Motor Cognition and Internal Models

1
Motor Cognition and Internal Models
Glenda Lassi University of Siena May 2004
2
ENERGY
SENSORY SYSTEM
NEURAL SIGNAL
MOTOR SYSTEM
CONTRACTILE FORCE
3
MOTOR CONTROL process of transforming sensory
inputs into consequent motor outputs

• The transformation
• from motor commands
• to sensory consequences
• is governed by the
• - physics of the environment
• - musculoskeletal system
• - sensory receptors.

• The transformation
• from sensory signals
• to motor commands
• is determined by
• - processes within the central
  
  
  
• nervous system.

Much of the complexity of our behavior arises as
a simple coupling of these two transformations.
INTERNAL MODEL the REPRESENTATION of
transformations
4
Hypothesis of the look up table the brain learns
single associationsbetween movements instances
and motor commands.
State and context hypothesis state changes
rapidly and continuously within a movement set
of activations of groups of muscles (synergies),
position and velocity of the hand Context
discrete and slow changes identity of a
manipulated object, mass of the limb. Wolpert
and Ghahrami, 2000
5
THE SENSORIMOTOR LOOP
Wolpert and Ghahramani, 2000
6
An internal model is used either- to predict
the movement consequences of a motor commands
(forward model)- to determine the motor
commands needed to achieve a desired movement
trajectory (inverse model).
7
Behavioral evidence psychophysics

• Motor Learning mastering and adapting
  sensorimotor transformations
• That can be broken into kinematic and dynamic
  transformations
• Kinematic transformation to transform a target
  position into a command to the skeletal system to
  move the hand i.e. to convert between coordinate
  systems
• Dynamic transformation relate motor commands to
  the motion of the system in the reaching task
  here considered, the forces applied changed the
  system without changing the kinematics.
• Shadmehr Mussa-Ivaldi, 1994.

8
Task center-out visually guided instructed
reaching movements. A handle attached to a
two-link robotic manipulandum had to be moved
through a planar workspace.
9
TASK TIMING
Seconds
Target hold
Movement
Delay
Center hold
10
Each experimental session consisted of three
behavioral epochs Baseline, Force field,
Washout.
Trajectories
Shadmehr Mussa-Ivaldi, 1994.
Aftereffects in the washout Hooks oriented in
opposite directions.
11
Correlation coefficient of the performance in the
Force field epoch.
Shadmehr Mussa-Ivaldi, 1994.
12
Velocity profiles
Baseline
Early Force Field
Late Force Field
Shadmehr Mussa-Ivaldi, 1994.
13
This study with human subjects show that- the
central nervous system initially specifies a
desired trajectory of the hand and then uses an
internal model of the limbs dynamics to produce
torques appropriate for moving the hand along
this desired trajectory. - new internal models
can be acquired when subjects adapt to a new
dynamic environment.- most of the development
of this internal model took place early in the
training period (as shown by the correlation
coefficient).
14
Evidence from Neurophysiology Further
investigation has been quested with monkeys in
order to study the neural correlate.
15
The same task
16
And the same psychophysics, e.g. trajectories
17
Cellular changes in the circuitry of the primary
cortical motor area of the monkey during the
acquisition of a motor skill.
Change in the mechanical environment with which
the hand interacted.
The neural representation of the arm has to
develop a new model.
18
Preferred directiontuning curve cells were
related to the direction of the arms movements
19
KINEMATIC CELL
BASELINE FORCE FIELD WASHOUT
Li et al., 2001
20
DYNAMIC CELL
BASELINE FORCE FIELD WASHOUT
Li et al., 2001
21
1. The newly tuned adapted cells may reflect the
formation of an internal model to handle the new
dynamics generated by the force fields.2.
Distinctive population of neurons are transformed
by the exposure to learning.Likewise other
areas have been shown learning dependent neuronal
activity in the supplementary motor cortex.The
comparison of the dynamics-related activity and
plastic changes in the supplementary motor area
(SMA) with M1 showed that neurons in SMA process
the dynamics of the upcoming movement during the
instructed delay at a time when only few M1
neurons are directionally tuned and before
dynamics-related activity can be recorded in M1
(Padoa-Schioppa et al. 2002, 2004).
22
Another way to analyze the neuronal
activityraster plots and perievent histograms
23
Mirror Neurons
Gallese et al., 1996
24
EMGs
Gallese et al., 1996
An interpretation of mirror neurons function is
that their discharge generates an internal
representation of the movement. This
representation may have different complementary
functions, among which motor learning and the
understanding of meaning of the observed action.
25
Reaching and grasping in primates depend on the
coordination of neural activity in large
frontoparietal ensembles. Nicolelis group have
demonstrated that primates can learn to reach and
grasp virtual objects by controlling a robot arm
through a closed-loop brain-machine interface
that uses multiple mathematical models to extract
several motor parameters (i.e. hand position,
velocity, gripping force, and the EMGs of
multiple arm muscles.)
Multichannel chronic implants
Implants areas included the dorsal premotor
cortex, supplementary motor area, the primary
motor cortex and parietal cortex.
26
TASK
Carmena et al., 2003
27
EMGs
Carmena et al., 2003
28
Implicit or explicit?
A well practiced motor skill is usually
considered procedural knowledge.In the
center-out task- Human subjects were not given
any instructions regarding the trajectories with
which the targets should have been reached. The
reward was monkeys motivation to learn.
Implicit learning?
29
Aknowledgements

• Bizzi lab, Brain and Cognitive Dept., MIT
• Prof. Castelfranchi and Prof. Carli for real and
  virtual conversations on the topic.