Exp Brain Res (1995) 104:144-152
ORIGINAL
9 Springer-Verlag 1995
PAPER
S. Kuriki. Y. Okita 9Y. Hirata
Source analysis of magnetic field responses from the human auditory cortex elicited by short speech sounds
Received: 30 May 1994 / Accepted: 7 December 1994
Abstract We made a detailed source analysis of the magnetic field responses that were elicited in the human brain by different monosyllabic speech sounds, including vowel, plosive, fricative, and nasal speech. Recordings of the magnetic field responses from a lateral area of the left hemisphere of human subjects were made using a multichannel SQUID magnetometer, having 37 fieldsensing coils. A single source of the equivalent current dipole of the field was estimated from the spatial distribution of the evoked responses. The estimated sources of an N l m wave occurring at about 100 ms after the stimulus onset of different monosyllables were located close to each other within a 10-mm-sided cube in the three-dimensional space of the brain. Those sources registered on the magnetic resonance images indicated a restricted area in the auditory cortex, including Heschl's gyri in the superior temporal plane. In the spatiotemporal domain the sources exhibited apparent movements, among which anterior shift with latency increase on the anteroposterior axis and inferior shift on the inferosuperior axis were common in the responses to all monosyllables. However, selective movements that depended on the type of consonants were observed on the mediolateral axis; the sources of plosive and fricative responses shifted laterally with latency increase, but the source of the vowel response shifted medially. These spatiotemporal movements of the sources are discussed in terms of dynamic excitation of the cortical neurons in multiple areas of the human auditory cortex.
Key words Auditory cortex 9Evoked magnetic field. Speech sound - Current dipole source - Human
S. Kuriki Research Institute for Physiological Sciences, Okazaki National Institutes, Okazaki 444, Japan S. Kurild ( ~ ) . Y. Okita 9Y. Hirata Research Institute for Electronic Science, Hokkaido University, N-12 W-6, Kita-ku, Sapporo 060, Japan; Fax no: +81-11-706-4976, e-maih sk@ ae.hines.hokudai.ac.jp
Introduction Various electrophysiological methods have been employed to study neural functions related to speech and language in different cortical regions of the human brain. From the deficits of object naming and short-term verbal memory evoked by electrical stimulation, it has been shown that multiple regions in the lateral temporal cortex (including Wernicke's area) of the dominant hemisphere are involved in the language function (Penfield and Roberts 1959; Fedio and Van Buren 1974; Ojemann 1978). In the sites identified by electrical stimulation as being involved with language, attenuation of alpha-band spectral density in the electrocorticogram has been observed during silent naming (Ojemann et al. 1989). Microelectrode recordings have revealed the existence of neural activities that are evoked by specific aspects of spoken language in the superior temporal gyms of both hemispheres (Creutzfeld et al. 1989). Neural activities in the superior temporal gyms during overt speech and in the anterior temporal lobe during silent speech have been observed in the microelectrode recordings (Ojemann et al. 1988). It is also reported that poststimulus inhibition of the neural activity occurs during visual word recognition in most cortical sites on the lateral surface of the frontal, temporal and parietal region (Bechtereva et al. 1992). Event-related potentials (ERPs) recorded on the scalp during tasks requiring the processing of speech sounds or words have been widely studied, using this noninvasive method. The ERPs elicited by verbal stimuli show more or less symmetrical scalp distribution, suggesting the participation of both hemispheres in language processing (Hillyard and Picton 1987). Lateralized distribution is observed, however, in some components of the ERPs associated with word processing (Kutas et al. 1988). During semantic processing of words, the ERP distribution extends to more a posterior part than that during verbal identification (Lovrich et al. 1988). As for the cortical generators of the scalp ERPs elicited by auditory stimuli, multiple sources in the superior
145 t e m p o r a l p l a n e and in the s u p e r i o r t e m p o r a l g y r u s were s u g g e s t e d (Vaughan et al. 1980; W o o d and W o l p a w 1982). However, little is k n o w n about the function o f the a u d i t o r y area in the s u p r a t e m p o r a l plane, m a i n l y b e c a u s e the s u p r a t e m p o r a l p l a n e lies d e e p within the s y l v i a n fissure and requires d e p t h e l e c t r o d e r e c o r d i n g for d e t a i l e d p h y s i o l o g i c a l analysis. T h e p r i m a r y a u d i t o r y area was l o c a t e d at H e s c h l ' s g y m s in the s u p r a t e m p o r a l p l a n e ( C e l e s i a 1976; L e e et al. 1984; L i e g e o i s - C h a u v e l 1991). N e u r o m a g n e t i c fields r e c o r d e d with a S Q U I D (superc o n d u c t i n g q u a n t u m i n t e r f e r e n c e device) over the scalp are b e l i e v e d to o r i g i n a t e f r o m electric currents that flow c o h e r e n t l y w i t h i n a p o p u l a t i o n o f cortical neurons in loc a l i z e d spots (Hari 1990). F r o m the m e a s u r e m e n t s o f auditory e v o k e d fields, it was s u g g e s t e d that their m a i n source e x i s t e d in the s u p r a t e m p o r a l p l a n e (Hari et al. 1980; E l b e r l i n g et al. 1982; S a m s et al. 1985). R e c e n t studies c o m b i n e d with m a g n e t i c r e s o n a n c e ( M R ) i m a g ing have c o n f i r m e d that the sources o f the e v o k e d fields b y pure tones (Reite et al. 1988; Pantev et al. 1990; Pap a n i c o l a o u et al. 1990) and s p e e c h sounds (Kuriki 1992) are l o c a t e d at the a u d i t o r y area in the s u p r a t e m p o r a l plane. T h e n o n i n v a s i v e n e u r o m a g n e t i c a p p r o a c h is suited for studying the cortical activities in the s u p r a t e m p o r a l plane, since the c o m p o n e n t o f the currents p a r a l l e l to the skull e x c l u s i v e l y contributes to the m a g n e t i c field outside o f the skull. T h e m a g n e t i c field is h a r d l y i n f l u e n c e d b y the e l e c t r i c a l c o n d u c t i v i t y o f the b r a i n tissue and the skull and is s p a t i a l l y l o c a l i z e d a r o u n d the skull near the source, w h i c h leads to h i g h spatial r e s o l u t i o n w h e n estim a t i n g the l o c a t i o n o f the source o f the currents f r o m the m e a s u r e d field distribution. Several reports have d e s c r i b e d the m a g n e t i c field responses w h i c h are e l i c i t e d b y speech sounds or c o m p l e x sounds ( K a u k o r a n t a et al. 1987; M a k e l a et al. 1988; Kuriki and M u r a s e 1989; H a y a s h i et al. 1992; K u r i k i 1992). A spatial difference has b e e n o b s e r v e d in the e s t i m a t e d source l o c a t i o n b e t w e e n the r e s p o n s e s to fricative/plosive c o n s o n a n t and v o w e l speech sounds, s u g g e s t i n g the sensitivity o f the u n d e r l y i n g neural activity in the suprat e m p o r a l p l a n e to the acoustic attribute o f the speech sounds ( K a u k o r a n t a et al. 1987; K u r i k i and M u r a s e 1989). D u r i n g the p a s t 5 years, l a r g e - s c a l e m u l t i c h a n n e l S Q U I D s y s t e m s have b e e n d e v e l o p e d , w h i c h a l l o w us to o b t a i n t i m e - c o h e r e n t m a g n e t i c field d a t a f r o m a w i d e area over the scalp and thus high s p a t i a l - t e m p o r a l resolution is assured. In the p r e s e n t study w e have m a d e a detailed source a n a l y s i s o f the m a g n e t i c field r e s p o n s e s that are elicited b y m o n o s y l l a b i c speech sounds. Spatial and t e m p o r a l b e h a v i o r s o f the equivalent current sources have b e e n a n a l y z e d to d e l i n e a t e the d y n a m i c characteristic o f the activity o f the a u d i t o r y cortex.
Materials and methods Subjects The present study was approved by the Ethics Committee for Human Research of the National Institute for Physiological Sciences
and performed under the regulations of the 1964 Declaration of Helsinki. Eight male subjects with normal hearing, from whom informed consent was obtained following an explanation of what the task involved, participated in the experiments. In two subjects, the amplitude of the magnetic field responses to speech sounds was extremely low over the left hemisphere. In the present study magnetic field data obtained from the other six subjects were analyzed. Their ages ranged from 27 to 37 years (mean 30.5, SD 4.7). Stimuli Monosyllabic speech sounds including vowel/a/, nasal-consonant vowel/na/, plosive-consonant vowel/ka/, and fricative-consonant vowel/ha/were used as stimuli. These speech sounds of 150 to 200 ms, which were pronounced by a broadcast announcer, were digitized at a 20-kHz sampling rate and stored in the memory of a personal computer. During the measurements these stimuli were read out from the memory in a pseudo-random order, without succession of identical stimuli at a constant onset-to-onset interval of 1.28 s. In addition to these, oddball stimuli of four monosyllables, which had 50-120 ms longer duration, were mixed randomly in such a way that the occurrence probability was about 3%. The stimuli were then amplified, filtered, transformed into sound with a transducer, which did not produce magnetic noise, and delivered to the subject binaurally through plastic tubes and earpieces. The passband of the sound at the output of the earpiece was 60- to 3800-Hz frequency. The acoustic level of the stimuli was 80-87 dB sound pressure level (SPL). Procedures Measurements of the auditory evoked magnetic fields were made in a magnetically shielded room. A 37-channel SQUID (BTi) was used, with 37 sensing coils covering a circular area of 14 cm diameter on a spherical surface of 12 cm radius. The sensing coils consisted of first-order gradiometers 2 cm in diameter, with a baseline of 5 cm. The subject lay on a bed while he received the binaural auditory stimuli. A temporal area centered over the sylvian fissure in the left hemisphere was measured. The subject was instructed to keep his eyes open and count the occurrence of the oddball stimuli. This procedure was adopted to keep the vigilance level of the subject constant. A single measurement-run contained 528 epochs of the response, in which about 128 epochs for different monosyllabic stimuli and about 16 epochs of the oddball response were included. In each subject the measurements were repeated in 1-2 days, to obtain data from five to seven runs. Data analysis The 528 epochs were selectively averaged on-line into four averaged responses to the monosyllables and the oddball response, where the averaged responses included 300- and 600-ms pre- and poststimulus periods, respectively. A mean of the signal during the 300-ms prestimulus period served as the baseline. The averaged responses were then digitally filtered in a bandwidth of 1-40 Hz. The oddball responses were discarded from the subsequent data analysis because of possible contamination with a P3 wave. The waveform and the distribution of the field in the measured area at selected latencies were examined for the averaged monosyllabic responses. Among the main deflections in the waveform of the response, the N l m wave, occurring at about 100 ms after the stimulus onset, was exclusively studied. This is because the N l m was the most reproducibly observed, high-amplitude, long-latency component. From the field data of the N l m wave obtained in repeated measurements, three to six runs (mean 4.4, SD 0.92) were selected for each stimulus, depending on the amplitude and the distribution of the field, that is, whether a dipolar pattern appeared in the measured area. Those selected multiple-runs data were used in the following source analysis.
146 In order to estimate the location of the underlying neural activity of the Nlm wave in the brain, a single equivalent current dipole (ECD) was assumed as the source of the magnetic field of the Nlm wave. The ECD was characterized, in a spherical conductor having a radius that best fitted the local curvature of the skull in the measured area, by three coordinates, the moment defined by the product of current and length, and the direction of current. These parameters were determined in iterative calculations to minimize the squared sum of the error field, i.e., Y,(Bm-Bc) 2, where Bm and B c are the measured field and calculated field, respectively, and the summation is done across the measurement points. A parameter search method (Oshiro et al. 1992) was used in these calculations. The locations of the Nlm source were obtained at 2-ms steps in a 30-ms period centered at the peak latency of the Nlm wave.
STIMULI WAVEFORMS Vowel
Nasal
/nal
Magnetic resonance images In three subjects MR images were taken with markers attached on the fiducial points at the nasion, right and left preanricular points, and Cz. The MR coordinates were transformed into neuromagnetic field coordinates using the locations of these markers in the MR images. Then, coronal and axial slices, which included the calculated Nlm sources, were selected. The source locations were registered on these MR images.
Plosive
/kal
.....,rlrrrrr,, Fricative
Results The monosyllabic speech sounds used in the experiments had specific waveforms at their consonant part, as shown in Fig. 1. Nasal In/ is characterized by an oscillatory wave, p l o s i v e / k / b y a burst of noise, and f r i c a t i v e / h / b y the continuity of low-amplitude noise. The acoustic frequencies included at the beginning of these monosyllables are highest in the plosive and fricative speech, due to the noise component, and lowest in the nasal speech, due to the slow oscillatory wave. In order to evaluate the latency of the N l m peak from the stimulus onset, the N l m peak was determined from the m a x i m u m of the root-mean-squares of the fields at 37 recording sites. The means and SDs of the peak latency across subjects were 99.4_+8.0, 112.6_+10.5, 91.8_+2.1, and 96.9-+13.7 ms for / a / , / n a / , / k a / , a n d / h a / r e s p o n s e s , respectively. That the plosive and fricative responses had latencies shorter than the latency of the vowel response suggests that the N l m wave was elicited by the onset of the low-amplitude high-frequency consonant, not by the onset of the subsequent high-amplitude vowel. The longest N l m latency of the nasal response may be related to the gradual increase in intensity of the low-frequency oscillatory wave at the onset of the consonant/n/. Typical waveforms of the responses recorded at a posterior location of the field m a x i m u m are shown in Fig. 2 for the six subjects. The vowel and nasal responses have similar waveforms in all subjects, consisting of a prominent positive N l m wave followed by a broader negative P2m wave. Here, positive response (upward deflection) indicates the field out of the skull, and the negative response (downward deflection) indicates the field into the skull. Such waveforms are similar to the responses elicited by bursts of pure tones (Sams et al. 1985) and noises (Hari et al. 1987).
/hal
'rlfTff,,' I
0
J
50 msec
1
0
100
J
200 msec
Fig. 1 Waveforms of the monosyllabic speech sounds used as stimuli in the experiments. Left column shows enlarged waveforms at the onset of the sounds
As can be seen in Fig. 2, variations of the waveforms among subjects were observed in the plosive and fricative responses, in which double peaks often appeared at about the N l m wave. The first peak, indicated by an arrow, is the N l m elicited by the consonant onset, while the second peak m a y be the response elicited by the onset of the vowel. It should be noted that the N l m wave in the plosive and fricative responses has comparable amplitude to that in the vowel response, though the stimulus amplitude of those consonants is much lower than the vowel part (Fig. 1). As an example of the waveforms of the responses recorded at different locations, Fig. 3 illustrates the nasal responses recorded in one subject (R.M.), where 37 waveforms are displayed according to their recording locations in the array of sensing coils. In the selected responses A, B, and C, the N l m wave has positive, minimal, and negative deflection, respectively. The minimum amplitude locations are also indicated by the broken line. In the recording area, magnetic fields of the N l m wave exhibit an overall polarity reversal between the superior-
147
AN
FT
JM
RM
TK
TN
/na/ 100fr I
I
I
0
I
I
0.5s
I ....
0
0.5s
I ....
0
Fig. 2 Example of the magnetic field responses to different monosyllables in six subjects (initials at top) recorded at a left posterior maximum location, where upward deflection and downward deflection indicate direction of field out of and into the skull, respectively. The peak of the N l m wave is indicated by an arrow in cases where double peaks appear
0.5s
I
0
0.5s
I
0
0.5s
[
0
0.5s
posterior part and the inferior-anterior part. This polarity reversal can be clearly seen when all 37 waveforms are superimposed on a single baseline. Figure 4 shows the superimposed waveforms of the responses to different monosyllables and isocontour field maps at the N l m peak latency indicated by a cursor. These data were obtained in a single recording in the same subject.
Fig. 3 Waveforms of the responses to nasal speech recorded with 37 sensing coils in one subject (R.M.), where representative waveforms A, B, and C are selected in the right col-
Ina/
umn
vV
C
+
V " I lOOfT I
0
I
0.6 see
148 Fig. 4 Superposition of 37 waveforms in the responses to different monosyllables obtained in a single recording in one subject (R.M.). Lower figures illustrate isocontour maps of the field distribution at the Nlm peak latency indicated by a cursor in the waveform. S o l i d l i n e s and d o t t e d l i n e s indicate the fields out of and into the skull, respectively
/a/ i ii '
*ii
/ka/
i li ii~
i
_2_..J_+.+
•
100fr
'_~_._i-~.~.~__.i___i_L._.i.__il
~i ' i
~
-.q--4--~--,---,
' I I i i i I i
-~--~
40reset , i-T-T-T~--+-~-.--b-i-
-~-i-~ ~ -4.-+--r
iiii:iiii!!iiii~'~ IlOOmsec
/~/'/
1~!msec~
i ~
10fT STEP
...... -::.,-:.,-.,,.,
/ha/
/ha/
100fT
'~--7-~ -~--4--~ ' '
-r~--x--~-i~---.~- ~.-~'.-~ .--' 4-L-~-.4...-L__2. . . . L ' i i i ~ r
40msec -7-TY-i-~ ~ ~ . , , ....
--
'!-?-i--!-i-~- i--i- i- ~.....
105msec
~
100msecx _ _ _ ~ / ~
.:,,-............., :,, '. \ ,
The isofield maps at the N l m peak latency exhibited clear dipolar patterns, such as those in Fig. 4, irrespective of the type of consonants of the monosyllables. Based on these observations, we applied a single-ECD analysis to the field data to estimate the source of the N l m wave. As described previously, we used the field data in multiple recording runs to obtain reliable results for the ECD parameters. The use of multiple-recorded field data is expected to reduce the effect of background spontaneous fields (Kuriki et al. 1994). Effects of intrasubject variability of the response waveforms, which were confirmed in the repeated recordings, may also be suppressed. The obtained source locations at the N l m peak latency in the responses to different monosyllables were very close to each other in the three-dimensional space. In
10fT STEP
Fig. 5 the source locations projected on the x - y (axial, top view) plane and the x-z (sagittal) plane are illustrated for six subjects, together with the direction of the dipole current indicated by a bar. Here, the origin of the x, y, z coordinates is the midpoint between left and right preauricular points. The x-axis is given by connecting the nasion and the origin, the y-axis is perpendicular to the xaxis and directed toward the left ear, and the z-axis is directed toward Cz perpendicular to the x - y plane. Figure 5 shows that the sources of different monosyllables are within a cube of about 10 mm in side-length for each subject. The dipole currents are directed inferoposteriorly and perpendicular to the sylvian fissure, suggesting that the ECDs represent the neuronal currents in the auditory cortex on the supratemporal plane. The sources at the N l m peak latency were registered
149 I 9 a ANT.
9 POST.
MED. 4
AN
ANT.
5
9-
7
.,~X
6
.kA.
6
7,\,-
INF,
3
2
1
0
5
-1
3
4
~
5
3
2
1
i
0
44
4
-1
2
1
0
1
0
-1
1
0
-1
r
3
7
5
~
6
Ik.
6
~
5
~';I x
7
3
. . . . 2 1 0
4
-1
4
7
5
6
6
9 ~'-*'-
73
2
1
0
4
-1
8
5
7
6
9 ,
t
1
,
0
2
.,~ 3
2
6 9 -1
i
.~ \
5
-2
'
2
4
1
0
-1
-2
0
-1
8
~ 6 7
3
5
4
7i 2
TN
0
1
i
TK
1
6
6 74
RM
2
7
5
JM
POST,
SUP. 8
LAT. 7
FT
l 9 ha ]
na 9 ka
v
~,~ m. 3
2
1
X (cm)
N
0
-1
k 6
5
3
2
1
X (cm)
Fig. 5 Estimated source locations at the Nlm peak latency in the responses to different monosyllables obtained in six subjects. These sources were projected on axial (top view, left) and sagittal (right) planes. The direction of the dipole current is indicated by a
bar on the M R images taken for three subjects. The results (in Fig. 6) show that on the coronal images the sources o f different monosyllables are located in the sylvian fissure, probably at the lower bank. On the axial images the sources are located at or near Heschl's gyri. These results indicate that all the sources o f monosyllabic responses lie in the auditory cortex on the supratemporal plane, more specifically, in a restricted area including the primary cortex and the surrounding subfields. Figure 7 shows the grand mean of the N l m source locations across subjects, where the averaged coordinates o f x, y, and z are plotted as functions o f the relative latency centered at the N l m peak. Standard errors at the N l m peak latency are indicated in the inset. Although the coordinates o f the sources for different monosyllables are close to each other, there seem to be slight but character-
Fig. 6 Magnetic resonance images for three subjects of the coronal (/eft) and axial (right) slices on which estimated Nlm sources of different monosyllables were registered istic differences, depending on the consonants. On the xaxis the sources are located in such an order that the fricative response source is most posterior on the average than the plosive response source and then the vowel and nasal response sources. Here, /a/ and /na/ sources are very close with a separation less than 1 mm, while the mean separation b e t w e e n / a , na/ and /ha/ sources in the analysis period was about 4 m m (P<0.005, paired t >3.69; 2-tail, df=-9), and the separation b e t w e e n / a , na/ a n d / k a / s o u r c e s was 2 m m (n.s.). On the y-axis at the beginning of the analysis period o f - ( 1 4 - 8 ) ms, the plosive and fricative response sources are slightly medial to the vowel and nasal response sources; the m e a n separation b e t w e e n / a , na/, a n d / k a , h a / s o u r c e s was 3 m m (n.s.), while separations b e t w e e n / a / a n d / n a / sources and between/ka/and/ha/sources were less than 1 mm. This
150 ANT.
Table 1 Mean coordinates x, y, z (in centimeters) at the peak latency across subjects and their displacements Ax, Ay, Az during a 30-ms period around the peak for the Nlm sources of different monosyllabic responses. The significance level of the slope of the regression line used to calculate the displacement is also indicated
sEIIIl a
na
ka
ha
(3
X
POST.
0
!
,
I
,
I
x Ax y Ay z Az
,
t"
/a/
/ka/
/ha/
Ina/
1.32 0.38* 5.83 -0.41 * 6.00 -0.71"
1.12 0.44* 5.88 0.54* 6.22 -0.15 "-s.
1.00 0.17"** 5.85 0.64* 6.16 -0.19"*
1.37 0.59* 6.01 -0.01 n.s. 6.22 -0.64*
LAT.
* P<0.001; ** P<0.005; *** P<0.05; n.s. nonsignificant 6 E 5
sE;[ a
MED.
4
I na
I
ka
ha
I
i
I
i
,
I
,
I
,
I
r
sEIIII
SUP.
a
na
ka
ha
o
INF.
5
t
-20
I
- 10
0 LATENCY (ms)
l0
20
Fig. 7 Grand mean of x, y, and z coordinates of the Nlm sources of different monosyllabic responses as functions of the latency, where the latency was offset to zero at the time of Nlm peak. Standard errors at the Nlm peak latency are indicated in the in-
es along x-, y-, and z-axes in the 30-ms period were obtained, together with the level of statistical significance of the slope (t-test, 2-tail) of the regression line. When the slope is significantly positive or negative, the displacement is incremental or decremental. The results in Table 1 indicate that on the x- and z-axes all the sources shift toward the anterior and inferior directions with latency, respectively. Here, the inferior shift may be the result of the anterior shift of the sources that exist in the auditory cortex on the supratemporal plane, since the sylvian fissure is inclined downward on the x - z plane. On the y-axis the spatiotemporal shifts are specific to the consonants. The plosive and fricative response sources shift laterally with latency, but the vowel response source shifts medially. The nasal response source does not show significant shift. The small shifts on the z-axis of the plosive and fricative response sources, compared with the shifts of the vowel and nasal response sources, seem to be related to the positive shifts on the y-axis. The magnitude of the mean shifts on the x-, y- and z-axes is 4 m m (SD 2.3 ram) in 30 ms.
set
Discussion
relative location between the plosive/fricative response sources and the vowel response source is reversed at the end of the analysis period of (8-14) ms, so that t h e / a / source was medial t o / k a , h a / s o u r c e s with a mean separation of 4 m m (P<0.05, paired t >2.26; 2-tail, df=-9). Figure 7 also shows that the sources exhibit variations of the coordinates as the latency increases in a manner that depends on the coordinate axis and the type of consonants. The main features of these spatiotemporal variations are summarized in Table 1. Here, we first calculated linear regression lines in the relation between the latency and the coordinate values of the sources in different subjects, assuming that the variations of the coordinates are smooth and can be approximated with straight lines. In this calculation, considering the difference in the brain anatomy of individual subjects, the mean coordinate value of x, y, or z during the 30-ms analysis period was shifted to the same value in all subjects. Finally, the values of the displacement (Ax, Ay, and Az) of the sourc-
We have applied a single ECD analysis to the N l m wave evoked by monosyllables. The validity of assuming a single ECD may be supported by the observation of a dipolar field distribution at the N l m peak latency. However, a simulation study has shown that distributed ECDs, for example, on a sheet with an area of less than a few square centimeters, which would represent a cortical activity of populational neurons, are hardly distinguished from a single ECD located in the center of the area (Okada 1985). It has also been shown that a moving single ECD cannot be differentiated in terms of the waveform and the field distribution from two closely separate ECDs that are activated with a time difference (Kobayashi et al. 1993). Therefore, apparent movement of the N l m sources estimated in the present study may represent either the cortical activation in a limited area that propagates through lateral connections or correlated activation of closely separated areas with a time difference. Along the mediolateral (y) axis at the beginning of the
151 analysis period, the N l m sources of the plosive and fricative responses were located medially to the sources of the vowel and nasal responses. This difference is consistent with what would be expected from the known tonotopic organization and with the fricative and plosive consonants containing high-frequency noise at their onset. As originally found in the steady state response of pure tones (Romani et al. 1982) and later confirmed in the transient response (Pantev et al. 1988), the sources of middle latency response and the N l m wave are located progressively medially as the frequency of the stimulus increases. In the time course of the N l m source location, selective movements along the mediolateral axis were found in different monosyllabic responses: the sources of the plosive and fricative responses shifted laterally as the latency increased, and the source of the vowel response shifted medially. These lateral and medial shifts are consistent with the spatial difference between the source locations of the plosive and vowel responses observed in our previous study (Kuriki and Murase 1989), where the time dependence of the source location was not analyzed. It seems that these lateral and medial shifts in the plosive/fricative and vowel response sources cannot be explained solely by the tonotopic organization. Along the anteroposterior (x) axis, posteriority of the N l m sources of the plosive and fricative responses to the sources of the vowel and nasal responses were observed during the analysis period. This difference of the source locations between the plosive/fricative and the vowel responses confirms the previous results obtained at the N l m peak latency (Kaukoranta et al. 1987; Kuriki and Murase 1989). An interesting observation is that all the sources of different monosyllables shifted toward the anterior direction as the latency increased. The extent of the displacement is of the order of 5 mm in 30 ms. Such an anterior shift of the N l m source has been observed in the responses to pure tones (Rogers et al. 1990) and vowels (Takeuchi et al. 1992), though variability across subjects exists. Therefore, it seems that the anterior shift is a general process that occurs irrespective of the acoustic structure of the stimulus sounds. The above results indicate the existence of spatial difference of the neural activities along the anteroposterior axis in the auditory cortex in response to monosyllabic speech sounds and of common and selective shifts of the activities with latency increase along the anteroposterior and mediolateral axis, respectively. From the structural organization found in the human and animal auditory cortices it is suggested that such spatiotemporal behaviors may be related to the activation of multiple fields within the auditory cortex. In the human auditory cortex in the supratemporal plane multiple areas having different cytoarchitectonic structures exist, circumscribing the primary auditory area at Heschl's gyrus (or gyri; Galburda and Sanides 1980; Seldon 1981). Similar organization of auditory areas in the sylvian fissure also exists in the monkey (Merzenich and Brugge 1973; Brugge and Reale 1985). Since the speech sound is a temporal signal
conveying information in the time domain, spatiotemporal neural excitation in multiple areas, performing sequential analyses, may be a crucial process in extracting the acoustic structure of the sound. A recent study using optical recording in the guinea pig has shown a spatiotemporal variation of the neural activation in the auditory cortex (Taniguchi and Nasu 1993). The excited cortical region shifts temporally along an axis that is orthogonal to the isofrequency bands of tonotopic organization, in addition to another shift of the excitation toward the neighboring auditory area. This temporal shift orthogonal to the isofrequency bands coincides with the anterior shift of the N l m activity in the present study. It is not clear at the present stage whether this anterior shift is associated with the neural function that extracts a general acoustic feature, such as sound intensity (Pantev et al. 1989; Taniguchi and Nasu 1993). To summarize, the present study has demonstrated that neuromagnetic analysis of the neural activities of the human auditory cortex can provide high spatiotemporal resolution. It indicated a dynamic neural excitation in the auditory cortex on the supratemporal plane, exhibiting common and specific spatiotemporal variations in response to monosyllabic speech sounds. Acknowledgements We wish to thank E Takeuchi and T. Kobayashi of Hokkaido University and A. Nambu, R. Matsuzaki, O. Nagata, and Y. Takeshima of the Research Institute for Physiological Sciences for their valuable help in various stages of the experiments. This work was partially supported by a Grand-in-Aid for Scientific Research from the Ministry of Education, Science and Culture of Japan (05NP0101).
References Bechtereva NP, Abdullaev YG, Medvedev SV (1992) Properties of neuronal activity in cortex and subcortical nuclei of the human brain during single-word processing. Electroencephalogr Clin Neurophysiol 82:296-301 Brugge JF, Reale RA (1985) Auditory cortex. Cereb Cortex 4:229271 Celesia GG (1976) Organization of auditory cortical areas in man. Brain 99:403-414 Creutzfeld O, Ojemann G, Lettich E (1989) Neuronal activity in the human lateral temporal lobe. I. Responses to speech. Exp Brain Res 77:451-475 Elberling C, Bak C, Kofoed B, Lebech J, Saermark K (1982) Auditory magnetic fields from the human cerebral cortex: location and strength of an equivalent current dipole. Acta Neurol Scand 65:553-569 Fedio R Van Buren JM (1974) Memory deficits during electrical stimulation of speech cortex in conscious man. Brain Lang 1:29-42 Galaburda A, Sanides F (1980) Cytoarchitectonic organization of the human auditory cortex. J Comp Neurol 190:597-610 Hari R (1990) The neuromagnetic method in the study of the human auditory cortex. In: Hoke M (ed) Auditory evoked magnetic fields and electric potentials. (Advances in audiology, vol 6) Karger, Basel, pp 222-282 Hari R, Aittoniemi K, J~'vinen M-L, Katila T, Varpula T (1980) Auditory evoked transient and sustained magnetic fields of the human brain. Exp Brain Res 40:237-240 Hari R, Pelizzone M, M~ikel~iJR Hallstrom J, Leinonen L, Lounasmaa OV (1987) Neuromagnetic responses of the human auditory cortex to on- and offsets of noise bursts. Audiology 26: 31-43
152 Hayashi M, Mashiko T, Imada T, Odaka K (1992) Brain magnetic fields evoked by five Japanese vowels. Abstracts of 14th International Congress of Acoustics Band 4:58-59 Hillyard SA, Picton TW (1987) Electrophysiology of cognition. In: Mountcastle VB, Plum F, Geiger SR (eds) Higher functions of the brain. (Handbook of physiology, sect 1, The nervous system, vol 5) Am Physiol Soc, Bethesda, pp 519-584 Kaukoranta E, Hari R, Lounasmaa OV (1987) Responses of the human auditory cortex to vowel onset after fricative consonants. Exp Brain Res 69:19-23 Kobayashi T, Kuriki S, Takeuchi F (1993) A simulation study of dynamic characteristics of neuromagnetic fields using singlemoving-dipole and two-fixed-dipoles models. Phys Med Biol 38:1271-1282 Kuriki S (1992) Localization of cortical neuromagnetic sources of the auditory evoked responses to short speech sounds. Biomed Res [Suppl 1113:39-43 Kuriki S, Murase M (1989) Neuromagnetic study of the auditory responses in right and left hemispheres of the human brain evoked by pure tones and speech sounds. Exp Brain Res 77:127-134 Kuriki S, Takeuchi F, Kobayashi T (1994) Characteristics of the background noise in multichannel-recorded magnetic field responses. Electroencephalogr Clin Neurophysiol 92:56-63 Kutas M, Van Petten C, Besson M (1988) Event-related potential asymmetries during the reading of sentences. Electroencephalogr Clin Neurophysiol 69:218-233 Lee YS, Lueders H, Dinner DS, Lesser RP, Hahn J, Klem G (1984) Recording of auditory evoked potentials in man using chronic subdural electrodes. Brain 107:115-131 Liegeois-Chauvel C, Musolino A, Chauvel P (1991) Localization of the primary auditory area in man. Brain 114:139-153 Lovrich D, Novick B, Vaughan HG Jr (1988) Topographic analysis of auditory event-related potentials associated with acoustic and semantic processing. Electroencephalogr Clin Neurophysiol 71:40-54 M~ikel~i JR Hari R, Leinonen L (1988) Magnetic responses of the human auditory cortex to noise/square wave transitions. Electroencephalogr Clin Neurophysiol 69:423-430 Merzenich MM, Brugge JF (1973) Representation of the cochlea partition on the superior temporal plane of the macaque monkey. Brain Res 50:275-296 Ojemann GA (1978) Organization of short-term verbal memory in language areas of human cortex: evidence from electrical stimulation. Brain Lang 5:331-340 Ojemann GA, Creutzfeld O, Lettich E, Haglund MM (1988) Neuronal activity in human lateral temporal cortex related to shortterm verbal memory, naming and reading. Brain 111:13831403 Ojemann GA, Fried I, Lettich E (1989) Electrocorticograhic (ECoG) correlates of language. I.Desynchronization in temporal language cortex during object naming. Exp Brain Res 73:453-463 Okada Y (1985) Discrimination of localized and distributed current dipole sources and single and multiple sources. In: Weinberg H, Stroink G, Katila T (ed) Biomagnetism: application and theory. Pergamon, Oxford, pp 266-272
Oshiro O, Takeuchi F, Kuriki S (1992) Analysis of errors in neuromagnetic localization of multiple current dipole sources. Phys Med Biol 37:845-852 Pantev C, Hoke M, Lehnertz K, Ltitkenh6ner B, Anogianakis G, Wittkowski (1988) Tonotopic organization of the human auditory cortex revealed by transient auditory evoked magnetic fields. Electroencephalogr Clin Neurophysiol 69:160-170 Pantev C, Hoke M, Lehnertz K, LUtkenh6ner B (1989) Neuromagnetic evidence of an amplitopic organization of the human auditory cortex. Electroencephalogr Clin Neurophysiol 72:225231 Pantev C, Hoke M, Lehnertz K, Ltitkenh~ner B, Fahrendorf G, Stober U (1990) Identification of sources of brain neuronal activity with high spatiotemporal resolution through combination of neuromagnetic source localization (NMSL) and magnetic resonance imaging (MRI). Electroencephalogr Clin Neurophysiol 75:173-184 Papanicolaou AC, Rogers RL, Baumann S, Saydjari C, Eisenberg HM (1990) Source localization of two evoked magnetic field components using two alternative procedures. Exp Brain Res 80:44-48 Penfield W, Roberts L (1959) Speech and brain-mechanisms. Princeton University Press, Princepton, N.J. Reite M, Teale R Zimmerman J, Davis K, Whalen J (1988) Source location of a 50 msec latency auditory evoked field component. Electroencephalogr Clin Neurophysio170:490-498 Rogers RL, Papanicolaou AC, Baumann SB, Saydjari C, Eisenberg HM (1990) Neuromagnetic evidence of a dynamic excitation pattern generating the N100 auditory response. Electroencephalogr Clin Neurophysiol 77:237-240 Romani GL, Williamson SJ, Kaufmann L, Brenner D (1982) Characterization of the human auditory cortex by the neuromagnetic method. Exp Brain Res 47:381-393 Sams M, H~m~il~iinen M, Antervo A, Kaukoranta E, Reinikainen K, Hari R (1985) Cerebral neuromagnetic responses evoked by short auditory stimuli. Electroencephalogr Clin Neurophysiol 61:254-266 Seldon HH (1981) Structure of human auditory cortex. I. Cytoarchitectonics and dendritic distributions. Brain Res 229: 277-294 Takeuchi F, Mukai M, Kuriki S, Hayashi M, Imada T (1992) Spatio-temporal source localization of neuromagnetic fields evoked by pure tones. In: Hoke M, Erne SN, Okada YC, Romani GL (ed) Biomagnetism: clinical aspects. Elsevier Science, Oxford, pp 181-184 Taniguchi I, Nasu M (1993) Spatio-temporal representation of sound intensity in the guinea pig auditory cortex observed by optical recording. Neurosci Lett 151:178-181 Vaughan HG Jr, Ritter W, Simon R (1980) Topographic analysis of auditory event-related potentials. Prog Brain Res 54:279-285 Wood CC, Wolpaw JR (1982) Scalp distribution of human auditory evoked potentials. II. Evidence for multiple sources and involvement of auditory cortex. Electroencephalogr Clin Neurophysiol 54:25-38