Chunking by intermittent motor control in music
Rolf Inge Gody
Department of Musicology, University of Oslo, Norway
r.i.godoy@imv.uio.no
ABSTRACT
1. Background
Various theories in music perception and cognition, from clas-
sical gestalt theory to more recent experimental work and
data-driven modeling, have contributed to our understanding
of chunking in musical experience (see Gody 2008 for a
summary). But our own research on music-related body mo-
tion has singled out intermittency in motor control, i.e. a basi-
cally discontinuous and point-by-point control scheme, as an
essential factor in chunking. Classical motor control theories
with claims of so-called closed loop continuous feedback have
in recent years been challenged by models suggesting inter-
mittent control, manifest in so-called open loop and prepro-
grammed motor commands, because continuous feedback
loops are thought to be too slow for many highly demanding
tasks (Loram et al. 2011, 2014). Various findings in human
motor control, i.e. the so-called psychological refractory
period, principles of posture-based motion control, of action
hierarchies, of goal-directed behavior, etc., and our own re-
search on music-related body motion, seem to converge in
suggesting the existence of intermittent motor control for
chunking at the short-term timescale of very approximately
0.5 seconds duration, all in all suggesting that there are what
could be called quantal elements in musical experience
(Gody 2013).
Our interest in chunking stems from several years of re-
search and reflections on what Pierre Schaeffer called sonic
objects (Schaeffer 1966), i.e. holistically perceived fragments
of musical sound in the approximately 0.5. to 5 seconds range,
and which we with Schaeffer regard as one of the most essen-
tial elements of musical experience (Gody 2006). However,
this also leads to questions of what are the criteria for the con-
stitution of such sonic objects, and of how they are perceived
in musical contexts. It seems that various features in the audi-
tory signal (e.g. shifts between sound and silence, contrasts in
pitch registers, contrasts in timbre, or the occurrence of re-
peated patterns, etc.) may be fairly robust cues for chunking,
but in many cases, these cues may be less salient (e.g. be
weak, competing, masked, or even temporarily absent), re-
quiring the use of other perceptual schemas. Adopting a
so-called motor theory perspective (Galantucci, Fowler, and
Turvey 2006) on music perception in general and chunking in
particular, we believe images of sound-producing body mo-
tion may be effective in chunking continuous streams of
sound into somehow meaningful units in musical experience
(Gody 2001, 2003, 2013, 2014).
The basic tenet here is that schemas of sound-producing
body motion are projected onto whatever musical sound it is
that we are hearing. This in turn means recognizing a number
of constraints of body motion and motor control, something
that suggests an unequal distribution of attention and effort in
musical experience. Although the underlying neurocognitive
processes of our organism may very well be continuous
(Spivey 2008), the main argument for intermittency is that
control takes time, and that continuous feedback is just not
feasible (Loram et al. 2011, 2014). The various constraints of
motor control also fit well with the idea of action gestalts as
optimal for our organism (Klapp and Jagacinski 2011).
This means that we need a novel theory of chunking based
on the idea of intermittency in motor control. Such a theory
should also be capable of accommodating the (seemingly pa-
radoxical) coexistence of continuity and discontinuity in mu-
sical experience, something we shall argue is possible by rec-
ognizing intermittent motor control of chunks, yet recognizing
that chunks unfold continuously in time, and furthermore, that
the concatenation of several chunks in succession may give
rise to subjective sensations of continuity in musical experi-
ence.
2. Aims
The aim of the presentation is to demonstrate a novel model
of intermittent motor control in musical performance and its
consequences for chunking in music perception, based on
convergent findings in several domains and our own research
on music-related body motion, in particular our findings on
coarticulation in music-related body motion, i.e. the fusion
and the contextual smearing of otherwise distinct motion and
sound events into holistically perceived chunks. The main
claims of the model are as follows:
That chunking is based on a confluence of production
constraints
That anticipatory cognition is manifest in sound-produc-
ing (and also sound-accompanying) body motion
That we have fusion by coarticulation in musical perfor-
mance
That we need a more rigorous analysis and differentiation
of timescales involved in the perception and production of
music, and notably so, of both sound and corresponding
body motion.
This last point means that we in particular need to differenti-
ate the features of the sub-0.5 seconds timescale, what could
be called the delta timescale, and which typically includes the
psychological refractory period, as well as coarticulation and
thresholds for so-called phase-transitions, meaning the fusion
of singular events into superordinate events (or conversely,
the fission of superordinate events into singular events), all
dependent on the rate (the speed or temporal distance) and du-
ration of events (Gody 2014).
Furthermore, the model should establish chunking in music
as conditioned by goal-directed behavior, similar to goal-di-
rected body motion in various everyday contexts, and as fo-
cused on certain salient postures (Rosenbaum et al. 2007). We
could say that chunking is focused on what we call key-pos-
tures, meaning salient shapes and positions of the effectors
(fingers, hands, vocal tract, etc.), and that there is continuous,
coarticulated motion between these key-postures. Furthermore,
we think these key-postures are found at salient moments in
time, typically at downbeats or other accented points in the
music, at what we call goal-points. In music, we think of these
key-postures at goal-points as surrounded by prefixes (tra-
jectories to the key-postures) and suffixes (trajectories from
the key-postures), with overlap of suffixes and prefixes in se-
quences of chunks producing the effect of continuous motion.
The basic tenet here is that all motion takes time, i.e. that
instantaneous displacement of the effectors is impossible, and
hence, that there always will be contextual smearing of mo-
tion leading to coarticulation within any chunk (Gody 2013,
2014, in press).
Also, we think we can detect these goal-points and key-
postures in motion capture data of musicians' sound-produc-
ing body motion, as points associated with maximal accele-
ration and velocity reversals (Gody 2013, 2014), points in
time that we in turn understand as the basis for intermittent
control.
3. Main Contribution
The main contribution of the presentation is a novel theory of
chunking in music based on constraints of human motor cog-
nition, in particular on the need for anticipatory motor control,
its manifestation in intermittent control, and the associated
coarticulation and contextual smearing of motion and sound
within any chunk. This will all come together in a conceptual
model that will include:
1. Intermittent motor control, focused on key-postures at
goal-points
2. Continuity, coherence, fusion, integration, etc. within any
chunk because of continuous motion trajectories with
coarticulation
One attractive feature of this model is that the duality of dis-
continuity and continuity is built into the model: the within
chunk coherence is due to a single discontinuous impulse, i.e.
to the intermittent, so-called serial ballistic (Loram et al.
2014), control impulse, focused on one goal-point. The evi-
dence for this model so far is in various other cognitive sci-
ence research on human motor control, as well as the inter-
pretation of our own empirical findings:
Motion capture data on sound-producing motion tra-
jectory shapes and their corresponding velocity and
acceleration features suggest the existence of goal-points
Motion capture data clearly indicate the coarticulatory
shaping of sound-producing motion trajectories.
4. Implications
One consequence of recognizing intermittent motor control is
to modify our basic understanding of chunking in music so
that we see chunking as just as much dependent on intermit-
tent motor control schemas as on purely acoustic features.
Specifically, such an understanding of chunking may:
Account for chunking difficult to induce by purely sig-
nal-driven schemes
In focusing on goal-points, shift our attention towards
chunk centroids from chunk boundaries, i.e. boundaries
are seen as secondary to centroids, and this model may
also accommodate/tolerate fuzzy boundaries.
Account for the internal coherence and cohesion within
the chunk as the result of one singular and intermittent
control impulse
But clearly, we also need to:
Develop experimental methods for more systematic
studies of intermittency in practice
EMG measurements to indicate temporal distribution of
effort in sound-producing body motion
Computational models (i.e. so-called reverse engineering)
of intermittent control in sound-producing body motion
5. Keywords
Intermittency, motor control, action gestalts, coarticulation,
contextual cohesion
REFERENCES
Galantucci, B., Fowler, C. A., and Turvey, M. T. (2006). The motor
theory of speech perception reviewed. Psychonomic Bulletin &
Review, 13 (3), 361-377.
Gody, R. I. (2001). Imagined Action, Excitation, and Resonance. In
R. I. Godoy and H. Jorgensen (Eds.), Musical Imagery (pp.
239-252). Lisse: Swets and Zeitlinger.
Gody, R. I. (2003). Motor-mimetic Music Cognition. Leonardo
36(4), 317-319.
Gody, R. I. (2006) Gestural-Sonorous Objects: embodied exten-
sions of Schaeffers conceptual apparatus. Organised Sound,
11(2), 149-157.
Gody, R. I. (2008). Reflections on Chunking in Music. In A.
Schneider (Ed.), Systematic and comparative musicology:
concepts, methods, findings (pp 117-132). Frankfurt: Peter
Lang.
Gody, R. I. (2013). Quantal Elements in Musical Experience. In. R
Bader (Ed.), Sound Perception Performance. Current Re-
search in Systematic Musicology, Vol. 1 (pp. 113-128). Berlin,
Heidelberg: Springer.
Gody, R. I. (2014). Understanding Coarticulation in Musical
Experience. In M. Aramaki, M. Derrien, R. Kronland-Martinet &
S. Ystad (Eds.), Sound, Music, and Motion. Lecture Notes in
Computer Science Volume 8905 (pp. 535-547). Berlin: Springer.
Gody, R. I. (in press). Key-postures, trajectories and sonic shapes.
In D. Leech-Wilkinson and H. Prior (Eds.), Music & Shape. Ox-
ford: Oxford University Press.
Klapp, S. T., and Jagacinski, R. J. (2011). Gestalt Principles in the
Control of Motor Action. Psychological Bulletin, Vol. 137, No. 3,
pp. 443462.
Loram, I. D., Gollee, H., Lakie, M., and Gawthrop, P. J. (2011).
Human control of an inverted pendulum: Is continuous control
necessary? Is intermittent control effective? Is intermittent
control physiological? The Journal of Physiology, 589, 2, pp.
307-324.
Loram, I. D., van De Kamp, C., Lakie, M., Gollee, H., and Gawthrop,
P. J. (2014). Does the motor system need intermittent control?
Exercise and Sport Science Review, Vol. 42, No. 3, pp. 117-125.
Rosenbaum, D., Cohen, R. G., Jax, S. A., Weiss, D. J., and van der
Wel, R. (2007). The problem of serial order in behavior:
Lashley's legacy. Human Movement Science, 26(4), 525-554.
Schaeffer, P. (1966). Trait des objets musicaux. Paris: ditions du
Seuil.
Spivey, M. (2008). The Continuity of Mind. New York: Oxford
University Press.