Attention to Speed of Motion, Speed Discrimination, and Task Difficulty:
An fMRI Study
Stefan Sunaert,* Paul Van Hecke,* Guy Marchal,* and Guy A. Orban†
*Afdeling Radiologie, UZ Gasthuisberg, B-3000 Leuven, Belgium; and †Laboratorium voor Neuro- en Psychofysiologie,
KULeuven Medical School, Campus Gasthuisberg, B-3000 Leuven, Belgium
Received December 6, 1999
We studied the functional neuroanatomy of atten-
tion to speed of motion using functional magnetic res-
onance imaging in eight healthy subjects, who per-
formed a speed discrimination (SID) task using a
random textured pattern moving at a reference speed
of 6 deg/s. During the control condition (DIM), with
retinal stimulation identical to that during SID, sub-
jects detected the dimming of the central fixation
point. Attention to speed (SID compared to DIM) acti-
vated mainly ventral V3 and V4, dorsal V3 and V3A.
Compared to a fixation control condition, speed dis-
crimination recruited a large visuomotor network, in-
cluding hMT /V5?. However, hMT /V5? was only mar-
ginally more active during speed discrimination than
during dimming detection. T hus hMT /V5? is involved
in speed discrimination, in line with the speed dis-
crimination impairments following hMT /V5? lesions,
but our results suggest that this activity simply re-
flects the processing of motion rather than attention
to speed. Manipulating the difficulty of the speed dis-
crimination task over a large range of the psychomet-
ric curve revealed that increasing difficulty linearly
increases activity in right frontal regions, as well as in
lateral occipital and dorsal parietal regions. A weak
effect of difficulty was also observed in dorsal V3.
© 2000 Academic Press
Key Words: vision; motion; speed discrimination;
functional magnetic resonance imaging.
Estimating speed of motion is an important aspect of
daily life. After all, pursuit and interception move-
ments depend heavily upon accurate speed judgments.
The question then arises as to where in the human
brain speed of motion is processed, a question that also
has clinical relevance. Impaired speed discrimination
has been reported in dyslexics (Demb et al., 1998) and
is associated with impaired smooth pursuit in schizo-
phrenics and their relatives (Chen et al., 1999). One of
the obvious candidate regions for speed processing is
the human homologue of the primate middle temporal
(MT/V5) area. Electrophysiological studies have shown
that a substantial portion of its neurons are tuned for
speed and/or for direction of motion (Maunsell and Van
Essen, 1983; Mikami et al., 1986; Lagae et al., 1993). In
addition, lesions of this area and its satellites produce
profound deficits in speed discrimination when random
dot patterns are used as stimuli (Orban et al., 1995).
Speed discrimination deficits have been linked to le-
sions of hMT/V5? in humans (Greenlee et al., 1995;
Vaina, 1989). Corbetta et al. (1991), through the use of
positron emission tomography (PET), was the first to
implicate hMT/V5? in attention tothe speed of motion
of a multidimensional stimulus. Beauchamp et al.
(1997) specifically studied featural attention to speed
and reported an enhanced MR signal over hMT/V5?
during attention to speed.
Although hMT/V5? is the best known motion-re-
sponsive area (Zeki et al., 1991; Watson et al., 1993;
Tootell et al., 1995), recent studies (Cheng et al., 1995;
Dupont et al., 1994; Tootell et al., 1997; Sunaert et al.,
1999) have revealed many more such areas in the
human brain. Furthermore, a recent PET study of our
group (Orban et al., 1998) has shown that lingual and
cuneal regions, rather than hMT/V5?, are involved in
speed discrimination. However, we could have missed
an attention to speed effect in hMT/V5? for two rea-
sons. In our PET study, speed discrimination was di-
rectly compared to a control condition without atten-
tion to speed, while Beauchamp et al. (1997) reported
that MR signal changes compared to a baseline were
different during attention to speed and to color.
Beauchamp et al. (1997) studied hMT/V5? in individ-
ual subjects, while this region was only defined for the
group in our study. Therefore, we reinvestigated the
processing of speed using whole brain fMRI, by com-
paring the speed discrimination task, used in the PET
study, to a visually matched control condition, detec-
tion of the fixation point dimming, and also to a lower
level control condition, passive viewing of the fixation
point. Compared to the PET study we made an impor-
tant addition by imaging speed discrimination over the
NeuroImage 11, 612–623 (2000)
doi:10.1006/nimg.2000.0587, available online at http://www.idealibrary.com on
Copyright © 2000 by Academic Press
All rights of reproduction in any form reserved.
whole psychometric range, hypothesizing that more
difficult discriminations would increase featural atten-
tion and hence the chance to observe an hMT/V5?
effect. Furthermore, we mapped the motion-responsive
(Sunaert et al., 1999) and retinotopic (Tootell et al.,
1995) regions in each subject.
TheMR scans wereperformed on eight young (23–26
years), right-handed subjects with normal or corrected
to normal vision and no history of neurological or psy-
chiatric disease. The ethical committee of the KULeu-
ven medical school approved the study, and subjects
gave their informed consent in accordance with the
Helsinki declaration. All subjects had their head move-
ments restricted by a bite-bar, viewed the stimuli bin-
ocularly, and were instructed tomaintain fixation on a
small red target in the center of the screen. Fixation of
gaze was monitored (Sunaert et al., 1999) using an
MR-compatible infrared eye movement-tracking device
(Ober2, Permobil Meditech AB, Sweden).
Stimuli were projected by means of an LCD projector
(Sharp GX-3800, 640 by 480 pixels, 60 Hz refresh) onto
a translucent screen positioned in the bore of the mag-
net at a distanceof 30 cm from thepoint of observation.
Stimuli were generated with a PC using a Tiga-dia-
mond (Salient AT3000) graphics card. In all experi-
ments, except the retinotopic mapping experiment, the
stimulus was a central, 3° diameter, circular random
textured pattern (RTP) consisting of 50% white dots
(5-minarc pixelsize) on a black background. The mean
luminancewas 79.4 cd/m2 and thecontrast 0.97. In the
main experiment (Fig. 1), subjects performed twotasks
under identical retinal input. During the first part of
the trial (duration 600 to 700 ms, randomly chosen),
the RTP moved uniformly to the right on a horizontal
axis at a speed lower or higher than the reference
speed of 6 deg/s. Small changes in speeds were ob-
tained by shifting the pattern not every frame but only
every two frames or every two out of three frames.
Additionally, in half the trials the central red fixation
point decreased in luminance during this part. During
the remainder of the trial (400 to 300 ms) the RTP
remained stationary and the fixation target main-
tained a constant luminance. Trial rate was 60/min. In
the speed discrimination task (speed identification -
SID) subjects judged the speed of motion by pressing a
right or left response key for speeds faster or slower
than the reference. Speed identification is one partic-
ular instance of speed discrimination, and this latter
term will be used for this throughout this report. We
only used one type of speed discrimination since in our
previous PET study (Orban et al., 1998) there was no
significant difference between identification and dis-
crimination except for a right fusiform region. In the
control task they signal the dimming of the fixation
point (DIM) by pressing both response keys simulta-
neously. Subjects weretested at fivedifficulty levels for
the speed discrimination task and at one level for the
DIM task. Speed differences were individually adapted
toproduce 62, 74, 84, 90, and 95% correct responses on
the SID task, and luminance decrease was adapted to
yield 84% correct responses in the DIM task. Stable
performance was reached after 5 to 6 daily training
F IG. 1.
Speed of the random textured pattern (RTP) and luminance of the fixation point are plotted as a function of time. Solid and dashed lines
represent different types of trial. Stippled lines indicate that motion and dimming duration lasted either 600 or 700 ms. These random
changes were introduced to uncouple motion speed and amplitude. Arrows indicate direction of motion, black dot the fixation target. Each
trial lasted 1 s.
Schematic representation of the stimulus presentation during a speed discrimination or dimming of fixation point detection trial.
SPEED DISCRIMINATION AND fMRI
sessions, in which an average of 900 SID trials and 125
DIM trials were presented.
A functional imaging series consisted of 90 gradient-
echo(GE) echoplanar imaging (EPI) whole brain scans
(Siemens Vision 1.5T) acquired every 4.5 s (TE 40 ms,
flip angle 90°, 64 ? 64 matrix, 200 ? 200 mm2FOV, 32
noncontiguous slices, 4-mm slicethickness, 1-mm gap).
In the main experiment the five difficulty levels of the
SID task, the DIM task, and a fixation only condition
(FIX - black background with red fixation dot only)
were alternated. Within one time-series (90 images),
the SID and DIM conditions were each presented once
for 45 s (10 images), interleaved with the FIX condition
for 22.5 s (5 images). The subjects were given auditory
instructions as towhat task toperform with the words
“speed,” “dimming,” or “stop” presented at the begin-
ning of each epoch. These series were repeated 12
times in two scanning sessions, with the conditions in
a different order, yielding a total of 120 images per SID
or DIM condition.
Twocontrol experiments were performed on all eight
subjects. In control experiment 1, two conditions were
alternated every 10 images within a time-series: the
UNI condition in which subjects passively viewed the
3° RTP moving coherently at 6 deg/s in eight randomly
ordered directions, and the STA condition in which
they viewed thestationary RTP. Twotime-series of 120
GE-EPI images were acquired, yielding 120 images per
condition. These conditions were identical to those
used in our earlier study (Sunaert et al., 1999) and
were used to identify motion-responsive regions, such
as hMT/V5?. In the second control experiment, a
coarse retinotopic mapping was performed by stimula-
tion of the horizontal and vertical visual field meridian
with a wedge-shaped alternating checkerboard (Tootell
et al., 1995). The horizontally (HOR) and vertically
(VER) oriented checkerboards were alternated every
10 images during 2 time-series of 120 GE-EPI scans.
Sagittal anatomical images were acquired prior to
functional scanning in each session (3-D MPRAGE,
TR/TE ? 11.4/4.4 ms, TI ? 300 ms, FOV 256 ? 256
mm2, 256 ? 256 matrix, 160-mm slab thickness, 128
After transfer of the data to a workstation, image
processing and statistical analyses were performed us-
ing SPM96 (Friston et al., 1994a,b). All EPI volumes
were realigned to the first volume and a mean image
was created of the realigned volumes. The sagittal
anatomical image was coregistered with this mean im-
age to ensure that the anatomical and the functional
images occupy the same space. Finally, the anatomical
image was spatially normalized to a standard brain
(Talairach and Tournoux, 1988), using an affine and
nonlinear transformation. This transformation, map-
ping the anatomical scan to the template, was applied
to the functional volumes. The data were smoothed
using a 5-mm full-width at half maximum (FWHM)
isotropic Gaussian kernel for the individual analyses
and a 10 mm FWHM kernel for thegroup analysis. The
use of a wider smoothing kernel for the group compar-
ison was chosen to compensate for residual variability
in gyral anatomy after spatial normalization. The
smoothness of theSPMs was estimated at 9 and 14 mm
for single-subject and group analysis, respectively.
The statistical data analysis was performed by mod-
eling the different conditions as a box car function
convolved with the hemodynamic response function
(implemented as a delayed Gaussian function), in the
context of the general linear model as employed by
SPM96. Global changes were adjusted by proportional
scaling and low frequency confounding effects were
removed by an appropriate high pass filter. Specific
effects were tested by applying appropriate linear con-
trasts to the parameter estimates for each condition,
resulting in a t statistic for each and every voxel. These
t statistics constitute a statistical parametric map
(SPM). The threshold was set at P ? 0.05 corrected for
multiple comparisons for activation height and to P ?
0.05 for activation extent.
In themain experiment, thefollowing contrasts were
used: (1) To isolate attention to speed, all difficulty
levels of speed discrimination were contrasted to de-
tection of the dimming of the fixation point (SID-DIM).
(2) To identify all brain regions involved in speed dis-
crimination, including early visual and motor regions,
all SID conditions were contrasted to fixation (SID-
FIX). (3) In order to identify brain regions that covary
linearly or quadratically with thedifficulty levels of the
speed discrimination task, four contrasts were defined:
increasing and decreasing first-order contrasts, as well
as “U”-shape and “inverted U”-shape second-order con-
trasts. In control experiments 1 and 2, the contrast
UNI-STA was performed toidentify motion-responsive
regions, the HOR-VER comparison to identify the hor-
izontal meridian projection, and VER-HOR to find the
vertical meridian projection.
Task Performance during MRI Scanning
All subjects maintained fixation well during scan-
ning as attested by the eye movement recordings: the
subjects maderelatively few saccades (fewer than 3 per
45-s epoch) and there were no significant differences
among the various treatment and control conditions.
At the time of scanning, the subjects were fully trained
(Fig. 2A) and had reached stable task performance at
all five difficulty levels of the speed discrimination task
(SID) for at least two consecutive training sessions.
During the main experiment, the average SID task
performance of the eight subjects equaled 96, 91, 83,
74, and 67% correct responses, for speed differences
(Weber fraction) averaging 23, 15, 10, 8, and 6% (Fig.
SUNAERT ET AL.
2B). The average performance level for the DIM task
was 85% correct.
Attention to Speed: Group Analysis
The comparison between speed discrimination and
detection of the dimming of the fixation point (SID-
DIM) isolates attention to speed, but also the decision
process associating a given speed with a motor re-
sponse. This contrast resulted in significant activation
(P ? 0.05 corrected for multiple comparisons) in left
and right lingual gyrus, in the cuneus bilaterally, and
in a midline region in the anterior cingulate gyrus,
which might correspond to SMA (Van Oostende et al.,
1997) (Table 1; Fig. 3A). Weaker activations (Pcorr ?
0.2) were found in an additional focus in the left cu-
neus, situated slightly dorsal and anterior to the first
one, as well as a focus in primary visual cortex. The
opposite subtraction (DIM-SID) yielded no significant
activation site. These group results agree with our
previous PET results (Orban et al., 1998) rather well.
Compared to a detection task, speed discrimination
yielded significant activation in right cuneus and left
anterior cingulate, with weaker activation sites in left
and right lingual gyrus.
Speed Discrimination Network: Group Analysis
When speed discrimination (SID) was compared to
FIX, in which subjects viewed only the fixation target
in an otherwise empty display, an extensive network
(Fig. 3B) was visualized, including early visual areas,
hMT/V5?, hV3A, sensory-motor cortices, basal gan-
glia, and cerebellum as well as right medial and infe-
rior frontal regions. The extent of the network is not
surprising given the many differences in visual stimu-
lation, task, difficulty and motor response between the
Motion-Sensitive Regions Compared to Attention to
Speed and to the Speed Discrimination Network:
A group analysis comparing passive viewing of a
uniformly moving RTP to a static RTP identified
hMT/V5?, hV3A, a bilateral lingual focus, and sev-
eral regions along the dorsal extent of the intrapari-
etal sulcus (Table 2) as regions that are responsive to
passive viewing of motion. This replicates our previ-
ous study (Sunaert et al., 1999) since 12 of the 13
motion-responsive regions identified in that study’s
group analysis are also significant in the present
group analysis. Furthermore, we observe bilateral
activation of the two parietal motion-responsive re-
gions which were significant only in the right hemi-
sphere in Sunaert et al. (1999; see Figs. 3–5 of this
paper for the precise anatomical localization of these
different motion responsive regions). Testing for
speed attention effects in the motion-responsive re-
gions confirm the very limited effect of this featural
attention in the motion-responsive regions (compare
Figs. 3A and 3C). E xcept for the lingual motion-
T ABL E 1
Group Analysis (N ? 8): Regions Significant
in the Subtraction SID-DIM
XYZZ scoreP value
Note. Bold: P ? 0.05 corrected for multiple comparisons, other P ?
0.2 corrected. See list of abbreviations for labels identifying the
activation foci. R and L indicate right and left hemispheres.
F IG. 2.
first training session (triangles) and at thefully trained stage, during
fMRI scanning (dots). (B) Mean speed discrimination performance of
eight subjects during scanning. Percentage correct responses are
plotted as a function of speed difference expressed as the Weber
fraction. Vertical error bars indicate standard deviations across tri-
als (A) or across subjects (B). Horizontal error bars indicate standard
deviations of Weber fractions used across subjects (B).
(A) Speed discrimination performance for subject 1 at
SPEED DISCRIMINATION AND fMRI
responsive regions and a weak effect in right hV3A
and left LOS/K O attention to speed has no effect on
the activity of motion-responsive regions (contrast
SID-DIM in Table 2). This also replicates our earlier
study (see Fig. 4 in Orban et al., 1998). More of the
motion-responsive regions are, however, integrated
into the speed discrimination network (contrast SID-
FIX in Table 2). It is noteworthy that while the
moving stimulus used in the speed discrimination
recruits occipital motion-responsive regions, includ-
ing hMT/V5?, this is much less the case for the
parietal motion regions (see also Fig. 3B).
F IG. 3.
all difficulty levels) and detection of the dimming of the fixation point (DIM). (B) Statistical parametric map showing the differ-
ence between speed discrimination (SID, all difficulty levels) and fixation only (FIX). (C) Statistical parametric map showing the
difference between passive viewing of moving RTP compared to passive viewing of stationary RTP. Only voxels with significance of P ?
0.001 are shown (colored red to yellow according to scale shown in insets), overlaid on transversal sections through the average
structural MRI of the subjects’ brains. Numbers indicate the level of sections below or above anterior commissure-posterior commissure
Group analysis (N ? 8). (A) Statistical parametric map showing the difference between speed discrimination (SID,
SUNAERT ET AL.
Speed Discrimination Difficulty Effects: Group
The group analysis of the main experiment yielded
significant first order (linear) effects of task difficulty
in several regions, listed in Table3. Linear increases in
brain activity with task difficulty were significant in
right frontal cortex (Fig. 4): one site was located in the
medial frontal gyrus, one in the middle frontal gyrus,
and another in the anterior portion of the right insula.
In addition, a region in right lateral occipital cortex
(LOC) and a region along the right dorsal intraparietal
sulcus (DIPS) exhibited significant increases (Fig. 4).
The increase in adjusted MR signal in these five acti-
vation sites is plotted in Figs. 5A–5E. Note that in all
regions the MR signal measured during the DIM task
approximates that of the intermediate SID condition
(SID3), in which the number of correct responses was
nearly equal tothat for DIM (85% correct responses). It
also deserves mention that in all sites, except lateral
occipital cortex, the MR signal in DIM exceeds that in
the baseline fixation condition, indicating that greater
difficulty results in increased activation of these re-
No significant linear effects of difficulty were found
in the regions activated by attention tospeed. It is only
in right dorsal V3 that the MR signal (Fig. 5F) showed
a slight but consistent increase with task difficulty,
which proved to be significant when we performed a
conjunction between linear increases of task difficulty
and thecomparison SID-DIM. It is noteworthy that left
T ABL E 2
Group Analysis (N ? 8): Z Score and Percentage MR Signal Increase (PC)
for the Comparisons SID-DIM, SID-FIX, and SID-DIM
CoordinatesUNI-STA SID-DIMSID-FIX DIM-FIX
XYZZ PCZ PCZ PCZ PC
Note. Values shown are for the most significant voxel of the motion-sensitive regions. Motion-sensitive regions were determined by the
comparison of uniformly moving RTP compared to static RTP. Significant (P ? 0.05 corrected for N ? 15 regions tested) Z scores are shown
in bold typeface. See list of abbreviations for labels identifying the activation foci. R and L indicate right and left hemispheres. *Corresponds
to region labeled SFS in Sunaert et al. (1999).
F IG. 4.
purple) activity with speed discrimination task difficulty. The SPMs are thresholded at P ? 0.001 uncorrected, and superimposed on the
three-dimensional surface reconstruction of the SPM96 template brain.
Group statistical parametric map (N ? 8) indicating voxels that show a linearly increasing (red-yellow) or decreasing (blue-
SPEED DISCRIMINATION AND fMRI
V3d just failed toreach significancein this conjunction.
This underscores the role of V3d in speed discrimina-
Regions that showed a significant linear decrease
with task difficulty (Table 3) were localized in the
cingulate gyrus (twosites in the anterior cingulate and
one in the posterior cingulate), left superior and infe-
rior frontal gyri, the precuneus, left inferior parietal
lobule, right postcentral gyrus, left superior temporal
gyrus, right temporal pole, medial temporal lobe bilat-
erally, cerebellum, and thalamus (Fig. 4). These re-
gions were generally deactivated—though not signifi-
cantly at P ? 0.05 corrected for multiple comparison—
during SID compared to DIM.
Second order, U and inverted U-shaped effects of
task difficulty did not reach the corrected significance
level. A nonsignificant trend was present in left cere-
bellum (coordinates ?9, ?48, ?12).
Retinotopic Mapping Compared to SID-DIM:
Single-subject analysis of speed discrimination com-
pared to detection of the dimming (SID-DIM) revealed
significant cuneal and lingual activation sites similar
tothose observed in the group analysis (Table 1). In an
effort to further identify the cuneal and lingual sites,
they were compared with the projections of the hori-
zontal and vertical meridians. Figure 6A shows two
neighboring coronal sections through the brain of sub-
ject 4 ontowhich the left lingual activation site (yellow
voxels) and the horizontal (green outlines) and vertical
(blue outlines) meridians are superimposed. Of partic-
ular interest is the second vertical meridian projection,
counting from the calcarine sulcus, which separates
V3v from V4v (Sereno et al., 1995; DeYoe et al., 1996;
Engel et al., 1997). The activation generated by atten-
tion tospeed in the lingual gyrus is located just lateral
to the second vertical meridian, indicating that it cor-
responds toV4v. Alsofor the cuneal activation, it is the
second vertical meridian projection (again counting
from the calcarine) which is of interest as it separates
V3d from V3A. This can be best visualized on horizon-
tal sections. Two adjacent horizontal sections through
the brain of subject 3 are shown in Fig. 6C. The left
cuneal attention to speed activation is located medial
to the second vertical meridian projection in this sub-
F IG. 5.
levels of difficulty of SID (SID1 to SID5, light hatching), DIM (no
hatching), and FIX (dark hatching) conditions plotted for the most
significant voxel of the regions that showed a significant linear
increase in brain activity with task difficulty. Localization of the foci
in A-E is tabulated in Table 3. The horizontal line indicates the
average adjusted MR signal in DIM. Vertical lines indicate SE.
Group average (N ? 8) adjusted MR signal for the five
T ABL E 3
Group Analysis (N ? 8): Regions That Show a Significant
Linear Increase or Decrease with Task Difficulty in the SID
R Ant Insula
L Lat POC
R temp pole
R Postcentral gyrus
Note. R and L indicate right and left hemispheres. See list of
abbreviations for labels identifying the activation foci.
SUNAERT ET AL.
ject, indicating that it corresponds to V3d. Figures 6B
and 6D show, in the same two subjects, the motion-
responsive voxels superimposed upon the meridian
projections on the sections used for visualizing the
attention to speed sites. This confirms, at the single-
subject level, the relative lack of overlap between mo-
tion-responsive regions and attention to speed effects.
The horizontal section in Fig. 6D also shows that the
comparison of motion-responsive sites with the atten-
tion to speed sites can assist in the discrimination of
these latter. Indeed the V3d attention to speed site is
located just medial to the hV3A region and lateral to
the site of a V2d motion response. Significant activa-
tion of the second vertical meridian projection was
obtained in most (6/8) subjects (Table 4), allowing the
lingual and cuneal activation sites to be identified in
these subjects. In the remaining two subjects most of
the attention to speed effects occurred in the cuneus
and their location with respect to hV3A, identified by
its motion response, was used for identification pur-
pose. Thus only the identification of the right lingual
activation in subject 1 as V3v, based on anatomical and
stereotactic location, can be considered tobe somewhat
tentative. The single-subject analysis therefore indi-
cates that attention to speed effects were localized in
V3v, V4v, V3d, and V3A. Comparison across subjects
revealed no systematic pattern, and neither was any
consistent pattern observed with respect todorsal-ven-
tral, left-right dominance. Table 4 indicates that at
least one ventral and dorsal region was activated in
every subject and that in most (6/8) subjects, V3d was
active in either the left or right hemisphere.
Given the large variability across subjects, we list in
Table 5 those regions which were significantly acti-
vated in SID-DIM in at least three of eight subjects. In
addition to V3d bilaterally, right V3A and left V3v,
significant activation was observed in the right fusi-
form gyrus, in right frontal cortex, in the cerebellum
bilaterally, as well
as in left postcentralgyrus
Involvement of hMT/V5? in Speed Discrimination:
We failed to observe a significant attention to speed
effect in hMT/V5? in the group analysis (Table 2).
Interindividual differences in thestrength of theatten-
tion effect and in the localization of hMT/V5? might
have blurred the group result. Therefore, we identified
hMT/V5? in each subject individually by comparing
uniform moving RTP minus its static counterpart and
tested the contrast SID-DIM in the most significant
voxel of hMT/V5?. There was a significant attention to
speed effect in only 3 of the16 hMT/V5? regions, in the
left hemisphere of subjects 6–8.
Up to now we have made only direct comparisons of
the MR signals in SID and DIM. Given the thresholds
used in thestatistical tests of SPM, small differences in
hMT/V5? activity between SID and DIM may have
been missed. The procedure of Beauchamp et al. (1997)
comparing changes in MR signals relative toa baseline
in conditions with and without attention tospeed may,
however, be more sensitive. Therefore, the MR-signal
increase in the most significant voxel of hMT/V5? was
compared to FIX for the SID and DIM conditions in
Fig. 7. It is plain that the distributions of MR signal
change are very similar for the two conditions (Fig.
7A). There is, however, a small difference as the MR
signal change is on average 0.1% larger in speed dis-
crimination than in dimming detection. This difference
is significant when tested with a paired Student’s t test
(P ? 0.013). Thus, hMT/V5? is only marginally more
active during SID than during DIM, while the large
activation in DIM and SID compared toFIX reflects the
T ABL E 4
Single-Subject Analysis. Hemispheres in Which the Differ-
ence SID-DIM Reached Significance in Four Retinotopic Re-
Subject V3vV4vV3d V3A
L ? R
L ? R
Note. R and L indicate right and left hemispheres, Y indicate
significant activation by the meridians in the retinotopic mapping
T ABL E 5
Single-Subject Analysis: Regions Significant
in the Subtraction SID-DIM
L Postcentral gyrus
Note. Prevalence (observed subjects/total subjects), median Ta-
lairach coordinates and median Z score are indicated. Only regions
significant in at least 3/8 subjects are listed. R and L indicate right
and left hemispheres. See list of abbreviations for labels identifying
the activation foci.
SPEED DISCRIMINATION AND fMRI
factor common to SID and DIM, i.e., the automatic
processing of motion.
Figure 7B plots the distributions in MR signal
changes in the most significant voxel of hV3A, which
could also be identified in both hemispheres of each
subject. Compared to the distribution of MR signal
increase for DIM, the distribution for SID is clearly
shifted toward stronger activations. The average dif-
ference in MR signal change equaled 0.42%. The dif-
ference between the distributions was indeed signifi-
cant: P ? 0.05 on a simple Student’s t test. This
analysis confirms that hV3A is moreactivein SID than
in DIM, as was also observed, to some extent, in the
group and single-subject analysis of the contrast SID-
Our results can be summarized in three points: (1)
Attention to speed has very restricted effects in the
human visual system, enhancing activity in V3d, V3A,
V3v, and V4v, but leaving that in hMT/V5? almost
unchanged. (2) Although hMT/V5? is scarcely affected
by attention tospeed, it is active during speed discrim-
ination, an activity reflecting mostly sensory process-
ing. (3) Increasing the difficulty of the speed discrimi-
nation moderately affected V3d, but engaged a number
of non-visual, mostly frontal regions. Each of these
findings is discussed below.
Attention to Speed of Motion
The results of subtraction SID-DIM, isolating both
attention to speed and a decision making component,
nicely replicate our previous PET findings using simi-
lar tasks and the same stimuli in a different group of
subjects (Orban et al., 1998). One could argue that our
F IG. 6.
maps of subject 4 showing the voxels differing at a level of P ? 0.001
(colored red-yellow) in the contrast SID-DIM (A) and the contrast
UNI-STA (B) overlaid onto two coronal sections at the levels indi-
cated. (C, D) Statistical parametric maps of subject 3 showing the
voxels differing at a level of P ? 0.001 (colored red-yellow) in the
contrast SID-DIM (C) and the contrast UNI-STA (D) overlaid onto
two transverse sections at the levels indicated. Blue and green out-
lines indicate the statistical parametric maps (Z ? 3.09) for the
horizontal and vertical meridians, respectively, obtained in the reti-
notopic mapping control experiment. V3A refers to hV3A, MT to
hMT/V5?, and STS to the superior temporal motion-responsive re-
gion (Sunaert et al., 1999).
Single-subject analysis. A&B: Statistical parametric
F IG. 7.
tive tofixation in the speed discrimination condition (light hatching)
and detection of fixation point dimming condition (dark hatching) for
the most significant voxels of individual hMT/V5? (A) and hV3A (B)
regions. These regions were defined by the contrast UNI-STA.
Frequency distributions of the MR signal increase rela-
SUNAERT ET AL.
subtraction SID-DIM isolates not only attention to
speed but also includes a minor spatial component.
This is highly unlikely, however, since the cuneal and
lingual activations were observed in contrasts in which
attention was directed to identical stimuli, as was the
case in the study of Orban et al. (1998) and in a com-
parison of attention to speed and attention to speed
changes (Sunaert S., Van Hecke P., Marchal G., Orban
G. A., unpublished). Both the present fMRI study and
the earlier PET study are in good agreement with the
PET study of Corbetta et al. (1991), using a multicom-
ponent visual stimulus in selective and divided atten-
tion tasks. As discussed in Orban et al. (1998), the foci
in or around area 17, reported by Corbetta et al. (1991),
as activated by attention to speed, correspond rela-
tively well to the cuneal and lingual activations found
in the present study. The minor differences may reflect
a number of methodological differences between the
studies such as a difference in stimulus sizes: 32° in
Corbetta et al. (1991) compared to 3° in ours, in the
training level of the subjects, and in the use of different
statistical procedures and stereotactic atlases. It
should be noted that the stimulus diameter used in the
present study is equally efficient in driving hMT/V5?
and other motion responsive regions than larger stim-
uli (see Fig. 10 of Sunaert et al., 1999). This is not
surprising given the large proportion of MT/V5 neu-
rons that have antagonistic surrounds (Raiguel et al.,
1995; Tanaka et al., 1986).
Beauchamp et al. (1997) reported a difference in MR
signal, averaged over a region of interest centered on
hMT/V5?, when subjects made a simultaneous speed
discrimination compared to a color discrimination of
coherently moving dots restricted to a peripheral an-
nulus and embedded in dynamic noise. Using a proce-
durevery similar tothat of Beauchamp et al. (1997), we
observed slight difference between SID and DIM in the
MR signals change relative to FIX. The reduction in
hMT/V5? responses in the absence of attention to
speed was smaller (20%) than the 36% reported by
Beauchamp et al. (1997). This difference might simply
reflect the different eccentricities used in the twostud-
ies. It might also relate to a possible confound in the
experimental design of the Beauchamp et al. study.
The feature to be discriminated seems to be con-
founded with the cue segmenting the peripheral annu-
lus. The use of motion as a segmentation cue might
explain the increase in hMT/V5? activity during speed
discrimination, as suggested by imaging (O’Craven et
al., 1997) and single-cell (Treue and Maunsell, 1996)
data. Finally one should note that, although they re-
strict their formal analysis tohMT/V5?, Beauchamp et
al. (1997) suggest that additional regions are involved
in speed discrimination, in accord with our main find-
ing, that visual regions other than hMT/V5? are more
strongly influenced by attention to speed.
Demb et al. (1998) in their recent study also impli-
cated hMT/V5? in speed discrimination by showing
that under conditions of low luminance, intended to
isolate the M pathway (Merigan and Maunsell, 1990),
the MR signals over hMT/V5? correlated with perfor-
mance during speed discrimination. One interpreta-
tion of this result is that M signals, indexed by the
hMT/V5? activity, must reach the regions that we
haveshown tobeactivated by attention tospeed. Demb
et al. (1998) observed little or no correlation between
theV3 or V4 activity and speed discrimination, but this
might be due to the small proportion of M signals
reaching these areas, as discussed by these authors.
Alternatively, the hMT/V5? activity might be critical
under their experimental conditions. The two most
obvious stimulus differences between the two studies
are the use of moving gratings rather than RTPs and
the lower luminance in their study. Only a direct map-
ping of the attention to speed effects under their con-
ditions could distinguish between these two interpre-
Interestingly, attention tospeed enhanced activity of
the second motion area, hV3A (Tootell et al., 1997).
This was somewhat evident in the group and single-
subject analysis of the comparison of speed discrimina-
tion with dimming detection. The different effect of
attention to speed in hV3A compared to hMT/V5? can
be best appreciated when distributions of MR signal
increases compared toFIX areplotted for SID and DIM
(Fig. 7). That this region is engaged by attention for
speed just as it is by attention to direction (Cornette et
al., 1998a) underscores its status as motion processing
Involvement of hMT/V5? in Speed Discrimination
Vaina (1989) reported speed discrimination deficits
after bilateral lesions involving the occipito-temporo-
parietal junction and concluded that this might reflect
the involvement of hMT/V5?. Greenlee et al. (1995)
alsoreported that lesions in patients impaired in speed
discrimination of slowly moving sine wave gratings
had in common a region just dorsal to hMT/V5?. In-
terestingly, the lesions in this group of patients also
overlapped in a second region corresponding rather
well to hV3d/V3A. Finally, in nonhuman primates, le-
sions involving MT/V5 and satelliteregions alsoimpair
speed discrimination over a broad range of speeds (Or-
ban et al., 1995). At first glancethevery weak attention
to speed effect in hMT/V5? reported here appears in-
consistent with these lesion results. However, the SID-
FIX subtraction (Fig. 3B) indicates that hMT/V5? is
highly active during speed discrimination (1.06% sig-
nal increase relative to FIX). Since the increase in
activity during speed discrimination is nearly as high
during the control condition DIM, the hMT/V5? activ-
ity must reflect sensory processing, not attention to
SPEED DISCRIMINATION AND fMRI
speed. Thereis noa priori reason why a region engaged
in sensory processing could not also be critical for a
discrimination. Hence, our imaging results do not con-
tradict thelesion data, nor thepossiblemotion process-
ing deficit in hMT/V5? of schizophrenics.
Task Difficulty Effects
The increasing activity of right dorsolateral and ven-
trolateral prefrontal cortex with task difficulty agrees
with another visual discrimination study in which dif-
ficulty was manipulated. Degrading the faces by static
noise corruption (Grady et al., 1996) decreased accu-
racy in a match to sample task and caused a linear
increase in activation of the right middle frontal focus
(26, 42, 32) close to the MFG site in our study. Simi-
larly, degrading letters (Barch et al., 1997) decreased
accuracy in a working memory task and caused in-
creases in activity in frontal regions, including a right
inferior frontal site (48, 19, 23) close to the IFG region
in the present study. These similarities suggest that
the increased activity in right frontal regions reflects
general task difficulty. This conjecture is further sup-
ported by the fact that frontal activity in this case was
independent of the task (Fig. 5). Increased difficulty
might activate right prefrontal cortex for several rea-
sons: increase in supervisory or executive activity, in
sustained attention, or in guessing, which all have
been shown to enhance right prefrontal activity (Carl-
son et al., 1998; Cornetteet al., 1998b; Paus et al., 1997;
Pardo et al., 1991; Elliot et al., 1999). Deactivations
with increasing task difficulty were observed mostly in
left-sided frontal and parietal sites. These regions
showed a trend of deactivation during SID compared to
DIM, thus this activity decreasing with task difficulty
merely reflects a deeper deactivation.
It is noteworthy that increases in difficulty, which
enhanced attention, had so little effect in the visual
regions. The exception was dorsal V3, which further
points to the importance of this region in speed dis-
crimination. The small effect of difficulty in the visual
regions is in agreement with single cell studies in the
monkey. Spitzer et al. (1988) reported stronger re-
sponses of V4 neurons during fine discrimination com-
pared to coarse discrimination similar to a detecting
task (R. Desimone, personal communication). While
large changes in stimulus difference do have an effect
small changes in stimulus difference, as used in the
present study, have no effect on the tuning curve of IT
neurons (Vogels and Orban, 1998).
The cortical activity in response toattention tospeed
of motion, as well as its modulation by the difficulty
level of the speed discrimination task, were described.
Further work is needed to discern not only how atten-
tion todifferent aspects of motion influence the numer-
ous motion-responsive regions (Dupont et al., 1994) of
the human brain, but also how different properties of
the moving stimulus, such as luminance or eccentric-
ity, may influence the attention to speed effects.
The authors are indebted to M. De Paep, P. Falleyn, P. Kayen-
bergh, G. Meulemans, and Y. Celis for technical assistance. The
authors thank Dr. K. Friston and Dr. R. Frackowiak (Functional
Imaging Laboratory, Queen Square, London), for making the SPM
software available, and Dr. S. Raiguel and Dr. J . Todd for critical
reading of an earlier version of the manuscript. This work was
supported by the regional research council (FWO G0202.99) and the
Queen Elisabeth Medical Foundation. S. Sunaert holds a junior
fellowship from the FWO.
Barch, D. M., Braver, T. S., Nystrom, L. E., Forman, S. D., Noll,
D. C., and Cohen, J . D. 1997. Dissociating working memory from
task difficulty in human prefrontal cortex. Neuropsychologia 35:
Beauchamp, M. S., Cox, R. W., and DeYoe, E. A. 1997. Graded effects
of spatial and featural attention on human area MT and associated
motion processing areas. J . Neurophysiol. 78:516–520.
Buchel, C., J osephs, O., Rees, G., Turner, R., Frith, C. D., and
Friston, K. J . 1998. The functional anatomy of attention to visual
motion. A functional MRI study. Brain 121:1281–1294.
Carlson, S., Martinkauppi, S., Rama, P., Salli, E., Korvenoja, A., and
Aronen, H. J . 1998. Distribution of cortical activation during visuo-
spatial n-back tasks as revealed by functional magnetic resonance
imaging. Cereb. Cortex 8:743–752.
Chen, Y., Nakayama, K., Levy, D. L., Matthysse, S., and Holzman,
P. S. 1999. Psychophysical isolation of a motion-processing deficit
in schizophrenics and their relatives and its association with im-
paired smooth pursuit. Proc. Natl. Acad. Sci. USA 96:4724–4729.
Cheng, K., Fujita, H., Kanno, I., Miura, S., and Tanaka, K. 1995.
Human cortical regions activated by wide-field visual motion: a
H215O PET study. J . Neurophysiol. 74:413–427.
Corbetta, M., Miezin, F. M., Dobmeyer, S., Shulman, G. L., and
Petersen, S. E. 1991. Selective and divided attention during visual
discriminations of shape, color, and speed: functional anatomy by
positron emission tomography. J . Neurosci. 11:2383–2402.
Cornette, L., Dupont, P., Rosier, A., Sunaert, S., Van Hecke, P.,
Michiels, J ., Mortelmans, L., and Orban, G. A. 1998. Human brain
regions involved in direction discrimination. J . Neurophysiol. 79:
Cornette, L., Rosier, A., Dupont, P., Bormans, G., Beatse, E.,
Schoups, A., Mortelmans, L., and Orban G. A. 1998b. Dissociation
of short-term and working memory: a 3-D PET study. Soc. Neuro-
sci. Abstr. 24:1897.
Demb, J . B., Boynton, G. M., and Heeger, D. J . 1998. Functional
magnetic resonance imaging of early visual pathways in dyslexia.
J . Neurosci. 18:6939–6951.
DeYoe, E. A., Carman, G., Bandettini, P., Glickman, S., Wieser, J .,
Cox, R., Miller, D., and Neitz, J . 1996. Mapping striate and extra-
striate visual areas in human cerebral cortex. Proc. Natl. Acad.
Sci. USA 93:2382–2386.
Dupont, P., Orban, G. A., De Bruyn, B., Verbruggen, A., and Mor-
telmans, L. 1994. Many areas in the human brain respond to
visual motion. J . Neurophysiol. 72:1420–1424.
SUNAERT ET AL.
Elliot, R., Rees, G., and Dolan, R. J . 1999. Ventromedial prefrontal Download full-text
cortex mediates guessing. Neuropsychologia 37:403–411.
Engel, S. A., Glover, G. H., and Wandell, B. A. 1997. Retinotopic
organization in human visual cortex and the spatial precision of
functional MRI. Cereb. Cortex 7:181–192.
Friston, K. J ., J ezzard, P., and Turner, R. 1994a. Analysis of func-
tional MRI time-series. Hum. Brain Map. 1:153–171.
Friston, K. J ., Worsley, K. J ., Frackowiak, R. S. J ., Mazziotta, J . C.,
and Evans, A. C. 1994b. Assessing the significance of focal activa-
tions using their spatial extent. Hum. Brain Map. 1:214–220.
Grady, C. L., Horwitz, B., Pietrini, P., Mentis, M. J ., Ungerleider,
L. G., Rapoport, S. I., and Haxby, J . V. 1996. Effect of task diffi-
culty on cerebral blood flow during perceptual matching of faces.
Hum. Brain Map. 407:227–239.
Greenlee, M. W., Lang, H.-J ., Mergner, T., and Seeger, W. 1995.
Visual short term memory of stimulus velocity in patients with
unilateral posterior brain damage. J . Neurosci. 15:2287–2300.
Lagae, L., Raiguel, S., and Orban, G. A. 1993. Speed and direction
selectivity of macaque middle temporal neurons. J . Neurophysiol.
Maunsell, H. R., and Van Essen, D. C. 1983. Functional properties of
neurons in middle temporal visual area of the macaque monkey. I.
Selectivity for stimulus direction, speed, and orientation. J . Neu-
Merigan, W. H., and Maunsell, J . H. 1990. Macaque vision after
magnocellular lateral geniculate lesions. Vis. Neurosci. 5:347–352.
Mikami, A., Newsome, W. T., and Wurtz, R. H. 1986. Motion selec-
tivity in macaque visual cortex. I. Mechanisms of direction and
speed selectivity in extrastriate area MT. J . Neurophysiol. 55:
O’Craven, K. M., Rosen, B. R., Kwong, K. K., Treisman, A., and
Savoy, R. L. 1997. Voluntary attention modulates fMRI activity in
human MT-MST. Neuron 18:591–598.
Orban, G. A., Saunders, R., and Vandenbussche, E. 1995. Lesions of
the superior temporal cortical motion areas impair speed discrim-
ination in the macaque monkey. Eur. J . Neurosci. 7:2261–2276.
Orban, G. A., Dupont, P., DeBruyn, B., Vandenberghe, R., Rosier, A.,
and Mortelmans, L. 1998. Human brain activity related to speed
discrimination tasks. Exp. Brain Res. 122:9–22.
Pardo, J . V., Fox, P. T., and Raichle, M. E. 1991. Localization of a
human system for sustained attention by positron emission tomog-
raphy. Nature 349:61–64.
Paus, T., Zatorre, R. J ., Hofle, N., Caramanos, Z., Gotman, J .,
Petrides, M., and Evans, A. C. 1997. Time-related changes in
neural systems underlying attention and arousal during the per-
formance of an auditory vigilance task. J . Cogn. Neurosci. 9:392–
Peuskens, H., Rosier, A., Dupont, P., Sunaert, S., Mortelmans, L.,
Van Hecke, P., and Orban, G. A. 1999. Heading judgments involve
hMT/V5?. Soc. Neurosci. Abstr. 25:1546.
Raiguel, S., Van Hulle, M. M., Xiao, D. K., Marcar, V. L., and Orban,
G. A. 1995. Shape and spatial distribution of receptive fields and
antagonistic motion surrounds in the middle temporal area (V5) of
the macaque. Eur. J . Neurosci. 7:2064–2082.
Sereno, M. I., Dale, A. M., Reppas, J . B., Kwong, K. K., Belliveau,
J . W., Brady, T. J ., Rosen, B. R., and Tootell, R. B. H. 1995. Borders
of multiple visual areas in humans revealed by functional mag-
netic resonance imaging. Science 268:889–993.
Sunaert, S., Van Hecke, P., Marchal, G., and Orban, G. A. 1999.
Motion-responsive regions of the human brain. Exp. Brain Res.
Talairach, J ., and Tournoux, P. 1988. Co-Planar Stereotaxic Atlas of
the Human Brain. Thieme, New York.
Tanaka, K., Hikosaka, K., Saito, H., Yukie, M., Fukada, Y., and Iwai,
E. 1986. Analysis of local and wide-field movements in thesuperior
temporal visual areas of the macaque monkey. J . Neurosci. 6:134–
Tootell, R. B. H., Reppas, J . B., Kwong, K. K., Malach, R., Born, R. T.,
Brady, T. J ., Rosen, B. R., and Belliveau, J . W. 1995. Functional
analysis of human MT and related visual cortical areas using
magnetic resonance imaging. J . Neurosci. 15:3215–3230.
Tootell, R. B. H., Mendola, J . D., Hadjikhani, N. K., Ledden, P. J .,
Liu, A. K., Reppas, J . B., Sereno, M. I., and Dale, A. M. 1997.
Functional analysis of V3A and related areas in human visual
cortex. J . Neurosci. 17:7060–7078.
Treue, S., and Maunsell, J . H. 1996. Attentional modulation of visual
motion processing in cortical areas MT and MST. Nature382:539–
Vaina, L. M. 1989. Selective impairment of visual motion interpre-
tation following lesions of the right occipito-parietal area in hu-
mans. Biol. Cybern. 61:347–359.
Van Oostende, S., Van Hecke, P., Sunaert, S., Nuttin, B., and Mar-
chal, G. 1997. fMRI studies of supplementary motor area and
premotor cortex. NeuroImage 7:145–151.
Watson, J . D. G., Myers, R., Frackowiak, R. S. J ., Hajnal, J . V.,
Woods, R. P., Mazziotta, J . C., Shipp, S., and Zeki, S. 1993. Area V5
of the human brain: evidence from a combined study using
positron emission tomography and magnetic resonance imaging.
Cereb. Cortex 3:79–94.
Zeki, S., Watson, J . D. G., Lueck, C. J ., Friston, K. J ., Kennard, C.,
and Frackowiak, R. S. J . 1991. A direct demonstration of func-
tional specialization in human visual cortex. J . Neurosci. 11:641–
SPEED DISCRIMINATION AND fMRI