Neuroscience · Unit 4: Motor systems

Motor Units, Reflexes and Locomotion

Movements of the limbs and trunk end in the spinal cord, where motor neurons turn synaptic input into muscle force. This lesson explains how the cord grades force by recruiting motor units in order of size and by changing their firing rates, how the stretch reflex and its inhibitory partner hold a joint, how gamma motor neurons keep muscle spindles in range, and how spinal circuits make a walking rhythm on their own. Afterwards you can predict a recruitment order from Ohm's law and trace which motor neurons a stretch excites and which it inhibits.

Watch first One Stretch, Two Opposite Commands · 2:28

Tap the tendon just below your kneecap and your leg kicks, through a path in the spinal cord with only one synapse between sensor and motor neuron. The same few segments of cord decide how hard a muscle pulls, keep a joint steady and can lay down the rhythm of a step. This lesson takes those jobs one at a time.

Directions here use the spinal-cord frame: dorsal means toward the back, ventral toward the belly. Ipsilateral means on the same side of the body, and contralateral on the opposite side.

Motor units and the size principle

The cells that command skeletal muscle are alpha motor neurons, whose cell bodies sit in the ventral horn of the cord and whose axons leave through the ventral roots. Each axon branches to many muscle fibres. A motor unit is one alpha motor neuron and all the fibres it innervates; a motor pool is all the motor neurons that supply one muscle. In mature muscle each fibre is contacted by one motor neuron, after an early period of multiple innervation that synapse elimination covers; how a motor neuron gets its identity is in the motor neuron code, and how its junction is built in the neuromuscular junction.

At the junction, acetylcholine released by the motor neuron depolarizes the fibre (synaptic transmission), and normally each motor-neuron spike gives every fibre of its unit an action potential and a twitch. So the unit is the smallest piece of force the cord can command. Units differ widely: small motor neurons supply few fibres and make weak units, large ones supply many and make strong units.

When the drive to a pool rises, which units start first? Under shared input, smaller motor neurons are recruited before larger ones, the size principle. The reason is Ohm’s law. Input resistance is the voltage change a cell shows per unit of steady current injected into it (the resting membrane and passive signals). A small cell has less membrane, so fewer channels in parallel and a higher input resistance. The same synaptic current therefore depolarizes it more:

ΔV = I × R_in, so the current needed to reach threshold is I = ΔV / R_in

How synaptic currents sum on the way is the subject of synaptic integration. The order is a tendency, first measured in cats, and it holds under common input. Differences in threshold, in persistent inward currents and in neuromodulation can shift it, and separate inputs to parts of a pool can break it.

Grading force: recruitment and rate

Worked problem. Three motor neurons, S, M and L, in one pool receive the same synaptic current, ramped up from 0 at 1 nA per second. These are illustrative round numbers, in the range reported for cat lumbar motor neurons, not readings from one recording. Input resistances are S 4.0 MΩ, M 2.0 MΩ and L 1.0 MΩ. Assume steady state (the ramp is slow compared with the membrane time constant) and that each cell must depolarize 10 mV from rest to fire. That last assumption is a simplification, since real thresholds also differ.

Step 1: current to threshold. A millivolt divided by a megohm is a nanoampere.

S: 10 mV / 4.0 MΩ = 2.5 nA; M: 10 mV / 2.0 MΩ = 5.0 nA; L: 10 mV / 1.0 MΩ = 10 nA

Step 2: time to threshold. At 1 nA per second, S fires at 2.5 s, M at 5.0 s and L at 10 s. The order is smallest first.

Step 3: force. Give the units forces of 0.05 N, 0.30 N and 1.50 N. As each joins, total force goes 0.05 N, then 0.35 N, then 1.85 N. The steps get bigger as force grows, so the smallest steps come when force is low, where a small error matters most. In a real pool of many units the same logic means each new recruit adds force roughly in step with the force already present.

Size principle — one ramp, three motor neurons⚙ original · interactive
Current now
—
Units firing
—
Total force
—
Recruitment time = threshold current ÷ ramp rate
—

—

In this sketch: steady state (the ramp is slow compared with the membrane time constant), every cell needs the same 10 mV depolarization to fire, each unit is simply on or off, and circle area is drawn in proportion to 1/Rin (relative sizes, not to scale). Round numbers, the lesson’s own.

I = ΔV / Rin: with the same current in every cell, the one with the highest input resistance reaches 10 mV first. In this sketch, with steady state and equal 10 mV thresholds, the input resistances are illustrative round numbers in the cat lumbar range; the forces are the lesson’s round numbers.

Recruitment is half the story. Rate coding is the other half: an active unit can fire faster. One spike gives a twitch. If the next spike arrives before the twitch has relaxed, the second twitch adds to the first, and at high rates the twitches fuse into a smooth contraction, a tetanus. So force rises both by adding units and by driving the active ones harder.

How firing rate rises with force differs from unit to unit. Avrillon and colleagues (2024) recorded the tibialis anterior and vastus lateralis of eight people per muscle with grids of 64 surface electrodes, during contractions up to 80% of maximal force, and sorted the signals into single motor units, about 129 units per participant in tibialis anterior. After an initial steep acceleration at low force, firing rates rose between force plateaus 10% of maximal force apart by 3.5 ± 1.7 pulses per second in tibialis anterior and 2.0 ± 1.1 in vastus lateralis. Trajectories varied across units, and most fitted a logarithm better than a straight line or an exponential. The input to each motor neuron was inferred from force, not measured inside the cell.

The stretch reflex and its inhibitory partner

A muscle spindle is a stretch sensor: a capsule of small, specialized intrafusal fibres lying parallel to the ordinary extrafusal fibres that make force. Stretching the muscle stretches the spindle. Its Ia (primary) afferents respond to both the speed and the amount of stretch; its group II (secondary) afferents mainly to the amount (relaxed human muscle; reviewed by Dimitriou, 2022).

A tendon tap gives a brief stretch, and the Ia afferents fire. They enter the cord through the dorsal root and branch.

  • One branch excites the motor neurons of the same muscle, the homonymous motor neurons, directly: one synapse, a monosynaptic path. The muscle contracts against the stretch.
  • Another branch excites a Ia inhibitory interneuron on the same side of the cord, which inhibits the motor neurons of the antagonist, the muscle that moves the joint the other way. That path has two synapses, so it is disynaptic, and it is ipsilateral. Sherrington called this pairing of excitation and inhibition reciprocal innervation.

Only the excitatory branch to the homonymous muscle is monosynaptic. The reflex as a whole is not.

The Golgi tendon organ sits at the junction of muscle and tendon, in series with the fibres, so it signals the force the muscle makes rather than its length. Its Ib afferents act through interneurons. The textbook pathway inhibits the muscle’s own motor neurons, but Ib effects change with task and are not simply a safety cutout against tearing.

The crossed-extensor reflex is the one that crosses. A painful stimulus to one foot excites, through several interneurons, the flexors of that leg, which withdraws, and through interneurons whose axons cross the midline, the extensors of the other leg, which takes the weight. It is polysynaptic and bilateral, where the stretch reflex is local and one-sided.

What descending pathways add, and the signs that follow damage to them, are taken up in the basal ganglia and the cerebellum; diseases of motor neurons are in neurodegeneration.

Gamma: keeping the spindle in range

A spindle has a problem. When the muscle shortens, the spindle shortens too, falls slack and stops reporting. Gamma motor neurons fix this. They supply muscle spindles only, ending on the contractile poles of the intrafusal fibres. When the poles contract, the centre of the spindle, where the Ia endings wrap, is pulled taut again.

Think of a tape measure that winds itself in as the muscle shortens, so it never goes loose. When alpha and gamma motor neurons fire together, alpha-gamma coactivation, the spindle can be kept from falling slack during a contraction.

Gamma drive is not simply locked to alpha drive. Papaioannou and Dimitriou (2021) recorded single Ia afferents from wrist and finger extensors of volunteers with microneurography during an instructed delay before a reach. About 80 ms after the target was cued, Ia firing fell when the coming reach would stretch the spindle-bearing muscle, while kinematics and surface EMG showed no systematic change with the target. The principal analysis used eight afferents from six people, and separate experiments with a robot measured whole-arm stretch reflexes. The authors interpreted the results as compatible with goal-directed changes in gamma drive. Gamma neurons were not recorded, so that drive is an inference.

Rhythm without a conductor

A central pattern generator is a neural network capable of producing rhythmic output without patterned descending or sensory input. Graham Brown’s 1911 cats, with the cord cut and the dorsal roots of the hind limbs severed, still alternated ankle flexors and extensors.

Most modern evidence comes from fictive locomotion: an isolated spinal cord, usually from a newborn rodent, bathed in drugs such as NMDA and serotonin, producing alternating bursts in the ventral roots that would drive flexors and extensors. Hägglund and colleagues (2013) used optogenetics in the isolated neonatal mouse cord to activate or inhibit selected populations, and found that locomotor-like bursting could be induced on one side alone. That shows a spinal network can make rhythm. It does not show a single pacemaker cell. A 2021 review of mouse work reports that rhythm persisted after silencing or ablating each individual genetically defined ventral population it considered, so the rhythm generator is still an open question.

Sensory input still matters. In the work of Li, Garcia-Ramirez and Dougherty (2019) on cords of mice one to three days old, ankle nerve stimulation between flexor bursts could start a new flexor burst early. The network makes the rhythm and feedback adjusts its timing. For people, evidence for a spinal pattern generator is indirect; these results come from neonatal rodents and cats.

Design problem

Design an experiment to show that the mammalian spinal cord can generate a locomotor rhythm by itself, and that sensory input can reset it. Say what you record, what you predict, your controls and what your result cannot show.

One way to do it

Isolate the spinal cord of a newborn mouse (one to three days old) with ankle nerves attached, so there is no descending input and no movement feedback. Evoke activity with NMDA and serotonin in the bath; Li and colleagues (2019) used 7 µM and 8 µM. Record a flexor-related lumbar ventral root and an extensor-related one on both sides. Li and colleagues used L1 for flexor-related and L4 for extensor-related output; many laboratories use L2 and L5.

Predict that the flexor-related and extensor-related roots on one side alternate, and that the left and right roots of each pair alternate. Controls: no rhythm in saline without the drugs, and loss of the rhythm on washout.

To test resetting, first transect the ventral roots proximal to the spinal nerve, so pulses cannot travel back into the cord along motor axons, then stimulate an ankle nerve with brief trains (for example 4 to 6 pulses at 20 Hz, onsets 50 ms apart) at several phases of the cycle. Plot the time to the next flexor burst against the phase of stimulation, using unstimulated cycles as the baseline. An early burst after stimulation between flexor bursts shows resetting.

Limits: this is a neonatal cord driven by drugs, not adult or human walking. Rhythm with no patterned input shows the circuit is enough under these conditions; it does not identify a single pacemaker cell or prove that normal walking dispenses with the brain or with feedback.

How we measure it

Ventral-root recording of fictive locomotion

The spinal cord of a newborn rodent is removed into oxygenated saline, and drugs such as NMDA and serotonin evoke bursts of motor output. Electrodes on lumbar ventral roots record flexor-related and extensor-related bursts; their alternation shows a locomotor pattern with no limb moving and no brain attached.

High-density surface EMG with decomposition

Grids of many electrodes on the skin record muscle electrical activity, and a blind source separation algorithm sorts the mixture into the spike trains of individual motor units. It lets you follow dozens of units in a person, though the input to each motor neuron is inferred, not measured.

Microneurography

A fine tungsten needle inserted into a peripheral nerve in an awake person records the spikes of single sensory axons, such as a muscle spindle afferent. It shows what the spindle reports during a task, but not the gamma drive that shapes it.

Intracellular and patch recording from spinal neurons

An electrode inside or sealed onto a motor neuron or interneuron measures its membrane potential and synaptic responses while afferent nerves are stimulated. Latency and jitter of the responses suggest how many synapses lie between the afferent and the cell.

Finish line

Test yourself

8 questions

Practice and MCAT-style, with explained answers.

Next · Lesson 16

Descending Control and the Basal Ganglia

A stroke deep in the left hemisphere weakens the right side. A cut in the right half of the neck weakens the right side too. Why?

Continue →

← Lesson 14 · Smell and Taste

Nicotinic acetylcholine receptor — turns each motor-neuron spike into a muscle response
Nicotinic acetylcholine receptor — turns each motor-neuron spike into a muscle response ● LIVE · RCSB PDB 2BG9
open on RCSB PDB ↗
Loading live structure 2BG9 from RCSB PDB…

This is the receptor type at the neuromuscular junction, here from the fish electric organ. Find the five subunits around the central pore; acetylcholine released by a motor neuron opens it and depolarizes every fibre of the motor unit.

Drag to rotate · scroll or pinch to zoom · right-drag to pan
Charles Sherrington

Charles Sherrington

Studied reflexes in decerebrate cats and described how excitation of one muscle is paired with inhibition of its antagonist; with Edward Liddell, characterized the reflex contraction of a muscle to its own stretch.

Sherrington worked on the reflexes of the spinal cord and brainstem for decades, mostly in cats. A useful preparation was the decerebrate cat, in which the brainstem is cut through at the midbrain, leaving the cord and lower brainstem working and the extensor muscles in a steady stiffness that reflexes can act on. In such animals he found that when a reflex excites one muscle, the muscle that opposes it at the same joint relaxes; he called the pairing reciprocal innervation, and he treated inhibition as an active process of the nervous system rather than an absence of excitation. In 1924 he and Edward Liddell published a study in the Proceedings of the Royal Society of the reflex contraction of an extensor muscle in response to stretching it, the stretch or myotatic reflex. Sherrington shared the 1932 Nobel Prize in Physiology or Medicine with Edgar Adrian, for discoveries about the functions of neurons, and gave his Nobel lecture on inhibition as a coordinating factor. His work made the spinal cord a place where circuits, not single paths, decide which muscles act.

✓ verbatim Liddell & Sherrington, Proc. R. Soc. Lond. B 96:212–242 (1924); Sherrington, Nobel Lecture, 'Inhibition as a Coordinative Factor', 12 December 1932.
Elwood Henneman

Elwood Henneman

Recorded motor axons in cats and reported that motor neurons are recruited in order of size, small before large, an observation later called the size principle.

When a muscle is driven harder, which of its motor neurons join in, and in what order? Henneman approached the question in cats by recording from the ventral roots, where the axons of individual motor neurons leave the spinal cord. The size of an extracellularly recorded axon spike tends to grow with the size of the axon, and larger axons belong to larger motor neurons. In his 1957 paper in Science he reported that as reflex input to a motor pool increased, units with small spikes began firing before units with large spikes. In 1965, with George Somjen and David Carpenter, he described the relation between cell size and excitability in more detail in the Journal of Neurophysiology. The order has since been tested in many muscles and species, and it holds as a strong tendency under shared input, with exceptions. It made recruitment predictable from the properties of the cells themselves.

✓ verbatim Henneman, Science 126:1345–1347 (1957); Henneman, Somjen & Carpenter, J. Neurophysiol. 28:560–580 (1965).
T. Graham Brown

T. Graham Brown

Showed in cats that the spinal cord, cut off from the brain and from muscle sensory input from the legs, could still produce rhythmic alternation of flexor and extensor muscles, and proposed that paired, mutually inhibiting half-centres generate the rhythm.

In the early twentieth century a common view held that walking is a chain of reflexes: each step produces sensory signals that trigger the next. Graham Brown, who worked in Sherrington's Liverpool laboratory, tested this in cats. He cut the spinal cord to remove input from the brain and also cut the dorsal roots that carry sensory signals from the hind limbs into the cord. If walking needed a chain of reflexes, nothing rhythmic should remain. Instead, he recorded alternating contractions of ankle flexor and extensor muscles. He reported the result in the Proceedings of the Royal Society in 1911 and proposed that two groups of neurons, one driving flexors and one driving extensors, inhibit each other so that activity switches back and forth between them: the half-centre idea. The account was little followed for decades, until work from the 1960s onward, including fictive locomotion in isolated cords, revived it. His experiment made rhythm a property that the spinal circuit can make by itself.

✓ verbatim Graham Brown, Proc. R. Soc. Lond. B 84:308–319 (1911), 'The intrinsic factors in the act of progression in the mammal'.
Test yourself
Question 1 of 8 Score 0 / 0

Which statement correctly defines a motor unit?
Sources & Integrity

Every structure, slide, simulation, and clip on this page is listed below with the source it came from. None of the 3D structures is drawn by hand or by AI: each one loads by accession ID from RCSB PDB or PubChem. This panel is generated from the lesson's own data file, so what it lists is what the page loads.

Molecular structures live data

  • Nicotinic acetylcholine receptor — turns each motor-neuron spike into a muscle response — RCSB PDB 2BG9 · coordinates CC0 · Torpedo marmorata electric-organ nicotinic receptor, cryo-electron microscopy of tubular crystals (Unwin, J. Mol. Biol., 2005)

Simulations

  • Small before large · Original interactive — Biochemistrypedia (Deppmann Lab).

Scientist profiles & quotes

  • Charles Sherrington — Liddell & Sherrington, Proc. R. Soc. Lond. B 96:212–242 (1924); Sherrington, Nobel Lecture, 'Inhibition as a Coordinative Factor', 12 December 1932.
  • Elwood Henneman — Henneman, Science 126:1345–1347 (1957); Henneman, Somjen & Carpenter, J. Neurophysiol. 28:560–580 (1965).
  • T. Graham Brown — Graham Brown, Proc. R. Soc. Lond. B 84:308–319 (1911), 'The intrinsic factors in the act of progression in the mammal'.

Quotes are verbatim from the cited source. Profile text is original to this site and was checked against the sources cited.

Animated explainer

One Stretch, Two Opposite Commands · narration: an AI narrator voice (Gemini, Charon) reading narration written from this lesson's text · animation: AI-generated in code (Manim) by Claude, a schematic with no drawn molecular structures, plots computed from their equations · captions from the narration script.

Images & figures

This lesson carries no raster figures. Every structure is rendered live from RCSB PDB or PubChem, and every plot is drawn from its equation in the browser.

integrity Machines check the mechanics. The build confirms that every local file exists, that every question's answer key points at a real choice, and that any AI-generated scientific figure names the person who checked it; a weekly job re-resolves every accession ID against its database. Questions are original; MCAT-style items are written for this site and are not real AAMC questions. Banned sources (the Tymoczko textbook, publisher decks, copyrighted figures) are not used.

Like I'm five

Your muscles are pulled by nerve cells in your back, inside the spinal cord. Each nerve cell pulls a small team of muscle threads. To lift something light, only small teams pull; for something heavy, bigger teams join, and every team pulls faster. The spinal cord can even keep your legs stepping in a rhythm without the brain counting the beats.
High school

A motor neuron and the muscle fibres it controls make a motor unit. Small motor neurons control few fibres and are the easiest to excite, so they switch on first; larger ones join as effort grows, and active ones can fire faster, so their twitches add up. Sensors called muscle spindles report stretch. A sudden stretch excites the same muscle through a single synapse and relaxes the opposing muscle through an inhibitory nerve cell, which is the knee jerk. Separate circuits in the spinal cord can produce the alternating rhythm of walking.
College

A motor unit is one alpha motor neuron and the fibres it innervates; the motor pool is all the motor neurons of one muscle. Under common synaptic input, small motor neurons have high input resistance, so a given current depolarizes them more (V = IR) and they are recruited first; they also make the weakest units. Force grows by recruitment and by rate coding, as twitches summate. Ia spindle afferents excite homonymous motor neurons monosynaptically and inhibit antagonist motor neurons through a Ia inhibitory interneuron on the same side. Tendon organs (Ib) signal force; the crossed-extensor reflex crosses the midline. Gamma motor neurons set spindle sensitivity, often coactivated with alpha motor neurons. Spinal central pattern generators produce locomotor rhythm without patterned input, and sensory input can reset them.
Graduate

Orderly recruitment follows from a common input current acting on motor neurons whose input resistance falls with size, though thresholds, persistent inward currents and neuromodulation also shape it, and the order is a tendency rather than a law. Human motor-unit decomposition now samples over a hundred units per participant in one muscle; in tibialis anterior and vastus lateralis, force-rate trajectories were heterogeneous and mostly fit best by a logarithm among the models compared, with net input inferred from force under a proportionality assumption. Fusimotor control is not fixed to alpha drive: during an instructed delay, human Ia firing fell for targets that would stretch the muscle, with no systematic EMG change, which the authors read as compatible with goal-dependent gamma drive. Open question: which spinal neurons generate the locomotor rhythm, given that silencing single genetically defined classes in mice has left rhythm in place, and how much of the mammalian picture, built on neonatal rodent cords, holds for adult human walking?
Walking without a brain or a sensory signal

If walking were a chain of reflexes, each step setting off the next, then removing sensory input from the legs should stop it. Graham Brown cut the spinal cord of cats and also cut the dorsal roots carrying sensory signals from the hind limbs. The ankle flexor and extensor muscles still contracted in alternation. He concluded that the rhythm arose within the cord and proposed two half-centres that inhibit each other. The experiment was largely overlooked at first; isolated spinal cords in a dish were later found to produce the same kind of alternating output.

— Graham Brown, Proc. R. Soc. Lond. B 84:308–319 (1911).
A spindle that gets ready before you move

Stylianos Papaioannou and Michael Dimitriou recorded single spindle afferents from wrist and finger muscles of volunteers with microneurography while they waited to reach toward a cued target. About 80 ms after the cue, Ia firing fell when the coming movement would stretch the muscle, while EMG and posture showed no systematic change with the target. The principal analysis rested on eight afferents from six people. The authors read the result as compatible with goal-directed changes in gamma drive; the gamma efferents themselves were not recorded.

— Papaioannou & Dimitriou, Science Advances 7:eabe0401 (2021).
Kicking the rhythm early

In spinal cords from mice one to three days old, kept in a dish with 7 µM NMDA and 8 µM serotonin, flexor-related and extensor-related roots fired in alternation. The team stimulated ankle nerves with brief trains of 4 to 6 pulses at 20 Hz at different points in the cycle. Delivered between flexor bursts, stimulation of either flexor-related or extensor-related ankle nerves could start a new flexor burst. The cord made the rhythm, and sensory input could reset it.

— Li, Garcia-Ramirez & Dougherty, Frontiers in Cellular Neuroscience 13:452 (2019).

 ↑ Top