Matching Two Imagined Clocks: the
Functional Anatomy of Spatial Analysis in
the Absence of Visual Stimulation
Luigi Trojano, Dario Grossi1, David E.J. Linden2,3, Elia
Formisano4,5, Hans Hacker5, Friedhelm E. Zanella5, Rainer
Goebel2 and Francesco Di Salle5,6
Do spatial operations on mental images and those on visually
presented material share the same neural substrate? We used the
high spatial resolution of functional magnetic resonance imaging to
determine whether areas in the parietal lobe that have been
implicated in the spatial transformation of visual percepts are also
activated during the generation and spatial analysis of imagined
objects. Using a behaviourally controlled mental imagery paradigm,
which did not involve any visual stimulation, we found robust
activation in posterior parietal cortex in both hemispheres. We could
thus identify the subset of spatial analysis-related activity that is
involved in spatial operations on mental images in the absence of
external visual input. This result clarifies the nature of top-down
processes in the dorsal stream of the human cerebral cortex and
provides evidence for a specific convergence of the pathways of
imagery and visual perception within the parietal lobes.
requiring only a verbal–semantic judgement, which did not rely
on spatial or imagery processes.
In the second experiment, we compared the activity during
the mental clock test (imagery) with that evoked by the same
operation performed on visually presented material (perception)
and by a different non-spatial control task (syllable counting),
the attentional load of which was comparable to that of the
imagery task.
From the lesion-based evidence on the cortical substrate
of spatial mental imagery we expected to see an increase of
neuronal activity, as ref lected by the blood oxygen level-dependent (BOLD) signal, in the parietal lobes bilaterally during the
mental clock test. In addition, our analysis included the entire
cortex in order to assess the coactivation of other brain areas.
Introduction
Materials and Methods
Focal lesions of cerebral cortex may lead to selective defects of
visual mental imagery (Farah, 1984; Trojano and Grossi, 1994).
This suggests that the ability to revisualize objects or events in
the mind relies on the integrity of specific neural structures.
One major issue in the search for the neural basis of mental
imagery is whether perception and imagery share a common
neural substrate. On the basis of studies suggesting that there are
two visual systems, one for identifying objects and one for
locating them (Ungerleider and Mishkin, 1982), it has been
proposed that visual object and spatial imagery are functionally
independent processes, which must rely on different underlying neural systems (Levine et al., 1985). Patients with selective
deficits in the performance of visual or spatial imagery tasks have
been reported in the literature (Farah et al., 1988; Luzzatti et al.,
1998). Several brain mapping studies (Mellet et al., 1998) tried to
identify the neural basis of spatial mental imagery in the absence
of visual stimulation, but none of them verified actual execution
of the imagery task on-line.
In the present study we used functional magnetic resonance
imaging (fMRI) to explore the neural correlates of a behaviourally controlled spatial imagery task. The experimental task
was derived from the mental clock test (Paivio,1978; Grossi et
al., 1989, 1993), a paradigm particularly suitable for the fMRI
investigation of mental imagery because it involves a behavioural
control that can be performed on-line during scanning. Subjects
are asked to imagine pairs of times that are presented acoustically and to judge at which of the two times the clock hands form
the greater angle.
The spatial imagery task was used in two experiments. In
the first experiment it was alternated with a control condition
© Oxford University Press 2000
Salvatore Maugeri Foundation, IRCCS, Center of Telese, Loc. S.
Stefano in Lanterria, I-82037 Telese Terme (BN), Departments
of 1Neurological Sciences, 4Electronic Engineering and
6
Radiological Sciences, Federico II University, Nuovo
Policlinico, Via S. Pansini 5, I-80131 Naples, Italy,
2
Max-Planck-Institut für Hirnforschung, Deutschordenstrasse
46, D-60528 Frankfurt am Main and Departments of
3
Neurology and 5Neuroradiology, Klinikum der Johann
Wolfgang Goethe-Universität, Schleusenweg 2–16, D-60528
Frankfurt am Main, Germany
Subjects
We recruited seven right-handed post-graduate students (four male, three
female; mean age 27 years; range 23–32), who gave their informed
consent to participate in the study. None of the subjects was taking any
medication or was affected by neurologic or psychiatric conditions. All of
them were unaware of the purposes and predictions of the experiment at
the time of testing. All seven subjects participated in experiment 1; four
of them also participated in experiment 2.
Magnetic Resonance (MR) Hardware and Sequences
The MR scanner used for imaging was a 1.5 T whole-body superconducting system (MAGNETOM Vision, Siemens Medical Systems, Erlangen,
Germany) equipped with a standard head coil, an active shielded gradient
coil (25 mT/m) and Echo Planar (EPI) sequences for ultra-fast MR imaging.
For functional imaging, we used a BOLD sensitive single shot EPI
sequence [echo time (TE) = 66 ms; f lip angle (FA) = 90°; matrix size =
128 × 128, voxel dimensions = 1.4 × 1.4 × 4 mm] with an interscan
temporal spacing of 5 s. Each functional time series consisted of 5
‘resting’ volumes (25 s, discarded from further analysis) followed either
by 100 (experiment 1) or 72 (experiment 2) acquisitions during which
the imagery condition was periodically alternated with a control condition every ten (experiment 1) or eight scans (experiment 2).
After the functional acquisitions, a high-resolution three-dimensional
data set covering the whole brain (referred to as 3-D anatomical image)
was collected for each subject with a 3-D magnetization-prepared rapid
acquisition gradient echo (MP-R AGE) sequence (TR = 9.7 ms, TE = 4 ms,
FA = 12°, matrix = 256 × 256, thickness = 1 mm, number of partitions =
170–180, Voxel dimensions = 1mm × 1mm × 1mm). For the four subjects
participating in experiment 2, we furthermore performed a 3-D T1
weighted fast low-angle shot measurement at the same resolution for
subsequent surface reconstruction and surface-based statistical analysis
(see Data Analysis).
Cerebral Cortex May 2000;10:473–481; 1047–3211/00/$4.00
Experiment 1
Experimental Paradigm
Subjects were asked to imagine two analogue clock faces based on the
times that were presented verbally by the examiner [e.g. 9.30 and 10.00;
inter-stimulus interval (ISI) = 1 s] and to judge at which of the two times
the clock hands form the greater angle (imagery). We selected 50 pairs of
times, involving only half-hours (i.e. 7.30) or hours (i.e. 9.00); in half of
the conditions, the correct answers corresponded to numerically greater
times (i.e. 3.00 versus 1.00), and in the other half to numerically smaller
times (i.e. 8.30 versus 11.00) in order to avoid subjects considering only
the numerically greater pairs. The clock faces were balanced for the side
on which the hands had to be imagined and presented in pseudo-random
order. Subjects had to push a button with their right index finger if the
hands of the first clock formed the greater angle, or their left index finger
for the second. Subjects’ responses were registered by an optic fibre
answer box and analysed for accuracy.
As a control task, we asked subjects to judge which of two times was
numerically greater. Materials, presentation and response modality were
the same as in the imagery task, but the control condition required
only a verbal–semantic judgement, which did not rely on the activation of
imagery processes.
The two tasks were alternated in blocks of ten trials (total number of
trials = 100); every trial block was preceded by the appropriate instruction. Trials were presented every 5 s (1 trial/volume).
In a pre-experimental session all subjects completed an angle comparison task on visually presented pairs of analogue clocks and several
practice trials of both the imagery and the control task, to familiarize
them with materials, tasks and response modality.
Following the common procedure for imagery paradigms, subjects
were asked to keep their eyes closed during the scanning session. To
exclude the possible confounding effect of eye movements, subjects were
also instructed to keep their eye position steady. Control scans with open
eyes and infrared eye-tracking were performed on two subjects (see
Results).
Data Analysis
Data analysis, registration and visualization were performed with the
fMRI software package BrainVoyager 2.0 (Goebel et al., 1998a,b).
Prior to statistical analysis, the time series of functional images were
aligned for each slice in order to minimize the signal changes related to
small motions of the subject during the acquisition. The realigned time
series were at first spatially filtered by convolving each EPI – image with
a bidimensional Gaussian smoothing kernel with full width at half maximum (FWHM) = 2 pixels, and then temporally filtered by convolving
each pixel’s time course with a smoothing Gaussian kernel with FWHM =
2 samples. Furthermore, the linear drifts of the signal with respect to time
were removed from each pixel’s time course.
After these pre-processing steps, the cerebral regions responding to
the stimulus were identified by means of a cross-correlation analysis
(Bandettini et al., 1993). The reference vector used in our analysis was a
‘box-car’ ideal vector which assumed a value of 0 during the control
period and 1 during the imagery period. In order to take into account the
variations in the haemodynamic delay of the activation signal throughout
the cortex, the cross-correlation coefficient was evaluated for each pixel
using the vector obtained by shifting the original reference vector by one
sample. The maximum of the two cross-correlation coefficients was then
considered in the activation map. Activated brain areas were selected in
the cross-correlation maps by imposing a conservative intensity threshold
for the cross-correlation coefficients (r > 0.5, corresponding to P < 10–7;
uncorrected).
Two-dimensional statistical maps were converted into polychromatic
images and incorporated into the 3-D MP-R AGE data sets through
interpolation to the same resolution (voxel size: 1.0 × 1.0 × 1.0 mm3).
This made it possible to superimpose 3-D statistical maps onto the
3-D anatomical data sets. Since the 2-D functional and 3-D structural
measurements were performed within the same recording session,
co-registration of the respective data sets could be computed directly
based on the Siemens slice position parameters of the T2*-weighted
measurement (number of slices, slice thickness, distance factor, Tra-Cor
angle, FOV, shift mean, off-centre read, off-centre phase, in plane resolu-
474 The Functinal Anatomy of Spatial Imagery • Trojano et al.
tion) and on parameters of the T1-weighted 3-D MP-R AGE measurement
(number of sagittal partitions, shift mean, off-centre read, off-centre
phase, resolution) with respect to the initial over view measurement
(scout).
For each subject the structural 3-D data sets were transformed into
Talairach space. The Talairach transformation was performed in two
steps. The first step consisted in rotating the 3-D data set of each subject
to be aligned with the stereotaxic axes. For this step the location of the
anterior commissure (AC), the posterior commissure (PC) and two
rotation parameters for midsagittal alignment had to be specified
manually in the 3-D MP-R AGE data set. In the second step the extreme
points of the cerebrum were specified. These points together with the AC
and PC coordinates were then used to scale the 3-D data sets into
the dimensions of the standard brain of the Talairach and Tournaux atlas
(Talairach and Tournaux, 1988) using a piecewise affine and continuous
transformation for each of the 12 defined subvolumes.
The individual Talairach 3-D maps were averaged across subjects and
superimposed on a normalized anatomical 3-D data set. Prior to averaging, the functional 3-D maps were smoothed with a Gaussian kernel of
5 mm FWHM. Activated brain areas were selected in the average map by
imposing lower overall correlation values (r > 0.25) to take into account
the interindividual spatial variability of activated clusters.
The high-resolution T1-weighted 3-D recording of one subject was
used for a surface reconstruction of both hemispheres. The white/grey
matter border was segmented using a region-growing algorithm. The
discrimination between white and grey matter was improved by several
manual interactions (e.g. labelling subcortical structures as ‘white
matter’). The white/grey matter border was finally tesselated in a single
step using two triangles for each side of a voxel located at the margin of
white matter. The tesselation of a single hemisphere typically consists
of roughly 240 000 triangles. The reconstructed surface is subjected
to iterative corrective smoothing (100–200 iterations). An interactive
morphing algorithm (Goebel et al., 1998a) was used to let the surface
grow smoothly into the grey matter. Through visual inspection, this
process was halted when the surface reached the middle of the grey
matter (approximately layer 4 of the cortex). The resulting surface was
used as the reference mesh for the visualization of functional data. The
iterative morphing algorithm was further used to inf late each hemisphere. An inf lated hemisphere possesses a link to the folded reference
mesh so that functional data may be shown at the correct position of
the inf lated representation. This link was also used to keep geometric
distortions during inf lation to a minimum with a morphing force that
keeps the area of each triangle of the inf lated hemisphere as close
as possible to the value of the folded reference mesh. This display of
functional maps on an inf lated hemisphere allows the topographic
representation of the 3-D pattern of cortical activation without loss of the
lobular structure of the telencephalon (Goebel et al., 1998a; Linden et al.,
1999).
Experiment 2
Experimental Paradigm
In this experiment, we compared the activity during the imagery
condition with that evoked by the same cognitive operation on visually
presented material (perception) and by a different non-spatial control task
(syllable counting).
The imagery condition was the same as in experiment 1, and
employed 48 pairs of times different from those of the previous experiment but counterbalanced in the same way.
In the perception condition we used 48 pairs of analogic clock faces;
the clock faces of each pair were generated on a computer screen and
projected one at a time (ISI = 1 s) in the central visual field. The subjects
had to decide at which of the two times the clock hands formed the
greater angle.
During the syllable counting condition, we asked subjects to count
the syllables of each of 48 auditorily presented pairs of times and to report
whether the total syllable number was odd or even.
During the imagery and perception condition, materials and response
modality were the same. In the latter case, however, the pairs of times
were presented visually to the subject while in the imagery condition
they had to be visualized mentally.
In the syllable counting condition, material, presentation and
response modality were the same as in the imagery task, but here a
verbal–phonological judgement was required that was matched for
response time with the imagery conditions.
Three sessions of 72 measurements were conducted for each subject.
Within each session, the three conditions were alternated in blocks of
eight trials (1 trial/measurement, 8 measurements/block) and separated
by blocks of four ‘resting’ measurements. The sequence of stimulation
conditions consisted of: imagery, perception, syllable counting. This
sequence was repeated twice per session. In all the conditions, the
inter-trial interval was 5 s (1 trial/scan), and every trial block was
preceded by the appropriate instruction.
In this experiment subjects had to press a button with their right
index or middle finger; responses were registered by an optic fibre
answer box and analysed for accuracy and response times. Mean correct
reaction times for each subject were used for the speed of performance
analysis. Because of the low number of subjects, statistical analysis was
performed by nonparametric methods (Mann–Whitney test for twogroup comparisons). Subjects were asked to keep their eyes open during
the scanning session and foveate a fixation cross in order to avoid eye
movements.
nonparametrically adjusting for multiple comparisons while preserving
from an excessive loss of statistical power of the conventional Bonferroni
approach (Goebel and Singer, 1999).
Statistical results were then visualized through projecting 3-D
statistical maps on folded and inf lated surface reconstructions of the
cortical sheet. For significantly activated voxels, the relative contribution
RC between two selected sets of conditions in explaining the variance of
a voxel time course were computed as RC = (bs1 – bs2)/(bs1 + bs2) where bsi
is the sum of the estimates of the standardized regression coefficients
of all conditions included in set si. The RC index was visualized with a
red–green pseudo-colour scale. An RC value of 1 (red) indicates that a
voxel time course is solely explained with predictor set s1 whereas an RC
value of –1 (green) indicates that a voxel time course is explained solely
with predictor set s2. An RC value of 0 indicates that a voxel time course
is explained with an equal contribution of both predictor sets. In the
contrast maps, only |RC| values >0.7 were visualized.
Data Analysis
As this experiment comprised three conditions, the statistical analysis
was based on the application of the multiple regression analysis to time
series of task-related functional activation (Friston et al., 1995). Furthermore, a method based on the reconstruction of a cortical sheet was used
for adjusting the significance values for multiple comparisons (Goebel
and Singer, 1999). These extended analytical tools were implemented in
BrainVoyager 3.0 (Goebel et al., 1998a,b; Dierks et al., 1999).
Talairach transformation (see experiment 1) was performed for the
complete set of functional data of each subject, yielding a 4-D data
representation (volume time course: 3 × space, 1 × time). Prior to
statistical analysis, the time series of functional images were aligned in
order to minimize the effects of head movements. The central volume of
the time series was used as a reference volume to which all other volumes
were registered, using a 3-D motion correction that estimates the three
translation and three rotation parameters of rigid body transformation.
Spatial and temporal filtering of the image time series was the same as in
experiment 1.
For each subject, the high-resolution T1-weighted 3-D recording was
used for the surface reconstruction of the grey/white matter boundary of
the brain. Since the surface reconstruction and the 4-D functional data set
were co-registered, it was possible to tag those voxels of the functional
data which were within 0 and 2 mm with respect to the reconstructed
cortical sheet and to limit the statistical analysis to these voxels. The
general linear model (GLM) of the experiment was then computed from
the 12 (four subjects, three sessions per subject) tagged and z-normalized
volume time courses. The signal values during the imagery, perception
and syllable counting conditions were considered the effects of interest.
The corresponding predictors, obtained by convolution of an ideal
box-car response (assuming the value 1 for the time points of task
presentation and the value 0 for the remaining time points) with a linear
model of the haemodynamic response (Boynton et al., 1996), were used
to build the design matrix of the experiment. The global level of the signal
time courses in each session was considered to be a confounding effect.
To analyse the effects of conditions compared with baseline and contrasts
between conditions, 3-D individual and group statistical maps were
generated by associating each tagged voxel with the F value corresponding to the specified set of predictors and calculated on the basis of
the least mean squares solution of the GLM. Effects were only accepted as
significant when, considering an F distribution with n1 and n2 degrees of
freedom (n1 = number of orthogonal predictors and n2 = number of time
samples – n1 –1), the associated P value yielded P ′ = P/N 10–3, where N
represented the number of independent statistical tests. In single-subject
analysis, N was chosen as the total number of voxels indexed from the
surface mesh (28 118, 27 267, 28 528 and 26 771 respectively for the four
subjects) and the uncorrected P threshold was adequately adjusted in
order to keep the corrected P ′ threshold at 10–3. In multi-subject analysis,
N was chosen as the number of voxels resulting from the logic OR of the
individual surface meshes. This procedure provided an effective means of
Behavioural Results
All subjects performed the imagery task without difficulty, and
none of them made more than three errors (mean correct
responses: 47.6 ± 1.3). The semantic task proved to be slightly
easier than the imagery task (mean correct responses: 48.9 ±
1.1). The difference in accuracy between the two tasks was not
significant (t = 2).
Results
Experiment 1
FMRI Results
The contrast between imagery and control conditions yielded
an increase of the BOLD signal in several parietal and frontal
regions.
The individual correlation maps (r > 0.5) showed a bilateral
activation in the posterior parietal cortex (PPC) [primarily
Brodmann area (BA) 7], in the fronto-basal region centred on the
ascending ramus of the lateral sulcus and on parts of the inferior
frontal gyrus (BA 45), and in the anterior insula in all subjects.
The middle frontal gyrus including the dorsolateral prefrontal
cortex (DLPFC, BAs 9, 46) was bilaterally activated in five of
seven subjects, the superior frontal gyrus including the supplementary motor area in four volunteers, and the frontal eye fields
(FEF) in 3 subjects. The inferior occipito-temporal region (BA
37) was also activated (4/7 left, 3/7 right).
The analysis of the averaged correlation map (n = 7, r > 0.25)
confirmed the main results of the individual analysis. The largest
clusters of activation were located bilaterally in the PPC, mainly
along the intraparietal sulcus IPS (Fig. 1). Smaller foci of activity
were also visible in the inferior frontal and perisylvian region of
both sides (Table 1). These areas of group activation corresponded to those that were found consistently in the individual
correlation maps: each centre of mass of a cluster in the average
map lay within two standard deviations of the distribution of the
centres of mass of the respective clusters in the individual maps
(Table 1). The DLPFC, FEF and occipito-temporal cortex, which
were activated only in some volunteers, did not appear in the
averaged map at the selected rigorous threshold.
Eye Movements
The experiment was repeated with two subjects, who were
requested to keep their eyes open while their eye movements
were monitored by the fMRI-compatible Ober2 (Permobil Meditech, Timra, Sweden) infrared eye tracking system (Aisenberg,
1996), sampling the horizontal and vertical positions of both
eyes at 120 Hz. This system has already been used for fixation
Cerebral Cortex May 2000, V 10 N 5 475
Figure 1. Experiment 1. 3-D correlation map of BOLD activation during imagery versus control averaged over seven subjects and superimposed on (A) a folded (axial cut at z = 44)
and (B) an inflated surface-rendered individual normalized 3-D anatomy. The white lines indicate the descending portions of the intraparietal sulci.
control in a recent study of attentional effects on fMRI activity in
human MT/MST (O’Craven et al., 1997). We could thus confirm
that subjects maintained fixation during the entire experiment.
The magnetic artefacts produced by the eye monitoring
device affected the quality of the EPI signal, but only in areas
anterior to the clusters of activation described above. The BOLD
activation pattern during eye monitoring did not differ significantly from that obtained without the eye tracking system.
the imagery (2517 ± 860 ms; 45.5 ± 6.4 correct responses) or
the perception conditions (1433 ± 618 ms; 47.1 ± 0.8 correct
responses). However, only the comparison between the perception and syllable counting conditions yielded significant
differences both in accuracy (Mann–Whitney test: z = 2.39; P =
0.02) and in reaction time (Mann–Whitney test: z = 2.30; P =
0.02), whereas the behavioural differences between the syllable
counting and imagery conditions were not significant.
Experiment 2
Behavioural Results
The syllable counting task proved to be the most difficult
condition, with higher reaction times (2940 ± 1039 ms) and
error rates (39.5 ± 4.9 correct responses/48 trials) than either
476 The Functinal Anatomy of Spatial Imagery • Trojano et al.
FMRI Results
All three tasks, compared with baseline, were accompanied by
an increased activation of left sensorimotor cortex (contralateral
to the hand of the button press response). The two tasks that
involved auditory stimulation (imagery, syllable counting) were
Table 1
Experiment 1: locations, extension and significance of activated brain areas during the mental clock task
Anatomical area
Brodmann area
Left posterior parietal cortex
7
Right posterior parietal cortex
7
Left inferior parietal lobule
Left perisylvian cortex
Right perisylvian cortex
Tailarach coordinates
39–40
45/insula
45/insula
X
Y
Z
–24 (–20 ± 4.3)
–22 (–23 ± 2.0)
22 (21 ± 2.5)
24 (25 ± 1.8)
–31 (–33 ± 3.1)
–31 (–33 ± 5.5)
44 (43 ± 4.5)
–70 (–74 ± 3.3)
–72 (–76 ± 4.7)
–76 (–75 ± 2.7)
–80 (–78 ± 2.7)
–48 (–46 ± 4.4)
23 (22 ± 3.0)
16 (17 ± 2.5)
44 (42 ± 4.2)
31 (33 ± 2.5)
43 (42 ± 0.8)
31 (33 ± 1.7)
38 (34 ± 2.7)
12 (8 ± 4.0)
4 (5 ± 1.7)
Vox.
r
243
53
261
52
77
39
96
0.73
0.63
0.72
0.63
0.64
0.58
0.57
The position of each area is given as the Tailarach coordinates of the centre of mass of the supra-threshold (r > 0.25) clusters in the 3-D-average map. The mean values (± SD of seven subjects) of the
Tailarach coordinates of the centre of mass of each cluster as selected in the individual correlation maps (r > 0.5; P < 10–7; uncorrected) are reported in brackets. Vox = number of voxels (1.4 × 1.4 ×
4 mm) of each area; r = cross-correlation coefficient of the cluster time-courses averaged across subjects.
Table 2
Experiment 2: locations of brain areas activated during the imagery, perception and syllable counting conditions
Anatomical area
Left posterior parietal cortex
Right posterior parietal cortex
Left inferior parietal lobule
Right inferior parietal lobule
Left perisylvian cortex
Right perisylvian cortex
Left superior frontal gyrus
Right superior frontal gyrus
Left precentral gyrus
Left gyrus of Heschl
Right gyrus of Heschl
Right prefrontal cortex
Left occipitotemporal cortex
Right occipitotemporal cortex
Left middle occipital gyrus
Right middle occipital gyrus
Brodmann area
7
7
39/40
39/40
45/ins
45/ins
6/8
6/8
4
41/42
41/42
46
37/19
37/19
19
19
Imagery versus baseline
Perception versus baseline
Syllable counting versus baseline
X
Y
Z
X
Y
Z
X
Y
Z
–17
24
–30
–
–31
38
–
4
–33
–57
63
38
–
–
–
–
–70
–69
–50
–
16
9
–
28
–26
–24
–16
36
–
–
–
–
46
46
38
–
15
3
–
39
55
9
6
22
–
–
–
–
–26
27
–26
32
–
–
–6
4
–31
–
–
–
–41
38
–35
32
–65
–65
–50
–45
–
–
4
–3
–26
–
–
–
–67
–60
–79
–80
46
55
41
39
–
–
46
46
54
–
–
–
–5
–13
3
2
–
–
–46
44
–35
46
–5
4
–41
–51
54
45
–
–
–
–
–
–
–48
–45
16
9
–16
–5
–31
–20
–17
25
–
–
–
–
–
–
44
42
3
5
61
54
55
9
–3
35
–
–
–
–
The position of each area is given as the Talairach coordinates of the centre of mass of the supra-threshold clusters (P ′ < 10–3, corrected) of the group analysis.
accompanied by activation of auditory cortex in the temporal
lobes bilaterally (Table 2).
The comparison imagery versus baseline yielded a prominent
increase in activation of the PPC in all four subjects. Increased
activity was also present in the left inferior parietal lobule, in the
right pre-frontal cortex, and in the inferior frontal and
perisylvian region, bilaterally. Moreover, a cluster of activation
was seen in the right superior mesial frontal region (Table 2).
The comparison between the perception and baseline
conditions showed an activation of the posterior and inferior
parietal cortex very similar to that observed during the imagery
condition. Increased activity was also found in the basal and
lateral surfaces of the occipital lobe. More consistently observed
in the left hemisphere, this activity extended anteriorly in the
basal surface of the temporal lobe. An activation was also seen in
the superior mesial frontal region bilaterally (Table 2).
In the syllable versus baseline contrast increased activation
was recognizable bilaterally in the inferior parietal lobule, in the
inferior frontal and perisylvian region, in the superior mesial
frontal region and in the right prefrontal cortex (Table 2). The
contrast between the imagery and syllable counting conditions
yielded, in all four subjects, a bilateral activation of the PPC,
mainly along the IPS (Table 3), closely corresponding to the
largest cluster of experiment 1 (Figs 1, 2 and 3a). During the
syllable counting task, activity increased bilaterally in the
inferior parietal lobules and right prefrontal cortex (Fig. 3a).
In the contrast between the imagery and perception conditions no activation was evident in the PPC. In the imagery
condition increased activity was only present in right prefrontal
cortex and in mesial frontal areas bilaterally. In the perception
condition, occipital and inferior temporal areas were activated
bilaterally (Fig. 3b, Table 3).
Discussion
The mental clock test requires the generation of multipart mental
images (Trojano and Grossi, 1994). These mental images served
as the substrate of a spatial comparison task, which all of our
subjects performed at a high level of accuracy.
The most striking results of our two experiments demonstrate
that cortical activation (as measured by an increase of the fMRI
BOLD signal) during the mental clock test was most prominent
in the posterior parietal lobes of both hemispheres.
The absence of imagery-related activation in early visual areas
in both experiments provides new data bearing on the debate
regarding the cortical representations of mental imagery,
especially because the present study employed an on-line
behavioural control to ensure the actual generation of mental
images. Our results therefore confirm the recent findings that
Cerebral Cortex May 2000, V 10 N 5 477
Figure 2. Experiment 2. The contrast maps for imagery versus syllable counting (P ′ < 10–3, corrected) superimposed on folded surface reconstructions showing the activation during
the imagery condition of the posterior parietal cortex in all four subjects (axial cuts at z1 = 51, z2 = 48, z3 = 48, z4 = 44). Left hemispheres are shown on the right side.
certain imager y conditions, particularly those which rely on
abstract patterns and schematic figures, produce increases in
activity in primary visual areas to only a small extent (Goebel et
al., 1998a) or not at all (Mellet et al., 1996).
The activation of the prefrontal cortex, which was observed
in a number of our subjects in the first experiment, has been
associated with the attentional demand involved in different
kinds of mental imagery (Mellet et al., 1996; Goebel et al.,
1998a) and with tasks of working memory (Ungerleider et al.,
1998). Accordingly, increased prefrontal activity was evident in
the second experiment only when a task was compared with the
baseline or when a more difficult task was compared with an
easier one. In the second experiment, activations of roughly
478 The Functinal Anatomy of Spatial Imagery • Trojano et al.
equal strength during the imagery and syllable counting
conditions were obser ved in superior mesial frontal cortex.
Activation of this area, unlike that of the PPC, was thus related to
the attentional demand of the task and not to its specific nature.
Likewise, the activation of perisylvian areas in experiment 1
should be considered not to be specific for mental imagery
because a similar activation pattern was found for syllable
counting versus imagery in experiment 2. The activation of the
inferior parietal lobule during the syllable counting task can be
explained by the requirement of phonetic segmentation and
short-term memory (Wildgruber et al., 1999).
Increased activity in the posterior parietal lobes, particularly
along the IPS, was observed in experiment 1 and was par-
Figure 3. Experiment 2. Multi-subject contrast maps (P ′ < 10–3, corrected) superimposed on inflated representations of the cortical sheet of an individual normalized 3-D anatomy.
(A) Imagery (green) versus syllable counting (red). (B) Imagery (green) versus perception (red). The white lines indicate the descending portions of the intraparietal sulci.
ticularly evident when we compared the imagery condition
of experiment 2 (execution of the mental clock test) with the
baseline. This activation maintained the contrast with a nonspatial control task that was performed on the same auditorily
presented material and matched for difficulty (imagery versus
syllable counting, Fig. 3a) and was consistently observed
between individual subjects (Fig. 2). This differential activation
of the posterior parietal lobes by the spatial task confirms the
specific role of the parietal cortex in spatial processing, rather
than an unspecific activation related to attentional demand.
Activation of posterior parietal areas has been observed in
conjunction with the spatial transformation of visually presented
stimuli (Cohen et al., 1996; Alivisatos and Petrides, 1997) and
the orientation discrimination of tactile stimuli (Sathian et al.,
1997). In a recent PET study of mental rotation (Alivisatos and
Petrides, 1997), the right PPC was seen to participate in the processing of mirror images of letters or digits. Activation of the
superior parietal lobule and of the IPS in both hemispheres has
been described in a recent fMRI study of the analysis of spatially
transformed words and phrases, which also distinguished this
spatial transformation-related activation from general attentional
effects (Goebel et al., 1998b). The construction of 3-D mental
images from auditory instructions, as studied by PET (Mellet et
al., 1996), involved a distributed system of frontal, occipital and
parietal areas. The parietal activation, however, was most prominent in the right precuneus and supramarginal gyrus and did not
involve the IPS region. This absence of IPS activation might
ref lect the specific nature of the task of Mellet et al. (Mellet et
al., 1996), which did not require the spatial comparison between objects and was not performed through mental rotation.
Cerebral Cortex May 2000, V 10 N 5 479
Table 3
Experiment 2: locations, extension and significance of contrasts between the imagery and control conditions
Anatomical area
BA
Imagery (I) vs syllables (S)
Imagery (I) versus perception (P)
Task
X
Y
Z
Voxel
R
Task
Left posterior parietal cortex 7
I
–
726
0.65
–
Left perisylvian cortex
45/ins
S
507
0.64
–
Right perisylvian cortex
45/ins
S
984
0.52
–
Left inferior parietal lobule
39/40
S
612
0.70
–
Right inferior parietal lobule
39/40
S
48
(50.5 ± 3.7)
43
(45.3 ± 2.2)
0
(–0.5 ± 3.3)
1
(1.5 ± 0.6)
44
(43.0 ± 2.0)
43
(44.8 ± 1.7)
0.76
I
–67
(–69.8 ± 5.6)
–64
(–68.5 ± 4.0)
15
(11.3 ± 3.0)
13
(10.5 ± 3.1)
–46
(–45.5 ± 4.2)
–45
(–45.0 ± 3.6)
1010
Right posterior parietal cortex 7
–19
(–16.3 ± 3.9)
17
(15.8 ± 4.6)
–49
(–46.5 ± 5.2)
44
(42.5 ± 3.4)
–45
(–43.5 ± 3.7)
52
(49.5 ± 4.4)
532
0.72
–
Left superior frontal gyrus
6/8
–
I
Right superior frontal gyrus
6/8
–
I
Right prefrontal cortex
9
S
Left occipitotemporal cortex
37/19
–
P
Right occipitotemporal cortex 37/19
–
P
Left middle occipital gyrus
19
–
P
Right middle occipital gyrus
19
–
P
43
(43.3 ± 2.9)
21
(13.5 ± 4.4)
39
(35.5 ± 4.7)
662
I
x
y
z
Voxel
R
–3
(–2.8 ± 1.7)
4
(5.3 ± 2.3)
53
(50.8 ± 4.8)
–45
(–44.3 ± 2.2)
44
(41.8 ± 4.5)
–35
(–37.5 ± 4.8)
32
(34 ± 3.4)
28
(31.3 ± 4.1)
28
(23.7 ± 6.7)
13
(13.8 ± 1.5)
–57
(–58.8 ± 4.3)
–65
(–61.5 ± 4.4)
–79
(–77.5 ± 5.8)
–80
(–81.5 ± 3.7)
37
(34.8 ± 3.8)
40
(38.7 ± 5.1)
26
(26.5 ± 6.6)
–9
(–7.3 ± 2.6)
–12
(–10.5 ± 1.7)
3
(0.3 ± 2.5)
2
(3.3 ± 2.2)
171
0.54
625
0.50
435
0.64
556
0.68
1181
0.70
515
0.56
401
0.54
The position of each area is given as the Tailarach coordinates of the centre of mass of the supra-threshold (P ′ < 10–3, corrected) clusters in the multi-subject maps. The mean values (± SD of four
subjects) of the Tailarach coordinates of the centre of mass of each cluster (P ′ < 10–3, corrected) as selected by the single-subject analysis are reported in brackets. Vox = number of voxels (1× 1 ×
1 mm) of each area; R = multiple regression coefficient of the cluster time-courses averaged across subjects.
Table 4
Meta-analysis of activated brain areas in the posterior parietal cortex during spatial tasks
Authors
Brodman
area
Tailarach coordinates
X
Y
Z
Goebel et al. (1998b) (mirror reading)
7L
7R
Alivisatos and Petrides (1997) (mental rotation)
Alivisatos and Petrides (1997) (mirror image)
Sathian et al. (1997) (orientation discrimination)
7R
7R
7/19 L
–23
–25
29
12
31
–20
–68
–71
–68
–69
–62
–71
47
25
28
51
45
32
The majority of earlier studies of mental rotation and related
tasks suffered from the possible confound of saccade-related
activation in the parietal lobe (Milner and Goodale, 1995). The
spatial transformation-related activation of the IPS region that
was observed by Goebel et al. (Goebel et al., 1998b) could be
separated from the activity related to overt (saccades) and covert
attention shift by additional control tasks and correlation analysis
of the BOLD signal time course. In the present study, we excluded the contribution of saccadic activity to the task-related
BOLD signal changes by controlling eye movements during the
scanning session.
Goebel et al. (Goebel et al., 1998b) identified a ‘spatial transformation area’ in the parietal lobe. Their spatial transformation
task, like those of Alivisatos and Petrides (Alivisatos and Petrides,
1997) and Cohen et al. (Cohen et al., 1996), was performed on
visually presented material. The present study shows that the
area in the superior IPS is active when the spatial task is
performed on mental images, in the absence of any visual stimulation. This similarity of activation patterns can be explained in
480 The Functinal Anatomy of Spatial Imagery • Trojano et al.
two ways. First, the superior IPS might be instrumental in the
computation of spatial transformations, regardless of whether
the material is present in the visual field or merely as a mental
image. Alternatively, any spatial transformation task, whether
it involves visually perceived or imagined material, or indeed
tactile stimulation (Sathian et al., 1997), might require the
implicit generation of mental visual representations (Kosslyn and
Sussmann, 1995).
In the second experiment the perception condition (spatial
matching of visually presented clocks), compared with the
fixation baseline, yielded only weak activation of primary visual
cortex (probably because of the low contrast of the visual
stimulus, and the TR of 5 s, which was relatively long compared
with the short presentation of the stimulus), but very prominent
bilateral activation of the inferior temporal and the inferior and
lateral occipital lobes. These areas, which include the fusiform
gyri, have been shown to be involved in the cortical processing
of visual objects (Ungerleider and Mishkin, 1982; Malach et al.,
1995). In the posterior parietal lobes, a strong activation of the
IPS was observed during this task. The direct contrast between
the perception and imagery conditions (i.e. the clock test performed on visually presented and mentally imagined material,
respectively) led to the disappearance of this activity. The comparison of the present results with the data on transformations
of visually presented material from our (Goebel et al., 1998a) and
other groups (Cohen et al., 1996; Alivisatos and Petrides, 1997)
suggested a close spatial overlap between the two sets
of paradigms (Table 4).The new evidence from visuospatial
perceptual and mental imagery tasks performed by the same
subjects in the same experimental session confirmed that IPS
activation is common to both conditions. Our results reveal a
striking similarity between brain activation during spatial comparisons of visually perceived and imagined material.
Conclusion
In the present study we used a behaviourally controlled
paradigm to verify the neural basis of visuospatial imagery in
the absence of visual stimulation and eye movements. Our highresolution fMRI findings demonstrate that the PPC is strongly
involved in the processing of spatially coded material in the
imagery domain. Whilst our study did not address the separation
of the single subcomponents of spatial mental imagery (Trojano
and Grossi, 1994), the comparison of our results with those
of recent studies of spatial transformations of visually presented
material indicates that the analysis of visual space in perception
and imagery has a common neural basis in the parietal lobes. It
is suggested that the neural networks involved in the processes
of spatial transformation might be shared by several cognitive
functions, including visuospatial imagery.
Notes
The authors are grateful to Jim Waltz and Lars Muck li for comments
on the manuscript. This paper is dedicated to Prof. Giuseppe Caruso,
unforgettable teacher.
Address correspondence to Dr Luigi Trojano, Salvatore Maugeri
Foundation, IRCCS, Center of Telese, Loc. S. Stefano in Lanterria, I–82037
Telese Terme (BN), Italy. Email: [email protected].
References
Aisenberg S (1996) Measurement and analysis of binocular rapid eye
movement, including saccades, saccadic prediction, fixations, nystagmus, and smooth pursuit. Invest Ophthalmol Vis Sci Suppl 37:S274.
Alivisatos B, Petrides M (1997) Functional activation of the human brain
during mental rotation. Neuropsychologia 35:111–118.
Bandettini PA, Jesmanowicz A, Wong EC, Hyde JS (1993) Processing
strategies for time-course data sets in functional MRI of the human
brain. Magn Reson Med 30:161–173.
Boynton GM, Engel SA, Glover GH, Heeger DJ (1996) Linear systems
analysis of functional magnetic resonance imaging in human V1. J
Neurosci 16:4207–4241.
Cohen MS, Kosslyn SM, Breiter HC, DiGirolamo GJ, Thompson WL,
Anderson A K, Brookheimer SY, Rosen BR, Belliveau JW (1996)
Changes in cortical activity during mental rotation. A mapping study
using functional MRI. Brain 119:89–100.
Dierks T, Linden DE, Jandl M, Formisano E, Goebel R, Lanfermann H,
Singer W (1999) Activation of Heschl’s gyrus during auditory hallucinations. Neuron 22:615–621.
Farah MJ (1984) The neurological basis of mental imagery: a componential analysis. Cognition 18:245–72.
Farah MJ, Hammond KM, Levine DN, Calvanio R (1988) Visual and spatial
imagery: dissociable systems of representation. Cogn Psychol 20:
439–462.
Friston KJ, Holmes AP, Worsley KJ, Poline JB, Frith CD, Frackowiak RSJ
(1995) Statistical parametric maps in functional imaging: a general
linear approach. Hum Brain Map 2:189–210.
Goebel R, Khorram-Sefat D, Muckli L, Hacker H, Singer W (1998a)
The constructive nature of vision: direct evidence from functional
magnetic resonance imaging studies of apparent motion and motion
imagery. Eur J Neurosci 10:1563–1573.
Goebel R, Linden DEJ, Lanfermann H, Zanella FE, Singer W (1998b)
Functional imaging of mirror and inverse reading reveals separate
coactivated networks for oculomotion and spatial transformations.
Neuroreport 9:713–719.
Goebel R, Singer W (1999) Cortical surface-based statistical analysis of
functional magnetic resonance imaging data. Neuroimage 9:S64.
Grossi D, Modafferi A, Pelosi L, Trojano L (1989) On the different roles of
the two cerebral hemispheres in mental imagery: the ‘O’Clock test’ in
two clinical cases. Brain Cogn 10:18–27.
Grossi D, Angelini R, Pecchinenda A, Pizzamiglio L (1993) Left imaginal
neglect in heminattention: experimental study with the o‘clock test.
Behav Neurol 6:155–158.
Kosslyn SM, Sussmann AL (1995) Roles of imagery in perception: or,
there is no such thing as immaculate perception. In: The cognitive
neurosciences (Gazzaniga M, ed.), pp. 1035–1042. Cambridge, MA:
MIT Press.
Levine DN, Warach J, Farah M (1985) Two visual systems in mental
imagery: dissociation of ‘what’ and ‘where’ in imagery disorders due
to bilateral posterior cerebral lesions. Neurology 35:1010–1018.
Linden DEJ, Prvulovic D, Formisano E, Völlinger M, Zanella FE, Goebel R,
Dierks T (1999) The functional neuroanatomy of target detection:
an fMRI study of visual and auditory oddball tasks. Cereb Cortex 9:
515–523.
Luzzatti C, Vecchi T, Agazzi D, Cesa-Bianchi M, Vergani C (1998) A
neurological dissociation between preserved visual and impaired
spatial processing in mental imagery. Cortex 34:461–469.
Malach R, Reppas JB, Benson RR, Kwong KK, Jiang H, Kennedy WA,
Ledden PJ, Brady TJ, Rosen BR, Tootell RB (1995) Object-related
activity revealed by functional magnetic resonance imaging in human
occipital cortex. Proc Natl Acad Sci USA 92:8135–8139.
Mellet E, Tzourio N, Crivello F, Joliot M, Denis M, Mazoyer B (1996)
Functional anatomy of spatial mental imagery generated from verbal
instructions. J Neurosci 16:6504–6512.
Mellet E, Petit L, Mazoyer B, Denis M, Tzourio N (1998) Reopening the
mental imagery debate: lessons from functional anatomy. Neuroimage
8:129–139.
Milner AD, Goodale MA (1995) The visual brain in action. Oxford:
Oxford University Press.
O’Craven KM, Rosen BR, Kwong KK, Treisman A, Savoy RL (1997)
Voluntary attention modulates fMRI activity in human MT-MST.
Neuron 18:591–598.
Paivio A (1978) Comparisons of mental clocks. J Exp Psychol (Hum
Percept) 4:61–71.
Sathian K, Zangaladze A, Hoffmann JM, Grafton ST (1997) Feeling with
the mind’s eye. Neuroreport 8:3877–3881.
Talairach J, Tournoux P (1988) Co-planar stereotaxic atlas of the human
brain: 3-dimensional proportional system: an approach to cerebral
imaging. Stuttgart: Thieme.
Trojano L, Grossi D (1994) A critical review of mental imagery defects.
Brain Cogn 24:213–243.
Ungerleider LG, Mishkin M (1982) Two cortical visual systems. In:
Analysis of visual behavior (Ingle DJ, Goodale MA, Mansfield RJW,
eds), pp. 549–586. Cambridge, MA: MIT Press.
Ungerleider LG, Courtney SM, Haxby JV (1998) A neural system for
human visual working memory. Proc Natl Acad Sci USA 95:883–890.
Wildgruber D, Kischka U, Ackermann H, Klose U, Grodd W (1999)
Dynamic pattern of brain activation during sequencing of word
strings evaluated by fMRI. Cogn Brain Res 7:285–94.
Cerebral Cortex May 2000, V 10 N 5 481