European Journal of Neuroscience, Vol. 19, pp. 2337±2344, 2004 ß Federation of European Neuroscience Societies

Lateral inhibition and habituation of the human auditory cortex

C. Pantev1,2, H. Okamoto2,4, B. Ross1,2, W. Stoll3, E. Ciurlia-Guy4, R. Kakigi5 and T. Kubo6 1Institute for Biomagnetism and Biosignalanalysis, MuÈnster University Hospital, Kardinal-von-Galen-Ring 10, 48129 MuÈnster, Germany 2Rotman Research Institute, Baycrest Centre for Geriatric Care, University of Toronto, Ontario, Canada 3ENT Department, MuÈnster University Hospital, Germany 4ENT-Department, Health Science Centre Sunnybrook, Toronto, Ontario, Canada 5Department of Integrative Physiology, National Institute for Physiological Sciences, Myodaiji, Okazaki, Japan 6Department of Otolaryngology and Sensory Organ Surgery, Osaka University School of Medicine, Osaka, Japan

Keywords: auditory cortex, auditory system, habituation, lateral inhibition, magnetoencephalography, MEG

Abstract The goal of this study was to compare the lateral inhibition and the habituation in the human auditory cortex, two important physiological effects during auditory processing that can be reliably measured by means of magnetoencephalography when recording auditory evoked ®elds. Applying 40-Hz amplitude-modulated stimuli allowed us to record simultaneously the slow transient evoked and the steady-state ®elds and thus to characterize the lateral inhibition and the habituation effect in primary and non-primary auditory cortical structures. The main ®nding of the study is that the lateral inhibition effect of non-primary auditory areas as measured on the major component of the slow transient auditory evoked ®eld (N1) is signi®cantly stronger than the corresponding habituation effect. By contrast, this effect was not observed for the 40-Hz steady-state ®elds, characterizing the activation of the primary auditory cortex in humans. The results might be interpreted as (i) evidence that the inhibition mediated by lateral connections is stronger than the habituation of excitatory neurons in the non-primary auditory cortex and (ii) the processing hierarchy in the human auditory cortex is demonstrated by the different behaviour of lateral inhibition and habituation in primary and non-primary auditory cortical structures.

Introduction The N1m component of the evoked response to an auditory stimulus is mechanism of diminishing the auditory evoked response when repe- known to decrease in amplitude (and increase in latency) if the time titive stimuli are applied. Lateral inhibition spanning a number of interval to the preceding auditory stimulus becomes shorter. This tonotopic channels has been documented in the inferior colliculus and phenomenon has different names, such as habituation (Thompson higher levels (Muller & Scheich, 1988; Vater et al., 1992). The model & Spencer, 1966; Ritter et al., 1968), adaptation (Tarkka et al., of the auditory lateral inhibition is derived from the classic visual 2002), sensory gating (Boutros & Belger, 1999) and refractoriness lateral inhibition scheme by von BeÂkeÂsy (1967). A neuron in the (Budd et al., 1998). All these terms imply assumptions regarding the central auditory pathway is characterized by its tuning curve deter- origin and the mechanisms of the observed amplitude decrement, mined by a characteristic frequency (CF) and it is surrounded by other although the exact mechanisms, the neuro-anatomical basis and the neurons that tonotopically span a range of CFs. If the neuron is functional signi®cance of the N1m decrement are still not fully activated from a lower level, it projects not only to higher levels, understood. The theoretical de®nition of habituation given by but also distributes inhibition via interneuron collaterals laterally to Thompson & Spencer (1966) includes response decrement, response adjacent neurons with higher or lower CFs of their tuning curves. The recovery and dishabituation, and an increased response after the inhibition effect depends upon the ®ring rate of the neuron and the insertion of a deviant stimulus. However, to make a clear determination number of collaterals. We propose that inhibition mediated by lateral of whether the N1m response decrement is associated with habituation connections can be assumed as an active mechanism that inervates or refractoriness is very dif®cult because they may be superimposed or inhibitory neurons and causes the decrement of the auditory evoked interactive (Picton et al., 1976). In this study we have adopted the name response. By contrast, the habituation can be understood as a passive habituation for the N1m decrement caused by a preceding sound in the inhibitory mechanism that causes the decrease of the auditory evoked spectral range of the stimulus. response through reduction of the sensitivity of excitatory neurons to Lateral inhibition in the central auditory pathway (Ehret & the applied auditory stimuli. Merzenich, 1988; Burrows & Barry, 1990; Suga, 1995) is another In normal hearing subjects, spectrally notched sound (a sound from which the power in a speci®ed frequency band was completely removed) simulates the type of reversible functional deafferentation Correspondence: Dr Christo Pantev, 1Institute for Biomagnetism and Biosignalanalysis, as (Pantev et al., 1999). In a magnetoencephalographic (MEG) experi- above. ment it has been demonstrated that after prolonged exposure to E-mail: [email protected] spectrally notched music, neurons in the auditory cortex responsive Received 6 June 2003, revised 3 February 2004, accepted 5 February 2004 to frequencies within the notch were strongly inhibited. This result was doi:10.1111/j.1460-9568.2004.03296.x 2338 C. Pantev et al. interpreted as re¯ecting the lasting effect of a lateral inhibitory process, amplitude modulated with a 40-Hz sinusoid, resulting in frequency i.e. the inhibitory in¯uence of the stimulated neurons on to those spectra as shown in Fig. 2b and c. This stimulus design allowed us to neighbouring neurons with CFs of their tuning curves within the investigate simultaneously the evoked responses from the primary and notched area. Recently, a psychoacoustic study (Norena et al., the non-primary auditory cortical areas (Engelien et al., 2000). The 2000) and an EEG experiment (Kadner et al., 2002) con®rmed our onset of the stimuli evoked an N1m response (the magnetic counterpart results (Pantev et al., 1999). of the slow auditory evoked potential with a peak amplitude around The goal of this study was to compare simultaneously in a single 100 ms after stimulus onset) originating mainly from non-primary experiment the decrease in activation of the human auditory cortex auditory structures (Pantev et al., 1995), whereas the 40-Hz rhythm of induced by habituation and lateral inhibition, respectively. the amplitude modulation evoked a 40-Hz steady-state response, which has its sources mainly in the primary auditory cortex (Pantev Materialsand methods et al., 1996). Both MSs were of 3 s duration including 20 ms rise and decay Subjects times. The durations of the CS and TS were 500 ms (12.5 ms rise and Ten right-handed subjects (six females, 35.6 Æ 7.7 years) with no decay times). The silent intervals between the CS and MS and history of otological or neurological disorders participated in this between the MS and TS, respectively, were 500 ms; the silent study. Their hearing thresholds were 15 dB hearing level (HL) or interval between a TS and the succeeding CS was 2.5 s, resulting better, in the frequency range from 250 to 8000 Hz as tested by means in a total duration of 7.5 s for the stimulus sequence. Two hundred of pure tone audiometry. The subjects consented to their participation sequences of both stimulus types (PB and SB) and the same MS were after they were completely informed about the nature of the study. The presented in a random order within one session having duration of Ethics Commission of the Baycrest Centre for Geriatric Care approved about 1 h. A second session on a different day was performed for the all experimental procedures, which are in accordance with the Declara- second MS (broad band noise, white noise) and was counterbalanced tion of Helsinki. between subjects. The two stimuli and the two masker sounds were adjusted separately Experimental design and stimulation for the same intensity above normative thresholds obtained from The schema given in Fig. 1 explains the design of the auditory subjective tests on a group of ten subjects. At the beginning of each stimulation. A masking sound (MS) was preceded by a control experimental session the individual hearing threshold for one of the stimulus (CS) and succeeded by a test stimulus (TS), which was stimuli was determined. All stimuli were presented binaurally at the identical to the CS. The masker sounds were two noise signals with intensity of 45 dB sensation level (SL). different spectral properties. The amplitude spectrum of the ®rst MS The stimuli were prepared as sound-®les and presented under (Fig. 2a) contained alternating series of pass- and stop-band sections of control of STIM software (NeuroScan Inc., El Paso, USA) using the same width on the logarithmic frequency scale. The centre ER30 transducers (Etymotic Research, Elk Grove, USA), re¯ection- frequencies of the pass-band sections were spaced by half an octave less plastic tubes of 2.5 m length, and silicon ear pieces, ®tting to the between 0.5 and 2.8 kHz. The suppression in the stop-bands was subject's ears. typically 36 dB. Because of the periodical structure of the spectrum, this MS was called comb-®ltered noise (CFN). It was obtained from a Data acquisition white noise signal by Fourier ®ltering. The spectral distribution of the Auditory evoked magnetic ®elds (AEFs) were recorded with a helmet- second MS is shown in Fig. 2d. It was a broadband noise signal that shaped 151-channel whole cortex neuro-magnetometer (OMEGA, was limited only by the transfer characteristic of the sound delivery CTF Systems Inc., Vancouver, Canada) in a quiet magnetically system. The CS and TS were complex sounds composed from ®ve shielded room. The subjects were placed in a comfortable seated spectral components each, corresponding either to the pass-band position. In order to keep them in an alert state, a self-chosen soundless sections of the CFN (0.7, 1.0, 1.4, 2.0, 2.8 kHz, pass-band stimulus, video movie was presented during the MEG measurement. The PB) or to the stop-band sections of the CFN (0.59, 0.83, 1.19, 1.66, magnetic ®eld signals were 200-Hz low-pass ®ltered and sampled 2.39kHz, stop-band stimulus, SB). Both PB and SB were 100% at a rate of 625/s.

TS CS MS TS

stimulus sequence = 7.5s

3 2 1 0 1 2 3 4 5 Time (s)

Fig. 1. Schematic representation of the stimulus sequence. Identical control (CS) and test stimuli (TS) of 500 ms duration and masking sound (MS) of 3 s duration. Sequences involving different stimulus types were presented in random order.

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 Lateral inhibition and habituation of the human auditory cortex 2339

0 a comb filtered noise −20

−40

−60 0 b PB--stimulus -20

-40

−60 0 c SB--stimulus −20

Amplitude Spectrum (dB) −40

−60 0 d wide band noise −20

−40

−60 0.2 0.5 0.7 1.0 1.4 2.0 2.8 5.0 Frequency (kHz)

Fig. 2. Amplitude spectra of the auditory stimuli. The sounds were recorded at the silicon ear piece ®tting to the subject's ear. (a) Spectrum of the comb-®ltered noise signal. The distances between the centre frequencies of the pass-band sections are of half octave in the range between 0.5 and 2.8 kHz. (b) Complex sound stimulus composed of ®ve spectral components corresponding to the centre frequencies of the comb-®ltered noise pass-band sections and therefore called pass-band stimulus (PB). Each component shows spectral peaks at its carrier frequency and at two sideband frequencies of 40 Hz below and above, due to the amplitude modulation frequency of 40 Hz. (c) Complex sound stimulus with frequency components corresponding to the stop-band sections (SB) of the comb-®ltered noise. (d) Wide band noise masker signal. All acoustic spectra re¯ect the low pass characteristic of the sound transmission system.

Data analysis subject (CS and TS under four conditions). Based on the median of the source coordinates and orientations across these eight values the For each of the four stimulus conditions (two MS and two CS/TS) method of source space projection (SSP) was applied to all averaged epochs of magnetic ®eld data beginning 300 ms before the onset of the magnetic ®eld data. The SSP combines the magnetic ®eld waveforms CS and ending 300 ms after the offset of the TS (total duration 5.6 s) obtained from each sensor weighted by the sensitivity of each sensor were averaged after artefact rejection (threshold 3.0 pT). After 20-Hz for a source at the speci®ed location into a single waveform of low-pass ®ltering and DC offset correction based on 200-ms intervals magnetic dipole moment. The dipole moment is an equivalent of before the CS and TS, respectively, an equivalent single dipole (spatio- the cortical activation. N1m amplitudes were measured as peak dipole temporal model solution) was approximated to the magnetic ®eld moment occurring about 100 ms after stimulus onset referenced to the distribution around the maximum of the global ®eld power (measured mean of a 200-ms interval preceding the stimulus. Because the as root-mean-square across all channels) at about 100 ms after stimulus absolute values of the N1m amplitudes differ strongly among subjects, onset. The dipole location and orientation in the left and in the right the effect on the N1m amplitude to TS and CS stimuli, which has been hemisphere were determined in a head-based Cartesian coordinate clearly observed in single subjects, would be washed out in the group. system with the origin set to the midpoint of the medial±lateral axis Therefore, in order to be independent from these individual amplitude

(y-axis) between the entrances of the left and right ear canals. The variations the ratio TSampl/CSampl rather than the absolute N1m posterior±anterior axis (x-axis) ran between the nasion and the origin, values was evaluated. The results obtained were statistically tested and the inferior±superior axis (z-axis) through the origin perpendicu- using a three-factor (stimulus type  noise type  hemisphere) larly to the (x±y-plane). Estimates of the source parameters were repeated-measures ANOVA. Differences were accepted as signi®cant accepted for further evaluation only if both the goodness of ®t of at P < 0.05. the ®eld of the estimated equivalent current dipole (ECD) to the The steady-state responses (SSRs) were analysed in a similar way. measured magnetic ®eld was greater than 90%, and the distance of The model of an equivalent current dipole was applied to the averaged the ECD to the mid-sagittal plane was greater than 3 cm. The source magnetic ®eld data, band-pass ®ltered between 30 and 50 Hz in the analysis resulted in eight estimates for the N1m sources for each 300-ms time interval beginning 200 ms after the onset of the CS and TS

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 2340 C. Pantev et al. until the end of the stimulus. For each subject the median of the eight N1m amplitude was 16.4% on the right and 21.8% on the left. In accepted SSR source estimations was used for the source space comparison, for the PB stimulus and CFN the N1m reduction was 8.1% projection. A Hilbert transform calculated the envelope of the SSR in the right hemisphere and 6.7% in the left hemisphere. When dipole moment waveforms, and the mean envelope in the 300-ms presenting the white noise, N1m amplitude reduction of about 8% interval beginning 200 ms after the stimulus onset served as a measure was found in both hemispheres for both stimuli. for the SSR amplitude (Ross et al., 2002). The SSR amplitude ratio, as The ANOVA of the N1m amplitude ratios R ˆ TSampl/CSampl resulted described above, was calculated and statistically analysed. in a signi®cant effect of the factors stimulus type (F1,9 ˆ 7.65, P ˆ 0.021), a tendency for the noise type (F1,9 ˆ 4.44, P ˆ 0.065) and a strong interaction between noise and stimulus type (F ˆ 29.3, Results 1,49 P < 0.00001). Multiple t-tests with Bonferoni correction of the P- Clearly identi®able auditory evoked responses were obtained from all values demonstrated a signi®cantly stronger N1m decrement for the subjects. The example of individual magnetic ®eld waveforms shown stop-band stimulus after comb ®ltered noise than for all other combi- in Fig. 3b demonstrates the N1m responses after onset of the stimulus nations of stimulus and noise types (SB/CFN > PB/CFN, P ˆ 0.00001; and masker signals as the most pronounced components of the AEF. In SB/CFN > SB/wide, P ˆ 0.00019; SB/CFN > PB/wide, P ˆ 0.0040). addition, the P1m and P2m waves preceding and following the N1m as The N1m amplitude ratio for the effect of the CFN on the SB stimulus well as the off responses are clearly seen in the magnetic ®eld was R ˆ 0.782 in the left hemisphere and R ˆ 0.836 in the right waveforms. The iso-contour plots of the magnetic ®eld distributions hemisphere. Under all other conditions the ratio was around at the times of the maximum global ®eld power of the N1m responses R ˆ 0.93 with no signi®cant difference between the combinations of to the CS and TS show the typical pattern of dipolar sources in both noise and stimulus types. A summary of the N1m amplitude decrement hemispheres (Fig. 3c). The dipole moment waveform, which resulted caused by exposure to the intervening noise is given in Fig. 7. The from a grand average across all ten subjects and is shown in Fig. 3d, error-bars in this ®gure denote the 95% con®dence limits of the group exhibits in detail the P1m±N1m±P2m responses to the onset of the CS, mean and demonstrate the signi®cant contrast in case of the CFN, MS and TS sounds. In addition, the decrement in the N1m amplitude which was absent when the broadband noise was applied. between CS and TS, which was compared between the various None of the SSR amplitude ratios was signi®cantly different from stimulus conditions, is evident in Fig. 3d. Furthermore, response signal 1.0. This different behaviour of the SSR as compared with the N1m details including the off response after the end of the MS sound, the responses is shown in the bar chart of the mean ratios in Fig. 7. The sustained responses during presentation of all stimuli, and the SSR error bars in this diagram denote the 95% con®dence limits, which responses during the CS and TS are visible in the grand averaged were in the same order of Æ5% for the differences in both N1m and source waveforms. SSR responses. Thus the sensitivity detecting an amplitude difference The magnetic ®eld waveforms of the SSR are displayed in Fig. 4b. was equal for both evoked response components. The 30- to 50-Hz band-pass ®ltered signals exhibit the onset response and the development of the SSR after CS and TS onsets. The selected Discussion waveforms of channels around the right hemispheric ®eld maxima show clear polarity reversal. Even at ®eld amplitudes, which are The decrements of the N1m response caused by a preceding masking smaller by a factor of 20 than those of the N1m, the iso-contour plots noise were related to the type of noise. In the case of the CFN and the of the magnetic ®eld distribution demonstrate the typical pattern PB stimulus the relevant stimulus frequencies were masked. Conse- resulting from dipolar sources. The grand averaged SSR waveforms quently, habituation is the main effect causing the decrement of the in Fig. 4d show the constant SSR envelope in the 200±500-ms interval N1m amplitude under this condition. In the case of the CFN and the SB after stimulus onsets. stimulus only frequency bands neighbouring the stimulus were in¯u- The grand averaged dipole source locations in the y±z-plane (med- enced. Here lateral inhibition caused the N1m decrement. Using ial±lateral, inferior±superior directions) and the y±x-plane (medial± amplitude-modulated stimuli allowed us to observe corresponding lateral, posterior±anterior directions) as shown in Fig. 5 demonstrate effects on the 40-Hz steady-state responses. Thus, we were able for the signi®cant separation of N1m and SSR sources (result of t-test the ®rst time to compare simultaneously the change in activation of the (d.f. ˆ 9) for the distances between N1m and SSR sources: Left primary and non-primary human auditory cortex due to habituation hemisphere: x: n.s., y: P ˆ 0.008, z: P ˆ 0.0008; Right hemisphere: and lateral inhibition, respectively. x: n.s., y: P ˆ 0.04, z: P ˆ 0.0001). In general the SSR sources were The main ®nding of this study is that the TS/CS amplitude ratio of found to be about 0.5 cm more medial and nearly 1.0 cm more superior the N1m response is signi®cantly smaller to SB stimuli preceded by in both hemispheres. Both the SSR and the N1m sources were found CFN than to any of the other stimulus conditions. This ®nding was more anteriorly located in the right hemisphere. obtained at an intensity level of 45 dB SL, where the frequency The comparison of the grand averaged AEF evoked by the CS and speci®city in the auditory system is still preserved. The result can TS for the four combinations of stimulus and masking sounds is shown be interpreted that the inhibitory effect based on lateral inhibition upon in Fig. 6. The N1m amplitudes to the SB stimulus were 5.7 nAm larger the activation of secondary auditory areas as measured by their N1m than the mean amplitude of 78.4 nAm in case of the PB stimulus. The evoked response was signi®cantly stronger than the corresponding N1m amplitudes were on average 11.5 nAm larger in the right than in habituation effect. At ®rst glance the observed difference in N1m the left hemisphere. The mean N1m amplitudes in response to the test amplitude reduction of about 14% may appear small and inconsider- stimuli were 10.1 nAm smaller than the control stimuli before noise able. However, taking into account the experimental results of Ross exposure. The mean amplitudes were not different with respect to the et al. (1999), the reduction of N1m amplitude at 45 dB by 14% is noise types. The 95% con®dence interval for these comparisons was equivalent to lowering the corresponding stimulus intensity from 45 to Æ5.6 nAm or smaller. For all combinations of noise and stimulus types 26 dB; it is thus highly relevant. the TS response was reduced as compared with the CS response. In contrast to the results obtained for N1m response to the stimulus However, the combination of the SB stimulus with the CFN shows the onset, the comparison of habituation and lateral inhibition did not largest effect in both hemispheres. The amplitude reduction of the reveal any signi®cant differences in their inhibitory effect on primary

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 Lateral inhibition and habituation of the human auditory cortex 2341

Fig. 3. MEG recording of the slow auditory evoked responses. (a) Stimulation sequence consisting of control stimulus (CS), conditioning masker stimulus (MS) and the test stimulus (TS). (b) Overlay of individual magnetic ®eld waveforms of selected MEG channels over the auditory cortices (band-pass ®ltered between 1 and 24 Hz). (c) The contour map representing the magnetic ®eld distribution at the maximum of the responses to the control stimulus (left) and to the test stimulus (right). The corresponding time points are denoted by arrows in (b). (d) Grand average across over the ten subjects investigated of the cortical source waveforms obtained from the source space projection approach (low-pass ®ltered at 0±80 Hz). The graphic indicates the amplitude difference between the responses to the test and control stimuli.

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 2342 C. Pantev et al.

Fig. 4. MEG recording of auditory steady-state responses. (a) Stimulus sequence. (b) Overlay of selected magnetic ®eld waveforms band-pass ®ltered between 30 and 50 Hz. The insets display the responses to the control and test stimuli on an enlarged time scale. (c) Contour map of the magnetic ®eld distribution at the time points of maxima of the oscillatory response to both amplitude-modulated stimuli. (d) Grand average of the cortical SSR waveforms obtained across all ten subjects after source space projection (band-pass ®ltered between 20 and 80 Hz). The inserts show the SSR to both stimuli on an enlarged time scale. The constant amplitude in the interval 250±500 ms after stimulus onset was compared between test and control stimulus.

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 Lateral inhibition and habituation of the human auditory cortex 2343

left hemisphere right hemisphere Left Hemisphere Right Hemisphere comb filtered wide band comb filtered wide band SSR SSR noise noise noise noise SB PB SB PB SB PB SB PB 7 1.0

× × × × × × 0.9 N1 N1 6 × 0.8 × Amplitude Ratio inferior -- superior inferior (cm) -- N1

3 1.1 SSR SSR N1

SSR × × ×

1.0 × × × ×

N1 × 2 Amplitude Ratio 0.9

posterior -- anterior (cm) posterior -- Fig. 7. Amplitude ratios between test and control stimuli. They were obtained 5 4 3 3 4 5 under the various experimental conditions for the left and the right hemisphere. left lateral -- medial -- right lateral (cm) The error bars denote the 95% con®dence intervals for the mean amplitude ratios. Fig. 5. Grand mean localization of N1 sources (square symbols) and SSR sources (circles). They are presented in the y±z-plane (inferior±superior direc- tion vs. medial±lateral direction) and y±x-plane (posterior±anterior vs. medial± lower spectral content may explain in general the larger response to the lateral direction). The orientations of the equivalent current dipoles are repre- SB stimulus. In the case of broadband noise it could be argued that both sented by the solid lines starting at each dipole location. The ellipses around the lateral inhibition and habituation effects in¯uence the decrease of SSR dipole denote the 95% con®dence limits for the distance between SSR and auditory cortical activation and thus in this condition the largest N1m N1 sources. amplitude decrements could be expected. However, this was not con®rmed by the obtained experimental data. Thus, it seems that for the broadband noise the excitatory and lateral inhibitory effects are Left Hemisphere Right Hemisphere counterbalanced and it could be assumed that by analogy to the enhancement of edges of an image in the visual system the lateral inhibition in the auditory system becomes especially effective only in SB SB the presence of spectral contrasts. The lateral inhibition results of our study most likely re¯ect the comb filtered noise comb filtered noise remodelling of lateral connections in the central part of the auditory PB PB system (Norena et al., 2000) and can be satisfactorily explained by the slightly modi®ed model proposed by these authors. According to this SB SB model the activity of neurons with characteristic frequencies of their tuning curves far away from the notch edges is mainly determined by white noise white noise Dipole Moment their excitatory inputs, and, in addition, by the lateral inhibition PB PB in¯uence that they receive from neighbouring neurons. The spectral notch causes a contrast and, respectively, an imbalance in the pattern of 50 nAm CS excitation and inhibition. Neurons with characteristic frequencies TS close to the notch edges, but still outside the notch, receive less inhibitory in¯uence from their neighbours because some of these 0 200 400 600 0 200 400 600 neighbours have characteristic frequencies that are located inside Time (ms) Time (ms) the notch and therefore are not excited. The consequence is an Fig. 6. Grand averages of dipole moment waveforms. They correspond to the unchanged or even increased activation of the neurons close to the response of the control stimulus (solid lines) and the test stimulus (dashed lines) cut-off frequency of the notch. By contrast, those neurons have strong for both hemispheres, the different types of masking signal and the two types of inhibitory in¯uence on the neurons inside the notch region, and, stimuli with different spectral contents. CS, control stimulus; TS, test stimulus; SB, response waveforms to the stimuli with a frequency spectrum correspond- because these are not excited owing to the spectral notch, their activity ing to the stop-band sections of the comb-®ltered noise; PB, responses to stimuli is strongly suppressed. The psychoacoustic results of Norena et al. spectrally corresponding to the pass-band frequencies. (2000) fully support this model and they are in line with the data obtained in the present experiment. However, the results of this auditory structures at least as characterized by the simultaneously experiment point out that this is the case for the auditory associative recorded SSR during the ongoing stimulus. areas, where the generators of the N1m are assumed, whereas this does The amplitude of the N1m to the SB stimulus was larger than the not seem to be the case for the primary auditory areas, which are response to the PB stimulus. All ®ve spectral components of the SB assumed to generate the SSR. Thus our results agree with the results of stimulus consist of lower frequencies on logarithmic scale as compared the functional magnetic resonance imaging study of Wessinger et al. with the corresponding components of the PB stimulus. Even when the (2001), demonstrating processing hierarchy in the human auditory intensity of both stimuli was adjusted with respect to the sensation, the cortex.

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344 2344 C. Pantev et al.

Acknowledgements Muller, C.M. & Scheich, H. (1988) Contribution of GABAergic inhibition to the response characteristics of auditory units in the avian forebrain. J. Neuro- We thank T. Picton and L. Roberts for critical comments on an early version of physiol., 59, 1673±1689. the manuscript and CIHR, Ontario Innovation Trust, and Canadian Foundation Norena, A., Micheyl, C. & Chery-Croze, S. (2000) An auditory negative after- for Innovation for support. image as a human model of tinnitus. Hear. Res., 149, 24±32. Pantev, C., Bertrand, O., Eulitz, C., Verkindt, C., Hampson, S., Schuierer, G. & Elbert, T. (1995) Speci®c tonotopic organizations of different areas of the Abbreviations human auditory cortex revealed by simultaneous magnetic and electric recordings. Electroencephalogr. Clin. Neurophysiol., 94, 26±40. AEF, auditory evoked ®eld; CF, characteristic frequency; CFN, comb-®ltered Pantev, C., Roberts, L.E., Elbert, T., Ross, B. & Wienbruch, C. (1996) Tonotopic noise; CS, control stimulus; HL, hearing level; MEG, magnetoencephalogra- organization of the sources of human auditory steady-state responses. Hear. phy; MS, masking sound; PB, pass-band stimulus; SB, stop-band stimulus; SL, Res., 101, 62±74. sensation level; SSP, source space projection; SSR, steady-state response; TS, Pantev, C., Wollbrink, A., Roberts, L.E., Engelien, A. & Lutkenhoner, B. (1999) test stimulus. Short-term plasticity of the human auditory cortex. Brain Res., 842, 192±199. Picton, T., Hillyard, S. & Galambos, R. (1976) Habituation and attention in the auditory system. In Keidel, WaNWe (Eds), Handbook of Sensory Physiol- References ogy. Springer, Berlin, pp. 343±390. Ritter, W., Vaughan, H.G. Jr & Costa, L.D. (1968) Orienting and habituation to von BeÂkeÂsy, G. (1967) Sensory Inhibition. Princeton University Press, Prince- auditory stimuli: a study of short term changes in average evoked responses. ton, NJ. Electroencephalogr. Clin. Neurophysiol., 25, 550±556. Boutros, N.N. & Belger, A. (1999) Midlatency evoked potentials attenuation Ross, B., Lutkenhoner, B., Pantev, C. & Hoke, M. (1999) Frequency-speci®c and augmentation re¯ect different aspects of sensory gating. Biol. Psychiatry, threshold determination with the CERAgram method: basic principle and 45, 917±922. retrospective evaluation of data. Audiol. Neurootol., 4, 12±27. Budd, T.W., Barry, R.J., Gordon, E., Rennie, C. & Michie, P.T. (1998) Ross, B., Picton, T.W. & Pantev, C. (2002) Temporal integration in the human Decrement of the N1 auditory event-related potential with stimulus auditory cortex as represented by the development of the steady-state repetition: habituation vs. refractoriness. Int. J. Psychophysiol., 31, magnetic ®eld. Hear. Res., 165, 68±84. 51±68. Suga, N. (1995) Sharpening of frequency tuning by inhibition in the central Burrows, D.L. & Barry, S.J. (1990) Electrophysiological evidence for the auditory system: tribute to Yasuji Katsuki. Neurosci. Res., 21, 287±299. critical band in humans: middle-latency responses. J. Acoust. Soc. Am., Tarkka, I.M., Lehtovirta, M., Soininen, H., Paakkonen, A., Karhu, J. & 88, 180±184. Partanen, J. (2002) Auditory adaptation is differentially impaired in familial Ehret, G. & Merzenich, M.M. (1988) Complex sound analysis (frequency and sporadic Alzheimer's disease. Biomed. Pharmacother., 56, 45±49. resolution, ®ltering and spectral integration) by single units of the inferior Thompson, R.F. & Spencer, W.A. (1966) Habituation: a model phenomenon for colliculus of the cat. Brain Res., 472, 139±163. the study of neuronal substrates of behavior. Psychol. Rev., 73, 16±43. Engelien, A., Schulz, M., Ross, B., Arolt, V. & Pantev, C. (2000) A combined Vater, M., Habbicht, H., Kossl, M. & Grothe, B. (1992) The functional role of functional in vivo measure for primary and secondary auditory cortices. Hear. GABA and glycine in monaural and binaural processing in the inferior Res., 148, 153±160. colliculus of horseshoe bats. J. Comp. Physiol. [A], 171, 541±553. Kadner, A., Viirre, E., Wester, D.C., Walsh, S.F., Hestenes, J., Vankov, A. & Wessinger, C.M., VanMeter, J., Tian, B., Van Lare, J., Pekar, J. & Rauschecker, Pineda, J.A. (2002) Lateral inhibition in the auditory cortex: an EEG index of J.P. (2001) Hierarchical organization of the human auditory cortex revealed tinnitus? Neuroreport, 13, 443±446. by functional magnetic resonance imaging. J. Cogn. Neurosci., 13, 1±7.

ß 2004 Federation of European Neuroscience Societies, European Journal of Neuroscience, 19, 2337±2344