Human cortical auditory motion areas are not motion selective.

Abstract

The existence of a specialized mechanism supporting auditory motion processing in humans is a matter of debate in the psychophysical literature. Recent functional neuroimaging data appear to have resolved the debate in favor of a specialized motion system in that several studies have found cortical regions that seem to be motion selective. While all these studies contrast some form of moving auditory stimulation with a stationary stimulus, none have adequately controlled for the possibility that these areas are simply computing sound-source location and not motion per se: a moving stimulus varies in spatial location as well as motion, and so a system computing spatial location (and not motion) would be activated in response to both a moving and stationary sound source. To control for this possibility, ten subjects were scanned while listening to moving stimuli and while listening to stationary stimuli that varied randomly in spatial location. Consistent with previous imaging studies, we found that a motion stimulus when contrasted with rest (scanner noise) activated STG/planum temporale (bilaterally) and right parietal lobe. However, stationary stimuli presented at varying locations activated these regions equally well, arguing against the existence of specialized motion-processing areas in human cortex.

Full text

Page 1

Human cortical auditory motion areas are
not motion selective
Kevin R. Smith, Kayoko Okada, Kourosh Saberi and Gregory HickokCA
Departmentof Cognitive Sciences,UniversityofCalifornia,Irvine,CA 92612,USA
CACorresponding Author:gshickok@uci.edu
Received1September 2003; accepted29 October 2003
DOI:10.1097/01.wnr.0000130233.43788.4b
Theexistenceof a specializedmechanismsupportingauditorymo-
tionprocessinginhumansis amatterofdebatein thepsychophysi-
calliterature.Recentfunctionalneuroimagingdata appear to have
resolved the debatein favor of a specializedmotion systemin that
several studieshave foundcorticalregions that seem tobemotion
selective. While all these studies contrast some form of moving
auditory stimulation with a stationary stimulus, none have ade-
quately controlled for the possibility that these areas are simply
computing sound-sourcelocation and notmotionperse: a moving
stimulusvariesinspatiallocationaswellasmotion,andsoa system
computing spatial location (andnotmotion) wouldbe activatedin
responsetobothamovingandstationarysoundsource.Tocontrol
for this possibility, ten subjects were scanned while listening to
moving stimuliandwhilelistening to stationary stimuli thatvaried
randomlyin spatiallocation.Consistentwithpreviousimaging stu-
dies, we found that a motion stimulus when contrasted with rest
(scanner noise) activated STG/planum temporale (bilaterally) and
rightparietallobe.However, stationary stimulipresentedat vary-
ing locations activated these regions equally well, arguing against
theexistenceofspecializedmotion-processingareasinhumancor-
tex. NeuroReport 15:1523^1526 ? c
Wilkins.
2004 Lippincott Williams &
Key words: Auditorymotion; fMRI; Imaging; Motion selection; Rightparietal; STG
INTRODUCTION
The existence of a specialized auditory motion processing
system in humans is still a matter of debate. Until recently,
the primary source of evidence favoring a specialized
auditory motion system came from animal models in which
single unit recordings have identified brainstem and cortical
neurons in a variety of species that respond preferentially to
a specific direction of auditory source movement, and are
nonresponsive to stationary sounds [1,2]. Psychophysical
evidence from human observers, however, has been
equivocal. Some studies have found parallels in the way
stationary and moving stimuli are processed (e.g., equiva-
lent spatial acuity thresholds), suggesting a common
mechanism [3,4]. These studies propose that auditory
motion is inferred from an analysis of position changes of
discretely sampled loci in space. Such a snapshot model
holds that computational mechanisms supporting sound
localization are used to compute sound-source movement.
Other psychophysical studies, however, point to findings on
the perception of velocity, acceleration, and Doppler-shift
phenomena [5,6] as evidence for a specialized motion
processing system.
In the last few years, several human functional imaging
studies have weighed in on the debate. These studies, which
have contrasted various sorts of moving vs stationary
auditory stimuli, have all identified human cortical regions
more responsive to moving vs stationary sounds (Table 1).
These regions include most prominently, the planum
temporale (bilaterally) [7–9], premotor cortex (bilaterally)
[10–12], and right parietal cortex [10–13]. However, none of
these studies have ruled out a snapshot-like account of
motion computation. For example, contrasting a moving
sound source with a stationary source at a single location
confounds motion per se with the presence of a sound source
in multiple locations. Thus, the increase in neural activity in
the motion condition may be a consequence of a specialized
motion processing system, or may result simply from a
spatial localization mechanism doing more work localizing
multiple vs a single sound-source position. The present
study sought to disentangle these possibilities by contrast-
ing a moving sound source with stationary sources that vary
randomly and discretely in location over time.
MATERIALS AND METHODS
Participants:
Ten subjects (four male) participated in this
study. Subjects gave informed consent under a protocol
approved by the Institutional Review Board at the Uni-
versity of California, Irvine.
Materials:
presented through electrostatic headphones (STAX SR-001)
at a sampling rate of 44.1kHz. Motion was simulated by
dynamically changing the stimulus interaural level differ-
ence (ILD). Noise bursts were linearly ramped such that the
waveforms to the two ears received opposite slopes, e.g. left
ear was ramped down while right ear was ramped up in
level simulating motion from left to right. The total level
Stimuli were 400ms bursts of Gaussian noise
AUDITORYANDVESTIBULARSYSTEMS
NEUROREPORT
0959-4965? c Lippincott Williams& Wilkins
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.
Vol 15 No 9 28 June 2004
1523

Page 2

change was 24dB in 400ms (starting 12dB higher in one ear
and ending 12dB higher in the other ear). These stimuli
produce a strong percept of intracranial motion along the
interaural axis. For the stationary condition, a fixed ILD was
randomly selected on each stimulus presentation from the
set ?12, ?9, ?6, ?3, 3, 6, 9, and 12dB (negative and
positive signs denote higher intensity at left and right ears
respectively). A trial consisted of 8s stimulation during
which noise bursts were presented eight times (400ms noise
followed by 600ms silence repeated eight times). The 8s
stimulus period was followed by a 12s rest period. Each run
consisted of 12 blocks of 8s stimulation/12s silence.
Design and procedure:
using a 1.5T Siemens Vision scanner using an EPI pulse
sequence (FOV 256mm, matrix 64764, size 474mm, TE 40,
slice thickness 6mm). For each subject, a high resolution
anatomical image was acquired with a magnetization
prepared rapid acquisition with gradient echo (MPRAGE)
pulse sequence.
To correct for subject motion artifacts, the image volumes
of each subject were aligned to the sixth volume in the series
using a 3D rigid body, six parameter model in the AIR 3.0
program [14]. The volumes were then co-registered to the
high resolution anatomical image. After alignment, each
volume was spatially smoothed (Gaussian spatial filter,
4mm FWHM) and the time course of the blood oxygen level
dependent (BOLD) signal was temporally filtered (bandpass
0.025–0.1Hz).
Analysis proceeded in two steps. Initially, we sought to
replicate sites of activation during motion perception as
found previously [7]. Regression analysis was performed
Sixteen axial slices were collected
separately on each subject using AFNI software. Two
predictor vectors, which represented the time course of
stimulus presentations convolved with a standard hemody-
namic response function, were entered into the analysis to
find the parameter estimates that best explain the variability
in the data at each voxel. At each voxel, an F-statistic was
calculated and an activation map was created illustrating
the areas of activation in each subject during motion
perception vs rest. These activation maps were thresholded
at po0.0001 (uncorrected). For the next step, we used t-
statistics for the motion vs stationary condition. To reduce
the number of Type II errors, activation maps were
thresholded at a higher p-value (po0.025, uncorrected). In
addition, data from individual subjects were transformed
into Talairach space using AFNI. Coordinates for areas of
interest were calculated in individual subjects.
RESULTS
The motion condition, relative to background scanner noise,
showed significant activation bilaterally in the STG/planum
temporale (10 subjects showed left activation; nine showed
right activation) and in the right parietal lobe (five subjects)
at po0.0001 (uncorrected), consistent with previous studies
of auditory motion perception. When the motion condition
is contrasted with the stationary condition, however,
activation in these locations is eliminated in all but one
participant, even at the more liberal threshold of po0.025
uncorrected. In fact, no consistently activated region was
identified in the motion vs stationary contrast (Fig. 1, Fig. 2).
To ensure further that the lack of a motion selective response
in these regions was not a thresholding error, we plotted the
timecourse of the top five activated voxels defined by
T able1.
Summaryofprevious auditorymotion studies.
Motion stimuliControl stimulus Where activated (Talairachcoordinatesgiven) Experiment type
Interaural time di¡erences
(ITD)(binauralbeats)
ScannernoisePremotorcortex (bilateral) FMRI12
Rightparietalcortex (no coordinatesgiven)
Right STS(62, ?26, 5)
Rightplanum (no coordinatesgiven)
Premotorcortex (bilateral) (?36,0, 58), (58,8,46)
Rightparietalcortex (32, ?42, 50)
Premotorcortex (bilateral) (?42, 30, 22), (44,44,14)
Rightparietalcortex (48, ?22, 52), (42, ?40,48)
PostSTG (bilateral) (?44, ?30,10), (50, ?20,4)
ILD (300Hz sq. wave)
ILD (AM tones)
ITD/ILD (500Hz tone)
Scannernoise
Single stationary stimulus
Cancellationmotion (ILD-ITD)
FMRI8
FMRI9
FMRI/PET10,13
Rotating sound ¢eld
(broadbandnoise)
Horizontal/vertical
Stationary sound ¢eldPET1 1
2 Stationary stimuli
(+/? 45 degrees)
FMRI7
T able 2. Talairachcoordinates for thevoxelofpeak activationfromeach area (either left STG, right STG, orrightparietal) foreach subject.
Sub
Num
LSTG Location RSTG Location RPARLocation
1
2
3
4
5
6
7
8
9
49, 25,11
49,43,8
52, 20,8
49,18, ?2
54, 27 , 20
54, 31, 3
41, 29, 2
52,16,1 1
40, 23,8
56,17 ,14
Left transverse temporalgyrus
Leftmiddle temporalgyrus
Left superior temporalgyrus
Left superior temporalgyrus
Leftpostcentralgyrus
Leftmiddle temporalgyrus
Left sub-gyralwhitematter
Left transverse temporalgyrus
Left superior temporalgyrus
Leftpostcentralgyrus
?46, 29,1 1
?45,41,10
?43, 33,1 1
?50,18, ?2
?56,17 ,10
?54, 23,6
?52,17 ,15
?49, 22,0
?45,15,6
?46,19,7
Right transverse temporalgyrus
Right superior temporalgyrus
Right superior temporalgyrus
Right superior temporalgyrus
Right transverse temporalgyrus
Right superior temporalgyrus
Rightpostcentralgyrus
Right superior temporalgyrus
Rightinsularcortex
Right superior temporalgyrus
?36, 53,48
?44, 39,45
?43,41, 38
?29, 51,40
?47 , 36, 36
Right superior parietallobule
Right superior parietallobule
Rightinferior parietallobule
Right sub-gyralwhitematter
Right supramarginalgyrus
?32, 31,43
?49,47 , 36
Right sub-gyralwhitematter
Right supramarginalgyrus
10
1524
Vol 15 No 9 28 June 2004
NEUROREPORT
K.R. SMITHETAL.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Page 3

the motion minus rest (scanner noise) contrast, against the
activation in the same five voxels in the stationary
condition. Note the nearly identical response for both the
motion and stationary stimuli in the planum and parietal
regions (Fig. 3). Talairach coordinates for the most activated
voxel in each of the three areas are presented for each
subject in Table 2.
DISCUSSION
When contrasted with background scanner noise, moving
auditory stimuli activated regions previously identified as
candidates for motion selective cortical fields. However,
non-moving but spatially varying stimuli activated these
regions to a similar degree, arguing against the view that
they are selective for auditory motion processing. Further-
more, no other areas were found consistently across subjects
that showed motion selectivity, suggesting that unlike
vision, there is not a dedicated region in human cortex
specialized for auditory motion processing.
While this finding is consistent with the hypothesis that
auditory motion is computed using the same computational
mechanisms as non-motion spatial localization processes, it
2468 10
TR
2468 10
TR
2468 10
TR
−2
−1
0
1
2
−2
−3
−1
0
1
2
3
4
Z score
Z score
−2
−3
−1
0
1
2
3
Z score
LSTG/Planum
RSTG/Planum
Right parietal
Fig.1.
porale areas.The top row shows activations for themotion condition (at
po0.0001).The second row shows activations for the stationary condi-
tion (atpo0.0001).Thebottomrow shows the activation associatedwith
the contrastmotion^stationary (po0.025).Each column corresponds to
a di¡erent subject.
Activation maps for three subjects in the left STG/planum tem-
Fig. 2.
porale and parietal areas.The top row shows activations for the motion
condition (at po0.0001).The second row shows activations for the sta-
tionary condition (at po0.0001). The bottom row shows the activation
associatedwith the contrastmotion^stationary (po0.025).Eachcolumn
corresponds to a di¡erent subject.
Activationmapsfor threesubjectsintherightSTG/planum tem-
Fig. 3. Group average timecourses from the top ¢ve voxels from each
subject from each of the three areas (motion¼red (solid), station-
ary¼blue (dotted)).
Vol 15 No 9 28 June 2004
1525
HUMANCORTICAL AUDITORYMOTIONPERCEPTION
NEUROREPORT
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Page 4

does not rule out the possibility of a specialized motion
processing system. For example, neurons that have been
found with motion-selective properties are typically inter-
spersed with neurons that do not have motion-selective
properties [15]. It is possible that in humans, sound
localization and motion perception systems are functionally
separate and specialized systems, but that they inhabit the
same cortical regions, thus producing indistinguishable
responses at the spatial resolution of fMRI.
REFERENCES
1. Spitzer MWand Semple MN. Interaural phase coding in auditory midbrain:
influence of dynamic stimulus features. Science 1991; 254:721–723.
2. Moiseff A and Haresign T. Response of auditory units in the barn owl’s
inferior colliculus to continuously varying interaural phase differences.
J Neurophysiol 1992; 67:1428–1436.
3. Grantham DW. Auditory motion perception via successive ‘snapshot’
analysis. Sound localization by human observers. Proc Natl Acad Sci USA
1989; p. 35.
4. GranthamDW. Discrimination
differences. J Acoust Soc Am 1984; 76:71–76.
5. Perrott DR. Are there motion detectors in the auditory system? Sound
Localization by human observers. Proc Natl Acad Sci USA 1989; 36.
6. Perrott DR, Costantino B and Ball J. Discrimination of moving events
which accelerate or decelerate over the listening interval. J Acoust Soc Am
1993; 93:1053–1057.
of dynamic interaural intensity
7. Pavani F, Macaluso E, Warren JD, Driver J and Griffiths TD. A common
cortical substrate activated by horizontal and vertical sounds movement
in the human brain. Curr Biol 2002; 12:1584–1590.
8. Lewis J, Beauchamp M and DeYoe E. A comparison of visual and
auditory motion processing in human cerebral cortex. Cerebr Cortex 2000;
10:873–888.
9. Baumgart F, Gaschler-Markefski B, Woldorff MG, Heinze HJ and
Scheich H. A movement sensitive area in auditory cortex. Nature 1999;
4000:724–726.
10. Griffiths T, Green G, Rees A et al. Human brain areas involved
in the analysis of auditory movement. Hum Brain Mapp 2000; 9:
72–80.
11. Griffiths T and Green G. Cortical
of a rotating wide-field acoustic stimulus. Neuroimage 1999; 10:
84–90.
12. Bremmer F, Schlack A, Shah N et al. Polymodal motion processing in
posterior parietal and premotor cortex: a human fMRI study strongly
implies equivalencies between humans and monkeys. Neuron 2001;
29:287–296.
13. Griffiths T, Rees G, Rees A, Green GG, Witton C, Rowe D et al. Right
parietal cortex is involved in the perception of sound movement in
humans. Nature Neurosci 1998; 1:74–78.
14. Woods RP, Grafton ST, Watson JD, Sicotte NL and Mazziotta JC.
Automated image registration: II. intersubject validation of linear and
nonlinear models. J Comp Assist Tomogr 1998; 22:153–165.
15. Ahissar M, Ahissar E, Bergman H and Vaadia E. Encoding of sound-
source location and movement. Activity of single neurons and
interactions between adjacent neurons in the monkey auditory cortex.
J Neurophysiol 1992; 67:203–215.
activation duringperception
Acknowledgements:Thiswork was supportedby NIHgrant #DC03681.
1526
Vol 15 No 9 28 June 2004
NEUROREPORT
K.R. SMITHETAL.
Copyright © Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.