Neural changes in the ventral and dorsal visual streams during pattern recognition learning

bunkietameAI and Robotics

Oct 20, 2013 (4 years and 8 months ago)


Neurobiology of Learning and Memory xxx (2005) xxx–xxx
1074-7427/$ - see front matter  2005 Elsevier Inc. All rights reserved.
Neural changes in the ventral and dorsal visual streams
during pattern recognition learning
Colline C. Poirier
, Anne G. De Volder
, Dai Tranduy
, Christian Scheiber
Neural Rehabilitation Engineering Laboratory, Université catholique de Louvain, Brussels, Belgium
Neurofunctional Imaging Nuclear Medicine—MRI, UMR-7004 ULP/CNRS, Strasbourg, France
Received 13 May 2005; revised 2 August 2005; accepted 4 August 2005
The learning process related to pattern and object recognition is diYcult to study because the human brain has a remarkable
capacity to recognise complex visual forms from early infancy. In the present study, we investigated on-going neural changes under-
lying the learning process of visual pattern recognition by means of a device substituting audition for vision. Functional MRI evi-
denced the gradual pattern recognition-induced recruitment of the ventral visual stream, bilaterally, from learning session 1 to
session 3, and a slight decrease in these activation foci from session 3 to session 4. The initial increase in activation is thought to
reXect the gradually enhanced visualisation of patterns in the subjects’ mind across sessions. By contrast the subsequent decrease
reported at the end of the training period is interpreted as the progressive optimisation of neuronal responses elicited by the task. Our
results, in accordance with previous observations, suggest that the succession of activation increase and decrease in sensori-motor
areas could be a general rule in sensory and sensori-motor learning.
 2005 Elsevier Inc. All rights reserved.
Keywords:Visual pattern recognition; Learning; fMRI
1. Introduction
Over the past 10 years, numerous studies have dealt
with experience-dependent changes in the nervous sys-
tem. These studies mainly investigated motor and visuo-
motor learning processes (e.g., Jueptner et al., 1997;
Kawashima et al., 2000; Sakai et al., 1998; Toni, Krams,
Turner, & Passingham, 1998; Toni, Ramnani, Josephs,
Ashburner, & Passingham, 2001). Perceptual learning
processes received less attention from the neuroimaging
community. Most of these studies focussed on orienta-
tion discrimination tasks (Schiltz et al., 1999; Schiltz,
Bodart, Michel, & Crommelinck, 2001), face recognition
(Gathers, Bhatt, Corbly, Farley, & Joseph, 2004), and
categorisation phenomena (Chao, Weisberg, & Martin,
2002; Little, Klein, Shobat, McClure, & Thulborn, 2004;
Little & Thulborn, 2005). Pattern and object recognition
learning processes are diYcult to study because the
human brain has a remarkable capacity to recognise
visual complex forms from early infancy. In the present
study, we investigated neural changes underlying the
learning process of simple pattern recognition through a
prosthesis substituting audition for vision referred to as
PSVA (Capelle, Trullemans, Arno, & Veraart, 1998).
Such a device, normally devoted to blind people,
provides visual information through the auditory
system: trained blindfolded sighted users are able to
recognise visual patterns (Arno, Capelle, Wanet-Defalque,
Catalan-Ahumada, & Veraart, 1999; Poirier, Richard,
Tranduy, & Veraart, 2005), to perceive depth cues
(Renier, Collignon, Tranduy, Vanlierde, & De Volder,
2004), and visual illusions (Renier et al., 2005).
Corresponding author. Present address: Centre de Médecine
Nucléaire, Hôpital Neuro-Cardiologique, 59 Bd Pinel, 69677 BRON
Cedex, France. Fax: +33472357345.
E-mail address: (C. Scheiber).
2 C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx
According to the users of this device, these behavioural
performances seem to be mediated through mental
imagery processes. The PSVA device was thus used in
the present study as an artiWcial but original tool to
study visual learning. It allowed investigating the learn-
ing process from a naïve state without concern about
previous experience and studying the progressive setting
of the neural bases underlying a completely new
Literature about skill learning does not allow
predicting how brain activation will evolve through
learning. Increased activation (Poldrack, Desmond,
Glover, & Gabrieli, 1998; Kawashima et al., 2000;
Poldrack & Gabrieli, 2001) and decreased activation
(Kassubek, Schmidtke, Kimmig, Lucking, & Greenlee,
2001; Schiltz et al., 2001) have both been observed in
parallel with increases in behavioural performance.
Such disparity could be explained, at least partially, by
the diVerent samplings used to investigate the learning
process. Indeed, most neuroimaging studies dedicated
to learning compared the performances of subjects
before and after learning (e.g., Kawashima et al., 2000;
Schiltz et al., 1999, 2001). However, this approach
raises the issue of how to determine when learning is
over. Depending on studies, either accuracy
(Kawashima et al., 2000) or the speed (Kassubek et al.,
2001) at which a task was performed was used as a cri-
terion. Moreover, learning and inherent neural changes
are known to occur even when performances and
response times have ceased to evolve (Jueptner et al.,
1997). Finally, transitory neural changes occurring
during the learning period cannot be investigated
in this way. In the present study, we decided to follow-
up the course of learning and, to this end, to scan
our subjects throughout all the training sessions,
fromthe Wrst time they used the device until
they had acquired near-perfect pattern recognition
2. Methods
2.1. Subjects
Six right-handed male volunteers (mean 23 years,
range 20–33 years) naïve as regards the PSVA device
took part in the experiment. All gave prior written con-
sent. This experiment was approved by the Local Ethical
Committee (CCPPRB Alsace no. 1, France).
2.2. Stimulation device
Sounds were delivered by an auditory stimulation sys-
tem (E.A.R.TONE 3A Insert Earphone, Aero Company
Auditory Systems, Indianapolis) constituted of trans-
ducers and its dedicated calibrated plastic conduits. The
plastic conduits were inserted in the subjects’ ears. Head-
phones were added for further isolation purposes.
PSVA sounds were provided by a software simulating
all the characteristics of the prosthesis developed by
Capelle et al. (1998). A demonstration of the original
device is available on the web site http://www.percep- Visual patterns,
constituted of bars and dots, were implemented in a
6£5-pixel frame (6 columns and 5 lines) and sent to a
Digital Signal Processor (DSP) to be processed. Pro-
cessed images of visual patterns were then translated
into sounds according to a pixel-to-frequency code. In
this code, a single sine wave was assigned to each pixel of
the image according to its position. Frequencies,
between 446 and 1987Hz, increased from left to right
and from bottom to top of the image in such a way that
beating sounds corresponded to horizontally adjacent
activated pixels and harmonic sounds correspond to ver-
tically adjacent activated pixels. Left–right discrimina-
tion was achieved through interaural level diVerences.
The patterns could be moved using a joystick. When a
pattern was static, the corresponding sound was con-
stant; when subjects moved the pattern, the correspond-
ing sound varied according to the PSVA code (e.g., if a
subject moved a pattern from the top of the frame down-
wards, the sounds dropped gradually to a lower pitch).
Parts of the pattern which were outside the frame did
not deliver sounds. Analysing the modiWcation taking
place in the sounds as a function of joystick movements
allowed the subjects to recognise the patterns.
2.3. Stimuli and experimental design
The experimental protocol consisted in a block-
design of three active conditions with a rest period in-
2.3.1. The noise condition
Noise consisted of various combinations of Wve sine
waves among those used by the device. The subjects had
to move the joystick, thereby making the combination of
Wve sine waves change randomly. A 1-s blank appeared
in a time window of 11§2s after the beginning of the
noise. The subjects’ task consisted in listening carefully
to the noise, and in signalling the detection of silence by
pressing a button on the joystick. This condition lasting
for 24s was repeated 12 times.
2.3.2. The element condition
The subjects had to identify a dot, a vertical or hori-
zontal bar. According to the PSVA code, dots consisted
into one sine wave and thus sounded as a pure tone; hor-
izontal and vertical bars consisted into Wve sine waves
and sounded respectively as a beating complex sound or
as a harmonic complex sound. This condition lasting for
24s was repeated 12 times.
C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx 3
2.3.3. The pattern condition
The subjects had to identify patterns, which consisted
in the association of one vertical or horizontal bar and
one dot (Fig.1). This condition lasting for 48s was
repeated six times.
During these three conditions, the subjects were not
permitted to move the joystick freely. They had to follow
a predetermined motor sequence of successive vertical
and horizontal movements, and including joystick but-
ton press. Pressing the joystick button allowed the sub-
jects placing the element or the pattern on the frame
centre. This procedure ensured that the motor compo-
nent was identical for each condition and for each sub-
ject, and did not require decisional processes.
During the experiment, the subjects were blindfolded.
Each element and pattern blocks were followed by a ver-
bal statement from the subjects (e.g., “that was a dot” or
“There was a vertical bar on the left and a dot at the top
on the right”) and by the verbal answer from the experi-
menter (e.g., “Yes, it was,” or “No, there was a vertical
bar on the left and a dot at the bottom on the right”).
This fMRI experiment was replicated four times (Ses-
sions S1, S2, S3, and S4) over 2 weeks. Because identiW-
cation performances increase as a function of the
number of patterns explored (Poirier et al., 2005) all the
subjects explored the same number of elements and pat-
terns during each session. To have the same statistical
power as regards the detection of activation across sub-
jects and sessions, we chose to set the exploration time of
elements and patterns to 24 and 48s, respectively. These
lapses of time were based on pre-tests designed to evalu-
ate the mean time necessary to identify elements and pat-
terns. When subjects happened to have Wnished
exploring an item, before the end of the block, they were
instructed to keep their attention constant and to verify
their answer until the time was over. The order of the
conditions was pseudo-randomised across the four ses-
sions. The elements and patterns explored were identical
in the four fMRI sessions but their presentation order
was changed.
The subjects underwent two 1-h training sessions the
week before the Wrst fMRI experiment. During these ses-
sions, the subjects became familiar with the joystick
manipulation and the noise condition. Extensive verbal
explanations about the PSVA code were provided but
subjects did not hear organised sounds of the device
before the Wrst fMRI experiment.
The behavioural performances of the subjects were
recorded during the experiments and the number of pat-
terns identiWed was computed for each subject at each
2.4. Image acquisition and data analysis
fMRI data were obtained in a 2-T MRI system
(Bruker, Karlsruhe, Germany) with BOLD contrast
echo planar imaging (Xip angle 90°, TED50ms,
TRD4.8s). 32 continuous slices covering the whole
brain were acquired. The voxel size was 4£4£4 mm.
Anatomical images required for the localisation of func-
tional responses were obtained using a RARE T
weighted sequence using the following parameters:
128£128 (80 slices, TED73.8 ms, TRD15000ms).
Pre-processing and statistical analyses were carried
out using the SPM99 software (http://www.Wl.ion.
uk/spm). For each subject, all functional volumes were
motion-corrected using SINC interpolation and normal-
ised in the Talairach stereotaxic system of coordinates.
Images were then spatially smoothed with an 8-mm
width Gaussian kernel. The three active conditions were
Wtted with a box-car function convolved with the hemo-
dynamic response function (hrf) plus a temporal deriva-
tive. The verbal interactions between the experimenter
and the subjects were Wtted as a supplementary condi-
tion using the same function (box-car function con-
volved by the hrf). The verbal interactions and the six
head movement-related parameters were then consid-
ered as regressors of no interest. Low-frequency tempo-
ral drifts were removed by applying a 105-s high-pass
A Wxed-eVect group analysis was performed. The pat-
tern recognition task consisted in element identiWcation
as well as the identiWcation of the spatial location of these
elements. The identiWcation of elements and patterns was
achieved by analysing modiWcations in the sounds, as a
function of joystick movements. Assuming that the iden-
tiWcation of visual elements and patterns is mediated by
visual mental imagery, these recognition tasks were thus
likely to induce visual imagery of objects, as well as visuo-
spatial imagery. We thus focussed our analysis on the
ventral and the dorsal visual streams, known to be
recruited, respectively, by the visual imagery of objects
and visuo-spatial imagery (e.g., De Volder et al., 2001;
Slotnick, Thompson, & Kosslyn, 2005; Vanlierde, De
Volder, Wanet-Defalque, & Veraart, 2003). Due to the
auditory nature of the stimuli and the cognitive aspect of
the task, we also looked at potential modiWcations in the
auditory cortex and the frontal lobes. Four regions of
interest (ROI) were deWned on the basis of anatomical
criteria: a ventral stream ROI, a dorsal stream ROI, an
auditory ROI, and a frontal ROI (Table 1).
Fig.1. The six patterns explored during each fMRI session.
4 C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx
Owing to the heterogeneity of the results reported in
the literature about skill learning (Jueptner et al., 1997;
Kassubek et al., 2001; Kawashima et al., 2000; Poldrack
et al., 1998; Poldrack & Gabrieli, 2001; Sakai et al.,
1998; Schiltz et al., 1999, 2001), the relationship between
the hemodynamic response and the training sessions
cannot be predicted. Moreover, this relationship may
vary from one brain region to another (Sakai et al.,
1998; Toni et al., 1998). In order not to narrow our anal-
ysis by any a priori hypothesis, we compared sessions
two-by-two, assessing the condition£session interac-
tion. To identify increases, we used the following con-
trast: [Pattern vs. Noise]
minus [Pattern vs. Noise]
The analysis was restricted to the areas more activated
by Pattern than by Noise in sessions n ([Pattern vs.
) by using the conjunction analysis developed
by Nichols, Brett, Andersson, Wager, and Poline (2005).
Finally, to disentangle between increases in Pattern
condition and decreases in Noise condition, we inclu-
sively masked the previous contrasts by [Pattern
]. The resulting contrast was thus [Pat-
tern vs. Noise]
minus [Pattern vs. Noise]
[Pattern vs. Noise]
inclusively masked by [Pattern
vs. Pattern
]. To identify decreases, we used the
inverse contrast: [Pattern vs. Noise]
minus [Pattern
vs. Noise]
AND [Pattern vs. Noise]
masked by [Pattern
vs. Pattern
]. Voxels with a
statistical signiWcance of p<.05 corrected for multiple
comparisons inside each ROI were considered to be sig-
niWcantly activated. Only clusters with an extent greater
than 30 voxels were considered.
3. Results
Since similar fMRI results were observed for element
and pattern recognition, we focussed on pattern recogni-
tion results.
The analysis of behavioural results revealed the learn-
ing process in the group of subjects (Fig.2). The percent-
age of correct recognition was signiWcantly higher
during S4 (78%) than S1 (36%) (ZD¡2.23, pD.026).
Subjects reported to use mental imagery to perform pat-
tern identiWcation.
In the right ventral stream, activation increased from
S1 to S2, from S2 to S3, and decreased from S3 to S4
(Table 2, Figs. 3 and 4). In the left hemisphere, we
observed no signiWcant modiWcation of activation from S1
and S2, increased activation from S2 to S3, and a decrease
in activation from S3 to S4. In the right dorsal stream,
activation increased from S1 to S2. In the left hemisphere,
activation increased from S1 to S2, decreased from S2 to
S3, and then increased again from S3 to S4.
In the auditory cortex, the only observed modiWca-
tion was an increased activation on the left side from
S3 to S4 (Table 3). In the frontal lobes, activation
increased bilaterally from S1 to S2 in the frontal eye
Weld. We also observed increased activation from S2 to
S3 in the right side then subsequent decrease from S3 to
S4 in both sides in more anterior frontal areas
(Table 3).
4. Discussion
This study showed that several neural changes
occurred during the element and pattern recognition
learning processes. These changes were evidenced by
comparing two active conditions. According to the cog-
nitive subtraction theory, all non-speciWc changes,
potentially due to experimental habituation and/or vari-
ations in signal-to-noise ratio as a function of time were
removed by subtracting activation related to the Noise
condition from activation related either to the Element
or Pattern condition. The neural changes evidenced here
are thus thought to be learning-speciWc.
Mainly, fMRI evidenced the gradual pattern identiW-
cation-induced recruitment of the ventral and dorsal
visual stream, followed by a subsequent decrease in these
activation foci.
Description of the ROI
Note. 19d, dorsal part of Brodmann area 19; 19v, ventral part of Brodmann area 19.
ROI Coordinates of the ROI centre ROI size (x, y, z, in mm) Brodmann areas
Dorsal visual stream +/¡28, ¡82, 32 56, 24, 24 19d, 39
Ventral visual stream +/¡34, ¡64, ¡10 56, 44, 28 19v, 37
Auditory cortex +/¡50, ¡22, 6 36, 74, 28 41, 42, 22
Frontal lobes +/¡32, 28, 20 64, 84, 88 6, 8, 9, 10, 11, 24, 44, 45, 46, 47
Fig.2. Percentage of pattern recognition in the four fMRI sessions (+SE).
C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx 5
Increases in activation have been frequently observed
in the inferior temporal lobe during the mirror reading
learning process (Poldrack et al., 1998; Poldrack &
Gabrieli, 2001) and in the occipital cortex during a
motor task requiring internal visual representation
(Kawashima et al., 2000). Increased activation is usually
interpreted as a more important recruitment of these
areas and/or as the recruitment of additional cortical
units to perform a task (Büchel, Coull, & Friston, 1999;
Poldrack, 2000). In the present study, increased activa-
tion of visual areas could reXect the gradually enhanced
visualisation of patterns as subjects engage deeper in
mental imagery. This interpretation is reinforced by the
increased activation observed in the frontal eye Weld
since mental imagery is known to induce eye movements
(De’Sperati, 2003; Mast & Kosslyn, 2002).
ModiWcations across sessions of the neural recruitment induced by pattern identiWcation in the ventral and dorsal visual streams
Note. L, left; R, right; S1, session 1; S2, session 2; S3, session 3; S4, session 4.
Ventral visual pathway Dorsal visual pathway
Coordinates Z score Coordinates Z score
x y z x y z
S1 vs. S2
Increases R Inf. temporal gyrus (37) 54 ¡66 0 5.12 R Sup. occipital gyrus (19) 36 ¡72 24 5.75
— — — — L Precuneus (19) ¡26 ¡78 34 5.19
Decreases — — — — — — — —
S2 vs. S3
Increases R Inf. temporal gyrus (37) 48 ¡58 ¡6 4.48 — — — —
L Mid. occipital gyrus (37) ¡46 ¡66 ¡8 5.07 — — — —
L Inf. occipital gyrus (19) ¡38 ¡82 ¡4 4.61
Decreases — — — — L Precuneus (19) ¡22 ¡84 44 3.99
S3 vs. S4
Increases — — — — L Precuneus (19) ¡44 ¡74 38 4.47
Decreases R Inf. temporal gyrus (37) 44 ¡52 ¡6 4.14 — — — —
L Inf. occipital gyrus (19) ¡36 ¡80 ¡4 5.51 — — — —
Fig.3. Brain activation foci elicited by the pattern recognition condi-
tion, contrasted to the random noise condition. The statistical para-
metric map for this comparison (group analysis) is superimposed on
the sagittal section (x D¡52 and +54) of an individual normalised
brain MRI, allowing the visualisation of brain activation variations in
the ventral visual stream. Only voxels exceeding a threshold of p <.05
corrected for multiple comparisons in the whole brain are displayed.
Fig.4. Evolution of the signal across the four sessions, illustrating the
modiWcations of the ventral stream activation, in Pattern and Noise
conditions. Size eVects were computed in the left and right clusters of
the ventral stream, found to be activated in the contrast [Pattern vs.
Noise] in session 3. These clusters corresponded to the ones depicted in
Fig.3 (third row). Size eVects were computed using the MarsBaR tool-
box (Brett et al., 2002).
6 C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx
Decreases in activation have been frequently
observed in the occipital areas during the learning pro-
cesses of orientation visual discrimination (Schiltz et al.,
1999, 2001) and mirror reading (Kassubek et al., 2001).
Decreased activation can be due to the shorter time
taken to perform a task and/or to the decrease in the
attentional demand (Poldrack, 2000). In the present
study, as previously explained, we chose to set the time
dedicated to the exploration of elements and patterns.
The same experiment had been previously conducted,
outside the magnet, with another group of subjects, and
unlimited exploration time. It turned out that during the
fourth session, it took subjects 50s on average to recog-
nise the patterns with a similar accuracy (78% of recog-
nition) as during the fourth fMRI experiment.
Moreover, during the fMRI experiments, subjects were
instructed to keep their attention constant: they were
requested to verify their answer until the time was over.
For these reasons, we think that decreased activation is
probably not due to a decrease in attentional load.
The decreased activation observed in the anterior fron-
tal areas usually engaged in intense executive control of
mental processing (Cohen, Braver, & O’Reilly, 1996) sug-
gests that the task became less eVortful, more automatic,
at the end of training (Cohen et al., 1996; Deiber et al.,
1997). This decrease combined to the increased activation
observed in the secondary auditory cortex could lead one
to interpret decreased activation in the visual streams as a
disengagement of mental imagery in the pattern identiWca-
tion process. However, the lack of decreased activation in
the frontal eye Weld (recruited by eye movements normally
induced by mental imagery) and the ultimate increased
activation observed in the dorsal visual stream from S3 to
S4 go against this interpretation.
The decrease in activation in the occipital areas is
more likely to arise from long-term repetition priming.
Neural responses have been shown to decrease further
to repeated exposure to identical stimuli, even when
repetition occurs after several days (Büchel et al.,
1999; Poldrack et al., 1998; Poldrack & Gabrieli,
2001). In our study, the subjects explored the same pat-
terns across the diVerent sessions. The activation
decrease is thought to reXect the sharpening of
responses in a particular neural network as a function
of experience. A minority of neurons would Wre
strongly in response to a particular stimulus whereas
the majority of neurons would Wre less (Büchel et al.,
1999; Poldrack, 2000).
A slightly diVerential pattern of brain modiWcations
was observed in both visual streams. Activation gradu-
ally increased in the ventral stream from S1 to S3 then
decreased from S3 to S4 while activation in the dorsal
stream increased from S1 to S2, decreased from S2 to S3,
and Wnally increased again from S3 to S4. This result
conWrms that the relationship between the hemodynamic
response and the training sessions may vary from one
brain region to another (Sakai et al., 1998; Toni et al.,
1998, 2001). Toni et al. (1998) also reported an initial
activation increase, a subsequent decrease and then a
Wnal increase. The authors did not provided any inter-
pretation of these results. Additional studies are required
to explain why brain activation increased from S3 to S4
in the present study.
In the present study, we observed a transient increase
and a subsequent decrease in activation in the ventral
and dorsal visual streams. It thus seems that increased
activation is associated with early stages of learning and
decreased activation with late stages of learning.
Increases and decreases in sensory-motor areas reported
in literature previously cited cannot be easily interpreted
in terms of early or late stages of learning. In these stud-
ies, neural networks were compared only before and
after training. The reported “before training” state is
often not comparable with our because of the high level
ModiWcations across sessions of the neural recruitment induced by pattern identiWcation in the auditory cortex and the frontal lobes
Note. L, left; R, right; S1, session 1; S2, session 2; S3, session 3; S4, session 4.
Auditory cortex Frontal lobes
Coordinates Z score Coordinates Z score
x y z x y z
S1 vs. S2
Increases — — — — R Mid. frontal gyrus (6) 26 2 42 4.86
— — — — L Precentral gyrus (6) ¡40 ¡4 34 5.74
Decreases — — — — — — — —
S2 vs. S3
Increases — — — — R Mid. frontal gyrus (9) 44 38 ¡2 4.81
Decreases — — — — — — — —
S3 vs. S4
Increases L Sup. temporal gyrus (22) ¡60 ¡48 16 5.37 — — — —
Decreases — — — — R Sup. frontal gyrus (10) 26 52 2 5.95
R Mid. frontal gyrus (9) 26 38 16 5.24
— — — — L Precentral gyrus (9) ¡38 10 32 4.80
C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx 7
of behavioural performances (Poldrack et al., 1998,
2001; Schiltz et al., 1999, 2001) (>70% of accuracy in
these studies vs. 36% in our study). Kawashima et al.
(2000) seems to have investigated rather early stages of
learning (percentage of correct responses evolved from 5
to 27%) and reported increased activation in the occipi-
tal cortex. Toni et al. (1998, 2001) have investigated the
time course of neural changes during motor and visuo-
motor learning within one fMRI sequence. They
reported increased activation and subsequent decreases
in the supplementary motor area (Toni et al., 1998) and
in the ventral visual pathway (Toni et al., 2001). More
recently, Little et al. (2004, 2005) have investigated
changing patterns of brain activation during visual cate-
gory learning over 4 days. These authors reported
increased activation from day 1 to day 2 then decreased
activation from day 2 to day 4 in visual areas. The suc-
cession of activation increase and then decrease in sen-
sori-motor areas could thus be a general rule in sensory
and sensori-motor learning.
The authors gratefully acknowledge Corinne
Marrer, Thierry Wijns, and Frédéric Blanc for their
technical assistance, and Nathalie Heider for
linguistic corrections. ADV is senior research associate
at the Belgian National Fund for ScientiWc
Research.This work was supported by FRSM
(3.4505.04), and FNRS Grants (Belgium) and Euro-
pean Commission Quality of Life contract (No. QLG3-
Arno, P., Capelle, C., Wanet-Defalque, M. C., Catalan-Ahumada, M.,
& Veraart, C. (1999). Auditory coding of visual patterns for the
blind. Perception, 28, 1013–1029.
Büchel, C., Coull, J. T., & Friston, K. J. (1999). The predictive value of
changes in eVective connectivity for human learning. Science, 283,
Brett, M., Anton, J.-L., Valabregue, R., & Poline, J.-B. (2002). Region
of interest analysis using an SPM toolbox. Neuroimage, 16.
Capelle, C., Trullemans, C., Arno, P., & Veraart, C. (1998). A real-time
experimental prototype for enhancement of vision rehabilitation
using auditory substitution. IEEE Transactions on Biomedical Engi-
neering, 45, 1279–1293.
Chao, L. L., Weisberg, J., & Martin, A. (2002). Experience-dependent
modulation of category-related cortical activity. Cerebral Cortex,
12, 545–551.
Cohen, J. D., Braver, T. S., & O’Reilly, R. C. (1996). A computational
approach to prefrontal cortex, cognitive control and schizophrenia:
Recent developments and current challenges. Philosophical Trans-
actions of the Royal Society of London Series B, 351, 1515–1527.
De’Sperati, C. (2003). Precise oculomotor correlates of visuospatial
mental rotation and circular motion imagery. Journal of Cognitive
Neuroscience, 15, 1244–1259.
De Volder, A. G., Toyama, H., Kimura, Y., Kiyosawa, M., Nakano, H.,
Vanlierde, A., Wanet-Defalque, M. C., Mishina, M., Oda, K., Ishiw-
ata, K., & Senda, M. (2001). Auditory triggered mental imagery of
shape involves visual association areas in early blind humans. Neu-
roimage, 14, 129–139.
Deiber, M.-P., Wise, S. P., Honda, M., Catalan, M. J., Grafman, J., &
Hallet, M. (1997). Frontal and parietal networks for conditional
motor learning: a positron emission tomography study. Journal of
Neurophysiology, 78, 977–991.
Gathers, A. D., Bhatt, R., Corbly, C. R., Farley, A. B., & Joseph, J. E.
(2004). Developmental shifts in cortical loci for face and object rec-
ognition. Neuroreport, 15, 1549–1553.
Jueptner, M., Stephan, K. M., Frith, C. D., Brooks, D. J., Frackowiak,
R. S., & Passingham, R. E. (1997). Anatomy of motor learning. I.
Frontal cortex and attention to action. Journal of Neurophysiology,
77, 1313–1324.
Kassubek, J., Schmidtke, K., Kimmig, H., Lucking, C. H., & Greenlee,
M. W. (2001). Changes in cortical activation during mirror reading
before and after training: an fMRI study of procedural learning.
Brain Research Cognitive Brain Research, 10, 207–217.
Kawashima, R., Tajima, N., Yoshida, H., Okita, K., Sasaki, T., Schor-
mann, T., Ogawa, A., Fukuda, H., & Zilles, K. (2000). The eVect of
verbal feedback on motor learning—a PET study. Neuroimage, 12,
Little, D. M., Klein, R., Shobat, D. M., McClure, E. D., & Thulborn, K.
R. (2004). Changing patterns of brain activation during category
learning revealed by functional MRI. Brain Research Cognitive
Brain Research, 22, 84–93.
Little, D. M., & Thulborn, K. R. (2005). Correlations of cortical activa-
tion and behavior during application of newly learned categories.
Brain Research Cognitive Brain Research, in press.
Mast, F. W., & Kosslyn, S. M. (2002). Eye movements during visual
mental imagery. Trends in Cognitive Science, 6, 271–272.
Nichols, T., Brett, M., Andersson, J., Wager, T., & Poline, J. B. (2005).
Valid conjunction inference with the minimum statistic. Neuroim-
age, 25, 653–660.
Poirier, C., Richard, M. A., Tranduy, D., & Veraart, C. (2005). Assess-
ment of sensory substitution prosthesis in minimalist condition of
learning. Applied Cognitive Psychology, in press.
Poldrack, R. A. (2000). Imaging brain plasticity: Conceptual and meth-
odological issues—a theoretical review. Neuroimage, 12, 1–13.
Poldrack, R. A., Desmond, J. E., Glover, G. H., & Gabrieli, J. D. (1998).
The neural basis of visual skill learning: An fMRI study of mirror
reading. Cerebral Cortex, 8, 1–10.
Poldrack, R. A., & Gabrieli, J. D. (2001). Characterizing the neural
mechanisms of skill learning and repetition priming: Evidence from
mirror reading. Brain, 124, 67–82.
Renier, L., Collignon, O., Tranduy, D., Vanlierde, A., & De Volder, A.
G. (2004). Depth perception with a sensory substitution system in
early blind subjects. In Annual meeting of the Belgian psychological
society (p. 36). Universal Press.
Renier, L., Laloyaux, C., Collignon, O., Tranduy, D., Vanlierde, A.,
Bruyer, R., & De Volder, A. G. (2005). The Ponzo illusion using a
auditory substitution of vision in sighted and early blind subjects.
Perception, 34, 857–867.
Sakai, K., Hikosaka, O., Miyauchi, S., Takino, R., Sasaki, Y., & Putz.,
B. (1998). Transition of brain activation from frontal to parietal
areas in visuomotor sequence learning. Journal of Neuroscience, 18,
Schiltz, C., Bodart, J. M., Dubois, S., Dejardin, S., Michel, C., Roucoux,
A., Crommelinck, M., & Orban, G. A. (1999). Neuronal mecha-
nisms of perceptual learning: changes in human brain activity with
training in orientation discrimination. Neuroimage, 9, 46–62.
Schiltz, C., Bodart, J. M., Michel, C., & Crommelinck, M. (2001). A pet
study of human skill learning: Changes in brain activity related to
learning an orientation discrimination task. Cortex, 37, 243–265.
8 C.C. Poirier et al. / Neurobiology of Learning and Memory xxx (2005) xxx–xxx
Slotnick, S. D., Thompson, W. L., & Kosslyn, S. M. (2005). Visual men-
tal imagery induces retinotopically organized activation of early
visual areas. Cerebral cortex, in press.
Toni, I., Krams, M., Turner, R., & Passingham, R. E. (1998). The time
course of changes during motor sequence learning: A whole-brain
fMRI study. Neuroimage, 8, 50–61.
Toni, I., Ramnani, N., Josephs, O., Ashburner, J., & Passingham, R. E.
(2001). Learning arbitrary visuomotor associations: temporal
dynamic of brain activity. Neuroimage, 14, 1048–1057.
Vanlierde, A., De Volder, A. G., Wanet-Defalque, M. C., & Veraart, C.
(2003). Occipito-parietal cortex activation during visuo-spatial
imagery in early blind humans. Neuroimage, 19, 678–709.