Proc. Nati. Acad. Sci. USA Vol. 88, pp. 8996-9000, October 1991 Neurobiology
Human auditory evoked gamma-band magnetic fields C. PANTEV*t, S. MAKEIGt, M. HOKE§, R. GALAMBOS¶, S. HAMPSON II, AND C. GALLENII *Naval Health Research Center, San Diego, CA 92186; institute of Experimental Audiology, Munster, Germany; 1Department of Neuroscience, University of
California at San Diego, La Jolla, CA 92093; and ItDepartment of Neuropharmacology, Scripps Clinic and Research Foundation, San Diego, CA 92037
Contributed by R. Galambos, June 14, 1991
ABSTRACT We have discovered a ca. 40-Hz transient magnetic oscillatory response, evoked in the human brain by the onset of auditory stimuli, consisting of four or more cycles locked in phase to stimulus onset in approximately the 20- to 130-ms poststimulus interval. The response originates in the supratemporal auditory cortex, some millimeters deeper and anterior to the source of the larger-amplitude slow-wave M100 component of the evoked magnetic field and moves in a posterior arcing trajectory 1 cm or more in length. The oscillatory cortical activation elicited by auditory stimuli may be similar to the gamma-band cortical oscillations elicited by olfactory and visual stimuli and may represent an essential component of auditory perceptual processing.
METHODS The auditory stimulus-evoked magnetic brain activity generated in the left temporal cortex of 20 right-handed adult subjects was recorded in a magnetically shielded room with a 37-channel biomagnetometer (Biomagnetic Technologies, San Diego, CA). The response of one subject was recorded in 30 blocks of 128 stimuli. Three to six blocks were recorded from each of the other subjects. During the measurements subjects lay on their right side with eyes open and were asked to remain alert. Stimuli were 1000-Hz tone bursts of 60 dB relative to normal hearing level (nHL) (80-dB nHL in one session) with 500-ms duration and 10-ms rise and fall times, presented to the right ear at interstimulus intervals of 4 s. Blocks of 128 stimulus-related epochs of 1000 ms were recorded with a 200-ms prestimulus baseline. Electroencephalographic epochs were recorded simultaneously from one derivation (C,-earlobe). The neuromagnetic field pattern was recorded over a circular area 144 mm in diameter above the auditory cortex through 37 axially symmetric first-order gradiometer pickup coils (diameter, 20 mm; baseline, 50 mm). Each coil was connected to a superconducting quantum interference device (SQUID) sensor that produced a voltage proportional to the magnetic field radial to the coil. The intrinsic noise level was 4 ° 5response are visible in each subject's wide-band data superOs Iea(cry) 6X t imposed on the slow-wave activity. (ii) The spectrum of the wide-band response contains a peak near 10 Hz, and a poS t (cry,) at tit second, clearly identifiable peak at 30-40 Hz. (iii) After separation into slow-wave and gamma-band responses, B equivalent current dipoles account for each data set almost completely. (iv) The source locations of the slow-wave and gamma-band activity differ significantly, although superimposition of the source estimates on MRI images shows that 6S / . both responses originate within the auditory cortex. (v) The posterior movement of the GBR source estimates is contrary to the anterior progression of the slow-wave response source E 6 ) GBR, , l¢ ; . locations. While the initial peaks of the magnetic GBR have been
o
.
observed in magnetic recordings and identified with peaks of the electric middle latency response (28, 29), the full temporal extent of the response has not been recognized previously. S T6 hi's may be in part because the relatively small, highfrequency GBR waveforms were regarded as noise and 4.5 > < H e m >( X Hefiltered out of previous magnetic recordings. In additional experiments on the same subjects, we have observed similar 4.S 0 GBRs to tone bursts of different frequencies, short Gaussian tone pips, or clicks. Since it is generated by a variety of 6 stimuli and occurs very early in auditory cortical information X (crT) the GBR appears to reflect the activity of a processing, ,p° (crt) fundamental auditory perceptual mechanism. Artll The observed trajectory of the magnetic GBR peakC location estimates, a posterior-moving arc across the auditory cortex, is amenable to at least four interpretations: (i) continuous moving dipole: the GBR is produced in a small 6 locus s ^ ffiof cortex ) < > | ~>whose location moves continuously across the cortex Lo g along a posterior arc; (ii) discrete moving dipole: > * If theg GBR peaks are produced by a string of sequentially 6sGBR 1lQ| activated cortical loci; (iii) multiple dipoles: the apparent 7 55 movement of the GBR arises from a continuous change in the relative amplitudes of several fixed dipoles located at or beyond the two ends and outside the point of maximum curvature of the observed arc; and (iv) mass action: the C. . Ss6311> fiarc _of movement of the source location of the |observed successive GBR peaks represents the movement of the
rnea..at
_
4.5 ; fl y ~ 0 5 0 0 0.
rri,
6 e e 1
(crrn)
Pnt' *fl
Ct
A
_0
-
5 I)
FIG. 5. Equivalent dipole source locations of the major components of the evoked slow-wave field and of the successive GBR
peaks. (A) Source location estimates of the evoked slow-wave field
5peaks dd
M100 and M200 for 30 blocks of data (n = 128) from one subject. The opposite orientation of the arrows illustrates the opposite polarity of the two peaks. (B) Source locations of the individual GBR peaks for the same subject. To increase the signal-to-noise ratio of the GBR data, 10 subaverages of 3 blocks each were derived from the 30 data blocks. Both grand mean (circles) and subaverage (dots) responses are shown. The numeric values in the bottom plane indicate the mean peak latencies. (C) Source locations for the GBR peaks averaged across seven subjects. The three-dimensional cross in the upper-right corner shows the standard deviation of the localization estimates across subjects. The numerical values in the bottom plane indicate the mean peak latencies.
9000
Neurobiology: Pantev et al.
A YZ- Plane (C oronal View)
C
YZ-Plane (Enlargement)
B
1)
XZ-Plane
Proc. Natl. Acad. Sci. USA 88 (1991) (Sagittal View)
YX-Plane (Axial Views
overlapping sum of the successive GBRs. The relative locations of the sources for the largest GBR peak and the M100 peak are also consistent with previous reports on the relative spatial locations of the M100 and the SSR (18, 19). Many other facts about the auditory cortex GBR, including confirmation of its spatio-temporal extent, await further measurements. In general, the auditory evoked GBR appears to reflect the general observation that perception in most or all sense modalities involves coherent rhythmic brain activity at gamma-band frequencies. We thank Drs. F. Bloom, C. Gray, F. Crick, W. Black, E. Hirschkoff, and B. Ldtkenhoner for reviewing earlier versions ofthis manuscript and S. Cobb for assistance in overlaying the MRIs. The work was supported by grants from the Deutsche Forshungsgemeinschaft (Klinische Forschergruppe "Biomagnetismus und Biosignalanalyse"), from the Armstrong MacDonald Foundation and the Pasarow Foundation, and by a grant to Dr. Makeig from the National Alliance for Research in Schizophrenia and Depression. Berger, H. (1937) Arch. Psychiatr. Nervenkrankh. 106, 165-187. Bressler, S. L. (1990) Trends Neurosci. 13, 161-162. Adrian, E. D. (1942) J. Physiol. (London) 100, 459-473. Freeman, W. (1975) MassAction in the Nervous System (Academic, New York). 5. Freeman, W. (1985) in Handbook of Electroencephalography and Clinical Neurophysiology, eds. Gevins, A. & Remond, A. (Elsevier, Amsterdam), Vol. 3A, part 2, chap. 18. 6. Freeman, W. & Skarda, C. (1985) Brain Res. Rev. 10, 147-175. 7. Gray, C. & Singer, W. (1989) Proc. Natl. Acad. Sci. USA 86,
1. 2. 3. 4.
FIG. 6. Coronal (A), sagittal (B), enlarged view of the coronal (including the Sylvian fissure) (C), and axial (D) MRI sections for the same subject as in Fig. 5 A and B with overlays showing the mean source-location estimates of several of the GBR peaks and the peak of the evoked slow-wave field M100 component (white dots at the tips of arrows). The sagittal and axial slices shown include the location estimate for the 60-ms GBR peak. The coronal slice includes the location estimates for the M100 peak (right arrow in C) and several peaks in the GBR (left arrow in C; compare Fig. SB).
centroid of a spatially distributed pattern of cortical activation. The continuous and discrete moving dipole models are straightforward interpretations of the data but have no support from direct observations in the cortex. A multiple dipole model has been suggested by Scherg et al. (29) to explain their data on the early GBR components. The model they fit assumes two dipoles oriented at right angles to each other, but their results also lack support from anatomical evidence. Finely sampled observations of the spatial distribution of gamma-band activity in olfactory cortex have been reported by Freeman (4, 5), who has observed that GBR activity in the olfactory bulb spreads out across the cortical surface rapidly in all directions from an initial focus with only a small phase lag across a recording array spanning several millimeters. Until direct observations with finer spatial resolution are available, the most plausible interpretation of our results is that GBR activity may also spread out from an anterior focus across the surface of the superior temporal plane, and the observed movement of the equivalent source represents movement of the centroid of this gamma-band activity. A conceivable function of a global GBR excitation could be to synchronize and dynamically link separate subregions of auditory cortex in order to combine separately represented auditory features into unitary auditory percepts-for example, the pitch of complex tones (30). The fact that electric and magnetic SSRs are largest for rates in the frequency range of the GBR (35-40 Hz) also suggests that the SSR to auditory stimuli is essentially the
1698-1702. 8. Gray, C., Konig, P., Engel, A. & Singer, W. (1989) Nature (London) 338, 334-337. 9. Eckhorn, R., Bauer, R., Jordan, W., Brosch, M., Kruse, W., Munk, M. & Reitboeck, H. J. (1988) Biol. Cybernet. 60, 121-130. 10. Crick, F. & Koch, C. (1991) Neurosciences 2, in press. 11. Barinaga, M. (1990) Science 249, 856-858. 12. Steriade, M. & Dimetrescu, M. (1962) Electroencephalogr. Clin. Neurophysiol. 14, 21-36. 13. Basar, E. (1980) EEG Brain Dynamics (Elsevier, Amsterdam). 14. Bouyer, J. J., Montaron, M. F. & Rougeul, A. (1981) Electroencephalographr. Clin. Neurophysiol. 51, 244-252. 15. Galambos, R., Makeig, S. & Talmachoff, P. J. (1981) Proc. Natl. Acad. Sci. USA 78, 2643-2647. 16. Galambos, R. & Makeig, S. (1988) in Dynamics of Sensory and Cognitive Processing by the Brain, eds. Basar, E. & Bullock, T.
(Springer, Berlin), pp. 103-122. 17. Makeig, S. (1985) Studies in Musical Psychobiology (University Microfilms, Ann Arbor, MI). 18. Makela, J. P. & Hari, R. (1987) Electroencephalogr. Clin. Neurophysiol. 66, 539-546. 19. Hari, R., Hamalainen, M. & Joustiniemi, S. L. (1989) J. Acoust. Soc. Am. 86, 1033-1039. 20. Ribary, U., Llinas, R., Suk, J. & Ferris, S. H. (1990) in Advances in Biomagnetism, eds. Williamson, S., Hoke, M., Stroink, G. & Kotani, M. (Plenum, New York), pp. 311-314. 21. Romani, G. L., Williamson, S. J. & Kaufman, L. (1982) Science 216, 1339-1340. 22. Makeig, S. & Galambos, R. (1989) Soc. Neurosci. Abstr. 15, 113. 23. Makeig, S. (1990) in Psychophysiological Brain Research, eds. Brunia, C., Gaillard, A. & Kok, A. (Tilburg Univ. Press, The Netherlands), Vol. 2, pp. 60-64. 24. Pantev, C., Gallen, C., Hampson, S., Buchanan, S. & Sobel, D. (1991) Am. J. EEG Technol. 31, 83-101. 25. Geselowitz, D. B. (1970) IEEE Trans. Mag. 6, 346-347. 26. Sarvas, J. (1987) J. Phys. Med. Biol. 32, 11-22. 27. Talairach, J. & Tournoux, P. (1988) Co-Planar Stereotaxic Atlas of the Human Brain (Thieme, Leipzig). 28. Pelizone, M., Hari, R., Makela, J. P., Huttunen, J., Ahlfors, S. & Himaidinen, M. (1987) Neurosci. Lett. 82, 303-307. 29. Scherg, M., Hari, R. & Hfmilainen, M. (1990) in Advances in Biomagnetism, eds. Williamson, S., Hoke, M., Stroink, G. & Kotani, M. (Plenum, New York), pp. 97-100. 30. Terhardt, E. (1989) Naturwissenschaften 76, 496-504.
Human auditory evoked gamma-band magnetic fields C. PANTEV*t, S. MAKEIGt, M. HOKE§, R. GALAMBOS¶, S. HAMPSON II, AND C. GALLENII *Naval Health Research Center, San Diego, CA 92186; institute of Experimental Audiology, Munster, Germany; 1Department of Neuroscience, University of
California at San Diego, La Jolla, CA 92093; and ItDepartment of Neuropharmacology, Scripps Clinic and Research Foundation, San Diego, CA 92037
Contributed by R. Galambos, June 14, 1991
ABSTRACT We have discovered a ca. 40-Hz transient magnetic oscillatory response, evoked in the human brain by the onset of auditory stimuli, consisting of four or more cycles locked in phase to stimulus onset in approximately the 20- to 130-ms poststimulus interval. The response originates in the supratemporal auditory cortex, some millimeters deeper and anterior to the source of the larger-amplitude slow-wave M100 component of the evoked magnetic field and moves in a posterior arcing trajectory 1 cm or more in length. The oscillatory cortical activation elicited by auditory stimuli may be similar to the gamma-band cortical oscillations elicited by olfactory and visual stimuli and may represent an essential component of auditory perceptual processing.
METHODS The auditory stimulus-evoked magnetic brain activity generated in the left temporal cortex of 20 right-handed adult subjects was recorded in a magnetically shielded room with a 37-channel biomagnetometer (Biomagnetic Technologies, San Diego, CA). The response of one subject was recorded in 30 blocks of 128 stimuli. Three to six blocks were recorded from each of the other subjects. During the measurements subjects lay on their right side with eyes open and were asked to remain alert. Stimuli were 1000-Hz tone bursts of 60 dB relative to normal hearing level (nHL) (80-dB nHL in one session) with 500-ms duration and 10-ms rise and fall times, presented to the right ear at interstimulus intervals of 4 s. Blocks of 128 stimulus-related epochs of 1000 ms were recorded with a 200-ms prestimulus baseline. Electroencephalographic epochs were recorded simultaneously from one derivation (C,-earlobe). The neuromagnetic field pattern was recorded over a circular area 144 mm in diameter above the auditory cortex through 37 axially symmetric first-order gradiometer pickup coils (diameter, 20 mm; baseline, 50 mm). Each coil was connected to a superconducting quantum interference device (SQUID) sensor that produced a voltage proportional to the magnetic field radial to the coil. The intrinsic noise level was 4 ° 5response are visible in each subject's wide-band data superOs Iea(cry) 6X t imposed on the slow-wave activity. (ii) The spectrum of the wide-band response contains a peak near 10 Hz, and a poS t (cry,) at tit second, clearly identifiable peak at 30-40 Hz. (iii) After separation into slow-wave and gamma-band responses, B equivalent current dipoles account for each data set almost completely. (iv) The source locations of the slow-wave and gamma-band activity differ significantly, although superimposition of the source estimates on MRI images shows that 6S / . both responses originate within the auditory cortex. (v) The posterior movement of the GBR source estimates is contrary to the anterior progression of the slow-wave response source E 6 ) GBR, , l¢ ; . locations. While the initial peaks of the magnetic GBR have been
o
.
observed in magnetic recordings and identified with peaks of the electric middle latency response (28, 29), the full temporal extent of the response has not been recognized previously. S T6 hi's may be in part because the relatively small, highfrequency GBR waveforms were regarded as noise and 4.5 > < H e m >( X Hefiltered out of previous magnetic recordings. In additional experiments on the same subjects, we have observed similar 4.S 0 GBRs to tone bursts of different frequencies, short Gaussian tone pips, or clicks. Since it is generated by a variety of 6 stimuli and occurs very early in auditory cortical information X (crT) the GBR appears to reflect the activity of a processing, ,p° (crt) fundamental auditory perceptual mechanism. Artll The observed trajectory of the magnetic GBR peakC location estimates, a posterior-moving arc across the auditory cortex, is amenable to at least four interpretations: (i) continuous moving dipole: the GBR is produced in a small 6 locus s ^ ffiof cortex ) < > | ~>whose location moves continuously across the cortex Lo g along a posterior arc; (ii) discrete moving dipole: > * If theg GBR peaks are produced by a string of sequentially 6sGBR 1lQ| activated cortical loci; (iii) multiple dipoles: the apparent 7 55 movement of the GBR arises from a continuous change in the relative amplitudes of several fixed dipoles located at or beyond the two ends and outside the point of maximum curvature of the observed arc; and (iv) mass action: the C. . Ss6311> fiarc _of movement of the source location of the |observed successive GBR peaks represents the movement of the
rnea..at
_
4.5 ; fl y ~ 0 5 0 0 0.
rri,
6 e e 1
(crrn)
Pnt' *fl
Ct
A
_0
-
5 I)
FIG. 5. Equivalent dipole source locations of the major components of the evoked slow-wave field and of the successive GBR
peaks. (A) Source location estimates of the evoked slow-wave field
5peaks dd
M100 and M200 for 30 blocks of data (n = 128) from one subject. The opposite orientation of the arrows illustrates the opposite polarity of the two peaks. (B) Source locations of the individual GBR peaks for the same subject. To increase the signal-to-noise ratio of the GBR data, 10 subaverages of 3 blocks each were derived from the 30 data blocks. Both grand mean (circles) and subaverage (dots) responses are shown. The numeric values in the bottom plane indicate the mean peak latencies. (C) Source locations for the GBR peaks averaged across seven subjects. The three-dimensional cross in the upper-right corner shows the standard deviation of the localization estimates across subjects. The numerical values in the bottom plane indicate the mean peak latencies.
9000
Neurobiology: Pantev et al.
A YZ- Plane (C oronal View)
C
YZ-Plane (Enlargement)
B
1)
XZ-Plane
Proc. Natl. Acad. Sci. USA 88 (1991) (Sagittal View)
YX-Plane (Axial Views
overlapping sum of the successive GBRs. The relative locations of the sources for the largest GBR peak and the M100 peak are also consistent with previous reports on the relative spatial locations of the M100 and the SSR (18, 19). Many other facts about the auditory cortex GBR, including confirmation of its spatio-temporal extent, await further measurements. In general, the auditory evoked GBR appears to reflect the general observation that perception in most or all sense modalities involves coherent rhythmic brain activity at gamma-band frequencies. We thank Drs. F. Bloom, C. Gray, F. Crick, W. Black, E. Hirschkoff, and B. Ldtkenhoner for reviewing earlier versions ofthis manuscript and S. Cobb for assistance in overlaying the MRIs. The work was supported by grants from the Deutsche Forshungsgemeinschaft (Klinische Forschergruppe "Biomagnetismus und Biosignalanalyse"), from the Armstrong MacDonald Foundation and the Pasarow Foundation, and by a grant to Dr. Makeig from the National Alliance for Research in Schizophrenia and Depression. Berger, H. (1937) Arch. Psychiatr. Nervenkrankh. 106, 165-187. Bressler, S. L. (1990) Trends Neurosci. 13, 161-162. Adrian, E. D. (1942) J. Physiol. (London) 100, 459-473. Freeman, W. (1975) MassAction in the Nervous System (Academic, New York). 5. Freeman, W. (1985) in Handbook of Electroencephalography and Clinical Neurophysiology, eds. Gevins, A. & Remond, A. (Elsevier, Amsterdam), Vol. 3A, part 2, chap. 18. 6. Freeman, W. & Skarda, C. (1985) Brain Res. Rev. 10, 147-175. 7. Gray, C. & Singer, W. (1989) Proc. Natl. Acad. Sci. USA 86,
1. 2. 3. 4.
FIG. 6. Coronal (A), sagittal (B), enlarged view of the coronal (including the Sylvian fissure) (C), and axial (D) MRI sections for the same subject as in Fig. 5 A and B with overlays showing the mean source-location estimates of several of the GBR peaks and the peak of the evoked slow-wave field M100 component (white dots at the tips of arrows). The sagittal and axial slices shown include the location estimate for the 60-ms GBR peak. The coronal slice includes the location estimates for the M100 peak (right arrow in C) and several peaks in the GBR (left arrow in C; compare Fig. SB).
centroid of a spatially distributed pattern of cortical activation. The continuous and discrete moving dipole models are straightforward interpretations of the data but have no support from direct observations in the cortex. A multiple dipole model has been suggested by Scherg et al. (29) to explain their data on the early GBR components. The model they fit assumes two dipoles oriented at right angles to each other, but their results also lack support from anatomical evidence. Finely sampled observations of the spatial distribution of gamma-band activity in olfactory cortex have been reported by Freeman (4, 5), who has observed that GBR activity in the olfactory bulb spreads out across the cortical surface rapidly in all directions from an initial focus with only a small phase lag across a recording array spanning several millimeters. Until direct observations with finer spatial resolution are available, the most plausible interpretation of our results is that GBR activity may also spread out from an anterior focus across the surface of the superior temporal plane, and the observed movement of the equivalent source represents movement of the centroid of this gamma-band activity. A conceivable function of a global GBR excitation could be to synchronize and dynamically link separate subregions of auditory cortex in order to combine separately represented auditory features into unitary auditory percepts-for example, the pitch of complex tones (30). The fact that electric and magnetic SSRs are largest for rates in the frequency range of the GBR (35-40 Hz) also suggests that the SSR to auditory stimuli is essentially the
1698-1702. 8. Gray, C., Konig, P., Engel, A. & Singer, W. (1989) Nature (London) 338, 334-337. 9. Eckhorn, R., Bauer, R., Jordan, W., Brosch, M., Kruse, W., Munk, M. & Reitboeck, H. J. (1988) Biol. Cybernet. 60, 121-130. 10. Crick, F. & Koch, C. (1991) Neurosciences 2, in press. 11. Barinaga, M. (1990) Science 249, 856-858. 12. Steriade, M. & Dimetrescu, M. (1962) Electroencephalogr. Clin. Neurophysiol. 14, 21-36. 13. Basar, E. (1980) EEG Brain Dynamics (Elsevier, Amsterdam). 14. Bouyer, J. J., Montaron, M. F. & Rougeul, A. (1981) Electroencephalographr. Clin. Neurophysiol. 51, 244-252. 15. Galambos, R., Makeig, S. & Talmachoff, P. J. (1981) Proc. Natl. Acad. Sci. USA 78, 2643-2647. 16. Galambos, R. & Makeig, S. (1988) in Dynamics of Sensory and Cognitive Processing by the Brain, eds. Basar, E. & Bullock, T.
(Springer, Berlin), pp. 103-122. 17. Makeig, S. (1985) Studies in Musical Psychobiology (University Microfilms, Ann Arbor, MI). 18. Makela, J. P. & Hari, R. (1987) Electroencephalogr. Clin. Neurophysiol. 66, 539-546. 19. Hari, R., Hamalainen, M. & Joustiniemi, S. L. (1989) J. Acoust. Soc. Am. 86, 1033-1039. 20. Ribary, U., Llinas, R., Suk, J. & Ferris, S. H. (1990) in Advances in Biomagnetism, eds. Williamson, S., Hoke, M., Stroink, G. & Kotani, M. (Plenum, New York), pp. 311-314. 21. Romani, G. L., Williamson, S. J. & Kaufman, L. (1982) Science 216, 1339-1340. 22. Makeig, S. & Galambos, R. (1989) Soc. Neurosci. Abstr. 15, 113. 23. Makeig, S. (1990) in Psychophysiological Brain Research, eds. Brunia, C., Gaillard, A. & Kok, A. (Tilburg Univ. Press, The Netherlands), Vol. 2, pp. 60-64. 24. Pantev, C., Gallen, C., Hampson, S., Buchanan, S. & Sobel, D. (1991) Am. J. EEG Technol. 31, 83-101. 25. Geselowitz, D. B. (1970) IEEE Trans. Mag. 6, 346-347. 26. Sarvas, J. (1987) J. Phys. Med. Biol. 32, 11-22. 27. Talairach, J. & Tournoux, P. (1988) Co-Planar Stereotaxic Atlas of the Human Brain (Thieme, Leipzig). 28. Pelizone, M., Hari, R., Makela, J. P., Huttunen, J., Ahlfors, S. & Himaidinen, M. (1987) Neurosci. Lett. 82, 303-307. 29. Scherg, M., Hari, R. & Hfmilainen, M. (1990) in Advances in Biomagnetism, eds. Williamson, S., Hoke, M., Stroink, G. & Kotani, M. (Plenum, New York), pp. 97-100. 30. Terhardt, E. (1989) Naturwissenschaften 76, 496-504.
