Download Magnetoencephalographic Investigation of Human Cortical Area V1

6, 47–57 (1997)
NI970273
NEUROIMAGE
ARTICLE NO.
Magnetoencephalographic Investigation of Human
Cortical Area V1 Using Color Stimuli
Fiona Fylan,1 Ian E. Holliday, Krish D. Singh,* Stephen J. Anderson,* and Graham F. A. Harding
Applied Psychology, Aston University, Aston Triangle, Birmingham B4 7ET, United Kingdom; and *Department of Psychology,
Royal Holloway College, University of London, Egham, Surrey TW20 OEX, United Kingdom
Received January 3, 1997
tion of the M pathway in macaque but was not abolished. Anderson et al. (1995) showed that the reversedmotion illusion in human peripheral vision is consistent
with spatial aliasing by the P-ganglion-cell mosaic. The
ability of the proposed motion center, V5 (Zeki, 1974;
Zeki et al., 1991), to utilize chromatic information has
been demonstrated in humans using positron emission
tomography (ffytche et al., 1995). Further evidence that
the P pathway conveys motion information comes from
psychophysical studies demonstrating a contribution
from color-opponent mechanisms to motion perception
(e.g., Cavanagh and Anstis, 1991). Some studies suggest that there may be more than one chromatic motion
system (Derrington and Henning, 1993; Gorea et al.,
1993a,b; Cropper and Derrington, 1994; Gegenfurtner
and Hawken, 1995), although there is no general
agreement on their number or degree of independence.
Our aim in this study was to determine the response
properties of the human visual cortex to chromatic
(red/green) stimuli using MEG and in so doing investigate further the role of the P pathway in processing
motion information (for reviews of MEG, see Hämäläinen et al., 1993; Harding, 1993). MEG cannot be
used to measure the spatio-temporal selectivity of
individual neurons, but rather measures activity generated by a population of cortical neurons. To establish
the selectivity of individual units would require a
measure of their spiking rate, although it is unclear
that such information provides an optimal measure of
neuronal processing. Recordings from area V1 of macaque (Givre et al., 1995) do not show a direct correspondence between changes in multiunit activity and transmembrane current and it was suggested that
postsynaptic current flow may occur over larger portions of the cell and therefore be more readily detectable than action potential current flow. MEG measures
currents associated with summated postsynaptic potentials; the spatial-frequency and temporal-frequency
tuning curves we describe in this study therefore
represent increases and decreases in the amount of
The aim of this study was to determine the response
properties of the human visual cortex to chromatic
stimuli using magnetoencephalography (MEG). Evoked
responses were recorded to isoluminant red/green sinusoidal gratings for a wide range of spatial and temporal frequencies. For each condition the response was
dominated by a single major component which was
well modeled by an equivalent current dipole. Coregistration of MEG and MRI data provided evidence that
the principal evoked cortical activity originated from
visual area V1. To investigate the chromatic response
properties of this area, the maximum global field power
of the evoked response was plotted as a function of
stimulus spatial and temporal frequency. The spatialfrequency tuning was lowpass and the temporalfrequency tuning was multimodal, with peaks at 0 and
4 Hz. The results demonstrate the use of MEG as a
technique for investigating activity from discrete regions of cortex. r 1997 Academic Press
INTRODUCTION
There are two main pathways from the retina to the
primary visual cortex in primates: the magnocellular
(M) and parvocellular (P) pathways. There is a general
consensus that the M pathway plays an important role
in motion processing, although it may also be involved
in visual attention (for reviews, see Lennie, 1993;
Merigan and Maunsell, 1993). There is firm evidence
that the P pathway processes color information and fine
spatial information. Recent physiological (Merigan et
al., 1991) and psychophysical (Anderson et al., 1995)
studies suggest that the P pathway may also subserve
some aspects of motion processing. Merigan et al.
(1991) demonstrated that the ability of primates to
perform motion tasks, such as direction and speed
discrimination, was impaired following chemical abla1 To whom correspondence should be addressed. Fax: (0121) 3334220. E-mail: [email protected].
47
1053-8119/97 $25.00
Copyright r 1997 by Academic Press
All rights of reproduction in any form reserved.
48
FYLAN ET AL.
synchronized neural activity in the localized regions
described.
Previous studies have described the evoked response
to chromatic stimuli of a range of spatial frequencies
(Rabin et al., 1994; Regan and He, 1996) but did not
determine the cortical area generating the response. In
this study evoked magnetic responses were recorded to
a wide range of stimulus parameters and the cortical
origin of the evoked response was determined by analyzing both the retinotopic organization and the topographic distribution of the recorded signals and by
coregistration of MEG and MRI data. Our results
provide evidence for at least two chromatic-sensitive
mechanisms in human vision, one responsive at low
temporal frequencies and the other at high temporal
frequencies, both located within cortical area V1.
METHODS
Stimuli
Sinusoidal stimuli were generated using a Cambridge Research Systems VSG2/2 grating generator
and displayed on an Eizo Flexscan T560i color monitor
with 14-bit luminance resolution at a frame rate of 100
Hz. The chromaticity coordinates of the display were
rx 5 0.625, ry 5 0.340, gx 5 0.280, gy 5 0.595, bx 5 0.155,
and by 5 0.070. The mean luminance of the display was
12 cd/m2, measured with a Minolta photometer, and the
gamma-corrected display was linear over the range of
contrasts used. Red and green gratings of 100% contrast (Lmax 2 Lmin/Lmax 1 Lmin) were generated independently and combined 180° out of phase to produce
physically isoluminant red/green gratings. For all observers there were no stimulus conditions for which the
point of perceptual isoluminance, determined using
standard flicker criteria, differed significantly from the
point of photometric isoluminance. The red and green
gratings were also combined in phase to produce yellow/
black gratings of the same mean hue and luminance.
This stimulus was used as a control to ensure that the
chromatic responses did not arise from luminance
artifacts.
The gratings were presented horizontally using a
temporal envelope of 1 s duration with onset and offset
ramps of 10 ms. The interstimulus interval was 500 ms,
during which the gratings were replaced with a homogeneous background of the same mean hue and luminance. The spatial frequency of the gratings was varied
between 0.25 and 8 c/d. The gratings were presented
either stationary or drifting with a temporal frequency
of up to 40 Hz. The stimuli were viewed binocularly at a
distance of 3 m through a system of front-silvered
mirrors and subtended 4° vertically by 6° horizontally.
To determine the effect of the number of cycles displayed, a control experiment was performed in which
the field size was doubled by reducing viewing distance
to 1.5 m. A 6-mm-diameter circle served as a fixation
mark.
Procedure
Two females (FF and KL) and two males (IH and KS)
volunteered for the experiment. Their ages ranged from
26 to 39 years and all had normal visual fields and color
vision and a normal or corrected-to-normal binocular
Snellen acuity of 6/6. The experiments were performed
binocularly. Responses were sampled at 1 kHz for 512
ms, and 100 trials were averaged for each stimulus
condition. The resultant signals were bandpass filtered
at 2–30 Hz. To establish the retinotopicity of the evoked
response, the stimuli were confined to a single quadrant of the visual field for any one set of trials. To
restrict input to a discrete region of cortex, and to
prevent activation of multiple regions of cortex due to
dual representation of the horizontal and vertical meridians, the stimuli were displaced 0.5° from the principal
meridians.
The visual evoked magnetic response was recorded
using a SQUID magnetometer with 19 channels in a
hexagonal array which extended 13 cm over a flat
dewar base with 3-cm channel spacing. Each channel
comprised a second-order axial gradiometer with 5-cm
baseline and a 1.5-cm coil diameter and the mean noise
level was 23 fT/œHz. For further details of this system,
see Matlashov et al. (1995). The magnetometer was
housed in a magnetically shielded room with low
photopic lighting. The position of each channel within a
standard coordinate system was predetermined using a
Polhemus Isotrak three-dimensional digitizing system.
The subjects were seated with their head prone and
stabilized using a bite bar and forehead rest. The dewar
was centred above EEG 10–20 electrode position Oz as
preliminary measures showed that this position yielded
strong responses to chromatic stimuli. To ensure comparability of measurements within each experiment, all
data were collected with the dewar in a constant
position. For each experimental session the position of
the subject’s bite bar, and therefore head, was digitized
and converted to the standard coordinate system. Four
bite-bar reference markers (oil-filled drill holes) appeared as small, white spots on the magnetic resonance
images (MRI) allowing accurate coregistration of MEG
and MRI data (Singh et al., 1997). The MRIs were
obtained using a 1.5-T Siemens magnetic resonance
scanner with 1 mm 3 1 mm 3 1.5 mm voxel size.
Data Analysis
Estimates of the noise magnitude, si, in each channel
i were obtained using a standard anti-average procedure in which successive trials are alternately added
and subtracted, canceling the evoked response while
49
MEG ANALYSIS OF HUMAN AREA V1
retaining the noise (Regan, 1989, p57). For each of the
averaged evoked responses, the global field power G(t)
was calculated as
spatial distribution of the localizations. The ellipsoids
were displayed on coregistered MR images to show the
likely region of activated cortex, located at some point
within the ellipsoid.
channels
G(t) 5
o
(Bi(t))2,
RESULTS
i51
where Bi(t) is the signal in channel i at time t. The
global field power is therefore a measure of the total
power in the detector array at time t. Peak latencies for
each response were identified by plotting G(t ) as a
function of time. The noise magnitude si provided an
estimate of the error in G(t).
We determined the chromatic response properties of
area V1 by plotting the maximum value of G(t) as a
function of spatial frequency and temporal frequency.
This procedure assumes that the evoked responses
reflect activity in the same cortical area and is appropriate given that all recordings for each measurement set
were made with the recording sensors in the same
location with respect to the underlying cortex. To
confirm that the responses arose from the same cortical
region, source localizations were performed to identify
the generating cortical volume for each stimulus parameter used. The source localization model used was a
single equivalent current dipole in a homogeneous
conducting sphere (Hämäläinen et al., 1993). The dipolefitting algorithm is a nonlinear minimization (Mosher
et al., 1992) of the weighted x2 statistic with respect to
the dipole parameters using Powell’s conjugate gradient method (Press et al., 1989), where
channels
x2(t) 5
o
i51
1
The evoked magnetic responses to stationary chromatic (red/green) modulated gratings of 2 c/d are shown
in Fig. 1a for subject FF. The response to chromatic
gratings was dominated by a single major component
peaking at 105 ms, as shown by the global field power
plot in Fig. 1b. The latency of this peak was invariant
with respect to the visual-field quadrant used and the
strongest signals were evoked when the stimulus was
positioned in the lower field. The results were similar
for all subjects; the latency of the major component
varied between 105 and 121 ms.
2
2
Bi(t) 2 B̂i(t)
si
Evoked Magnetic Response to Isoluminant
Chromatic Gratings
and B̂i (t) is the calculated field from the postulated
dipole. Dipole solutions were accepted as being valid if
two criteria were met: first, that the correlation coefficient between the measured and the calculated fields
exceeded 0.95; and second, that the probability that the
final x2 value occurred by chance (the gammaQ statistic, see Press et al., 1989) exceeded 0.01.
Monte-Carlo analysis (Medvick et al., 1990) was used
to estimate the confidence volume of each source localization. With this method sample datasets were generated by adding simulated noise to the field from the
calculated dipole such that the noise in channel i was
assumed to be normally distributed with a magnitude
of si. Each of these sample datasets was then inverted
using the same dipole-fitting algorithm to yield a
spread of localizations. An iterative algorithm was then
used to fit a three-dimensional ellipsoid (Singh and
Harding, 1996) at the 95% confidence level for the
FIG. 1. (a) Magnetic field strength (fT) evoked by an isoluminant
(red/green) horizontal sinusoidal grating of spatial frequency 2 c/d as
a function of time (ms). The stimulus subtended 6° horizontally by 4°
vertically, presented with 10-ms onset and offset ramps for 1000 ms
in the right lower quadrant of the visual field (subject FF). Each trace
is the average of 100 trials and represents a temporal window of 512
ms. Traces 1–15 are the corresponding active magnetometer channels in the detector array (c). Four of the 19 channels were not
operating within acceptable parameters and were therefore excluded
from all measurements and analyses. Channel 1 was positioned
directly above International 10–20 position Oz. (b) The global magnetic field power for the responses shown in (a).
50
FYLAN ET AL.
FIG. 2. Pseudo-colored maps of the evoked magnetic field as a function of stimulus location in the visual field (A, left upper quadrant; B,
right upper quadrant; C, left lower quadrant; D, right lower quadrant). Stimulus parameters were as in Fig. 1. The data are for subject KL.
The inset color scale shows the magnetic field strength (positive values corresponding to field emerging from the head). The equivalent current
dipole source solution is shown as a white circle with a short line segment indicating the dipole orientation.
Figure 2 shows, for subject KL, pseudo-colored maps
of the magnetic field evoked by chromatic stimuli; the
maps were calculated at the time of maximum global
field power. Results are shown for stimulation of each
quadrant of the visual field [left (Fig. 2A) and right (Fig.
2B) upper quadrants; left (Fig. 2C) and right (Fig. 2D)
lower quadrants]. The data were well modeled by a
single equivalent current dipole (Table 1), the orientation of which is shown in each field map. For each
subject the calculated dipole was located on or near the
midline, pointing toward the contralateral hemisphere
with an orientation that was dependent on the quadrant stimulated. The dipole orientations vary in a way
that is consistent with the cruciform model of primary
visual cortex (Jeffreys and Axford, 1972a,b).
Figure 3 shows the 95% confidence regions for the
location of the calculated dipoles superimposed on
sagittal and coronal MR images for subject FF; the
inferior ellipsoid is for upper quadrant stimulation and
the superior ellispsoid is for lower quadrant stimulation. The different x2 values (see Table 1) and ellipsoid
sizes are consequences of the different signal-to-noise
ratios; upper-field stimulation evokes lower fieldstrength signals which produce larger confidence ellipsoids. The confidence regions lie on the midline (upper
visual-field stimulation) or displaced slightly toward
the contralateral hemisphere (lower visual-field stimulation). The active region of cortex is likely to lie
somewhere within this ellipse so that the data does not
imply that both hemispheres are activated by the
MEG ANALYSIS OF HUMAN AREA V1
TABLE 1
Statistics of the Equivalent Current Dipole Solutions for
the Responses to 1 c/d Isoluminant (Red/Green) Gratings
Presented in Each Quadrant (Left Upper, Left Lower, Right
Upper, Right Lower) of the Visual Field for Subjects FF, KL,
KS, and IH
Subject
FF
KL
KS
IH
Quadrant
Latency
(ms)
Correlation
coefficient (r)
r2
x2
gamma-Q
LU
LL
RU
RL
LU
LL
RU
RL
LU
LL
RU
RL
LU
LL
RU
RL
93
88
91
94
94
94
93
94
102
104
101
101
102
110
106
105
0.96
0.99
0.98
0.99
0.97
0.98
0.95
0.99
0.87
0.98
0.87
0.98
0.95
0.98
0.96
0.99
0.92
0.98
0.96
0.98
0.94
0.96
0.90
0.98
0.76
0.96
0.76
0.96
0.90
0.96
0.92
0.98
14
20
9.8
18
3.7
6.1
13
13
27
2.9
7.7
11
5.9
11
7.7
3.9
0.387
0.040
0.548
0.079
0.885
0.637
0.111
0.150
0.002
0.983
0.565
0.327
0.823
0.410
0.061
0.973
Note. Correlations less than 0.95 were regarded as nonsignificant.
51
stimulus. The lower visual field projects to the upper
surface of the calcarine fissure while the upper visual
field projects to the lower surface of the fissure. The
systematic shift in dipole orientation with stimulus
location and the source localizations shown in Fig. 3 for
the principal chromatic evoked reponses lead us to
conclude that the activated cortex lies within area V1.
Spatial-Frequency Characteristics of the Evoked
Magnetic Response
Figure 4 shows the variation in maximum global field
power as a function of the spatial frequency of stationary isoluminant chromatic gratings presented in the
right lower visual field for four subjects. The tuning
function measured for each subject has bandpass characteristics, peaking at 1–2 c/d and falling to the noise
level at 6–8 c/d. Figure 5 shows the latency of the
response increases with increasing spatial frequency.
This latency ranged from 87 ms (subject FF, 0.25 c/d) to
163 ms (subject KS, 6 c/d), with a mean increase of 27
ms when the spatial frequency increased from 0.5 to 4
c/d.
To examine the effect of the number of stimulus
cycles displayed on the magnitude of the evoked response, the experiment was repeated for subject FF
using field sizes of 4 3 6° and 8 3 12° (Fig. 6). At any
FIG. 3. Sagittal and coronal MR images of subject FF with the 95% confidence intervals for the location of the equivalent current dipole
solutions for the chromatic evoked response to right upper (inferior confidence interval) and right lower (superior confidence interval)
stimulation at the time of maximum global field power. The confidence ellipsoids lie on the lower (upper-field stimulation) and upper
(lower-field stimulation) banks of the calcarine fissure, on the midline (upper-field stimulation) or displaced slightly to the contralateral
hemisphere (lower-field stimulation).
52
FYLAN ET AL.
FIG. 4. Global magnetic field power (fT) as a function of the
stimulus spatial frequency (c/d) for isoluminant chromatic (red/
green) gratings presented in the right lower quadrant of the visual
field (four subjects). Other conditions as in Fig. 1. Error bars are
calculated from the anti-averaged response data.
given spatial frequency the larger field size (Fig. 6,
triangles) contained twice the number of cycles as the
smaller field size (Fig. 6, circles). The larger field size
yielded a stronger response, consistent with the increase in the area of activated cortex. In Fig. 6 the
results are normalized with respect to maximum global
field power. Bandpass spatial-frequency tuning curves
were obtained with both field sizes (Fig. 6a). Note that
above 1.5 c/d the tuning functions are coincident,
indicating that the field power is dependent on spatial
frequency. However, below 1.5 c/d the response is
stronger for the large-field condition, suggesting that
the limited number of cycles displayed for the low
spatial-frequency gratings contributes towards the observed low-frequency attenuation. Figure 6b plots the
normalized magnetic field power as a function of the
number of cycles displayed. The functions are coincident for stimulus displays containing less than six
cycles of sinusoid and diverge for greater cycle numbers. This indicates that when a small number of cycles
are present in the stimulus, the magnetic field power of
the response is limited by the number of cycles displayed.
FIG. 5. The latency (ms) of the peak global magnetic field power as a function of stimulus spatial frequency (c/d) for the data shown in Fig. 4.
MEG ANALYSIS OF HUMAN AREA V1
53
FIG. 6. Global field power as a function of (a) the spatial
frequency and (b) the number of displayed cycles of isoluminant
chromatic (red/green) grating stimuli presented in the right lower
quadrant of the visual field (subject FF). Results are shown for two
field sizes: 4 3 6° (circles); 8 3 12° (triangles). For each spatial
frequency, the larger field size contained twice as many cycles of
sinusoid as the smaller. To aid comparison the results are shown
normalized as a percentage of the maximum global field power
obtained in each case.
Temporal-Frequency Characteristics of the Evoked
Magnetic Response
Figure 7a shows the variation in global field power as
a function of the temporal frequency of 2 c/d isoluminant chromatic gratings presented in the right lower
visual field for four subjects (Fig. 7, filled circles). The
function is broadly lowpass, falling to the noise level at
40 Hz. However, note that the magnitude of the response is reduced for temporal frequencies near 1–2 Hz
and then increases to a peak near 4 Hz. This pattern is
confirmed by the trends in maximum global field power
as a function of temporal frequency averaged across the
four subjects (Fig. 7b). The latency of the chromatic
evoked response increased gradually with temporal
frequency up to approximately 10 Hz and then declined
rapidly to a level below that obtained for stationary
patterns (Fig. 8). To exclude the possibility that the
responses arose from luminance mechanisms responding to chromatic aberrations within the stimuli, the
experiment was repeated using 100% contrast luminance-modulated control stimuli for subjects FF and IH
(filled squares in Fig. 7). The control stimulus yielded
low-field-strength signals, typically less than 25% of
those obtained with chromatic gratings. Note also that
the latency of the response to the luminance modulated
control stimulus was independent of temporal frequency and at least 10 ms different from the chromatic
response. Therefore it is reasonable to conclude that
the responses we obtained with chromatic stimuli
cannot be attributed to luminance mechanisms.
Coregistration of MEG and MRI Data
The region of cortex generating the chromatic evoked
magnetic responses was determined as described under
FIG. 7. (a) The closed circles show global magnetic field power
(fT) as a function of the stimulus temporal frequency (Hz) for
isoluminant chromatic (red/green) gratings presented in the right
lower quadrant of the visual field (four subjects). Other conditions as
in Fig. 1. The filled squares show global magnetic field power for
100% contrast luminance-modulated (yellow/black) gratings of the
same mean chromaticity and luminance as the chromatic gratings
(two subjects). (b) Group-averaged data. Error bars indicate the
standard deviation.
Methods and confidence regions for the location of the
dipole generators were coregistered with both sagittal
and coronal MR images. Figure 9a shows the confidence volumes for stationary gratings of spatial frequency 0.5–2 c/d (subject F.F.). Figure 9b shows the
confidence volumes for a single test frequency of 2 c/d,
with drift rates of 0–16 Hz (subject IH). Note that in all
54
FYLAN ET AL.
FIG. 8. The difference in evoked response latency for 2 c/d isoluminant chromatic (red/green) drifting gratings presented in the right lower
quadrant of the visual field (four subjects) with respect to the latency for stationary stimulus presentation.
images the confidence volumes overlap and their positions are consistent with activity within area V1.
DISCUSSION
The evoked magnetic response to isoluminant chromatic (red/green) gratings was dominated by a single
major component (Fig. 1), the peak latency of which
depended on the spatio-temporal parameters of the
stimulus. Analysis of this major component showed
that the evoked responses were well modeled by a
single equivalent current dipole (Table 1) located in
cortical area V1 (Fig. 9), with the upper field projecting
to the lower surface of the calcarine fissure and the
lower field projecting to the upper surface (Fig. 3). We
show that luminance artifacts could not give rise to
these responses (filled squares in Fig. 7) and conclude
that the recorded activity arises from chromaticsensitive units within V1. Later components of lower
field power were evident in the evoked responses and
are thought to arise from extrastriate activity, possibly
V4; these components form the subject of a separate
investigation (Fylan et al., 1995). We determined chromatic response properties of area V1 by plotting the
maximum global field power of the evoked response as
a function of grating spatial and temporal frequency.
The spatial-frequency tuning function peaked at 1–2
c/d, with some attenuation occurring in field power at
lower spatial frequencies. The temporal-frequency tuning function was broadly lowpass, though the decline in
field power was nonmonotonic, which provides evidence
for the existence of at least two different types of
temporally tuned chromatic-sensitive units.
The responses recorded following stimulation of each
quadrant of the visual field with chromatic stimuli
were well described by equivalent current dipoles with
orientation predicted by the cruciform model of striate
cortex. Such adherence to the cruciform model was not
found by Aine et al. (1996) in their MEG investigation of
the retinotopic organization of striate cortex. However,
Aine et al. (1996) used luminance-modulated stimuli
which evoke complex response morphology and activate widespread regions of visual cortex over a prolonged time period (Fylan et al., 1995). This spatial and
temporal difference in cortical activation suggests that
chromatic and luminance information is processed
differently in area V1.
The spatial-frequency tuning response recorded here
showed some low-spatial-frequency attenuation, in contrast to the lowpass function determined psychophysically for red/green patterns (Mullen, 1985). However,
our experimental setup required a long viewing distance, precluding the use of more than one cycle of
sinusoid for the lowest spatial frequency displayed. It is
well documented that sensitivity to gratings varies
with the number of visible cycles (e.g., Howell and
Hess, 1978; Anderson and Burr, 1987) and this may
have caused a reduction in field power at the lower
spatial frequencies. This was investigated in a control
study by recording evoked responses using two different field sizes (see Fig. 6). The results provide evidence
that the reduction in field power at low spatial frequencies can be attributed to the small number of cycles
displayed, with less than six cycles of sinusoid resulting
in attenuation of the response. This may also explain
the reduction in visual evoked potential amplitudes at
low spatial frequencies with chromatic stimuli reported
by Rabin et al. (1994). Similarly, the magnetic evoked
responses to chromatic gratings reported by Regan and
He (1996) also show low spatial-frequency attenuation
MEG ANALYSIS OF HUMAN AREA V1
55
FIG. 9. Sagittal and coronal MR images with the 95% confidence intervals for the location of the equivalent current dipole solutions at the
time of maximum global field power for chromatic evoked responses to right lower quadrant stimulation. (a) Dipole locations for stationary
gratings of spatial frequency 0.5, 1.0, 1.5, and 2.0 c/d (subject FF). (b) Dipole locations for 2.0 c/d gratings drifting at 0, 1, 2, 4, 6, 8, and 16 Hz
(subject IH). Note that in each image the confidence intervals overlap.
in one subject, although this was not commented upon
and spatial frequencies below 1 c/d were not investigated. We conclude, as did Regan and He (1996), that
chromatic evoked responses are lowpass with respect to
spatial frequency. In addition, we suggest that the
chromatic units we have identified reside within area
V1 of the human visual cortex.
The temporal-frequency tuning response was broadly
lowpass, with a strong response occurring for stationary gratings (Fig. 7), but was nonmonotonic. Above 1
Hz the field power increases, reaching a maximum near
4 Hz before declining to the noise level at 32–40 Hz. A
similar pattern of temporal-frequency tuning for visual
evoked potentials was reported to isoluminant yellow/
56
FYLAN ET AL.
blue (Rabin et al., 1994) and red/green (M. A. Crognale,
personal communication) gratings. The form of the
temporal-frequency tuning function determined with
either electrical or magnetic measurements leads us to
suggest the existence of at least two different populations of chromatic-sensitive units, one sensitive to
stationary or slow moving patterns and the other
sensitive to fast moving patterns. Psychophysically
determined sensitivity functions for chromatic gratings
do not show a peak at 4 Hz (Kelly, 1983) although
recent psychophysical investigations of detection- and
direction-discrimination thresholds for chromatic
stimuli are broadly consistent with at least two chromatic-sensitive motion mechanisms (Derrington and
Henning, 1993; Gorea et al., 1993a,b). It has been
suggested (Kulikowski et al., 1996) that peaks in the
temporal-frequency tuning function of the response to
chromatic stimuli may arise from luminance artifacts
within the stimulus. However, while this may account
for the data described by Kulikowski (1996) in which
the electrical responses to chromatic and luminance
stimuli were of similar amplitude, the large difference
in field power between the chromatic and luminance
responses recorded here (Fig. 7) suggests that the
chromatic response could not have been generated by
the same mechanism as that underlying the luminance
response. As discussed by Kulikowski (1996), chromatic
aberrations make isoluminant stimuli difficult to
achieve. Furthermore, differences in the isoluminant
point of different units within the visual cortex (Zrenner, 1983) make it unlikely that the luminance pathway can be silenced. However, the response of the
luminance pathway to chromatic stimuli is not greater
than its response to high-contrast luminance stimuli of
the same spatial and temporal frequencies. Our data
therefore demonstrate that the peaks observed in the
temporal-frequency tuning curves are generated by the
same population of cortical neurons that gives rise to
the response to stationary chromatic gratings.
The principal response to chromatic stimulation
showed changes in latency with temporal frequency
that were similar for all four subjects (Fig. 8). The
latency increased with temporal frequency up to 8–16
Hz and thereafter decreased to values earlier than
those recorded using stationary gratings. This is suggestive of a transition from a low-temporal-frequency
selective mechanism to a high-temporal-frequency selective mechanism. The responses at high frequencies
must arise from chromatic units as the 100% contrast
luminance response is of significantly lower field-power
than the chromatic response at these temporal frequencies (Fig. 7). These results are indicative of a third
population of chromatic units, sensitive to high velocity
targets. This may be why in two subjects the temporalfrequency tuning curves show evidence of an additional
peak at 10–20 Hz. Examination of Fig. 9 shows that the
response to all temporal frequencies originate from a
single visual area, namely V1.
We have demonstrated that the principal component
of the evoked magnetic response to chromatic stimuli,
recorded from over the occipital cortex, arises from
neural activity within cortical area V1. We further
show that, for red/green patterns, this area is maximally responsive at low spatial frequencies (,4 cpd)
and shows a multimodal temporal-frequency response,
peaking at 0 and 4 Hz. Our results support the hypothesis that chromatic-sensitive units in the human visual
system play an important role in processing motion
information.
ACKNOWLEDGMENTS
The experimental work on this project was supported by the Dr.
Hadwen Trust for Humane Research, Aim Project A2020 Magnobrain, Fight for Sight (London), and a Wellcome Travel Grant to F.F.
REFERENCES
Aine, C. J., Supek, S., George, J. S., Ranken, D., Lewine, J., Sanders,
J., Best, E., Tiee, W., Flynn, E. R., and Wood, C. 1996. Retinotopic
organization of human visual cortex: Departures from the classical
model. Cerebral Cortex. 6:354–361.
Anderson, S. J., and Burr, D. C. 1987. Receptive field size of human
motion detection units. Vision Res. 27:621–635.
Anderson, S. J., Drasdo, N., and Thompson, C. M. 1995. Parvocellular
neurons limit motion acuity in human peripheral vision. Proc. R.
Soc. London B 261:129–138.
Cavanagh, P., Anstis, S. 1991. The contribution of color to motion in
normal and color-deficient observers. Vision Res. 31:2109–2148.
Cropper, S. J., Derrington, A. M. 1994. Motion of chromatic stimuli:
First-order or second-order? Vision Res. 34:49–55.
Derrington, A. M., Henning, G. B. 1993. Detecting and discriminating
the direction of motion of luminance and colour gratings. Vision
Res. 33:799–811.
ffytche, D. H., Skidmore, B. D., Zeki, S. 1995. Motion-from-hue
activates are V5 of human visual-cortex. Proc. R. Acad. Sci. B
260:353–358.
Fylan, F., Holliday, I. E., Singh, K. D., Anderson, S. J., Harding,
G. F. A. 1995. Cortical origins of visual evoked magnetic responses
differ in humans for chromatic and luminance stimuli. Hum. Brain
Mapping (Suppl.) 1:51.
Gegenfurtner, K. R., and Hawken, M. J. 1995. Temporal and chromatic properties of motion mechanisms. Vision Res. 35:1547–1563.
Givre, S. J., Arezzo, J. C., Schroeder, C. E. 1995. Effects of wavelength
on the timing and laminar distribution of illuminance-evoked
activity in macaque V1. Vis. Neurosci. 12:229–239.
Gorea, A., Papathomas, T. V., Kovacs, I. 1993a. Motion perception
with spatiotemporally matched chromatic and achromatic information reveals a ‘‘slow’’ and a ‘‘fast’’ motion system. Vision Res.
33:2515–2534.
Gorea, A., Papathomas, T. V., Kovacs, I. 1993b. Two motion systems
with common and separate pathways for color and luminance.
Proc. Natl. Aca. Sci. USA 90:11197–11201.
Hämäläinen, M., Hari, R., Ilmoniemi, R., Knuutila, J., Lounasmaa,
O. 1993. Magnetoencephalography—Theory, instrumentation and
applications to noninvasive studies of the working human brain.
Rev. Mod. Phys. 65:413–496.
MEG ANALYSIS OF HUMAN AREA V1
Harding, G. F. A. 1993. How surgeons could navigate the brain. New
Scientist 140; 1903:28–31.
Howell, E. R., Hess, R. F. 1978. The functional area for summation to
threshold for sinusoidal gratings. Vision Res. 18:369–374.
Jeffreys, D. A., Axford, J. G. 1972a. Source locations of pattern
specific components of human visual evoked potentials. I. Components of striate cortical origin. Exp. Brain Res. 16:1–21.
Jeffreys, D. A., Axford, J. G. 1972b. Source locations of pattern
specific components of human visual evoked potentials. II. Components of extrastriate cortical origin. Exp. Brain Res. 16:22–24.
Kelly, D. H. 1983. Spatiotemporal variation of chromatic and achromatic contrast thresholds. J. Opt. Soc. Am. 73:742–750.
Kulikowski, J. J., Robson, A. G., McKeefry, D. J. 1996. Specificity and
selectivity of chromatic visual evoked potentials. Vision Res. 36:
3397–3401.
Lennie, P. 1993. The role of M and P pathways. In Contrast
Sensitivity, Proceedings of the Retina Research Foundation Symposia (R. Shapley and D. Man-Kit, Eds.), Vol. 5, pp. 201–214. MIT
Press, Cambridge.
Matlashov, A., Slobodchikov, A., Bakharev, A., Zhuravlev, Yu, and
Bondarenko, N. 1995. Biomagnetic multichannel system built with
19 cryogenic probes. In Biomagnetism: Fundamental Research and
Clinical Applications (C. Baumgartner, L. Deecke, G. Stroink, and
S. J. Williamson, Eds.), pp. 493–496. Elsevier, Amsterdam.
Medvick, P., Lewis, P. S., Aine, C. J., and Flynn, E. R. 1990. Monte
Carlo analysis of localization errors in magnetoencephalography.
In Advances in Biomagnetism (S. J. Williamson, M. Hoke, M.
Kotani, and G. Stroink, Eds.), pp. 543–546. Plenum, New York.
Merigan, W. H., Maunsell, J. H. R. 1993. How parallel are the primate
visual pathways. Ann. Rev. Neurosci. 16:369–402.
Merigan, W. H., Byrne, C. E., and Maunsell, J. H. R. 1991. Does
primate motion perception depend on the magnocellular pathway?
J. Neurosci. 11:3422–3429.
Mosher, J. C., Lewis, P. S., Leaky, R. M. 1992. Multiple dipole
57
modeling and localization from spatio-temporal MEG data. IEEE
Trans. Biomed. Eng. 39:541–557.
Mullen, K. T. 1985. The contrast sensitivity of human colour vision to
red-green and blue-yellow chromatic gratings. J. Physiol. 359:381–
400.
Press, W. H., Flannery, B. P., Teukolsky, S. A., and Vetterling, W. T.
1989. Numerical Recipes in C. Cambridge Univ. Press, Cambridge.
Rabin, J., Switkes, E., Crognale, M., Scheck, M. E., Adams, A. J. 1994.
Visual evoked potentials in three-dimensional color space: Correlates of spatio-chromatic processing. Vision Res. 34:2657–2671.
Regan, D. 1989. Human Brain Electrophysiology. Evoked Potentials
and Evoked Magnetic Fields in Science and Medicine. Elsevier,
New York.
Regan, D., He, P. 1996. Magnetic and electrical brain responses to
chromatic contrast in human. Vision Res. 36:1–18.
Schiller, P. H. 1991. The color-opponent and broad-band channels of
the primate visual system. In From Pigments to Perception (A.
Valberg and B. B. Lee, Eds.), pp. 127–132. Plenum, New York.
Singh, K. D., Holliday, I. E., Furlong, P. L., Harding, G. F. A. 1997.
Evaluation of MRI-MEG/EEG co-registration strategies using
Monte-Carlo simulation. Electroenceph. Clin. Neurophysiol. 102:
81–85.
Singh, K. D., Harding, G. F. A. Monte-Carlo analysis and confidence
region ellipsoids for equivalent current dipole solutions to EEG/
MEG data. Proceedings of the 10th International Conference on
Biomagnetism, in press.
Zeki, S. M. 1974. Functional organisation of a visual area in the
posterior bank of the superior temporal sulcus in the rhesus
monkey. J. Physiol. London 236:549–573.
Zeki, S., Watson, J. D. G., Lueck, C. J., Friston, K. J., Kennard, C.,
Frackowiak, R. S. J. 1991. A direct demonstration of functional
specialisation in human visual cortex. J. Neurosci. 11:641–649.
Zrenner, E. 1983. Neurophysiological Aspects of Color Vision in
Primates. Springer-Verlag, Berlin.