Lecture 7 - Theories and Models of Sensorimotor Control

Question 1

Describe reflex theory

Answer 1

Movements are due to reflex chaining (sequences of reflexes), reflexes are building blocks for behaviours
stimulus-response

Question 2

What are the limitations of reflex theory?

Answer 2

1) Reflex theory can’t account for voluntary movement - no external stimulus
2) How does it describe very fast movements –> not enough time for sensory feedback to play role
3) Can’t account for learning new movements –> perform different actions with different stimuli based on already learned rules
4) Can’t account for different responses to same stimulus depending on context (overriding reflexes, scaling to meet task demand)

Question 3

Describe motor program theory; define parameters, invariant features, and schema

Answer 3

Motor program theory says we don't have stored motor commands for every action. Have generalized motor programs that generate motor commands. Motor programs are effector independent, can use sensory feedback to correct movement without affecting higher level program, errors reflected in movement can alter program and parameters. Motor program at lower level converts output from higher level to commands to motoneurons. Some programs govern discrete actions, complex movements then use programs to recruit these discrete programs
Motor program controls class of actions (group of similar motor skills), performs same action in different circumstances
Motor program --> set of rules for generating motor commands
Parameters --> features that change from one performance to another, change parameters = change behaviour
Invariant features --> stay the same throughout class of actions regardless of parameters, defines motor program
Schema = set of rules to provide basis for decision, representation of rules governing performance of skill in certain situation, facilitates selection of parameters

Question 4

What is the evidence for motor program theory?

Answer 4

Typing keys = relatively equal percentage regardless of speed
Scaling movement of reaching = equal velocity curves regardless of distance –> shows extent of movement are preplanned

Question 5

How can motor programs use sensory feedback?

Answer 5

If the movement duration is long enough, sensors can correct small perturbations without altering higher level motor program

Question 6

How does motor program theory explain the walk-to-run transition?

Answer 6

Walk and running are two different classes of actions, with different invariant features. The transition equates to a change in the motor program. Person chooses to change motor programs

Question 7

Describe Dynamic Action theory

Answer 7

Movement is self-organizing, based on interaction between individual and environment. New movements arise from a change in the control parameter that changes which attractor state is optimal

Question 8

What are attractors, control parameters, and order parameters?

Answer 8

stable behavioural steady states –> minimal variation, optimal energy
Control parameters are variables that increase or decrease and change the value of the order parameter (independent variable)
Order parameters are values that are attached to attractor states that define the overall behaviour

Question 9

Provide the evidence for dynamic action theory?

Answer 9

Bilateral finger movements
Order parameter = the degrees (180 or 360)
Attractor states = in phase or out of phase (defined by order parameter)
Control parameter = speed of finger movements
As control parameter increases, order parameter changes, and new stable behavioural state created

Question 10

How does dynamic action theory discuss the walk-to-run transition?

Answer 10

speed is the control parameter
relative phase of legs (flexion/extension) is the order parameter
attractor states = walking or running
gait change is the result of competition between the two attractor states

Question 11

What is the role of sensory feedback and what are the issues of sensory feedback?

Answer 11

Inherent delays = takes time to receive and process and then act on sensory feedback
Noise –> information not perfectly accurate (going from 3D to 2D to 3D)
Sensory feedback must be transformed into terms understood by motor systems
Needs a reference signal, only indicates extent of movement deviation from intended performance

Question 12

What are 3 types of noise, and 2 causes of noise, and the result of noise

Answer 12

sensor noise –> inaccurate sensory receptors, uncertainty in location
Motor noise –> noise in motor commands, variable movements
Sensorimotor noise –> noise from combining sensory feedback and transformation to motor coordinates
Causes of noise: spontaneous action potentials generated by receptors/neurons, muscle fatigue = change in discharge
Results in lower accuracy and precision

Question 13

What are the three types of internal models?

Answer 13

Cognitive model –> how objects work based on prior experience
Inverse model –> calculates motor commands based on desired trajectory and knowledge of limb state
Forward model –> calculates expected sensory feedback and limb state based on copy of motor command, predict behaviour of body, model causal relationship between actions and consequences

Question 14

What are the two components of the inverse internal model?

Answer 14

Kinematic transformation –> Calculate limb trajectories
Dynamic transformation –> based on calculated trajectory, what are the joint trajectories (torques/muscle activations) required to perform the movement

Question 15

What is the evidence for the forward internal model?

Answer 15

Grip force control prediction

Change the amount of lift force and hand acceleration based on perceived/expected slipperiness

Question 16

Describe the entire internal model

Answer 16

Start with estimation of where you are based on combination of actual and predicted sensory feedback, where the target is
This creates a difference vector
Plan limb and joint trajectories to reduce vector to 0, create a motor plan and convert to motor commands. Send an efference copy to the forward model, one to the muscles. Efference copy then predicts the sensory information it should receive, compares it to the actual sensory information, then can use the error in signal/expected signal to update the model. Sensory feedback then leads to new knowledge of limb state, can plan new movement

Question 17

Discuss how state estimate is created

Answer 17

weighted average of actual and predicted sensory feedback. If sensors were 100% accurate, wouldn’t need sensory feedback because prediction would be perfect
Temporary errors result in re-weighing, constant error requires update to the model

Question 18

What is the evidence for internal models?

Answer 18

Force perturbation, prism glasses cause dynamic/kinematic adaptation and then we see aftereffects

Question 19

How does internal models explain how the brain distinguishes between internal/external muscle stretch

Answer 19

Using predicted sensory feedback. External muscle stretch would lead to error between predicted and actual feedback, can then decide to act on it