cortex 45 (2009) 35–43

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/cortex

Special issue: Research report

The primal role of the vestibular system in determining

musical rhythm

Laurel J. Trainora,b,*, Xiaoqing Gaoa, Jing-jiang Leia, Karen Lehtovaaraa and

Laurence R. Harrisc
a
Department of Psychology, Neuroscience & Behaviour and the McMaster Institute for Music and the Mind,
McMaster University, Hamilton, Ontario, Canada
b
Rotman Research Institute, Baycrest Hospital, Toronto, Ontario, Canada
c
Centre for Vision Research, York University, Toronto, Ontario, Canada

article info abstract

Article history: Previous studies have indicated that physical movement on either every second or on every
Received 19 January 2007 third beat of an unaccented auditory rhythm pattern can disambiguate whether it is
Reviewed 30 May 2007 perceived in duple time as a march or in triple time as a waltz. Here we demonstrate that
Revised 20 June 2007 this disambiguation can also be accomplished by direct galvanic stimulation of the
Accepted 4 October 2007 vestibular system. The galvanically induced sensation, without any actual movement, that
Published online 7 November 2008 the head moved from side to side on either every second or on every third beat of the
ambiguous auditory rhythm pattern strongly biased whether adults perceived it as being in
Keywords: duple or in triple time. These results imply that the vestibular system plays a primal role in
Vestibular stimulus the perception of musical rhythm.
Galvanic stimulation ª 2008 Elsevier Srl. All rights reserved.
Musical rhythm
Vestibulo-auditory interaction

1. Introduction to music with different rhythmic patterns (Cutietta, 2001) and

more advanced students in Dalcroze Eurhythmics classes
Physical movement and the perception of musical rhythm learn to internalize rhythmic patterns through physical
seem to be closely bound. Historically, music and dance have movement in time and space (Jaques-Dalcroze, 1920;
evolved together, and across all cultures, people move in Juntunen and Hyvönen, 2004). Adults and children can readily
synchrony to musical rhythms (Arom, 1991; Clarke, 1999; tap to the strong beats of a rhythmic pattern, regardless of
Cross, 2003; Molinari et al., 2003; Todd, 1995; Wallin et al., musical training (Drake et al., 2000) and tapping behaviour
2000). Body movement is commonly used in music education reflects their auditory representations of the rhythm pattern
in order to teach rhythmic patterns, further suggesting a link (Repp, 2005, review). Clearly there is a deep connection
between movement and the perception of rhythm. For between movement and rhythm but how might this have
example, in Kindermusik classes infants are passively moved come about?

* Corresponding author. Department of Psychology, Neuroscience & Behaviour and the McMaster Institute for Music and the Mind,
McMaster University, Hamilton, ON L8S 4K1, Canada.
E-mail addresses: ljt@mcmaster.ca (L.J. Trainor), harris@yorku.ca (L.R. Harris).
0010-9452/$ – see front matter ª 2008 Elsevier Srl. All rights reserved.
doi:10.1016/j.cortex.2007.10.014
36 cortex 45 (2009) 35–43

Perhaps the most common form of movement with

a tempo is walking. That locomotion involves regularly spaced
movements of the arms, legs, and the head has been observed
for a long time. Curiously, however, the tempo range of loco-
motion movements turns out to correspond closely to the
range over which a beat can be felt – pulse separations of
between about 300 and 900 msec (e.g., Fraisse, 1982; Clarke,
1999). At tempos slower than this, auditory events are not
connected into a perceptual pattern and at faster tempos,
events cannot be sufficiently individuated. Furthermore an
individual’s preferred auditory beat rate correlates with
anthropometric variables such as height, leg length, shoulder
breadth, and body weight (Todd et al., 2007) that are in turn
Fig. 1 – The auditory rhythm. Thin vertical lines represent
related to an individual’s locomotion tempo. These correla-
unaccented snare drum beats. Thick vertical lines
tions provide circumstantial evidence for a link between the
represent accented snare drum beats. Oblique lines
cues generated by locomotion and the auditory perception of
represent rests. The top line represents the ambiguous
tempo. But musical rhythm can be much more complex than
rhythm used during familiarization. The other two lines
the simple repetitive frequency of walking.
represent the rhythms used during the test phase. The
One of the remarkable abilities of humans is that we can
middle line represents the rhythm with auditory accents in
entrain and feel musical beats at several different tempos.
duple (march) time and the bottom line the rhythm with
Although many non-human animals engage in rhythmic
auditory accents in triple (waltz) time.
behaviours (e.g., apes will drum and certain birds will peck
rhythmically), only a few species entrain their movements to
different externally defined tempos (e.g., cockatoos, Patel
et al., 2008). Indeed, it appears that only species that engage in completely different, just as the two interpretations of
vocal learning exhibit rhythmic entraining (Schachner et al., the Ruben figure look completely different. And just as only
2008). Being able to produce different tempos synchronized one interpretation of an ambiguous visual pattern can be
with other players is a likely prerequisite for human musical perceived at any one time (either a vase or two faces), only one
behaviour (Trainor, 2007). Humans can extract a complex interpretation of the ambiguous auditory rhythm could be
metrical hierarchy from a rhythm pattern beyond that directly perceived at any one time.
present in the stimulus (e.g., Lerdahl and Jackendoff, 1983). For A physical bouncing movement on either every second
example, the tactus level (tempo at which you would tap your beat or on every third beat of the ambiguous auditory rhythm
foot) of a rhythm pattern is readily extracted, even from pattern biased which interpretation was perceived in both
patterns that may not always contain sound events on the infant (Phillips-Silver and Trainor, 2005) and adult subjects
beats that are perceived as strong (i.e., accented). A tactus (Phillips-Silver and Trainor, 2007). Visual information was not
level can also be derived when the tempo is not steady or even necessary for the effect as the results were similar when the
if the rhythm contains accented ‘‘off-beat’’ sound events, as in subjects were blindfolded. On the other hand, physical
the syncopations of jazz. Furthermore, the perceived metrical movement of the subject was critical as passively observing
structure is hierarchical. For example, the beats of the tactus the experimenter moving did not bias whether the ambiguous
can be subdivided into groups of two sub-beats (as in a march) rhythm sounded like a march or a waltz in either infants or
or three sub-beats (as in a waltz), producing a fast tempo that adults. These studies demonstrated the primal role of move-
can be perceived at the same time as the ongoing slower ment in determining musical rhythm but could not indicate
tactus. Tempos can also be slowed down by grouping. For which aspect(s) of movement were critical in determining the
example, every second or every third tactus-level beat may be perceived metrical interpretation.
more strongly accented, producing a slower rhythm. What correlate of movement might establish the correla-
Metrical structure is normally derived from the pattern of tion between physical movement and auditory rhythm?
physically accented beats – e.g., sound events that are longer, Candidates include motor planning (efferent copy of motor
louder, and/or higher in pitch. However, physical movement signals), sensory feedback arising from the consequences of
can also influence the metrical interpretation of an ambiguous the movement such as tactile or visual information, or
rhythm pattern (Phillips-Silver and Trainor, 2005, 2007). Just as sensory feedback directly about the movement itself arising
ambiguous visual stimuli can be created that can be perceived from proprioceptive sources. A major source of proprioceptive
in two distinct ways, such as the famous Ruben drawing of information concerning whole body movement comes from
a vase, the edges of which can alternatively be perceived as the vestibular system. Passive movement of the legs on either
two inward-looking faces, Phillips-Silver and Trainor created every second or third beat of an ambiguous auditory rhythm
an ambiguous auditory stimulus. It consisted of a repeating did not bias whether adults perceived a duple or triple
six-beat drum pattern with no accented notes (see Fig. 1). If metrical structure, but passive movement of the head on
every second beat was ‘‘heard’’ as strong, the rhythm took on these beats did (Phillips-Silver and Trainor, 2008). Given that
a march-like quality (duple meter) whereas if every third beat both the legs and the head are strongly involved in locomo-
was ‘‘heard’’ as strong, the rhythm took on a waltz-like quality tion, sensorimotor planning, tactile input, and proprioceptive
(triple meter). The two interpretations of the pattern sounded input, but only movement of the head generates a vestibular
 cortex 45 (2009) 35–43 37

signal, these results suggest that vestibular input may be 326.5–653 msec (where the first sound coincided with the
crucial to the multisensory interaction between movement background downbeat) resulting in a rhythm of sound-rest-
and auditory rhythm. The hypothesis of vestibular involve- sound-sound-sound-rest (see Fig. 1). Note that this rhythm
ment is consistent with the early emergence of movement/ pattern is ambiguous because it can be perceived either as
auditory interactions in infancy and the early maturation of consisting of three groups of two beats, with every second
the vestibular system, as illustrated by the delight with which beat accented as in a march (SOUND-rest-SOUND-sound-
premobile infants greet bouncing, rocking, and swooping SOUND-rest), or as two groups of three beats, with every third
stimulation. beat accented, as in a waltz (SOUND-rest-sound-SOUND-
In this paper, we dissociate the vestibular signal from sound-rest), even though these accents are not physically
aspects of movement by stimulating the vestibular system present. The sounds can be heard at http://psycserv.
directly in the absence of physical movement using galvanic mcmaster.ca/ljt/research.htm.
stimulation (see Buys, 1909; Goldberg et al., 1984; Mars et al., The two auditory test stimuli were identical to the training
2005; Wardman et al., 2003; Zink et al., 1997). Subjects listened rhythm described above, except that the rhythm pattern of
to an ambiguous rhythm pattern while we stimulated the interest was disambiguated by physically accenting some
vestibular nerve in such a way as to create the sensation of sounds relative to others by playing them louder (60 vs. 55 dB).
a side-to-side movement of the head, timed to occur on every Specifically, for the duple rhythm, every second beat was
second beat of the auditory rhythm for half of the subjects and physically accented as in a march (SOUND-rest-SOUND-
on every third beat for the other half. We provide evidence sound-SOUND-rest), and for the triple rhythm, every third
that a vestibular signal alone is sufficient to bias the metrical beat was physically accented as in a waltz (SOUND-rest-
interpretation of an auditory rhythm pattern. sound-SOUND-sound-rest). In all cases, the beat onset-to-
onsets fell within the optimal range for tempo discrimination
(Fraisse, 1982; Baruch and Drake, 1997).
2. Method
2.2.2. Vestibular stimulus
2.1. Participants The vestibular stimulation consisted of a small current
applied to electrodes on the mastoid process behind the ears
Participants were 23 (6 male, 17 female) university under- (see Section 2.3). The current waveform consisted of
graduate students (aged 18–25 years, mean age ¼ 19 years) Gaussian-shaped pulses with a peak amplitude of 1 mA and
enrolled in an introductory psychology course who received a standard deviation of 100 msec. Positive and negative pulses
course credit for participation. All of the participants had no were presented alternately such that they were out of phase in
known hearing deficits and were unaware of the purpose of the two ears (Fig. 2). Three tempos were used for the peak-to-
the study. Half of the participants were randomly assigned to peak time interval between stimulations, 653 msec (corre-
the experimental condition and half to the control condition. sponding to the duple auditory stimulus rate), 979.5 msec
Two additional subjects were excluded, one due to equipment (corresponding to the triple auditory stimulus rate), and
failure and one for producing data more than 3 standard 816 msec (midway between the duple and triple tempos, used
deviations from the mean. The experimental group was to calibrate individual phase shifts between vestibular stim-
comprised of 11 subjects; the control group was comprised of ulation and perceived head movement, see Section 2.5,
12 subjects. Subjects had between 0 to 14 years of music below). During pilot testing, we examined different wave-
lessons (mean ¼ 4.3 years for the experimental group and 4.1 forms for the vestibular current stimulation, including
years for the control group). Procedures were approved by the impulse waves, triangular waves, and sine waves, and found
McMaster University Ethics Board and subjects gave written that the Gaussian waveform produced the clearest sensation
consent to participate. of the head moving from side to side.

2.2. Stimuli 2.3. Apparatus

2.2.1. Auditory stimuli The vestibular stimulus was generated by a Good Vibrations
Both the familiarization and test stimuli were similar to those Engineering Ltd. Galvanic Vestibular Stimulation system
of Phillips-Silver and Trainor (2005) and were presented in (GVS). For the experimental group, an electrode was attached
a sound-attenuating chamber over a noise floor of 25 dB(A). to the mastoid process under each of the participant’s ears.
The vestibular stimulation was presented with pulses every For the control group, an electrode was attached to each of the
25.1 msec, so the tempos of the auditory rhythms were chosen participant’s elbows (Fig. 3). The electrode was a 2-inch round
to be multiples of this number. The training stimulus consisted Proflex CC carbon conductor electrode (Canadian Medical
of a snare drum timbre downbeat background presented every Products Ltd, F2020PF). This electrode material and shape was
1959 msec at 60 dB(A), a slapstick timbre microbeat background selected during pilot testing as providing the most comfort
presented every 326.5 msec at 50 dB(A) such that 6 microbeats and the best sensation of head movement.
occurred within each period defined by the downbeats, and an The auditory stimuli were created using Cakewalk with the
ambiguous familiarization rhythm pattern of interest at snare drum (#229) and slapstick (#244) voices on a Roland 64-
60 dB(A) that was superimposed on the background beats. The Voice Synthesizer Module, digitized with Cool Edit 2000 on
rhythm pattern of interest consisted of four snare drum a personal computer using an AOpen AW-840 4-channel PCI
sounds with stimulus onset asynchronies (SOAs) of 653–326.5– sound card, and presented by a Power Macintosh 7300/180
38 cortex 45 (2009) 35–43

Fig. 2 – The galvanic stimulus delivered to the vestibular system. The left panels represent the Gaussian-shaped electrical
stimuli at the 653 msec tempo (duple time) and the right panel the 979.5 msec tempo (triple time). For each panel the stimuli
are out of phase across the right (top panels) and left (bottom panels) ears.

computer through a Denon PMA-480R amplifier to two audi- and at 979.5 msec intervals (corresponding to the triple audi-
ological GSI speakers located inside a large Industrial Acous- tory stimulus rate) in a second condition. Subjects were asked
tics Co. sound-attenuating booth with a sound floor of to tap along with their perceived side-to-side head movement
25 dB(A). The sound booth was set up so that the participant on a response pad (EGI 200) at the precise times at which their
sat between the two speakers, and auditory stimuli were head was maximally displaced to the right and to the left.
always presented from both speakers. Subjects were monitored to ensure that they did not make any
The experiment was controlled from the GVS system, with overt head movements. Fig. 4 shows histograms of the onset-
software running on an IBM Thinkpad 760ED laptop. At the to-onset times of tapping intervals for two individual subjects
same time that a vestibular stimulus was sent to the subject, for the 979.5 tempo. As can be seen, responses centred on the
a trigger was sent to the Power Macintosh through a custom- vestibular inter-stimulus separation times, indicating that the
built interface consisting of an opto-isolator. Sounds were head was perceived to move at the same rate as the vestibular
presented from the Macintosh using a custom software stimulation. Fig. 4 also shows the phase offset between
program that ensured accurate timing. vestibular peak stimulation and the timing of the subjects’
taps. As can be seen, this phase varied considerably between
2.4. Pilot test of phase relations between vestibular subjects. Because we did not want to fatigue the vestibular
stimulation and perceived head movement responses of subjects by running them through a long pilot
procedure, a short test was developed to determine the
Pilot testing revealed that with vestibular stimulation of
alternating positive and negative Gaussian waveforms that
were out of phase between the ears, subjects experienced
a side-to-side movement of the head, even though the head
actually remained stationary. However, the phase relation
between the vestibular stimulation and the time at which the
head was perceived to be maximally displaced to the right and
to the left varied considerably from person to person. It was
essential that the time of perceived maximal head displace-
ment corresponded to the onsets of auditory beats in the main
experiment, so it was necessary to characterize the extent of
these individual differences and to compensate for them on
an individual basis. In order to do this, 5 pilot subjects were
given vestibular stimulation alone (i.e., no concurrent sound) Fig. 3 – Vestibular and control conditions. In the vestibular
in an isochronous rhythm sequence, with the peaks of the condition, galvanic stimulation was delivered to the
vestibular stimulation occurring at 653 msec intervals in one vestibular nerve (left panel) and in the control condition it
condition (corresponding to the duple auditory stimulus rate) was delivered to the elbows (right panel).
 cortex 45 (2009) 35–43 39

Fig. 4 – Distribution of tapping responses synchronized to illusory perceived head movements evoked by galvanic
stimulation illustrated by the responses of two representative subjects. Bin widths are 30 msec. Stimuli were delivered
using the pattern illustrated in Fig. 2 with a separation of 979.5 msec. The distributions of inter-tap intervals (left panels)
were centred close to the vestibular stimulation rate at 978 msec (SD [ 72) and 983 msec (SD [ 80) for each subject. Phase
relations (delay in msec between the stimulus and tap, right panels) showed variability across subjects, peaking at 200 msec
(SD [ 102) and 245 msec (SD [ 71) in subjects A and B, respectively.

approximate phase relation for each individual subject as auditory pattern as a waltz, and choose the triple auditory
described below. stimulus as sounding most like what they heard during
familiarization. Each of these phases is described in detail
2.5. Procedure below.

Subjects were first given a pre-test to determine their indi- 2.5.1. Individual phase shift measurement
vidual phase relation between vestibular stimulation and The electrodes were applied behind the ears in the experi-
their maximal perceived side-to-side head movement. They mental group and to the elbows in the control group. Subjects
were then given a questionnaire about their musical back- in the experimental group were told that they might feel their
ground. This was followed by a familiarization phase in which head move from side to side while they heard a repeating
the ambiguous auditory rhythm pattern was presented tone. Subjects in the control group were told that they might
concurrently with vestibular stimulation on every second feel a tingling in their elbows. Experimental group subjects
(duple familiarization condition) or on every third (triple were to determine whether the tone came before or after the
familiarization condition) beat of the ambiguous auditory time at which they perceived their head to be maximally
rhythm. In this phase, the control subjects experienced displaced to the right and left. Control group subjects were to
stimulation of their elbows instead of their vestibular nerve. determine whether the tone came before or after the tingling
Subjects were monitored to ensure that they did not make any they felt in their elbows. A 100 msec 60 dB pure tone at 500 Hz
overt head movements. Finally, subjects were given an audi- was played repeatedly with an SOA of 816 msec (half way
tory-alone test with the disambiguated duple and triple between the two possible stimulation rates used during the
auditory stimuli (no vestibular stimulation) and asked to experiment that followed) and the vestibular/elbow stimula-
chose which sounded most like what they had heard during tion was applied at the same rate for 30 sec. Initially, the tone
the familiarization phase. The entire experiment lasted less was presented 200 msec offset from the peak of the electrical
than 1 h. stimulation. Subjects were asked to indicate whether the
All subjects heard exactly the same auditory stimuli sound came before or after the point of maximal perceived
throughout the experiment. If there were multisensory head displacement/elbow tingling or whether the two stimuli
interactions between the vestibular and auditory systems, were concurrent. The phase was then adjusted until the
subjects who experienced duple vestibular stimulation sound and perceived head movement/elbow tingling were
should perceive the ambiguous auditory pattern as reported to be concurrent. Five possible phase shift values
a march, and hence choose the duple auditory stimulus as were used, 150, 170, 200, 230, 260, and 290, covering the range
sounding most like what they heard during familiarization. of phase values measured in the pilot study. Phase shifts used
On the other hand, those experiencing triple vestibular ranged between 170 and 260 msec with a mean of 205 msec
stimulation would be expected to perceive the ambiguous (SD ¼ 33.3).
40 cortex 45 (2009) 35–43

2.5.2. Familiarization
All subjects listened to one minute of the ambiguous auditory
familiarization rhythm. Subjects in the experimental condi-
tion received concurrent vestibular stimulation while those in
the control group received concurrent elbow stimulation.
Within each group, half received duple vestibular/elbow
stimulation (i.e., with an SOA of 653 msec, individually phase
shifted to be perceived as concurrent with every second beat
of the auditory pattern) and half received triple vestibular/
elbow stimulation (i.e., with an SOA of 979.5 msec, individu-
ally phase shifted to be perceived as concurrent with every
third beat of the auditory pattern). All subjects were tested
with eyes closed. Although it is possible that the vestibular
stimulation caused some nystagmus, such eye movements
are unlikely to influence perceived auditory rhythm directly.
The brain largely monitors such eye movements via an Fig. 5 – Results. Percentage of times subjects identified the
efference copy of the vestibular driving signal rather than feed ambiguous auditory pattern as duple for each of the four
forward proprioceptive input from eye muscles (Bridgeman conditions. Open bars after stimulation at the duple
and Stark, 1991). If the vestibular stimulation we used were to frequency, shaded bars after stimulation at the triple
have an effect on the interpretation of auditory rhythm frequency. Left pair of bars after galvanic stimulation of the
patterns, it would be even more remarkable if it were medi- vestibular system (see Fig. 3, left panel), right pair of bars
ated by eye movements than if it were mediated directly after control stimulation of the elbows (see Fig. 3, right
through the vestibular system. The control condition with the panel). The horizontal dotted line represents chance
electrodes on the elbows would also likely cause some small responding (50%). Error bars represent the standard error of
muscular activity. A lack of effect in this condition would the mean.
suggest that small muscle movements are insufficient to
influence auditory pattern perception.

2.5.3. Testing auditory-alone rhythm was chosen as most similar to what

Immediately following the training phase, the participant was was heard during familiarization, revealed only a significant
given eight two-alternative forced-choice trials, with auditory interaction between experimental condition and familiariza-
presentation alone. Each trial contained a duple and a triple tion stimulation, F(1, 19) ¼ 11.3, p < .003. As can be seen in
test stimulus. Both of the duple and triple test stimuli were Fig. 5, stimulation of the elbows (control condition) at either
identical to the auditory familiarization stimulus, except that frequency during familiarization did not affect whether
physical accents (see Section 2.2) were present on every subjects perceived the ambiguous auditory stimulus in duple
second beat for the duple test stimulus, and on every third or triple form in any systematic way, p > .4. On the other hand,
beat for the triple. The two stimuli on each trial were pre- vestibular stimulation during familiarization had a significant
sented in random order for one cycle or 1959 sec each, sepa- effect on whether subjects perceived the ambiguous auditory
rated by 1000 msec. On the first trial, the order was stimulus in duple (march) or triple (waltz) form, t(10) ¼ 6.3,
counterbalanced, so that half of the participants in each p < .0001. There were no significant correlations between
condition heard the duple rhythm first, and half heard the musical training and performance for either group.
triple rhythm first. Participants were instructed to choose
which of the two stimuli was the same as, or most similar to,
the rhythm they had heard in the training phase, and their 4. Discussion
verbal responses were recorded by the experimenter.
Previous studies have shown that physical movement plays
a critical role in entraining and disambiguating a musical
rhythm (Phillips-Silver and Trainor, 2005, 2007). The present
3. Results study shows that this effect is mediated through the vestibular
system. Galvanic stimulation of the vestibular system caused,
After vestibular entrainment with 653 msec intervals (corre- on average, 79% of perceptual judgements to move towards
sponding to the beats of the duple interpretation), 87.5 percent the auditory interpretation entrained by the vestibular stim-
of judgements of the rhythm of the ambiguous auditory ulation, whereas control stimulation caused no such shift in
stimulus were for the duple interpretation; after entrainment auditory pattern perception. It is well known that music
with 979.5 msec intervals (corresponding to the beats of the makes us move, but these studies show that the act of feeling
triple interpretation), 70.0 percent of judgements were for the a rhythm is an interactive process: hearing a rhythm evokes
triple interpretation. An ANOVA, with the independent vari- physical movement and the resulting vestibular stimulation
ables experimental condition (vestibular, elbow) and electrical also influences the auditory interpretation of the rhythm.
familiarization stimulation (duple, triple) and the dependent The lack of correlation between years of musical training
variable the proportion of responses in which the duple and size of the vestibular influence on audition, in
 cortex 45 (2009) 35–43 41

conjunction with previous findings of movement–auditory and Young, 2004), and the posterior parietal cortex (PPC)
interactions in infants (Phillips-Silver and Trainor, 2005), (Bremmer, 2005; Colby et al., 1993; Lewald et al., 2002; Lewis
suggests that vestibular influence on auditory processing and Van Essen, 2000; Schlack et al., 2005, 2002). Models of
might arise early in development and not depend on any rhythmic movement have largely focused on interactions
special experience. The vestibular system is a very primitive between PPC, cerebellum and prefrontal cortex (e.g., Todd
system that emerges early in both phylogeny and ontogeny et al., 2002). The cerebellum in particular has been implicated
and that determines the organization and development of the in the processing of auditory rhythm (Penhune et al., 1998;
other senses. Given the primal role of the vestibular system it Parsons, 2003; Griffiths, 2003; but see Molinari et al., 2005). The
is perhaps not surprising that it has such a fundamental cerebellum is a major recipient of vestibular input (e.g., Suzuki
influence on auditory perception. Phylogenetically the and Keller, 1982). The role of the cerebellum in timing has
vestibular system was the first sensory system to develop in been conceptualized not as a clock or counter but simply as
evolution (Walls, 1962) and ontogenetically it is the first the structure that provides the necessary circuitry for the
system to develop in the womb (Romand, 1992), suggesting sensory systems to extract temporal information and for
that a sense of orientation in the gravitational field is more the motor system to learn to produce a precisely timed
fundamental to perception than is vision and hearing. Indeed response (Penhune et al., 1998). The effect of vestibular stim-
vestibular experience appears to be fundamental even for ulation of auditory rhythm, then, is likely to provide a similar
social–emotional development in primates. Vestibular input enhancement to the auditory sensory signal on selected
through movement experience can partially ameliorate the strong beats as would an increase in auditory intensity.
detrimental effects of maternal separation in infant rhesus The ventral intraparietal area (VIP) of the PPC is of partic-
monkeys raised by cloth-covered surrogate mother cylinders ular interest as well because single cell recordings in the
(Mason and Berkson, 1975), as in Harlow’s famous experi- Macaque monkey show that it responds to both auditory
ments (Mason and Harlow, 1958). (Schlack et al., 2005) and vestibular (Colby et al., 1993; Brem-
An important role for the vestibular system, in addition mer et al., 2001; Schlack et al., 2002) input as well as to visual
to its fundamental role in providing orientation information, and somatosensory stimulation. The view of the PPC as an
arises because it is dynamically sensitive in the range of bio- integrator of acoustic and vestibular cues, together with
logical rhythms, especially those in the frequency range of evidence of area VIP as a site of multimodal neurons that code
locomotion (Wilson and Jones, 1979). Biological rhythms are for spatial perception and self-motion, offers an account of
pervasive and exist at many different temporal scales. Two of the auditory–vestibular connections that may underlie our
the most likely sources of biological rhythm at the time scale findings of multisensory interactions between movement and
of musical rhythm are locomotion and heart beat. The tempo the perception of auditory rhythm (see Phillips-Silver and
range of locomotion movements corresponds closely to the Trainor, 2008).
range over which a beat can be felt – pulse separations of The locus of vestibular–auditory interactions could be
between about 300 and 900 msec (e.g., Fraisse, 1982; Clarke, determined in a number of ways. fMRI studies comparing
1999) and heart rate varies from about 60–150 beats/min, strengths of activation across different regions for auditory
corresponding to a very similar range of 400–1000 msec. Both alone, vestibular alone, and combined auditory–vestibular
of these rhythms can be directly sensed by the vestibular stimulation would be informative. Studies of whether patients
system. The large physical motions associated with walking with lesions in different auditory–vestibular convergence sites
and running of course excite the vestibular system, but even experience the influence of vestibular stimulation on auditory
the tiny movements of the head caused by the pulse are also rhythm processing would also give information about which
detected (Crawford, 1952). Developmentally, both of these brain regions are critical. Finally, if vestibular–auditory effects
biological rhythms are experienced by the fetus as the mother are cortically mediated, no multisensory effect would be
locomotes through the environment. But at what stage of expected in infants two months of age and younger, as the
development, and where in the nervous system does vestib- auditory cortex is not mature enough at this stage to support
ular input connect with auditory rhythmic experience? complex processing (Moore and Guan, 2002).
Although the auditory and vestibular end organs are
anatomically close and functionally similar, sounds must be
at least 95 dB before they directly stimulate the vestibular
system (Todd and Cody, 2000). As the sounds in this experi- 5. Conclusions
ment are much quieter than this, the auditory–vestibular
interaction that we have demonstrated must occur more Musical rhythm patterns elicit physical movement from head
centrally. There is a potential concern that our galvanic bobbing and foot tapping to all-out dancing. Our previous
stimulation might have had a direct effect on the auditory work showed that movement of the body can influence the
portion of the VIII nerve (Bucher et al., 1998). However, this is auditory perception of the metrical structure of rhythm. The
very unlikely because the currents used were very small and present paper demonstrates that this effect is mediated by
none of our subjects reported hearing any sounds. The rele- vestibular stimulation and can be recreated in the absence of
vant auditory–vestibular interactions are more likely to occur physical movement by artificially stimulating the vestibular
more centrally. nerve. Most likely, several regions of vestibular–auditory
Recent evidence indicates that auditory and vestibular convergence are involved in some way in rhythm processing
information may interact at many levels of the nervous and it will be for future research to discover the specific role of
system, including the dorsal cochlear nuclei (DCN) (Oertel each region.
42 cortex 45 (2009) 35–43

lobe of the macaque monkey. Journal of Computational

Acknowledgements Neurology, 428: 112–137, 2000.
Mars F, Vercher JL, and Popov K. Dissociation between subjective
This research was supported by grants from NSERC (Canada) vertical and subjective body orientation elicited by galvanic
to Laurel Trainor and Laurence Harris. We thank Jessica vestibular stimulation. Brain Research Bulletin, 65: 77–86, 2005.
Mason WA and Berkson G. Effects of maternal mobility on the
Phillips-Silver for discussions of the stimuli and Lisa Hotson
development of rocking and other behaviors in rhesus
for organizational assistance.
monkeys: a study with artificial mothers. Developmental
Psychobiology, 8: 197–211, 1975.
Mason WA and Harlow HF. Formation of conditioned responses
references in infant monkeys. Journal of Comparative and Physiological
Psychology, 51: 68–70, 1958.
Molinari M, Leggio MG, De Martin M, Cerasa A, and Thaut M.
Arom S. African Polyphony and Polyrhythm. Cambridge: Cambridge Neurobiology of rhythmic motor entrainment. Annals of the
University Press, 1991. New York Academy of Science, 999: 313–321, 2003.
Baruch C and Drake C. Tempo discrimination in infants. Infant Molinari M, Leggio MG, Filippini V, Gioia MC, Cerasa A, and
Behaviour and Development, 20: 573–577, 1997. Thaut MH. Sensorimotor transduction of time information is
Bremmer F. Navigation in space- the role of the macaque ventral preserved in subjects with cerebellar damage. Brain Research
intraparietal area. The Journal of Physiology, 566: 29–35, 2005. Bulletin, 67: 448–458, 2005.
Bremmer F, Schlack A, Duhamel J-R, Graf W, and Fink GR. Space Moore JK and Guan YL. Cytoarchitectural and axonal maturation
coding in primate posterior cortex. Neuroimage, 14: S46–S51, 2001. in human auditory cortex. Journal of the Association for Research
Bridgeman B and Stark L. Ocular proprioception and efference in Otolaryngology, 2: 297–311, 2002.
copy in registering visual direction. Vision Research, 31: 1903– Oertel D and Young ED. What’s a cerebellar circuit doing in the
1913, 1991. auditory system? Trends in Neuroscience, 27: 104–110, 2004.
Bucher SF, Dieterich M, Wiesmann M, Weiss A, Zink R, Yousry TA, Parsons LM. Exploring the functional neuroanatomy of music
et al. Cerebral functional magnetic-resonance-imaging of performance, perception, and comprehension. In Peretz I, and
vestibular, auditory, and nociceptive areas during galvanic Zatorre RJ (Eds), The Cognitive Neuroscience of Music. New York,
stimulation. Annals of Neurology, 44: 120–125, 1998. USA: Oxford University Press, 2003: 247–268.
Buys E. Beitrag zum studium des galvanischen nystagmus mit Patel AD, Iversen JR, Bregman MR, Schultz I, and Schulz C.
hilfe der nystagmographie. Mschr Ohrenheilk, 43: 801–803, 1909. Investigating the human-specificity of synchronization to
Clarke EF. Rhythm and timing in music. In Deutsch D (Ed), The music. Presented at the International Conference on Music
Psychology of Music. 2nd ed. San Diego: Academic Press, 1999: Perception and Cognition, Sapporo, Japan, August 2008.
473–500. Penhune VB, Zattore RJ, and Evans AC. Cerebellar contributions to
Colby CL, Duhamel JR, and Goldberg ME. Ventral intraparietal motor timing: a PET study of auditory and visual rhythm
area of the macaque: anatomic location and visual response reproduction. Journal of Cognitive Neuroscience, 10: 752–765,
properties. Journal of Neurophysiology, 79: 126–136, 1993. 1998.
Crawford J. Living without a balancing mechanism. The New Phillips-Silver J and Trainor LJ. Feeling the beat: movement
England Journal of Medicine, 246: 458–460, 1952. influences infant rhythm perception. Science, 308: 1430, 2005.
Cross I. Music, cognition, culture and evolution. In Peretz I, and Phillips-Silver J and Trainor LJ. Vestibular influence on auditory
Zatorre R (Eds), The Cognitive Neuroscience of Music. New York: metrical interpretation. Brain and Cognition, 67: 94–102, 2008.
Oxford, 2003: 42–56. Phillips-Silver J and Trainor LJ. Hearing what the body feels:
Cutietta RA. Raising Musical Kids: A Guide for Parents. New York: auditory encoding of rhythmic movement. Cognition, 105:
Oxford University Press, 2001. 533–546, 2007.
Drake C, Penel A, and Bigand E. Tapping in time with Repp BH. Sensorimotor synchronization: a review of the tapping
mechanically and expressively performed music. Music literature. Psychonomic Bulletin and Review, 12: 969–992, 2005.
Perception, 18: 1–23, 2000. Romand R (Ed), Development of Auditory and Vestibular Systems.
Fraisse P. Rhythm and tempo. In Deutsch D (Ed), The Psychology of New York: Elsevier, 1992.
Music. New York: Academic Press, 1982: 149–180. Schachner A, Brady TF, Pepperberg IM, and Hauser MD.
Goldberg JM, Smith CE, and Fernandez C. Relation between Entrainment to auditory rhythms in vocal learning bird
discharge regularity and responses to externally applied species. Presented at Neurosciences and Music III, Montreal,
galvanic currents in vestibular nerve afferents of the squirrel Canada, 2008.
monkey. The Journal of Neurophysiology, 51: 1236–1256, 1984. Schlack A, Hoffman KP, and Bremmer F. Interaction of linear
Griffiths TD. The neural processing of complex sounds: Musical vestibular and visual stimulation in the macaque ventral
brain substrates. In Peretz I, and Zatorre RJ (Eds), The Cognitive intraparietal area (VIP). European Journal of Neuroscience, 16:
Neuroscience of Music. New York, USA: Oxford University Press, 1877–1886, 2002.
2003: 168–177. Schlack A, Sterbing-D’angelo SJ, Hartung K, Hoffmann K-P, and
Jaques-Dalcroze E. Method of Eurhythmics: Rhythmic Movement, vol. 1. Bremmer F. Multisensory space representations in the
London: Novello and Co, 1920. macaque ventral intraparietal area. The Journal of Neuroscience,
Juntunen ML and Hyvönen L. Embodiment in musical knowing: 25: 4616–4625, 2005.
how body movement facilitates learning with Dalcroze Suzuki DA and Keller EL. Vestibular signals in the posterior
Eurhythmics. British Journal of Music Education, 21: 199–214, 2004. vermis of the alert monkey cerebellum. Experimental Brain
Lerdahl F and Jackendoff R. A Generative Theory of Tonal Music. Research, 47: 145–147, 1982.
Cambridge: MIT Press, 1983. Todd NP, Cousins R, and Lee CS. The contribution of
Lewald J, Foltys H, and Töpper R. Role of the posterior parietal anthropometric factors to individual differences in the
cortex in spatial hearing. The Journal of Neuroscience, 22: 1–5. perception of rhythm. Empirical Musicology Review, 2, 2007.
RC207, 2002. Todd NP, Lee CS, and O’Boyle DJ. A sensorimotor theory of
Lewis J and Van Essen DC. Corticocortical connections of visual, temporal tracking and beat induction. Psychological Research,
sensorimotor and multimodal processing areas in the parietal 66: 26–39, 2002.
 cortex 45 (2009) 35–43 43

Todd NPM. The kinematics of musical expression. Journal of the Walls GL. The evolutionary history of eye movements. Vision
Acoustical Society of America, 97: 1940–1949, 1995. Research, 2: 69–79, 1962.
Todd NPM and Cody FW. Vestibular responses to loud dance Wardman DL, Taylor JL, and Fitzpatrick RC. Effects of galvanic
music: a physiological basis of the ‘‘rock and roll threshold’’? vestibular stimulation on human posture and perception
Journal of the Acoustical Society of America, 107: 496–500, 2000. while standing. The Journal of Physiology, 551: 1033–1042, 2003.
Trainor LJ. Do preferred beat rate and entrainment to the beat Wilson VJ and Jones GM. Mammalian Vestibular Physiology. New
have a common origin in movement? Empirical Musicology York: Plenum, 1979.
Review, 2: 17–20, 2007. Zink R, Steddin S, Weiss A, Brandt T, and Dieterich M. Galvanic
Wallin NL, Merker B, and Brown S. The Origins of Music. vestibular stimulation in humans – effects on otolith function
Cambridge, MA: MIT Press, 2000. in roll. Neuroscience Letters, 232: 171–174, 1997.