Cerebral Cortex May 2008;18:1139--1147

doi:10.1093/cercor/bhm145

Advance Access publication August 23, 2007

A Frontoparietal Network for SpatialAttention Reorienting in the AuditoryDomain: A Human fMRI/MEG Study ofFunctional and Temporal Dynamics

M. Brunetti1,2, S. Della Penna1,2, A. Ferretti1,2, C. Del Gratta1,2,

F. Cianflone1,2, P. Belardinelli1,2,3, M. Caulo1,2, V. Pizzella1,2,

M. Olivetti Belardinelli3,4 and G. L. Romani1,2

1Institute of Advanced Biomedical Technologies, 2Department

of Clinical Sciences and Biomedical Imaging, University ‘‘G.

D’Annunzio’’of Chieti, Via dei Vestini, 33, 66013 Chieti (CH),

Italy, 3ECONA (Interuniversity Centre for Research on

Cognitive Processing in Natural and Artificial Systems) and4Department of Psychology, University of Rome ‘‘La Sapienza’’,

via dei Marsi, 78, 00185 Rome (RM), Italy

Several studies have identified a supramodal network critical to thereorienting of attention toward stimuli at novel locations and whichinvolves the right temporoparietal junction and the inferior frontalareas. The present functional magnetic resonance imaging(fMRI)\magnetoencephalography (MEG) study investigates: 1) thecerebral circuit underlying attentional reorienting to spatiallyvarying sound locations; 2) the circuit related to the regular changeof sound location in the same hemifield, the change of soundlocation across hemifields, or sounds presented randomly atdifferent locations on the azimuth plane; 3) functional temporaldynamics of the observed cortical areas exploiting the comple-mentary characteristics of the fMRI and MEG paradigms. fMRIresults suggest 3 distinct roles: the supratemporal plane appearsmodulated by variations of sound location; the inferior parietallobule is modulated by the cross-meridian effect; and the inferiorfrontal cortex is engaged by the inhibition of a motor response.MEG data help to elucidate the temporal dynamics of this networkby providing high-resolution time series with which to measurelatency of neural activation manipulated by the reorienting ofattention.

Keywords: auditory processing, fMRI, MEG, reorienting response,vertical meridian

The neural correlates of spatial attention have long been studied

in the visual domain. More recently, investigations are beginning

to focus on spatial attention in the auditory modality. In

a previous study, using a passive listening paradigm, we found

that sound localization processing involves the supratemporal

plane and supramarginal gyrus with different activation patterns

for sounds from fixed locations on the frontal plane, and sounds

presented randomly at different locations (Brunetti et al. 2005).

The different activation in the 2 above conditions could be due

to: 1) different pathways for orienting attention toward fixed

versus moving sound locations; 2) distinct covert motor

responses elicited by fixed and moving sounds; 3) different

pathways depending on whether the sounds cross the vertical

meridian plane. In regard to the first hypothesis, Corbetta and

Shulman (2002), reviewing experimental evidences about the

control of visual attention, proposed a 2-system model

composed of: 1) a dorsal network involving temporoparietal

and dorsal frontal regions for the intrinsic direction of attention

toward behaviorally relevant events and 2) a ventral network

involving the temporoparietal junction and ventral frontal

cortex implicated in the extrinsic reorienting of attention by

the environment. The latter circuit, recruited to detect un-

attended or low-frequency events, has been shown to be

independent of sensory modality. In regard to the second

hypothesis, Downar et al. (2000), using multisensory stimula-

tion, found activation in the temporoparietal junction, in the

inferior frontal gyrus and in the insula, lateralized to the right

hemisphere. In particular, the authors stress that the activation

in the frontal region, corresponding to Brodmann area (BA) 44

and normally involved in planning motor responses, was

observed in subjects that were not performing a motor task

during the experiment. They attributed this activation to the

involuntary planning of motor responses to changing stimuli,

even if these motor responses were not executed. Macaluso

et al. (2002), in an functional magnetic resonance imaging

(fMRI) study, demonstrated that the dorsal frontoparietal circuit

for spatial attention, previously observed to be activated in the

visual modality, was also observed in the tactile and auditory

modality; furthermore, in the auditory modality, the activity was

increased with increasing binaural coherence (Zimmer and

Macaluso 2006). Finally, in relation to our third hypothesis,

Rizzolatti et al. (1987) found that a large ‘‘cognitive’’ cost is paid

when a stimulus appears at an unattended location in the

hemifield opposite the attended one—that is, across the vertical

or horizontal meridian plane (meridian effect). The authors

postulate that the meridian effect is related to the way in which

eye movements are programmed and suggest that overt and

covert orienting of attention are controlled by common

mechanisms. Moreover, recent studies have demonstrated the

existence of an auditory vertical meridian (Ferlazzo et al. 2002)

and suggested that the meridian effect could be a consequence

of the interaction between visual and auditory modalities

(Olivetti Belardinelli et al. 2005, 2007).

The present fMRI/magnetoencephalography (MEG) study

aims to investigate the cerebral circuits that are engaged in the

orienting of attention toward different sounds locations. In

particular, we study: 1) the cerebral circuit activated by the

orienting of attention toward spatially varying sound locations,

providing a counterpart to earlier neuroimaging studies of

attentional reorienting in the visual modality; 2) differences in

the investigated circuit related to: b1 regular changes of sound

location within the same hemifield or across hemifields and b2

irregular sound source change on the azimuth plane; 3) the

functional temporal dynamics of these areas. With these aims,

we combine fMRI and MEG data to examine the effect of

reorienting attention on the waveform and latency of neural

activation in the involved regions disclosing the sequence of

activation in attention control areas.

Materials and Methods

SubjectsA group of 10 healthy volunteers (mean age 24 ± 2 years), all females,

participated in the fMRI data acquisitions, after providing written

� The Author 2007. Published by Oxford University Press. All rights reserved.

For permissions, please e-mail: journals.permissions@oxfordjournals.org

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

informed consent. The experimental protocol was approved by the local

institutional ethics committee. Out of these 10 subjects, 6 were studied

with MEG as well. All subjects were right handed according to the

Edinburgh handedness questionnaire (Oldfield 1971). The volunteers had

no anamnestic and clinical auditory impairment; they had normal hearing

thresholds at pure-tone audiometry, and they had no previous history of

neurological or psychiatric illness. Before the experiment, the sound

localization abilities of each subject were assessed by delivering the same

stimuli used during the experiment and asking the subjects to report on

the position of sounds aloud.

StimuliDuring both the fMRI and MEG sessions, stimuli were delivered via

a nonmagnetic and MRI compatible sound system (Commander XG

MRI Audio System) with a frequency range from 100 Hz to 25 kHz. The

electric signals generated by the computer audio board were amplified

and sent through the shielded room penetration panel to an electro-

pneumatic transducer. The sounds reached the subject’s headset via

flexible plastic tubes.

The stimulus usedwas an audio recordingof a knife tapping on anempty

glasswith a duration of 500ms. The frequency band of the stimulus ranged

from 1200 to 7200 Hz with principle components before and after the

critical threshold of 1500 Hz. These values lie within the frequency band

range of both the computer audio card and the nonmagnetic headset used

in theexperiment (formoredetails about the stimulus choice rationale, see

Brunetti et al. 2005). The output of each channel was calculated from the

primitive sound by convolution with the proper head-related transfer

functions (HRTFs) in order to simulate the incoming direction of the

sound. For each position of the sound source, the Matlab convolution

function was used to convolve the source stimulus with the HRTF

corresponding to that position. The HRTFs were downloaded from the

Massachusetts Institute of Technology Web page (http://xenia.media.mit.

edu/~billg/). The calculations have been made for a standard head.

A passive listening paradigmwas used in which subjects were asked to

localize the position of the incoming stimuli in each of 4 conditions.

Three conditions consisted of sounds regularly alternating between 2

locations in a 40-degree slice of space centered on the subject at

a distance of 1 m (+90� and +50�; –90� and –50�; –20� and +20�; see Fig. 1).These conditions were termed RIGHT, LEFT, and CENTRAL depending

on whether they were confined to one hemifield or crossed the

meridian. The fourth condition, termed MIXED, consisted of a random

sequence of sounds projected from 5 locations, the 4 locations used in

the RIGHT and LEFT fixed conditions and a fifth location at the meridian

(±90�, ±50�, and 0�). The stimuli were presented in blocks of each

condition, each block consisting of 8 events with an interstimulus

interval of 1.5 s. The 4 conditions were presented in pseudorandom

order so that a total of 35 blocks of each type were presented to each

subject.

This paradigm allowed us to verify the cerebral circuit activated by

the orienting of attention toward regularly varying sound locations

either in the same or in different hemifields and toward sounds

presented randomly at different locations on the azimuth plane.

fMRI Data Acquisition and AnalysisfMRI was carried out with a SIEMENS MAGNETOM VISION scanner at

1.5 T. A standard head coil was used, and the subject’s head was held in

place with foam pads to reduce involuntary movements. Blood oxygen

level--dependent (BOLD) contrast functional images were acquired by

means of T �2 -weighted echo planar imaging free induction decay

sequences with the following parameters: time echo (TE) 60 ms, matrix

size 64 3 64, field of view (FoV) 256 mm, in-plane voxel size 4 3 4 mm,

flip angle 90�, slice thickness 4 mm, and no gap. To avoid the

interference of the scanner intense bursts of noise, the ‘‘sparse’’

sampling technique proposed by Hall et al. (1999) was used. Following

the procedure outlined by these authors, functional volumes at the end

of the stimulation periods and of the baseline (silent) periods were

acquired. Each stimulus sequence was preceded by a 2 s and followed

by a 14-s silent period (REST). The number of silent periods was 140, so

that a total of 280 functional volumes were acquired (35 3 4 total

sequences + 140 silent periods). The interval between 2 consecutive

acquisitions (time repitition [TR]) was set at 14 s to allow the

hemodynamic response to return to baseline before the following image

acquisition (see above in the Stimuli paragraph). Functional volumes

consisted of 18 transaxial slices parallel to the anterior commissura-

posterior comminssura line covering the cortical regions of interest

(ROIs). A high-resolution structural volume was acquired at the end of

the session via a 3-dimensional (3D) magnetization prepared rapid

gradient echo (MP--RAGE) sequence with the following features: sagittal,

matrix 256 3 256, FoV 256 mm, slice thickness 1 mm, no gap, in-plane

voxel size 1 3 1 mm, flip angle 12�, TR = 9.7 ms, and TE = 4 ms.

Raw data were analyzed by means of the Brain Voyager software

version 4.9 (Brain Innovation, The Netherlands). Preprocessing of

functional scans included motion correction and the removal of linear

trends from voxel time series. The preprocessed functional volumes of

each subject were recorded in the same session as the corresponding

structural data set from the same session. The coregistration trans-

formation was determined using the slice position parameters of the

functional images and the position parameters of the structural volume.

After visual inspection, this transformation was slightly adjusted, when

necessary, to account for subject movement between functional and

anatomical scans. Structural and functional volumes were normalized to

the Talairach space (Talairach et al. 1988) using a piecewise affine and

continuous transformation. Statistical activation maps were generated

by means of a t-test comparing voxel by voxel images of a given

stimulation condition (LEFT, RIGHT, CENTRAL, or MIXED) to those of

the REST condition (baseline condition). These statistical maps were

thresholded at P < 0.0004 at the voxel level, and a cluster size of at least

4 voxels was required. These thresholds and an estimate of the spatial

correlation of voxels (Forman et al. 1995; 3dFWHM routine of AFNI

package, Cox 1996) were used as input in a Monte-Carlo simulation

(AlphaSim routine of AFNI package, Cox 1996; Forman et al. 1995), in

order to assess the overall significance level (which is the probability of

a false detection of a cluster in the entire functional volume). A

corrected P value of 0.05 was thus obtained. Thresholded statistical

maps were then overlaid onto the subject structural scan to localize

significantly activated areas. A statistical group analysis was also

performed. The time series obtained from all subjects were z

normalized and concatenated prior to the computation of the group

statistical activation maps. The group activation map was then

superimposed onto the Talairach transformed structural scan of one

of the subjects.

For each subject, ROIs were determined on the basis of the individual

activation maps. Specifically, for each subject, ROIs were determined by

superimposing activation maps for each of the 4 conditions (consid-

ering the Boolean OR among the 4 conditions). Rather than defining

ROIs on the basis of group activation, the above mentioned method was

used in order to take into account interindividual anatomical variability,

so that BOLD signals were not underestimated due to a mismatch

between a mean ROI and individual activation. The procedure was

consistent because the functional area was always clearly defined. A

common ROI (based on group data) was defined only when individual

Figure 1. Stimuli are delivered from 5 different spatial locations: the vertical head-centered meridian (0�), the right angles to the meridian axis on both sides (left �90�,right þ90�), and 40� to the vertical head-centered meridian (left �50�, right þ50).

1140 Cortical dynamics of auditory spatial reorienting d Brunetti et al.

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

activations were not observed in all subjects. Subject responses in each

stimulation condition were characterized by evaluating the BOLD

signal intensity variation in each ROI. The strength of the activation in

a given experimental condition was expressed as the mean relative

change with respect to the baseline of the BOLD signal of the voxels

belonging to a given ROI. Heschl’s gyrus was identified by reviewing the

coronal aspect of the MP--RAGE passing through the temporal lobes and

locating the prominent mid-superior temporal features of the gyrus. We

then located the anterolateral and posteromedial boundaries of the

gyrus for each individual. It has been reported that many superior

temporal gyri may contain multiple Heschl formations. In accordance

with the work of Galaburda and Sanides (1980), we identify only the

most anterior transverse gyrus as Heschl’s.

A regional comparison of activation was then performed by means of

a repeated-measures analysis of variance (ANOVA). The dependent

variable of the ANOVA analysis was the relative variation of the BOLD

signal between the stimulation and silent condition, and the factor was

the experimental condition (RIGHT, LEFT, CENTRAL, and MIXED).

MEG Data Acquisition and AnalysisMEG recordings were used to reveal high-resolution timing of the

activation patterns observed for attention control in the fMRI sessions

by localizing the neural components in time using Equivalent Current

Dipole MEG source localization. Auditory evoked fields were recorded

using the whole-head neuromagnetic system at the University of Chieti

(Pizzella et al. 2001), which was developed in collaboration with

Advanced Technologies Biomagnetics srl (Pescara, Italy).

In order to determine the position of the subject’s head with respect

to the sensors the magnetic field generated by 4 coils placed on the

scalp was recorded before and after each measurement session. The

positions of the coils on the subject head were digitized by means of

a 3D digitizer (Polhemus, 3Space Fastrak), together with 4 anatomical

landmarks (nasion, left/right ear, and vertex), to be used for the MRI-

fMRI-MEG coregistration. To this end, spherical oil capsules were

applied at the nasion and left/right ear landmarks for structural MRI

acquisition. About 40 additional points were digitized on the scalp and

subsequently used for fitting the spherical volume conductor to the

subject head.

During MEG recordings, a similar experimental paradigm as in fMRI

was used. The overall magnetic noise of the stimulation apparatus was

below the noise level of the sensor. A total of 1120 stimuli were

presented from the 4 conditions (RIGHT, LEFT, CENTRAL, and MIXED

as described above) with a time lag of 3 s between sequences. The

auditory evoked magnetic fields were sampled at 1 kHz and bandpass

filtered between 0.16 and 250 Hz. After artifact rejection, trials were

averaged for each condition from –100 ms to +700 ms relative to

stimulus onset. For each channel, the baseline was calculated as the

mean field value from 10 to 20 ms after stimulus onset. Source analysis

was performed using BESA (MEGIS Software, Germany) multiple source

analysis based on the Equivalent Current Dipole source model and

a homogenously conducting spherical volume to model the subject’s

head.

Source localization was performed on the averaged data in the

MIXED condition because the total signal power was the largest in this

condition. These localizations were then used to estimate the strength

and orientation of the sources for the other conditions. Inspection of

the data time course showed components of activity in 5 distinct time

intervals. According to the components observed in the time course,

the data were fitted in the above time intervals using a source model

consisting of 5 moving equivalent current dipoles (ECDs), 2 pairs of

which were constrained to be symmetric.

In order to explain the residual field (i.e., the measured field minus

the field generated by the localized ECDs) after the localization of the

first 5 ECDs, 2 additional fixed dipoles were introduced at locations

constrained by the fMRI activation (see table 1 for the source time

intervals). The coordinates of these 2 ECDs were bounded to a 6-mm

cube centered on the ‘‘center of mass’’ of the corresponding fMRI

activations in the parietal and frontal areas (see Results) (concerning

this method, see also Ahlfors et al. 1999). A complete 7-dipole

configuration was accepted when the relative residual variance was

less than 15%.

Results

fMRI—Group Analysis

Group analysis showed activation in the bilateral supratemporal

plane, in the right inferior parietal lobule (IPL) and in the

right prefrontal cortex (PFC). Bilateral activation was larger in

the right hemisphere. Statistical activation maps during the

different experimental conditions are superimposed on the

Talairach-transformed inflated cortex obtained from one of

the subjects (Fig. 2, Table 1). Three clusters of activation were

revealed in the supratemporal plane of the right hemisphere:

the medial (BA 42) and lateral regions (BA 41) of Heschl’gyrus

(MHg and LHg, respectively) and posterior superior temporal

gyrus (PSTg, BA 22). The latter activation showed a larger

posterior extension (up to y = –55 mm in Talairach coordi-

nates) during the MIXED condition with respect to the other

experimental conditions. In the left hemisphere, 2 peaks of

activity were observed in the supratemporal plane: the MHg

(BA 42) and LHg regions (BA 41).

The right IPL (BA 40) activation was larger during the

MIXED condition as compared to the other conditions. The

activation in the right PFC (BA 44, 9) was larger during

the MIXED than during the RIGHT or the CENTRAL condition.

It should be noted that due to intersubject variability, activation

in the right hemisphere during the LEFT condition was below

the statistical significance threshold. However, statistically

significant activation was observed in every single subject.

Activation was also observed in the premotor area (BA 6),

suggesting some covert motor response; however, analysis did

not reach statistical significance due to large variability across

subjects and therefore was not considered for further analysis

Table 1Group results (P\ 0.05 corrected): Talairach coordinates and Z scores of the peak activity in

brain areas activated during the different experimental conditions

Experimentalconditions

Cerebralregions

x y z Z score

MIXED Right LHg 49 �19 11 8.13Left LHg �58 �25 19 8.31Right MHg 44 �29 11 7.79Left MHg �40 �35 19 9.06Right PSTg 59 �39 11 8.49Right IPL 34 �48 35 6.36Right PFC 40 26 28 4.70

RIGHT Right LHg 56 �20 14 5.05Left LHg �58 �21 12 6.42Right MHg 39 �20 14 5.47Left MHg �40 �30 15 6.60Right PSTg 61 �40 12 4.86Right IPL 45 �40 40 3.22Right PFC 37 29 39 5.62

LEFT Right LHg 50 �24 14 7.52Left LHg �60 �21 12 3.78Right MHg 36 �20 14 6.73Left MHg �44 �16 8 5.97Right PSTg 58 �32 14 5.61Right IPL 50 �41 31 2.57Right PFC 33 25 31 3.39

CENTRAL Right LHg 49 �20 16 6.29Left LHg �49 �29 16 6.06Right MHg 50 �24 16 6.48Left MHg �37 �33 16 6.42Right PSTg 58 �27 21 5.49Right IPL �36 �52 34 3.89Right PFC 37 37 33 3.88

Cerebral Cortex May 2008, V 18 N 5 1141

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

in this study. Talairach coordinates and peak Z scores of

activated areas are shown in table 1.

fMRI—Individual Subject Analysis

The ROIs for the individual subject analysis were consistent

with the 3 active regions of the supratemporal plane (the LHg

and MHg, and the PSTg) and with the activation in the IPL and

the PFC. The individual activation (relative variation of the

BOLD signal between each experimental condition and the

baseline condition) for each ROI was analyzed by a repeated-

measures ANOVA in order to assess possible effects of the

experimental conditions and of hemispheric asymmetry (Fig. 3).

In the right hemisphere, the ANOVA on the LHg and MHg

across experimental conditions revealed a stronger activation

during the MIXED condition than during the other conditions

(P < 0.05 and P < 0.01, respectively). The activation in the PSTg

showed a significant difference across conditions (P < 0.01). A

post hoc analysis (Duncan test) revealed a stronger activation

during the MIXED condition than during the other conditions

for these 3 temporal regions (see table 2). The contrast

CENTRAL versus RIGHT + LEFT in the 3 temporal regions

revealed a statistical difference only in MHg (P < 0.05). Finally,

the contrast MIXED versus CENTRAL also revealed stronger

activation during the MIXED condition in the 3 temporal

regions.

A stronger activation in the MIXED condition than in the

others was also observed in the IPL (P < 0.05). Specifically, post

hoc analysis revealed a trend toward significance (P = 0.05) in

the contrast MIXED versus CENTRAL in IPL and a statistical

difference (P < 0.01) in the contrast CENTRAL versus RIGHT +LEFT (more details in table 2). The PFC activation tended to be

significantly different across the experimental conditions with

a stronger activation during the MIXED condition than during

the 3 other condition. In the left hemisphere, activation in LHg

and MHg was not significantly different across the experimen-

tal conditions. Furthermore, for these regions, a 2-way ANOVA

with the experimental condition and the hemisphere as factors

did not reveal any significant difference between hemispheres.

The MEG

In all subjects, the evoked magnetic field measured over the

helmet showed 3 components with reproducible latencies of

40/60 (double peak), 100, and 180 ms. Additionally, 2

components with smaller amplitudes occurred at about 240

and 350 ms. At these latencies, a dipolar pattern over the right

and left temporal region, the right IPL, and the right PFC was

observed. Figure 4 shows the butterfly plot of the evoked

signals (a) and the global field power (b) for a representative

subject (n�5). The time intervals used for the dipole fit are

marked on the butterfly plot (see further on). In Figure 5, field

maps and isofield contours are shown for subject n�5. We

identified 7 sources according to the following procedure:

First, a middle-latency bilateral component, characterized by

a W-shaped waveform with 2 similar positive peaks at

approximately 40 and 60 ms (P40/60) and a mean peak

amplitude of 9 and 11 nAm for the right and left hemisphere,

respectively, was localized in MHg anterior to primary auditory

cortex. Second, a distinct, long-latency bilateral source was

identified in the primary auditory cortex (LHg) and peaking at

about 100 ms after the stimulus presentation (N100) with

a mean amplitude of 34 and 27 nAm for the right and left

hemisphere, respectively. A late component was also observed

peaking at about 180 ms after the stimulus onset (P180, peak

amplitude 23 nAm) and was localized in the PSTg in the right

hemisphere. A left hemisphere source was also localized for the

180-ms component in 2 subjects, but further statistical analysis

was not performed on this source. The position of these 5

ECDs was consistent with the centroids of the fMRI activations

found bilaterally in MHg, LHg, and in right PSTg. In Figure 6, the

dipole positions localized in each subject are superimposed on

the individual fMRI maps for the MIXED condition. For each of

the 5 sources in the right hemisphere, the grand average across

subjects of the ECD waveform is shown in Figure 7 (upper, left

side). The amplitude of the N100 peak in the right hemisphere

was larger than the amplitude of the corresponding peak in the

left hemisphere (P = 0.05). During the MIXED condition, the

amplitude of the P40/60 was significantly larger (P < 0.05) than

in the other conditions. The amplitudes of the other 4 sources

localized in the supratemporal plane were larger in the MIXED

condition than in the other ones, even if this comparison was

not statistically significant.

Due to the fact that the 5 ECDs located in the supratemporal

plane accounted for an average of 82% of the total variance for

latencies larger than 220 ms, we decided to use fMRI results to

seed additional sources in an attempt to more fully represent

the data. Activation in fMRI suggested the addition of 2 sources

in the right hemisphere. We therefore seeded 2 ECDs in the IPL

and the PFC of the right hemisphere at locations corresponding

Figure 2. Results from the group analysis (P\ 0.05, corrected) superimposed onthe inflated cortex of a representative subject: activated areas during auditorystimulation in the 4 conditions: MIXED (a), RIGHT (b) LEFT (c), and CENTRAL (d). Inthe MIXED condition, activation in the bilateral Heschl’s gyrus and in the right PSTgwas more extended than in the other conditions. The PSTg activation extends dorsally(y 5 �55 in the right hemisphere). A right activation in the IPL is also observable.Activation in the PFC in the right hemisphere is present in the MIXED, RIGHT, andCENTRAL conditions (note that the group statistical maps shows activation in theright frontal cortex in these 3 conditions, whereas the individual subject analysisshows the same region activated also during the LEFT condition [see Fig. 3]).

1142 Cortical dynamics of auditory spatial reorienting d Brunetti et al.

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

to the fMRI activation in these regions. The ECD located in the

IPL (sixth source) showed a component peaking at about 240

ms (P240): After adding the IPL dipole, the explained variance

increased to 91.4% at this latency. The ECD located in the PFC

(seventh source) exhibited a component peaking at about 350

ms after the stimulus onset (P350). After adding the PFC dipole,

the explained variance increased to 89.7% at this latency. (See

table 3 for a summary of ECDs definition and related fit

interval). In general, the explained variance increased on

average by 9.9% due to the IPL dipole and by 7.4% after adding

the PFC dipole. The mean amplitude of these latter compo-

nents was 11.12 and 13.82 nAm, respectively. The P350

amplitude differed across the experimental conditions. Specif-

ically, it was stronger during the MIXED condition with respect

to the other conditions (P < 0.05). The grand average across

subjects of the waveform of these 2 sources is shown in Figure

7 (upper, right side). A one-way ANOVA for repeated measure-

ments was then performed on the latency of each component

of the right hemisphere for the CENTRAL versus RIGHT + LEFT

contrast. For the components in the temporal and the parietal

regions, a delay was observed during the CENTRAL condition,

but the comparison was not significant.

Furthermore, one-way ANOVA for repeated measure was

performed to verify statistically significant differences among

the mean peak latencies of the 5 dipoles located in the right

hemisphere (in MHg, LHg, PSTg, IPL, and PFC). The analysis

revealed a statistical difference in peak latency (P < 0.01), with

further post hoc analysis confirming that each component

latency was significantly different from the others (P < 0.01).

Figure 7 (lower) shows the mean latencies of the 5 sources

together with the statistical significance levels.

Discussion

In the literature, the neurophysiological correlates of orienting

attention in space have been extensively studied, mainly with

regard to visual attention. Both universal and modality-specific

systems have been hypothesized for attention processes

(Bushara et al. 1999). Further, a 2-system attentional model

for distinct stimulus characteristics: a posterior system

Table 2Significant differences in activation in the ROIs as result by post hoc analysis (Duncan test)

Contrasts LHg MHg PSTg IPL PFC

MIXED versus CENTRAL P\ 0.01 P\ 0.01 P\ 0.01 P 5 0.05 P 5 0.06MIXED versus CENTRALþ RIGHT þ LEFT

P\ 0.01 P\ 0.01 P\ 0.01 P\ 0.05 P 5 0.01

CENTRAL versus RIGHTþ LEFT

— P\ 0.05 — P\ 0.01 —

Figure 3. Activated areas in subject n�5 during MIXED condition. Activation map is superimposed on the segmented cortex (we decided to show the activation during theMIXED condition just for a representative purpose). In the box on the right side, activation map is superimposed on an axial section passing through supratemporal plane. Viewgraphics: relative variation of the BOLD signal and standard error in the LHg and MHg, in the PSTg, in the supramarginal gyrus (IPL), and in the PFC across the 4 experimentalconditions. The ANOVA analysis shows that the relative variation of the BOLD signal is higher in the MIXED condition than in the other conditions (see also table 2).

Cerebral Cortex May 2008, V 18 N 5 1143

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

monitoring stimulus position and an anterior system attending

to the selection of stimulus features (Posner and Petersen

1990) has been described.

In a previous paper (Brunetti et al. 2005), we found that the

reorienting of attention to auditory stimuli at unpredictable

locations takes place in a circuit involving the supratemporal

plane. More specifically, the posterior portion of the superior

temporal gyrus and the supramarginal gyrus. Moreover, we

found that the activation related to regularly moving stimuli

were different from those produced by stimuli randomly

presented at different locations. Starting from these consid-

erations, we developed a novel paradigm to compare, under

passive conditions, reorienting responses to spatially varying

auditory targets. We took care to ensure that the reorienting

response did not contain the execution of any motor response.

The main advantage of our paradigm is that it allows us to

manipulate the meridian effect by means of a contrast between

the CENTRAL condition and the RIGHT and LEFT conditions.

Using a completely passive paradigm, in which it was possible

to control the place of sound appearance, we investigated the

involvement of the temporoparietal junction and the ventral

frontal areas (BA 44) in the reorienting of covert attention in

the auditory modality. Particularly, we tried to answer 3

questions concerning the ventral frontoparietal network. We

wanted to know if the network exhibited: a spatial variation

effect, distinct covert motor responses for still and moving

sounds, and an effect for whether or not attention had to cross

the vertical meridian plane.

Activation of Superior Temporal Regions

fMRI results indicated 3 clusters of activation (BA 41, 42, 22) in

the supratemporal plane during passive listening to sounds

from different locations in the same hemifield and in different

hemifields. Activation in MHg and LHg (BA 41, 42), comprising

the primary auditory cortex, was observed bilaterally, in

accordance with previous results (Rao et al. 1997; Nakai et al.

2005). An additional region in the PSTg was also activated.

Analysis of MEG data revealed different sources in the supra-

temporal plane: 2 sources with 2 identical positive peaks at 40

and 60 ms (P40/60), 2 additional bilateral components peaking

at 93 ms (N100), and a fifth source peaking at 182 ms (P180). The

sources were localized in the primary auditory cortex (bilateral

LHg and MHg) and in the right PSTg. Previous studies (Itoh

Figure 4. Butterfly plot of the evoked signals (a) and the global field power (b) fora representative subject. Time intervals used for dipole fit are shown. See also table 3for further details.

Figure 5. Field maps and isofield contours for a representative subject.

1144 Cortical dynamics of auditory spatial reorienting d Brunetti et al.

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

Figure 7. Upper: The seven ECDs modeling the evoked field are superimposed on the individual structural MRI—left side: 3 sources in the supratemporal plane and the grandaverage of the waveform of the 3 sources in the right hemisphere; right side: 2 sources in the supramarginal gyrus and in the PFC and the grand average of the waveform of the2 sources in the right hemisphere (the waveform color corresponds to the dipoles color). Lower: Mean latencies and standard error of the 5 sources in the right hemisphere:MHg and LHg, PSTg, supramarginal gyrus (in the IPL), and PFC. The ANOVA revealed a significant difference between the latencies of the 5 sources in the right hemisphere, withP\ 0.01.

Figure 6. Individual dipole positions superimposed on the individual fMRI maps for the MIXED condition are shown for each subject.

Cerebral Cortex May 2008, V 18 N 5 1145

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

et al. 2000; Yvert et al. 2001) found a middle-latency magnetic

component in the supratemporal plane at a different latency

and with constrasting results to hemispheric dominance. Some

difference was found in the amplitude of the sources localized

in the temporal region suggesting a response to predictability;

however, this result was statistically significant only for the

sources localized in LHg. The lack of statistical significance for

the contrast in the other regions could be due to the limited

number of subjects. MEG analysis also indicated a statistical

trend for the response latency of left MHg, which was larger

during the MIXED condition than during the other conditions.

Essentially, there is some evidence that left hemisphere

response is slower during the MIXED condition than during

more regular conditions. This result suggests that the left

hemisphere response in MHg could be affected by a spatial

variation effect with a larger cognitive cost demanded when it

is necessary to reorient attention to stimulus positions that

randomly change in space, rather than toward regularly

changing stimulus position. The involvement of the PSTg was

in agreement with previous studies, further demonstrating the

role of this area in the sound localization process (Rauschecker

et al. 2000; Brunetti et al. 2005). These results suggest that the

temporal regions (MHg, LHg, and PSTg) could be responsive to

auditory location change detection, in accordance with pre-

vious studies that stress the role of the temporoparietal

junction in the detection of stimulus salience over the

processing of primary stimulus features (Corbetta and Shulmann

2002; Shulman et al. 2003).

Activation of IPL

Activation in the right inferior parietal lobule (Ba 40) was

observed with some response to the reorienting of attention in

space and across the vertical meridian plane. The involvment of

the right parietal cortex in the attentional orientation process

was observed in several clinical and neuroimaging studies

(Karnath et al. 2001; Brunetti et al. 2005). The activation in this

area (which together with the supratemporal plane forms the

temporoparietal junction) is consistent with the ventral

frontoparietal network described by Corbetta and Shulmann

(2002). Furthermore, our MEG results suggest a trend in the

latency of P240, the component localized in the IPL, which may

be interpreted as the influence of the meridian crossing effect

on IPL activity. This result, together with fMRI data, indicates

that the right IPL is a region responsive to the auditory

meridian crossing effect.

Activation of Frontal regions

The fMRI activation that we found in the frontal areas,

specifically in the PFC (BA 44), define a circuit that precisely

overlaps with the ventral frontoparietal network implicated in

detecting unattended or low-frequency events. Nonetheless,

the activation in PFC is quite different compared with that

observed in the parietal region: PFC appears only marginally

involved in the meridian crossing effect or the spatial variation

effect. A possible interpretation of this dual association is that

our paradigm does not completely eliminate the possibility of

head movement or oculomotor response. Although our pro-

tocol explicitly requests movement suppression in the sub-

jects, we do not completely bar movement through restraint,

and this could affect the activation in prefrontal region. In fact,

activation in the right PFC, together with activation in IPL, was

observed in a number of studies using a no-go task, suggesting

that inferior/ventral right prefrontal regions are involved in

response inhibition or no-go behavior (Garavan et al. 2002).

More recently, Heinen et al. (2006) found ventral prefrontal

activation in an oculomotor no-go task.

The observed pattern of activation is different from the one

shown in our previous study, in which frontal activation was

not observed in a similar paradigm based on fixed sounds

(Brunetti et al. 2005). Results from the present study

demonstrate the response of temporal regions to the detection

of sounds that change location in space, more intensive when

the sounds change location randomly. Furthermore, the differ-

ence in activation intensity of the temporo parietal junction

during listening to sounds that cross or do not cross the vertical

meridian plane allows us to consider the hypothesis of a specific

cognitive cost for this situation and suggest the involvement of

the IPL in this process. Finally, theweak responses in PFC to both

spatial variation and meridian crossing could be explained as

response inhibition more than the covert motor response

hypothesized above.

In terms of lateralization, the literature contains evidence

suggesting that this cortical circuit seems to be strongly

lateralized to the right hemisphere (Knight et al. 1998;

Arrington et al. 2000; Daffner et al. 2000). Our results partially

confirm this data because the parietal and frontal activations

were observed in the right hemisphere only. However, the

bilateral activation of the supratemporal plane was not

significantly stronger in the right hemisphere than in the left.

All together, these data suggest that the auditory attentional-

reorienting response starts bilaterally in the auditory cortex

before moving to the right PSTg and later projecting to the right

inferior frontal regions through the right inferior parietal cortex.

In summary, our multimodal study reveals activation in the

IPL and in the PFC, together with activation in the supra-

temporal plane. These activations were similar when the sound

position changed in the same hemifield at regularly varying

locations but were significantly stronger when the sound

position changed between the 2 hemifields. These results allow

us to describe a network, with different involvement of the

component regions, for the reorientation of attention toward

changing auditory stimuli locations. Additionally, our MEG

results demonstrate the temporal dynamics of the regional

activations within the described network: The activation of

Heschl’s gyrus was observed 95 ms after stimulus, the peak

latency of the PSTg occurred at 182 ms, the IPL peaked at 245

ms, and the frontal region peaked at 343 ms. This sequence of

neural activation, integrated with the fMRI results, highlights

a bottom-up progression from the temporal to the frontal

cortex and shows the way in which neural activity propagates

through the attentional reorienting network.

Table 3Definition of ECDs and related fit interval

Fitted ECDsSeeded fromfMRI

Fit interval(ms)

1\2 Symmetric (P40/60) No 40--603\4 Symmetric (N100) No 80--1205 Right hemisphere (P180) No 120--2106 Right hemisphere (P240) Yes 220--2807 Right hemisphere (P350) Yes 330--360

1146 Cortical dynamics of auditory spatial reorienting d Brunetti et al.

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from

Notes

The authors would like to thank Professor Maurizio Corbetta,

Washington University School of Medicine, for his invaluable sugges-

tions, and Chris Lewis, Institute of Advanced Biomedical Technologies,

University ‘‘G. D’Annunzio of Chieti, for a careful editing of the

manuscript. The authors would also like to thank the anonymous

reviewers for helpful comments on the manuscript. Conflict of Interest:

None declared.

Address correspondence to Marcella Brunetti. Email: mbrunetti@

itab.unich.it.

References

Ahlfors SP, Simpson GV, Dale AM, Belliveau JW, Liu AK, Korvenoja A,

Virtanen J, Huotilainen M, Tootell RB, Aronen HJ, et al. 1999.

Spatiotemporal activity of a cortical network for processing visual

motion revealed by MEG and fMRI. J Neurophysiol.

82(5):2545--2555.

Arrington CM, Carr TH, Mayer AR, Rao SM. 2000. Neural mechanisms of

visual attention: object-based selection of a region space. J Cogn

Neurosci. 12:106--117.

Brunetti M, Belardinelli P, Caulo M, Del Gratta C, Della Penna S,

Ferretti A, Lucci G, Moretti A, Pizzella V, Tartaro A, et al. 2005.

Human brain activation during passive listening to sounds from

different locations: a combined fMRI/MEG pilot study. Hum Brain

Mapp. 26(4):251--261.

Bushara KO, Weeks RA, Ishii K, Catalan MJ, Tian B, Rauschecker JP,

Hallett M. 1999. Modality-specific frontal and parietal areas for

auditory and visual spatial localization in humans. Nat Neurosci.

2(8):759--766.

Corbetta M, Shulman G. 2002. Control of goal-directed and stimulus-

driven attention in the brain. Neuroscience. 3(3):201--215.

Cox RW. 1996. AFNI: software for analysis and visualization of

functional magnetic resonance neuroimages. Comput Biomed Res.

29:162--173.

Daffner KR, Mesulam MM, Scinto LF, Acar D, Calvo V, Faust R,

Chabrerie A, Kennedy B, Holcomb P. 2000. The central role of the

prefrontal cortex in directing attention to novel events. Brain.

123:927--939.

Downar J, Crawley AP, Mikulis JD, Davis KD. 2000. A multimodal

cortical network for the detection of changes in the sensory

environment. Nat Neurosci. 3(3):277--283.

Ferlazzo F, Padovani T, Couyoumdjian A, Olivetti Belardinelli M. 2002.

Head-centered meridian effect on auditory spatial attention orient-

ing. Q J Exp Psychol. 55(3):937--963.

Forman SD, Cohen JD, Fitzgerald M, Eddy WF, Mintun MA, Noll DC.

1995. Improved assessment of significant activation in functional

magnetic resonance imaging (fMRI): use of a cluster-size threshold.

Magn Reson Med. 33:636--647.

Galaburda A, Sanides F. 1980. Cytoarchitectonic organization of the

human auditory cortex. J Comp Neurol. 190:597--610.

Garavan H, Ross TJ, Murphy K, Roche RAP, Stein EA. 2002. Dissociable

executive functions in the dynamic control of behaviour:

inhibition, error detection and correction. Neuroimage. 17(4):

1820--1829.

Hall DA, Haggard MP, Akeroyd MA, Palmer AR, Summerfield AQ,

Elliot MR, Gurney EM, Bowtell RW. 1999. Sparse temporal sampling

in auditory fMRI. Hum Brain Mapp. 7:213--223.

Heinen SJ, Rowland J, Lee BT, Wade AR. 2006. An oculomotor decision

process revealed by functional magnetic resonance imaging. J

Neurosci. 26(52):13515--13522.

Itoh K, Yumoto M, Uno A, Kurauchi T, Kaga K. 2000. Temporal stream

of cortical representation for auditory spatial localization in human

hemispheres. Neurosci Lett. 292:215--219.

Karnath HO, Faber S, Himmelbach M. 2001. Spatial awareness is

a function of the temporal not the posterior parietal lobe. Nature.

411:950--953.

Knight RT, Scabini D. 1998. Anatomic bases of event-related potentials

and their relationship to novelty detection in humans. J Clin

Neurophysiol. 15:3--13.

Macaluso E, Frith CD, Driver J. 2002. Supramodal effects of covert

spatial orienting triggered by visual or tactile events. J Cogn

Neurosci. 14(3):389--401.

Nakai T, Matsuo K, Ohgami Y, Oishi K, Kato C. 2005. An fMRI study of

temporal sequencing of motor regulation guided by an auditory cue:

a comparison with visual guidance. Cogn Process. 6:128--135.

Oldfield RC. 1971. The assessment and analysis of handedness: the

Edinburgh Inventory. Neuropsychologia. 9:97--113.

Olivetti Belardinelli M, Santangelo V. 2005. The head-centred meridian

effect: auditory attention orienting in conditions of impaired visuo-

spatial information. Disabil Rehabil. 27(13):761--768.

Olivetti Belardinelli M, Santangelo V, Botta F. 2007. Are vertical

meridian effects due to audio-visual interference? A new confirma-

tion with deaf subjects. Disabil Rehabil. 29(10):797--804.

Pizzella V, Della Penna S, Del Gratta C, Romani GL. 2001. SQUID systems

for biomagnetic imaging. Supercond Sci Technol. 14:79--114.

Posner MI, Petersen SE. 1990. The attention system of the human brain.

Annu Rev Neurosci. 13:25--42.

Rao SM, Harrington DL, Haaland KY, Bobholz JA, Cox RW, Binder JR.

1997. Distributed neural systems underlying the timing of move-

ments. J Neurosci. 17:5528--5535.

Rauschecker JP, Tian B. 2000. Mechanisms and streams for processing

of ‘‘what’’ and ‘‘where’’ in auditory cortex. Proc Natl Acad Sci USA.

97(22):11800--11806.

Rizzolatti G, Riggio L, Dascola I, Umilta C. 1987. Reorienting attention

across the horizontal and vertical meridians: evidence in favor of

a premotor theory of attention. Neuropsychologia. 25:31--40.

Shulman GL, McAvoy MP, Cowan MC, Astafiev SV, Tansy AP, d’Avossa G,

Corbetta M. 2003. Quantitative analysis of attention and detection

signals during visual search. J Neurophysiol. 90(5):3384--3397.

Talairach J, Tournoux P. 1988. Co-planar stereotaxic atlas of the human

brain. New York: Thieme.

YvertB,CrouzeixA,BertrandO, Seither-PreislerA,PantevC.2001.Multiple

supratemporal sources of magnetic and electric auditory evoked

middle latency components in humans. Cereb Cortex. 11:411--423.

Zimmer U, Macaluso E. 2006. Processing of the binaural sound

coherence in the human brain. Cogn Process. 7(Suppl 1):109--110.

Cerebral Cortex May 2008, V 18 N 5 1147

by guest on Decem

ber 19, 2015http://cercor.oxfordjournals.org/

Dow

nloaded from