Sensorimotor Control

Stefan Glasauer
Zentrum fr Sensomotorik, Klinische Neurowissenschaften
Ludwig-Maximilians-Universitt Mnchen

History
the spirits proceed into the nerves. Once there they have the power to
change the shape of the muscles the nerves end up in, and in this way to make all
the parts of the body move.
its muscles and tendons are like the engines and springs used to move them
Ren Descartes (1596-1650), Treatise On The Human Being (posthumous)
at the same instant, in which the spark shot, in each limb really violent muscle
contractions were expressed, as if the animal preparation was affected by tetanus.
Luigi Galvani (1737-1798), De viribus electricitatis (1791)
There is the coordination which a reflex action introduces when it makes an effector
organ responsive to excitement of a receptor
Charles Scott Sherrington (1857-1952), The Integrative Action of the Nervous
System (1906)
Coordination is overcoming excessive degrees of freedom of our movement organs,
that is, turning the movement organs into controllable systems.
Nikolai Alexandrowitsch Bernstein (1896-1966)

History
Man muss also von dem Auge nicht ganz dieselbe Prcision erwarten, wie von
einem physikalischen Apparate, wenn auch normale Augen unter gewhnlichen
Bedingungen ziemlich genau dem Dondersschen und Listingschen Gesetze
folgen.
Hermann von Helmholtz, Handbuch der Physiologischen Optik (1867)
Wir fragen nicht nach der Beziehung zwischen einer gegebenen Afferenz und der
durch sie bewirkten Efferenz, also nach dem ,,Reflex", sondern gehen umgekehrt
von der Efferenz aus und fragen: was geschieht im ZNS mit der yon dieser
Efferenz ber die Effektoren und Rezeptoren verursachten Afferenz, die wir
die ,,Reafferenz" nennen wollen?
... das Auge unterscheidet eigene und Umweltbewegungen zwar nicht, aber das
Tier ... unterscheidet sie wohl.
Erich von Holst & Horst Mittelstaedt, Das Reafferenzprinzip (1950)

The most appropriate discussion is in the form of a theory which accounts at once
for all the reported observations. Such a theory is without merit if its predictions
are not quantitatively accurate so the schema must be represented by a system
of differential equations ...
David A. Robinson, The mechanics of human saccadic eye movement (1964)

Movements

Kandel et al. Fig. 33-5

Kandel et al. Fig. 33-4

Invariant features and motor programs

Kandel et al. Fig. 33-10

Curvature and Velocity

V(t)=kg(R(t))=kR(t)1/3
with A(t)=V(t)/R(t)
and C(t)=1/R(t)
A(t)=kf(C(t))=kC(t)2/3
V: tangential velocity
R: instantaneous radius
A: angular velocity
C: curvature

Lacquaniti, Terzuolo & Viviani 1983

The two-third power law

Trajectories and profiles

Flash & Hogan 1985

Trajectories are straight lines, velocity profiles are bell-shaped

Minimum-jerk model
Find the function x(t) that minimizes the cost
General solution

etc.

Symmetric velocity profiles, straight trajectories (Flash & Hogan 1985)

Feedback and feedforward control

Kandel et al. Fig. 33-1

Kandel et al. Fig. 33-2

Motor actions mostly require both types of control

Spinal feedback control

Purves et al. Fig 16.9
Shadmehr & Wise 2005

The local spinal feedback control circuit

Theory of motor control

Krding & Wolpert 2006

How do we move and why do we move the way we do?

Controller

Krding & Wolpert 2006

Many movements are ballistic: feed-forward control

Minimum-variance model
Harris & Wolpert 1998

Harris & Wolpert 1998

Shadmehr & Krakauer 2008

neural commands have signal-dependent noise

minimize the total positional variance during the immediate post-movement period
total movement duration is pre-fixed

Fitts law

Shadmehr & Krakauer 2008

Movement time, speed, and accuracy are related

Kandel et al. Fig. 33-10

Predicting Fitts law

Minimum-variance model (Harris & Wolpert
1998)

Harris & Wolpert 1998

1) neural commands have signal-dependent noise

whose standard deviation increases linearly
with the absolute value of the neural control
signal
2) the optimal trajectory is the one that
minimizes the total positional variance during
the immediate postmovement period, given a
prefixed movement duration

Constrained minimum-time model

(Tanaka, Krakauer & Qian 2006)

Tanaka, Krakauer, Qian 2006

1) neural commands have signal-dependent noise

whose standard deviation increases linearly
with the absolute value of the neural control
signal
2) minimize movement duration under the
constraint of meeting an accuracy criterion
(bounded variance over postmovement
period)

Minimum-Variance and Minimum-Time model prediction

Estimator

visual

proprioceptive

Krding & Wolpert 2006

Where is my limb?

Optimal integration
Bays & Wolpert 2007

Optimal integration in the estimation of hand position

Optimal integration
Krding & Wolpert 2006

Optimal integration using a-priori knowledge

Estimation and active movement

visual

proprioceptive

Krding & Wolpert 2006

Where is my limb?

Reafference principle

Estimate of
external influence

Bays & Wolpert 2007

Efference copy is used to distinguish

self-generated sensory input from external perturbations
(von Holst & Mittelstaedt 1950)

Internal models

Davidson & Wolpert 2005

Glasauer 2007

Internal models mimicking the physical properties of real world systems

can be used to predict sensory feedback
or to generate motor commands

Theory of motor control

Krding & Wolpert 2006

Together: optimal feedback control

Utility or Cost

Bays & Wolpert 2007

What to do for minimal cost?

Theory of motor control

Scott 2008

Together: optimal feedback control

Brain control of movement

Summary
Kandel et al. Fig. 18-9

Computational Neuroanatomy

Shadmehr & Krakauer 2008

A schematic model for generating goal directed movements

Computational Neuroanatomy
Shadmehr & Krakauer 2008

A schematic model for generating goal directed movements

Own work: Combined eye-head gaze shift

Slow motion (10x)

Natural goal-directed gaze shifts involve eye and head

movements
Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Stereotypical movements

eye

head

gaze

Observation: stereotypical movements

Hypothesis: the choice is optimized
Question: if thats the case, what is the cost function?
Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Optimization criteria
Possible optimization criteria
Minimum jerk

Minimum torque-change

(Flash & Hogan 1985)

(Uno, Kawato, Suzuki 1989)

Minimum variance

Constrained minimum-time

(Harris & Wolpert 1998)

(Tanaka, Krakauer, Quian 2006)

Reward discounting

Minimum effort

(Shadmehr et al. 2010)

(e.g., Kardamakis & Moschovakis 2009)

Minimum consequence of motor noise

(van Beers 2008)

Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Duration-dependent variability

The mix of signal-dependent and constant motor noise

explains why there is an optimal movement duration
Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Optimization criteria
Harris & Wolpert 2006

Van Beers 2008

Shadmehr et al. 2010

Predictions

Minimizing the consequences of signal-dependent and constant motor noise

explains the features of normal eye-head gaze shifts
Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Test: changing head plant inertia

model prediction

experimental data

Model simulates eye-head gaze shifts with increased head moment of inertia
Saglam M, Lehnen N, Glasauer S: Optimal control of natural eye-head movements minimizes the impact of noise. J Neurosci 2011

Adapting to changes of head plant inertia

no weight
optimal command

with weight
old command

with weight
optimal command

Unmodified head command leads to major head oscillations

Head oscillations, even when compensated for, increase
post-movement variability

Combined gaze shifts in patients

healthy

bilateral vestibular deficit

cerebellar atrophy

The vestibular sensors and the cerebellum

are critically involved in head motor control
Lehnen N, Bttner U, Glasauer S: Head movement control during head-free gaze shifts. Prog Brain Res 171: 331-34, 2008

Adapting to the weight

bilateral vestibular deficit

cerebellar atrophy

cost

healthy

Healthy subjects rapidly adapt to the weighted condition.

Bilateral vestibular patients and patients with cerebellar atrophy
show no or only slow adaptation.

Summary
Combined eye-head movements minimize the
consequences of noise
Stereotypic movements as a result of optimization
Note: this does not imply missing flexibility, but
context-dependence
Short-term flexibility of sensorimotor control to
adapt to the properties of the world/body
Adaptation requires sensory feedback
and an intact cerebellum