Motor Systems - University of Sunderland

Download Report

Transcript Motor Systems - University of Sunderland

Motor Systems
(Lecture 11)
Harry R. Erwin, PhD
COMM2E
University of Sunderland
Resources
• Nicholls et al. (well-referenced, possibly the
best for neuroscientists)
• Kandel et al. (medical school textbook)
• Shepherd (general textbook)
• Johnston and Wu (emphasizes neurophysiology,
I don’t have)
• Avis Cohen (UMd Bio 708C course notes)
• Bower and Beeman, The Book of Genesis, ch 8.
Outline
•
•
•
•
•
•
Motor System Architecture
The Muscle
Central Pattern Generators
Motor Systems in the CNS
Biologically-Inspired Motor System Models
Book of Genesis, Chapter 8
Motor System Architecture
• Hierarchical
–
–
–
–
–
–
Cortex
Cerebellum
Brainstem
Spinal cord
Motoneurons
Muscles
Muscles
• Muscles are springs and correct for errors. As for any spring,
oscillation can be an issue, but the muscle controls it. The
muscle thinks!
• A muscle is made up of multiple muscle fibers—multinucleate
cells in mammals that contain myosin and actin (elastic). These
are excitable cells like neurons.
• In higher vertebrates, each fiber is innervated by a single
motoneuron, but a single motoneuron can innervate many fibers
of a single type. Fine motor skills can involve one fiber/one
neuron, though more usually about 20. (That makes it
interesting that H. erectus had a significantly smaller spinal cord
diameter than H. sapiens.)
Muscle Fibers
• Several types of muscle fibers (the first three listed are
important):
– SS use oxidative metabolism, are weak, do not appear to
fatigue, have a role in maintaining posture.
– FR are fatigue-resistant, use both oxidative and nonoxidative enzymes, are stronger, and their motoneurons have
intermediate input resistance and rheobase (current threshold
for initiating a spike).
– FF fatigue rapidly, are non-oxidative (glycolysis), are
strongest with low input resistance and high rheobase.
– SS are recruited first, followed by FR, and finally FF.
– S are slow and have a slow relaxation time.
Muscle Contraction Mechanisms
• Individual fibers ‘twitch’. Muscle contraction uses multiple
muscle fibers twitching in a pattern. This is non-stochastic—
action potentials always work.
• Myosin and actin are connected by protein bridges. The angles
of these bridges define the force that can be exerted, depending
on the type of myosin.
• Muscles maximize force when stretched. Going beyond that
maximum length (‘pulled’), they have no force.
• Muscle fibers respond to action potentials (ACh) allowing entry
of Ca++. Cramps reflect a calcium deficit.
Muscle Efferents
• Axon terminals synapse on the fiber. The AP
must invade entire fiber.
• ACh is the neurotransmitter. Vesicle release
leads to diffusion across the junction, where it
binds to receptors and is hydrolyzed by AChE.
Binding opens cationic (non-specific) channels
and the membrane potential collapses.
Spindle Systems
• Some muscle fibers are stretch receptors, rather than
force producers.
• Lack contractile elements.
• Types include primary and secondary muscle spindles.
• Primary spindles report that the fiber they’re
monitoring is carrying force. Measure rate of change,
allowing you to control velocity.
• Secondary spindles measure tension directly.
• Not servo control!—the muscle turns on first, control
is later.
The Spinal Cord
• Organized functionally
• Characterized by fully coordinated rhythmic
movements.
• Consists of fused dorsal and ventral roots surrounding
a central canal
– Dorsal roots are sensory (stretch receptors, pain, touch, joint
position)
– Ventral roots contain motoneurons
– Organized locally into central pattern generators (CPGs)
• Ends at the level of the kidneys.
Rhythmic Movements
• Questions for research:
– How do they happen?
– What do they mean?
– Where do they come from?
• Reflex chain?
• Sequential pattern of activation?
• Reverberatory circuits?
• Cutting the spinal cord removes the inhibition of
flexion/extension movements (Brown, 1914).
• These issues are also present in the CNS, but the two
communities do not interact much.
Central Pattern Generators (CPGs)
• A Central Pattern Generator is a system of neurons that
can generate a stereotyped rhythmic movement without
sensory afference or somatic feedback.
• It can be activated/sustained by a triggering stimulus
(either tonic or phasic), but requires no modulation of
the input to generate the basic pattern.
• Lundberg and Grillner had a nasty argument on
whether CPGs are present in the motor system. This is
Avis Cohen’s specialty.
Research into CPGs
• To demonstrate the existence of a CPG:
– The stereotyped movement must not be extinguished by the removal of
varying sensory afference.
– The movement must not be extinguished by the removal of somatic
feedback.
• Experiments beginning in 1960s produced evidence for CPGs.
Russian studies of decorticated cats showed they could maintain
walking motion without a cortex.
–
–
–
–
Hence the cortex turns walking on.
Strength of stimulation controls power, not frequency.
Gait changes are automatic.
Limbs controlled as a whole.
• Primitive mammal-like reptiles give insight here. (discuss)
CPGs and the Spine
• CPG models have been effective in describing how
coordinated rhythmic movements might be generated in the
spinal column.
• Involves interneurons (Renshaw cells) in the spinal cord.
However, these can be turned off and the animal still walks.
• Most motor actions are indirectly managed using opposing
pairs of muscles controlled by a CPG. Motor cortex neurons
synapse on the spinal interneurons (and directly on the
motoneurons used in delicate finger movements).
• The Renshaw cells are driven hard.
CPGs in Context
• CPGs seem to generate body shape, not force
commands.
• An acute spinalized curarized deafferented cat still
walks.
– CPG does not require sensory feedback
– CPG does not require descending control
• Reflex loops do not operate during locomotion. The
spinal cord decides whether you step on something
sharp. Corrections and adjustments to ground features
are all handled by the CPG.
Motor Systems in the CNS
•
•
•
•
Motor cortex (and related cortices)
Basal ganglia
Cerebellum
Brain-stem nuclei
Issues
• How are CPGs controlled by the CNS?
– Do as I do?
– Do as I say?
– Do as I suggest?
• Is the response
– An act? (time-dependent)
– A place? (autonomous)
– A combination?
Hypothesis
• Motor commands are suspected of specifying
“an image of attainment”.
– Passive—posture
– Active—an action
Mechanisms of Coordination
•
•
•
•
•
•
•
Preservation of phase relationships
Non-linear
Developmentally tuned
Phase differences are fixed
CNS provides drive
Spine returns periodic signal to the CNS
Mutual entrainment of spine and CNS
Motor Cortex
• Probably uses dense (vector) rather than sparse coding.
• Specifies terminal position of movement in worldcentered coordinates.
• The spinal cord seems to work in body-centered
coordinates. Giszter et al. claim that there are only four
degrees of freedom in the spinal cord of frog.
• Cerebellum may be responsible for the change of
coordinates.
Cerebellum
• Seems to be a sensory system (Bower)
• Receives motoneuron and sensory copies, via separate
pathways.
• Outputs a periodic inhibitory signal to the spinal cord
• Very fast response
• Extremely large primary (Purkinje) cells.
• Need to be modeled below the ion channel level. 5000
compartments are typical.
Basal Ganglia
•
•
•
•
Play a role in willed (rather than stimulus-triggered) acts.
No convergence in the striatum. Multiple parallel modules.
Convergence at the Globus pallidus (GP).
Feedback loop:
– Cortices (PM, SuppM, M, SS)
– Putamen (part of the striatum, hard to excite, hence sensitive to
synchronization, input from Substantia nigra—Parkinson’s)
– GPext (feedback relationship with the SubThalNuc, inhibited by
Putamen)
– GPinf and SNreticulata (excited by GPext and inhibited by Putamen)
– Thalamus (inhibited by GPinf and SNR, excites the prefrontal and
premotor cortex).
Biologically-Inspired Motor System
Models
• Two-way street all the way down. The cortex should
command a CPG, which then drives the cortex in
return, signals the motoneurons, and interacts with
sensory afferents.
• This is a universal picture, using comparisons at all
levels. See Rodney Brooks’ robot models.
• Note that if feedback is specific, cell-to-cell-back-tooriginal cell (as in the inferior colliculus), this supports
a back-propagation model for training (not just
reinforcement learning!).
Back to Central Pattern Generators
• The neuronal circuits that support rhythmic
muscle contractions are referred to as central
pattern generators (CPGs)
• These circuits can generate this activity in
isolation.
• The ability to switch between CPGs relies on
feedback from proprioceptors and higher CNS
control.
Two-Neuron Oscillators
The Math of the System
•
•
•
•
•
•
•
di(t) = i (in modular terms)
i(t) = (i + i(0) ) (mod 2)
When coupled, we get:
d1(t) = 1 + h12(1, 2)
d2(t) = 2 + h21(2, 1)
(t) = 1(t) - 2(t)
This describes the phase lag of oscillator 2 relative to
oscillator 1.
Calculating
• d(t)/dt = d1(t)/dt - d2(t)/dt
= (1- 2)+(h12(1, 2) - h21(2, 1))
• Now assume hij is a function of 1-2, and is zero at
zero. This is called diffusive coupling.
• For the phase lag to remain constant,
– d(t)/dt = 0.
• If hij is proportional to the sin of the difference, we get
different solutions depending on the constant of
proportionality, depending on the ratio of 1 - 2 to
the sum of the constants of proportionality.
Why is this interesting?
• It suggests that the typical behaviour of CPGs
should move from phase drift to phase-locking
and potentially ‘oscillator death’ for large
networks.
Four Neuron Oscillators
• Show similar but more complex behaviour.
• Representative to fish swimming (and by
implication of tetrapod walking)
• Gives some insight into how gaits might be
modelled.
• Gives us confidence that our models allow us to
understand more complex CPGs.