ch 7- sensorimotor control models Flashcards
Motor program theory
motor program= NOT a sequence of commands
–> if it were a sequence it would require a search through the “database” of possible sequences to match it to the task
motor command
specific info about timing and force sent from the brain to muscles
Motor program
a procedure/formula for generating motor commands that produce related movements: “class of actions”
-ex. throwing can be done underhand, overhand, sidearm etc. (all have the same goal)
A generalized motor program (GMP) controls
a CLASS of actions rather than specific movements/sequences
Components of the GMP
- Invariant features= features of the command that remain almost the same each time
- Parameters= ways that the command can be varied
Invariant features
-features remain almost the same.
-all motor skills within a class of action will have similar features
-ex. saccades: direction, amplitude may vary but kinematic profiles remain the same
Parameters
-variants
-can be added to invariant features to meet specific movement demands of a situation
Motor schema response concept
provides rules that decide the performance of a skill in a given situation/context
-selects parameters for motor program
-ex. you have an idea of what walking entails (moving your legs)
schema
a rule/set of rules that provide the basis for a decision
-identifies the class of action
-ex. you have seen many dogs, they range in appearance, you have developed a schema/set of rules for identifying a dog
Motor equivalence
varying parameters produce the same output
motor equivalence:
writing name w different body parts example
parameter= using different body parts
invariant features= aspects of the written text (the shape of letters, the general shape of words)
does the nervous system store sequences of commands to be sent out to spinal circuitry?
NO!!! there are wayyy too many sequences to store.
A motor program (rules for generating motor commands) can be stored.
motor programs are planned before the movement. can they be adjusted if the task requires it?
YES! sensory feedback can be used
-sensory feedback: stretch receptors
-visual feedback:
Limitations of sensory feedback
-Delays of sensory feedback: processing and integrating
-Noise: the info is not always accurate
-Encoding issue: needs common encoding (sensorimotor integration problem)
-Distinguishing between reafference and exafference
Delays
(sensory feedback)
the longer the path of travel, the longer the sensory info is delayed
Noise
(sensory feedback)
- Sensor noise= inaccurate/noisy sensory receptors
- Motor noise= noise in motor commands, results in variable movements
- Sensorimotor noise= noise from combining sensory feedback signals and transforming info to motor system
Potential causes of sensory noise
-Ion channel properties (random opening and closing)
-Spontaneous action potentials generated by receptors/neurons
-Muscle fatigue
Reafference
self generated motor behaviour
Exafference
external events of stimuli
How is your vision not blurry when you move your eyes to look at something (saccade)
nervous system uses predictive signalling (from efference copy signals)
Internal Models
-understand how the body works
-processes occur in the brain
-different brain regions responsible for different aspects
-argued to exist in the spinal cord
-develop over time (through experience)
3 types of internal models
- Cognitive model
- Inverse model
- Forward model
cognitive internal model
-how external objects work based on prior experience with the object/similar object (we know a ball bounces)
-you act differently depending on what the object is
-you choose a goal depending on the object or hazard in environment
inverse internal model
calculates motor commands needed from desired trajectory information
Inverse model subcomponents
- Inverse kinematic transformation/model= calculation of set joint angles than can reach an endpoint
- Inverse dynamic transformation/model= calculation of torques at each joint needed to achieve the motion
Describe how an inverse model (dynamic and kinematic) works when reaching for a cup
-locate your hand and the cup
-form a motor plan
-the plan is sent to the inverse kinematic model (determines joint angles needed)
-the inverse dynamic model takes the joint angle info and determined joint torques needed
What is the output of this inverse internal model example?
the motor command that is sent to the arm muscles to reach the cup
The motor plan to move is based on
- The goal (to reach the cup): can be influences by the cognitive model, we want to reach the cup because we are thirsty
- State of the limb: state estimate, position/velocity of limb
- Target (the cup): distance between limb and the target= difference vector
Forward internal model
-models the relationship between actions and consequences
-2 components= forward dynamic model and forward sensory model
-PREDICTS state estimation and sensory consequences from a copy of the motor command
What is a copy of the motor command called?
Efference copy
State estimation
Combining predicted sensory feedback with actual feedback to form a belief about the state of our body/world/environment
Describe how the forward model and state estimation works: goal= move finger to face
-A previous state estimate (where is the finger approximately?) is the basis for a new state estimate
-A motor command is generated using the state estimate. command to move the arm to face.
-An efference copy is sent to the forward dynamic model which predicts the current state of the finger
-The forward sensory model predicts the sensory feedback
-Sensory prediction error is used to correct the estimate of the current finger position
-The final estimate of the finger position has less error and become the new estimate for other movements
-Cycle repeats
can you form a motor command without info about the state of your limb?
NO!!
If the model is really good…
the nervous system can use the predicted state to plan to reach for a cup (for example) without waiting for sensory feedback
Predicted feedback is determined by
the forward sensory model
the forward sensory model uses
the predicted state from the forward dynamic model
how do internal models develop over time?
change with experience, growth, injury etc..
Why are movements sometimes not very accurate?
-accuracy of motor command is only as good as the internal model and its inputs
-noise
How does the nervous system know the state of the limb if sensory information is delayed????
the FORWARD internal model and STATE ESTIMATION
Difference vector
The state estimate of the hand is compared to the estimated cup location
(distance between the two)
Movement:
1. reach and grasp cup
2. bring cup to mouth
Describe how nervous system would compare sensory info (two quick sequential movements)
Incoming sensory feedback is compared to the predicted sensory feedback (from forward sensory model) and used to correct the sate estimate
When you grasp the cup with your hand, you have reduced the difference vector to
zero
Sensory-to-motor transformation
–> which model?
Inverse model
-converts sensory-based info related to a motor plan to a motor command
Motor-to-sensory transformation
Forward model
-converts motor based info about movement to a sensory-based prediction
What does it mean when there is a sensory prediction error?
predicted and actual sensory feedback don’t match
What does the nervous system do when there is a temporary sensory prediction error?
adjust the weighting between predicted vs. actual sensory feedback
What should the nervous system to if there is a constant sensory prediction error?
update its internal models
(could be caused by growth, injury etc.)
Credit Assignment problem
nervous system needs to assign a cause of the error
-is it body/limb based or world based?
-internal or external factors?
Examples of internal sources of error
factors within the body; altered sensorimotor mapping, injury, fatigue etc.
Examples of external sources of error
environmental factors; wind, object bumping into you, object that you are manipulating
Credit assignment problem is the opposite of the — problem
DOF
How are the credit assignment and DOF problems opposite?
credit assignment= determine which or the many neurons, muscles, limbs was the problem
DOF problem= choose and control different neurons, muscles, limbs for a particular movement
Two common paradigms to study motor learning
- Force-field perturbations
- Prism glasses or visuomotor rotation
Force field perturbations cause
dynamic adaptation over repeated trials
Prism glasses or visuomotor rotation cause
kinematic adaptation over repeated trials
task error in visuomotor rotation
difference between the end cursor position and the target position
sensory prediction error in visuomotor rotation
difference between predicted and actual sensory feedback (cursor doesn’t move in expected way)
Dynamic adaptation
a force field pushes a person’s arm off course so that initial trajectories start in the WRONG direction and curve to correct the error
Dynamic adaptation: what happens with practice?
-the person learns to adapt to the force field and produce a straight reaching movement
-learns a new relationship between the motor command and the motion of the limb
Kinematic adaptation
person learns a new relationship between between coordinate systems (such as arm joint angle/observed hand position relationship)
Wearing prism goggles cause…
(kinematic adaptation)
cause a visual shift so that the hand and target positions are perceived at different locations
-with practice, reaching movements change from being curved to being straight
aftereffect
removing a perturbation (such as prism goggles) the person inappropriately compensates with a reaching error in the opposite direction
The presence of the aftereffect is an indication of
updating of the person’s internal models
Visuomotor rotation paradigm usually involves
-moving a cursor on a screen
-person sees the start position, target location and a cursor on a screen
-can’t see their hand
-goal= move cursor to target as accurately as possible
The brain becomes — sensitive to errors with greater uncertainty
less
explanation for slower adaptation in the noise conditions?
the only way the brain knows there is a foot placement error is through visual feedback.
NOISE –> brain trusts forward model predictions more than vision
-uses visual feedback less to adjust the state estimate= slowed rate of adaptation
How can sensory prediction error drive a corrective response?
Example of modifying grip and load force to lift an object
-object is too heavy for your grip/slipping from fingers
-sensory prediction error drives corrective response (increase/decrease grip)
If cutaneous receptors discharge earlier or later than normal (or predicted based on previous experience)…
the object you are holding is too heavy for your grip and is slipping from your finger
–> the mismatch in timing is a sensory prediction error
Experiment: object is heavier than expected, cutaneous receptor activity measured.
Real life example= ?
Open fridge to get a carton of milk and expect it to be empty, but someone replaced it with a full carton so it is much heavier than expected
Heavier object than expected; how is the corrective response triggered?
-Nervous system is expecting the cutaneous afferent to discharge at a specific time
-BUT heavier object= longer to develop good grip and load force to lift object up
-this means that the cutaneous afferent discharge occurs later
-absence of expected cutaneous receptor event signalling at the predicted time triggers a corrective response
Forward models distinguish between — and — generated movement
internally and externally
Muscle stretch: internally generated and externally generated examples
-muscle stretch is detected by changes in muscle spindle length
-internally generated= voluntary= you move your own muscles
-externally generated= involuntary= someone or something moves your muscle
Forward model allows the nervous system to determine if a — error is present
sensory predicition
If a sensory prediction error is present, this suggests — generated movement
externally
If mismatch is not constant, should model be updated?
NO!!! no need to update bc it could be a one-time external event such as someone bumping into you or a gust of wind etc.
voluntary movement: how does a muscle contraction occur?
the corticospinal tract sends a motor command to alpha motor neurons in the spinal cord, which leads to a muscle contraction (and change in muscle length)
How is the resulting sensory feedback of the movement predicted?
An efference copy is sent to the forward model which predicts the resulting sensory feedback
Sensory feedback is compared to actual muscle spindle feedback which enters the — from the —
cerebellum from the dorsal spinocerebellar tract
muscle spindle feedback is also sent to the primary somatosensory cortex through
primary somatosensory cortex through the dorsal column medial lemniscus tract to inform the person of the movement
If there is a match between actual muscle spindle feedback and sensory feedback…
a signal from the cerebellum is sent to reduce the signal that reaches other parts of the brain
Why suppress activity in the cortex?
you don’t need to be consciously aware that you moved in a certain way if it was done as INTENDED
If there is a mismatch between actual muscle spindle feedback and sensory feedback…
the inhibitory influence is reduces/absent
when would a mismatch between actual and sensory feedback occur?
if someone or something caused a change in muscle length
You get a — activity in the cortex when the movement was — generated
larger
externally generated
