi_o.lo

Table 4 presents the test-retest correlations of the median-slopes of the radial and tangential dipoles. An excellent test-retest reliability can be found for the intensity dependence of the tangential dipoles. The reliability of the radial dipole is not satisfactory. Therefore, only the median-slope of the tangential dipole, based on peak-amplitude, will be considered in further analysis. Since no interhemispherical differences were observed in this parameter, the mean slope value for right and left hemisphere will be used for further analysis. A test-retest correlation of r = 0.88 is found for this parameter. When the test-retest reliability is calculated independently for the Nl- and PZcomponent, correlation coefficients of r = 0.67 (Nl) and r = 0.92 (P2) are found. 3.4. Intensity dependence of the tangential and effects of age and gender

dipole

and radial dipoles

Radial

Run 7 ((1 = 36)

+ P < 0.001.

of NI/P2-epoch-am-

Tangential

= 40)

left

(median-slope

correlations

Effects of age and gender on the intensity dependence (median-slope) were studied by analysis of variance (run 1). The subjects were divided at the median in two age groups. In the old group (mean age: 48.6 y; range: 42-62 y> were 9 males and 11 females, in the young group (mean age: 29.6 y; range: 22-40 y) were 10 males and 10 females.

U. Hegerl et al. /International

Journal of Psychophysiology 17 (1994) I-13

intensity dependence ~VeffilO dB] 1.8r

0.8

A

A

A

A

A

A

0.2 -_-_t_j___I 0 c __i__ _+-,___+ ~_-_t__+_i 15 20 25 30 35 40 45 50 55 60 65 age [years] Fig. 5. Relationship ence (median-slope)

between age and the intensity dependof the tangential dipole (run I).

Analysis of variance (factor 1: male,female; factor 2: old,young) revealed that the intensity dependence (median slope) of the tangential dipole is negatively related to age (F = 11.0; df: 1,39; p = 0.002; Pearson correlation coefficient: r = -0.48: p = 0.002; Fig. 5). No gender effects effects (F = 2.0; df: 1,39; p = 0.17) or interaction between ‘age’ and ‘gender’ were observed (F = 0.95; df: 1,39; p = 0.34). The findings are similar, when effects of age and gender are studied independently for the intensity dependence of Nl and P2. Age effects were significant both for Nl (F = 8.0; df: 1,39; p = 0.008) and P2 (F = 10.1;df: 1,39; p = 0.003). Neither gender effects nor interactions between ‘age’ and ‘gender’ were observed for these components.

4. Discussion 4.1. Basic dipole model In line with findings in the literature sults confirm that the auditory evoked components measured with 32 channels

the reNl/P2can be

0

explained almost completely by the activity of two dipoles per hemisphere: a tangential dipole, representing activity in the superior temporal plane (planum temporale and transverse gyri) and a smaller radial dipole, representing activity of secondary auditory areas in the lateral temporal cortex. When retested after 3 weeks a nearly identical dipole configuration and dipole activity was found confirming the stability of this dipole solution. Most of the activity is explained by the tangential dipoles (92% of the variance). The location and orientation of this dipole is in line with other studies showing that the superior temporal plane with the primary auditory cortex are the main generators of the Nl /P2-component [ 1.5,16,371. Concerning the Nl-component a combinded analysis of neuromagnetic source localization and magnetic resonance imaging showed that the tangential source projected exactly onto the transverse temporal gyri [27], in which the primary auditory cortex is located. In the generation of later activity secondary auditory areas with a more anterior location in the superior temporal plane may also be involved [27,28,37,38]. The stability of the radial dipoles, which explain a small percentage of the variance of the scalp data is considerably lower especially concerning location. The orientation of the radial dipoles suggests that they reflect at least in part activity of lateral temporal cortex (secondary auditory area>, although the location is more medial than expected. The high percentage of explained variance in the basic dipole model corresponds to the high signal-to-noise ratio (SNR) of the data. High SNR were already observed in the individual Nl/P2potentials and the SNR of the data for the basic dipole model was further enhanced by averaging additionally over intensities and subjects. The l-2% of variance unexplained by the dipole model therefore corresponds to the noise level in the grand average data. 4.2. Intensity dependence dipoles The tangential stronger intensity

of radial and tangential

dipoles show a significantly dependence than the radial

IO

I/. He&

rt ul. /International

Journal

dipoles. This lends support to the assumption that radial and tangential dipoles reflect different physiological processes, generated by different cortical structures. Further support stems from the observation that the dipole Nl/P2-activity of the radial dipole occurs about 40 ms later than that of the tangential dipole. It can be assumed that the tangential dipoles reflect to a great part activity of the primary auditory cortices. A stronger intensity dependence of the response from the primary compared to the secondary auditory cortex has also been observed with epidural recordings in monkeys [21]. corresponding nicely with our results. Less augmenting response of AEP with temporal compared to central leads has been reported by Prescott et al. [22]. This is also in line with our data, because acitivity of the superior temporal area, including the primary auditory cortex, is best recorded with central leads referenced to mastoids or car lobes. In studies on cross-modal correlations VEP and AEP were recorded from Cz. It has now to be assumed that in these studies evoked responses from primary auditory cortices were compared to those from unspecific or secondary areas in the visual modality. This may be the reason for the lack of cross-modal correlations of the intensity dependence in the literature. A reevaluation of this issue should be performed.

4.3. Test-retest pendence

stuhility

of the dipole intensity

de-

The test-retest stability of the intensity dependence of the tangential dipole activity was clearly enhanced (Y = 0.88) when compared to findings based on scalp potentials (r = 0.71-0.78 [25]. This is due to the fact that dipole source analysis disentangles overlapping subcomponents and combines the information of all channels. This represents an important progress, if the intensity dependence is to be used as a predictor of clinically relevant aspects in psychiatry (e.g., see Refs. 9 and 10). Within the paradigm used in this study, the radial dipole activities are neither prominent nor

of Psychophysiology reliable ter.

enough

17 (1994) I-I.7

to be a useful

research

parame-

4.4. Effects of uge and gender- on the intensity dependence of the tangential dipole Effects of covariablcs on the intensity dependence of the tangential dipole should be known, if this parameter is to be used for psychiatric rcsearch. In line with previous results [25], no gcnder effects were found. Age turned out to be a relevant covariable with a negative relationship to the intensity dependence of the tangential dipole. Such a relationship was not observed in a previous study, relying on scalp potentials only [25]. For VEP negative as well as positive relationships have been reported between age and the intensity dependence [4,39]. 4.5. Independent

analysis for Nl and P2

Nl- and P2-components differ in several aspects, e.g., in their intensity dependence as found in this study. This raises the question whether or not the generating structures of these two components can be represented by a common dipole. Several magnetoencephalographic studies suggest that the Nl- and the P2-components are generated by cortical structures located in the superior temporal plane. The P2-dipoles were found about 5 mm anterior to the Nl-dipoles by [38], and somewhat larger distances (l-2 cm) have been reported by others [40,41]. Considering these results it can be assumed that the activity of both components can be represented by the activity of a common equivalent dipole, because such relatively small location differences do not affect the dipole solution in a relevant manner. More important than location differences would be differences in dipole orientation. However, the direction of the P2-source has been reported to be almost 180 degree opposite to that of the Nlsource, corresponding to the finding that both lie in the superior temporal cortex. We have focused on the intensity dependence of Nl/P2-amplitudes and not independently on the intensity dependence of Nl- and P2-ampli-

U. Hegerl et al. /International

Journal of Psychophysiology I7 (1994) 1-13

tudes, because the relationship to personality factors or lithium response is closer for the combined parameter (unpublished results). Furthermore, the reliability of Nl/P2-amplitudes was higher than the reliability of Nl- and P2-amplitudes, measured with reference to a prestimulus baseline (unpublished results). Correspondingly the reliability of the intensity dependence of Nl was quite low (r = 0.67) in this study. For these reasons it seemed justified to use a parameter combining Nl and P2. Effects of covariables are similar for both the intensity dependence of Nland P2-dipole activity.

as noninvasive function.

de-

Our results are of interest for concepts on physiological processes underlying individual differences in the intensity dependence of sensory evoked potentials. Animal research indicated that these differences occur at the cortical rather than already at the subcortical level and that unspecific activating brainstem systems modulate the response pattern of the cortical evoked potentials [42]. However, this brainstem system has not been specified. We have presented in detail empirical and theoretical arguments for the hypothesis that the intensity dependence of auditory evoked responses depends on the central serotonergic system [lo]. A low serotonergic preactivation of the auditory cortex is supposed to result in a strong intensity dependence and vice versa. Our finding that tangential and radial dipoles differ in their intensity dependence is in line with this hypothesis because modulating serotonergic effects have to be expected especially for the primary auditory cortices, represented by the tangential dipole: The highest concentrations of cortical serotonin and the highest synthesis rates have consistently been found in primary sensory cortices. The radial dipole is not supposed to reflect serotonergic effects, because serotonergic innervation is low in secondary sensory areas [43-481. If the intensity dependence of the tangential dipole reflects modulating serotonergic effects, then this parameter can become of considerable clinical importance

of central

serotonergic

Acknowledgement This work was supported by ‘Deutsche Forschungsgemeinschaft’ 1680/2-l).

a

grant (DFG

of He

References ill Buchsbaum,

4.4. Serotonergic modulation of the intensity pendence of auditory ecoked response?

indicator

11

MS. and Silverman, J. (1968) Stimulus intensity control and the cortical evoked response. Psychoporn. Med., 30: 12-22. 121Stenberg, G., Rosen, I. and Risberg, J. (19881 Personality and augmenting/reducing in visual and auditory evoked potentials. Pen. Indilid. Differ., 9: 571-579. MS., Haier, R.J. and Johnson, J. (19X3) [31 Buchsbaum, Augmenting and reducing: Individual differences in evoked potentials. In A. Gale and J.A. Edwards (Eds.1, Physiological Correlates of Human BehaGour, Vol. III. Academic Press, London, pp. 117-138. study of visual [41 Von Knorring, L. (1978) An experimental averaged evoked responses (V.AER) and pain measures (PM) in patients with depressive disorders. Biol. Pbychol.. 6: 27-38. L., Monakhov, K. and Perris, C. (19781 bl Von Knorring, Augmenting/reducing: an adaptive switch mechanism to cope with incoming signals in healthy subjects and psychiatric patients. Neuropsychobioloby, 4: 150-179. M.T. (19921 ERP augmenting/ l61 Carrillo-de-la-Pena, reducing and sensation seeking: a critical review. ht. J. Psychophysiol., 12: 21 l-220. I., Ulrich. G. and Miiller-Oerling[71 Hegerl, U., Prochno, hausen, B. (19X9) Sensation seeking and auditory evoked potentials. Biol. Psychiatry, 25: 179-190. U., Ulrich, Cl. and Miiller-Oerlinghausen, B. [81 Hegerl, (19871 Auditory evoked potentials and response to lithium prophylaxis. Pharmacopsychiatry, 20: 213-216. U., Wulff, H. and Miiller-Oerlinghausen, B. [91 Hegerl, (1992) Intensity dependence of auditory evoked potentials and clinical response to prophylactic lithium medication: a replication study. Psychiatry Res.,44: 181-191. I101 Hegerl, U. and Juckel, G. (1993) Intensity dependence of auditory evoked potentials as an indicator of central serotonergic neurotransmission - Arguments for a new hypothesis. Biol. PAychiatty, 33: 173-187. [ill Connolly, J.F. and Gruzelier. J.H. (19881 Persistent methodological problems with evoked potential augmenting/reducing. Int. J. Psychophysiol., 3: 299-306. [121 Petrie, A. (1960) Some psychological aspects of pain and the relief of suffering. Ann. N.Y. Acad. Sci., X6: 13-27.

I2 [13]

[14]

[15]

]I61

[I71

[I81

U. Hegerl et al. /Intemutional

Davis, C., Cowles, M. and Kohn, P. (10X3) Strength of the nervous system and augmenting-reducing: a paradox lost. Pers. Indir~id. Differ. .4: 4Yl-1YX. Buchsbaum, MS. and Pfefferhaum, A. (lY7ll Individual differences in stimulus intensity response. P.s~‘(.h~~plly.~io/os. 8: 600-61 I. Scherg, M., Vajsar. .I. and Picton, T.W. (1989) A source analysis of the late human auditory evoked potentials, J. Cogn. Neurmci., I: 336-355. Scherg, M. and Von Cramon, D. (IYYO) Dipole source potentials of the auditory cortex in normal subjects and on patients with temporal lobe lesions. Arlf,. Au&l.. 6: 165-193. Zuckcrman, M., Simona, R.F. and Como. P.G. (198X) Sensation seeking and stimulus intensity as modulators of cortical, cardiovascular and electrodermal response: A cross-modality study. Pm. lndid. LX@. , 0: 361-372. Raine, A., Mitchell, D.A. and Venables, P.H. (19X1) Cortical augmenting reducing - modality specific‘? !%ychophysiolo,qy,

[I91

[21]

[22]

[23]

[24]

[251

18: 700-708.

Kaskey, G.B., Salzman, L.F., Klorman. R. and Pass, H.L. (19X0) Relationship between stimulus intensity and amplitude of visual and auditory event-related potentials. Rio/. Pyychol..

[20]

Journal

of Psychophysiology

Scherg. M. (1990) Fundamentals of dipole source potential analysis. Adr,. Audwl.,6: 40-69. [27] Pantev, C., Hoke, M.. Lehnertz. K.. Liitkenhoner. B.. Fahrendorf, G. and Stiiber. U. (1990) Identification of sources of brain neuronal activity with high spatiotemporal resolution through combination of neuromagnetic

and

magnetic

Clin.

resonance

Neurophysiol..

75:

173-184. [2X] Rogers,

R.L., Papanicolaou. A.C., Baumann, S.B.. Saydjari, C. and Eisenberg. II.M. (1990) Neuromagnetic evidence of a dynamic excitation pattern generating the NIOO auditory response. Eiectroenceph. Clin. Neurophysiol.. 77: 2377240.

[29]

Sams, M., Kaukoranta, E., Hamalainen, M. and NZitanen, R. (1991) Cortical activity elicited by changes in auditory stimuli: different sources for the magnetic NlOOm and mismatch responses, P.~ychophysiology. 2X: 21-2’).

[3O] Arezro,

J., Pickoff, A. and Vaughan, Jr., H.G. (1975) The sources and intracerebral distribution of auditory evoked potentials in the alert rhesus monkey. Bruin Rex.90: s7773. [3l] Knight, R.T.. Scabini, D., Woods, D.L. and Clayworth. c’. (19X8) The effects of lesions of superior temporal gyrus and inferior parietal lobe on temporal and vertex components of the human AEP. Electroenwph. Clim Neurophysid. .70: 4YY%SOY. [32] Woods,

D.L., Clayworth, C.C.. Knight, R.T.. Simpson. G.V. and Naeser. M.A. (1987) Generators of middle- and long-latency auditory evoked potentials: implications from studies of patients with bitemporal Icsions. Electroetz-

ceph. Clim Neurophy.siol.,

6X: I32-

14X.

Makehi, J.P. and Hari, R. (1990) Long-latency auditory evoked magnetic fields. Adr Nemo/., S4: 177-191. [34] Vaughan, Jr.. H.G. and Arezzo. J.C. (1988) The neural basis of event-related potentials. In T.W. Picton (Ed.),

[33]

Hundbook

of Ele~troertce~~hal(~~rapily

physiolo&~,

Rev. .Ser.2 Vol. S, Hurnan

ticzb. Elsevier,

[35]

[36]

[37]

[3X]

329%330. [26]

I-13

source localization (NMSL) imaging (MRI). Electrocnceph

IO: 115-125.

Connolly, J.F. (1987) ERPs suggest the importance of subcortical mechanisms in activities typically associated with cortical functions. In R. Johnson Jr., J.W. Rohrbaugh and R. Parasuraman (Eds.). Current Trends iti El,etlt-related Potential Research (EEG .suppl. 40). Elsevier. Amsterdam, pp. 635-644. Pineda, J.A., Holmes, T.C. and Foote. S.L. (1991) Intensity-amplitude relationships in monkey event-related potentials: parallels to human augmenting-reducing responses. Electroenceph. Clin. Neurophysiol., 78: 456-465. Prescott, J., Connolly, J.F. and Gruzelier. J.H. (1984) The augmenting/reducing phenomenon in the auditory evoked potential. Biol. P.sychiutp. IY: 31-44. Connolly, J.F. and Gruzelier, J.H. (19x2) Amplitude and latency changes in the visual evoked potential to different stimulus intensities. P.&wphysiology, 19: S999608. Naatlnen, R. and Picton, T.W. (1987) The NI wave of the human electric and magnetic response to sound: a review and an analysis of the component structure. PQchophysiology , 24: 375-425. Hegerl, U., Prochno, 1.. Ulrich, G. and Miiller-Oerlinghausen. B. (1988) Are auditory evoked potentials suitable for predicting the response to lithium prophylaxis’? A study on the effects of repeated measurement, age, gender, and personality on the amplitude/stimulus intensity function in healthy volunteers. Phurmucops)chiutr,21:

I7 (1994)

[3Y]

[40]

and Clinical Ewzt-Related

NeuroPoterz-

Amsterdam, pp. 45-96. Scherg, M. and Picton. T.W. (1991) Separation and identificaton of event-related potential components by brain electric source analysis. In C.H.M. Brunia, G. Mulder. M.N. Verbaten (Eds.), Elwzt-related Brairz Reserrrch (EEG Suppl. 42), Elsevier, Amsterdam, pp. 24-37. Mocks, J.. Gasser. T. and Kiihler, W. (198X) Basic statistical parameters of event-related potentials. .I. Psychophysiol.. 2: 6 I-70. Pantev, C.. Hoke, M., Lehnertz, K., Liitkenhoner. B., Anogianakis, G. and Wittkowski, W. (198X) Tonotopic organization of the human auditory cortex revealed by transient auditory evoked magnetic fields. Electroenceph. Clin. Neurophysiol.. 69: I ho- 170. Papanicolaou, A.C., Rogers, R.L.. Baumann, S., Saydjari. C. and Eisenberg. t I.M. (1990) Source localization of two evoked magnetic field components using two alternative procedures. ~q~. Brain Rex. 80: 44448. Buchsbaum, MS., IIenkin, R.I. and Christiansen. R.L. (1974) Age and sex differences in averaged evoked responses in a normal population, with observations on patients with gonadal dysgenesis. Electroenceph. Clim Neurophy.viol.. 37: 137-144. Pelizzone. M.. Williamson, S.J. and Kaufman, L. (1985)

U. Hegerl et al. /International

[41]

[42]

[43]

[44]

Journal of Psychophysiology 17 (1994) l-13

Evidence for multiple areas in the human auditory cortex. In H. Weinberg, G. Stroink, T. Katila (Eds.), Biomagnetism: applications and theory. Pergamon Press, New York, pp. 326-330. Hari, R., Pelizzone, M., Makela, J.P., Hallstrom, J., Leinonen, L. and Lounasmaa, O.V. (1987) Neuromagnetic responses of the human auditory cortex to on- and off-sets of noise bursts. Audiology, 26: 31-43. Lukas, J.H. and Siegel, J. (1977) Cortical mechanisms that augment or reduce evoked potentials in cats. Science, 198: 73-75. Azmitia, E.C. and Gannon, P.J. (1986) The primate serotonergic system: a review of human an animal studies and a report on macaca fascicularis. Adu. Neural., 43: 407468. Lewis, D.A., Campbell, M.J., Foote, S.L. and Morrison, J.H. (1986) The monoaminergic innervation of primate neocortex. Hum. Neurobiol.. 5: 181-188.

13

[45] Campbell, M.J., Lewis, D.A., Foote, S.L. and Morrison J.H. (1987) Distribution of choline acetyltransferase-, serotonin-, dopamine-P-hydroxylase-, tyrosine hydroxyase-immunoreactive fibers in monkey primary auditory cortex. J. Comp. Neural.. 261: 209-220. [46] Takeuchi, Y. and Sane, Y. (1983) Immunohistochemical demonstration of serotonin nerve fibers in the neocortex of the monkey (Macaca fuscata). Anat. Embryol., 166: 155-168. [47] Brown, R.M., Crane, A.M. and Goldman, P.S. (1979) Regional distribution of monoamines in the cerebral cortex and subcortical structures of the rhesus monkey: concentrations and in vivo synthesis rates. Brain Res.,168: 133-150. [48] Papadopoulos, G.C. and Parnavelas, J.G. (1991) Mono amine systems in the cerebral cortex: evidence for anatomical specifity. Prog. Neurobiol., 36: 195-200.