Nociception is not pain, and the difference matters

One more piece of vocabulary has to be right before those three questions can be answered, because getting it wrong makes the rest of the paper look like it is describing something else.

Nociception is the signal a tissue sends when something is wrong with it. Specialized nerve endings called nociceptors detect damage, or the chemical conditions that go with damage, and they fire. That firing travels to the spinal cord whether or not the person notices anything.

Pain is what a person experiences. It is produced in the brain, after a great deal of processing, and it does not track nociception reliably in either direction. A tissue can be sending nociception with no pain reported at all. A person can be in severe pain with very little nociception arriving.

So a stimulus can be nociceptive without hurting. This paper depends on that distinction the whole way through. When it says a second nociceptive input, it does not mean hurting the patient, and the patient does not have to feel anything for the effect to appear.

The fibers carrying this traffic out of muscle and joint are the small ones, the group III and group IV afferents named earlier in the numbering scheme. Group III endings respond mainly to mechanical loading. Group IV endings respond to the chemical byproducts of tissue activity.

Question one: what makes a muscle test weak

Start with what is intact.

Under experimental pain, produced in the laboratory by injecting saltier-than-blood saline into a muscle, maximal voluntary force falls while the muscle's evoked twitch is unchanged, so the contractile machinery is working normally [10]. Evoked twitch means the researcher stimulates the muscle's own nerve electrically, bypassing the person's intention entirely, and measures the resulting contraction. If the muscle's own machinery had failed, that twitch would shrink. It did not. The same fall in strength under experimental knee pain has been measured with an instrumented dynamometer rather than a twitch [11].

Schabrun and colleagues found the same [12]. During experimental muscle pain, stimulating the motor cortex produced a smaller muscle response. Stimulating the muscle's own nerve produced the same response as before the pain. The muscle and its nerve were fine, but the brain and cord were driving it less.

Farina and colleagues raised the nociceptive signal coming from the tibialis anterior in three steps, using saltier-than-blood saline, while subjects held a light contraction. The same motor units fired a little slower at each step, and their firing tracked how much pain the subjects reported. Force was held at the target and the muscle fibers did not change [37]. The more signal the tissue sends, the further motor unit firing falls.

What produces the fall is a tissue generating nociception. Group III and IV afferents from that tissue enter the back portion of the spinal cord, called the dorsal horn, and drive inhibitory interneurons onto the motorneuron pools serving that region [13, 29]. An interneuron is a short local nerve cell that connects one cell to another inside the cord. An inhibitory one makes the cell it connects to less likely to fire.

Now the gain matters, and this is where the two halves of the paper meet.

Shutting the persistent inward current off requires active inhibition arriving at the cell, and once inhibition does arrive the current gives way easily. A small fraction of the cell's inhibitory synapses can nearly abolish it [14,4]. That is a strange and important asymmetry. It is hard to build the gain up, because that needs monoamines delivered from the brainstem, and it is easy to knock it down, because a little inhibition arriving at the right place does it.

Low-threshold stimulation of a peripheral nerve carrying flexion reflex afferents can suppress it completely in an extensor motorneuron [1]. Flexion reflex afferents are the fibers that drive the withdrawal reflex, the one that pulls a hand off a hot stove. Nociceptive afferents are flexion reflex afferents. The same thing has been seen in people. Revill and Fuglevand held a steady electrical stimulus on the sural nerve, a skin nerve on the outside of the foot that carries flexion reflex afferents, while subjects slowly raised and lowered the force of tibialis anterior, the muscle that lifts the foot. The stimulus ran at about eleven times the level at which subjects first felt it (probably nociceptive even though nociception was not mentioned). Without the stimulus, motor unit firing climbed steeply just after each unit switched on, the signature of the persistent inward current. With the stimulus, that steep climb disappeared in some units and was much flatter in the rest, so most of the amplification was gone during the phase when force was being built. Every subject reported that the same force took more effort with the stimulus on, which is what a suppressed amplifier would require [32].

Hug and colleagues then measured the human estimate of the gain itself during experimental pain. Saltier-than-blood saline was injected into the fat pad below the kneecap, tissue that is not muscle, while subjects held a moderate contraction of the thigh. In the vastus lateralis the gain estimate fell about 12 percent while the pain lasted and firing returned to normal when the pain passed [36]. The firing-pattern measures of the serotonin and noradrenaline supply held steady, so the authors attribute the drop to inhibition [36]. Nociception knocked the gain down without touching its supply, which is the asymmetry this paper argues: the gain is hard to build up and easy to knock down. The drop in gain tracked the drop in firing (r = 0.61) [36]. Losing gain meant losing output.

The pain did not come from the muscle. It came from the fat pad behind the kneecap, and it was only moderate, about 4 out of 10. Even so, the thigh muscle lost gain [36]. That is the clinical picture: the source is usually a joint, tendon, or other tissue rather than the muscle itself, and the patient may report little or no pain. Weakness through the nervous system does not need an injured muscle or severe pain.

Steffens and Schomburg, recording in a decapitated high-spinal cat, delivered a nociceptive input first, which is what conditioning means, and found it reduced the direct one-synapse input running from the muscle's stretch receptors to the alpha motorneurons [13]. Nothing in that circuit requires a brain, and nothing requires the patient to feel anything.

That preparation is worth pausing on, because it settles something a reader will otherwise keep wondering about. The animal had no brain in the circuit at all. There was no consciousness, no attention, no expectation and no pain in any meaningful sense. Nociceptive input still turned motor output down. Whatever is happening here, it is not the patient deciding not to push.

The motor unit studies above held a set target force during pain, and a manual muscle test still grades weak. Martinez-Valdes and colleagues explain the gap. They tracked the same tibialis anterior motor units during saltier-than-blood saline at a light and at a hard contraction. At the light one the units slowed. At the hard one the small units held their rate while the large units fired faster and switched on earlier, carried by extra drive from the brain. The authors conclude that this compensation cannot reach maximum force, since high forces during pain can be reached only as far as voluntary drive can be raised [41]. A 5/5 grade asks for maximal resistance, the force the authors say this compensation cannot reach. In Hug's study, about 15–25% of motor units did not slow during pain [36, 41]. The pool spreads the work to hold a set force, which is why a lab task can look normal while a maximal manual muscle test grades weak.

A weak grade is a motorneuron pool that has lost its gain.

The same nociceptive signal that suppresses the PIC at the spinal cord continues upward, increasing inhibition in the motor cortex [15,16]. This reduces the motor command descending from the cortex to the motorneuron pool. Similar changes have been found in people with musculoskeletal injury [17,18].

A standing nociceptive tissue acts on its own region while the patient sits still. It produces the baseline weak grade found on the first pass, and it does not reach across the body on its own [19,20]. That last point becomes important in a moment.

The exception is a muscle with no drive left to restore. Denervation, myopathy, motor neuron disease. In an outpatient orthopedic caseload it is rare [21], and a clinician taking a good history up front helps flag it.

Question two: why loading that muscle weakens everything else

Two steps, and the first is easy to miss.

Contraction turns the existing source up. The tissue with the condition in it is the tissue being loaded. Group III endings are mechanically sensitive and group IV endings respond to the metabolic byproducts of contraction, and sensitization drops the threshold of both. Sensitization means the endings have become easier to set off than they were when the tissue was healthy. So ordinary movement now fires endings that used to stay quiet. When a patient bends forward as if doing dishes, the painful low back is being loaded, and its afferent output climbs. Using an inhibited muscle raises the output of the nociceptive driver that was already sitting there.

The same small fibers are part of why a muscle's own motor units slow during ordinary sustained use. In healthy people holding a steady contraction for four minutes, motor unit firing in the tibialis anterior drifted down as the hold went on. With saltier-than-blood saline in the muscle, the units started about 0.7 pulses per second slower from the first seconds and slowed little further, and after about ninety seconds the two conditions were firing alike [38].

This is the step that makes the whole thing counterintuitive. Asking a patient to use the weak muscle is not a neutral act of measurement. It is an intervention that increases the very signal causing the primary weakness.

Above a certain traffic level the effect stops being regional. The spinoreticular tract carries nociceptive input to the reticular formation, a network in the brainstem, and the reticulospinal system carries inhibitory commands back down from there to motorneuron pools throughout the spinal cord. Its reach is broad rather than confined to the body region the input initially came from. This is what the reticular formation does with the nociceptive traffic arriving there. That is why a nociceptive input from one region of the body is felt well past that region, which is what a body-wide drop in previously strong muscles looks like [22,23,24].

The reticulospinal system does more than one job, and the jobs look different. One set of reticulospinal neurons helps coordinate voluntary movement. Recorded in monkeys during reaching, single neurons of this kind helped the flexors and held back the extensors of one arm, and did the reverse in the other arm [42, 43]. That pattern coordinates the two limbs. A separate inhibitory region in the medial medulla suppresses motor activity across the whole body, in all four limbs at once [44]. Inhibition of the same kind, delivered onto motorneurons as glycine and gamma-aminobutyric acid (GABA), is what produces the paralysis of REM sleep [45]. The drop of previously strong muscles under an active nociceptive input has the shape of this second output: every strong muscle, in all four limbs.

The same drop has been measured in the motor units of a muscle with no source of its own. Saltier-than-blood saline was injected into the shin muscle, or into the hamstrings, while subjects held a steady knee extension. In both cases the quadriceps motor units fired slower than they did without pain, although nothing the quadriceps did could have unloaded the shin [40].

Two published pieces sit close to this. Sidhu showed fatiguing contraction raising inhibition inside the motor cortex through group III and IV feedback, which is the afferent class doing exactly what the model needs, and the measurement was taken in the muscle doing the work [25]. Simis recorded motor cortex inhibition at a hand muscle in knee osteoarthritis patients, scaling with the knee's cartilage grade rather than with reported pain [26]. That one is remote inhibition from a joint source, measured. A knee problem changed muscle activation at a hand.

Question three: why a second nociceptive input turns the weak muscle on

This is the part that sounds impossible, and it has a published mechanism.

Nociceptive afferents from anywhere on the body converge on the subnucleus reticularis dorsalis (SRD) in the caudal medulla, one of the relays for this effect sitting above the spinal cord. The medulla is the lowest part of the brainstem, where it joins the spinal cord. This nucleus receives nociceptive traffic from the entire body rather than from one region, which is the unusual thing about it.

It projects back down to every spinal segment and depresses the convergent dorsal horn neurons, the ones that take nociceptive and ordinary input together. Le Bars, Dickenson and Besson measured depressions of sixty to a hundred percent in those neurons in the rat, outlasting the stimulus by minutes [27,9]. Sixty to a hundred percent is close to a shutdown, and it happens at the first synapse, before the signal has gone anywhere.

So the effect of adding nociception somewhere else is that the cord stops passing along nociception generally, including the traffic that was already running from the original tissue.

The contact changes nothing in the first tissue. Those nociceptors keep firing at the rate they were firing beforehand. Less of that firing crosses the first synapse [28]. Fewer inhibitory interneurons get driven. The gain switches back on and the muscle grades strong.

Read that sequence again, because it is the answer to Part 2 of the opening. Nothing was repaired. The shoulder that graded 4-/5 is in exactly the condition it was in a second earlier. What changed is how much of its nociceptive traffic is reaching the motorneuron pool, and with less inhibition arriving, the gain comes back.

The same pattern has been measured as force. In knee osteoarthritis the quadriceps already fails to activate fully. Rafsanjani and colleagues stimulated the common peroneal nerve at about twice tibialis anterior motor threshold while the person pushed. Force rose and the activation failure shrank while that input was on [31]. Stimulation at that level is most likely strong enough to recruit the nerve's group III fibers, offered here as an inference. An already-inhibited pool, a second input held during the contraction, voluntary force going up.

There is a second route running at the same time. The SRD is also wired into the reticulospinal neurons, and the descending damping system releases serotonin, noradrenaline and endogenous opioids. Those are the transmitters that set the gain. So one contact opens the gate and raises the gain setting, and the projection carrying the gain has no capacity to aim. Agonist and antagonist rise together [9,1].

Both halves of that have now been measured together in the same people, in a pilot. Healthy men held a foot in ice water for two minutes. Afterward, the brain's response to a painful pulse at the finger was smaller in every subject, and in the two whose motor units were recorded, the gain estimate in a leg muscle rose, though not by enough to reach statistical significance. The authors, who include Heckman, attribute both changes to more serotonin and noradrenaline coming down from the brainstem [39].

Why the same contact does the opposite thing to a strong muscle

Here is the objection a careful reader should already be forming. If the contact raises gain everywhere, why does anything get weaker?

Because two things arrive at every muscle, not one.

Afferents from the added input also drive their own inhibitory interneurons, and that traffic reaches whatever pools it maps to. Two influences arrive at every muscle at once, and their ranking is known. Inhibition outweighs neuromodulation at the motorneuron, because so few inhibitory synapses abolish the current [1]. Neuromodulation here means the serotonin and noradrenaline that set the gain, as against the inhibition arriving at the cell directly.

Follow both cases. A muscle carrying its own standing source has a large suppression waiting to be lifted, and lifting it dominates everything else happening to it. A muscle with no standing source has nothing available to lift, so the arriving inhibition is the only change it sees. Same contact, opposite outcomes, decided by what each pool was already carrying. No part of the circuit has to identify a target for this to work.

That last sentence matters more than it looks. Nothing in this account requires the nervous system to know which muscle is the injured one, or to select a target, or to do anything clever. One input does one thing everywhere. The difference in outcome comes entirely from the state each pool was in beforehand.

Back to the two puzzles

Part 1. The right shoulder tests 4-/5 because there is a nociceptive source in that region, sending traffic that drives inhibitory interneurons onto that pool and knocks its gain down. The left tests 5/5 because it carries no such source.

Then a weight goes into the right hand with the arm held out. The affected side is now being loaded, its sensitized group III and IV endings fire harder, and its afferent output climbs. Above a certain level that traffic stops being a local matter, because the spinoreticular route carries it to the brainstem and the reticulospinal route carries the consequence back out to pools everywhere. The left shoulder, which had nothing wrong with it, drops to 4-/5.

Nothing happened to the left shoulder. It got caught in a body-wide change produced by loading the right one.

Part 2. A tender spot on the right side of the neck is a second nociceptive source. Palpating it adds nociceptive input, and the palpation is held while each retest is made.

Retest the right shoulder with the contact held and it grades 5/5. The added input engaged the medullary damping system, less of the original neck-and-shoulder traffic is crossing the first synapse in the cord, fewer inhibitory interneurons are being driven onto that pool, and the gain comes back.

Palpate the same spot, hold it, and retest the left shoulder, and it drops to 4-/5. That pool had no standing suppression to lift. For it, the added input is simply new inhibitory traffic arriving, and inhibition outweighs the gain effect at the motorneuron.

One contact. Two muscles. Opposite results, from the same mechanism, decided by what each side was carrying before the finger landed.

Four things this accounts for

1. The contact does not have to hurt, since the steps that produce the reversal run below the cortex.

2. The speed follows from the same anatomy, because conduction and synaptic gating take milliseconds and nothing has to circulate or remodel.

3. The strength lasts as long as the input does. The medullary damping runs while the input is running, and the original source transmits throughout. Take the input away and the muscle grades weak again. The weakness was never removed, only outvoted.

4. A wrong spot costs nothing. An input landing where there is no condition in the tissue drives the large myelinated touch fibers and generates no nociceptive traffic in the group III and IV fibers, so it sets off neither the inhibition that turns a pool down nor the medullary damping.

The third of those deserves emphasis, because it is the honest limit on the whole finding. The reversal is not a treatment and it does not fix anything. The source is still there, transmitting, the entire time. The damping system is simply outvoting it for as long as the input is there.

What is published, and what is not

Every link above is published. The chain has not been run end to end.

Adding a second input has been measured as force once, in the knee osteoarthritis study described above. Quadriceps force rose while the common peroneal nerve was stimulated [31]. Taking nociception away has been measured the same way, with force numbers, in studies of subacromial anesthetic injection and of draining an effused joint [50, 51].

Adding it has also been measured as inhibition, and the evidence there is a pair of experiments from Schabrun's group using the same muscle and the same injection of saltier-than-blood saline, differing in one thing. In the 2013 experiment a second input, electrical stimulation of the ulnar nerve at the wrist, ran throughout the motor recording, and motor output was not suppressed while the person was in pain; the suppression appeared only after the pain had resolved [30]. In the 2015 experiment no such input ran, and the suppression appeared during the pain itself [12]. The outcome, force rising in an already-inhibited muscle during a second input, is established. A human estimate of the gain fell about 12% during moderate experimental pain from tissue outside the muscle [36]. What remains unmeasured is whether the rise in force during a second input runs through the gain coming back. The first attempt to measure the gain after a remote painful input points that way [39].

Two parts of the account above go beyond what the literature has established. That the dorsal horn damping and the restored gain are the same event in a patient has never been measured at both ends. And the route from a held load to a body-wide drop, spinoreticular in and reticulospinal out, is the path the anatomy provides rather than an experiment anyone has run.

The mechanism is established physiology. That it produces a change in MMT grade has not been measured by anyone.

What would test this, and what would break it

The physiology is published. Of the clinical measurements, only the rise in force in an already weak muscle has been made [31]. These are the results the account requires.

A dynamometer, not a hand grade. One already-weak muscle and one already-strong muscle, same person, same session. Hold the second input while force is recorded.

• A held second nociceptive input raises force in the weak muscle while it is held, and force returns to baseline when it is released. The muscle itself is unchanged.

• The same contact lowers force in the strong muscle.

• A non-nociceptive contact does neither.

• Reported pain is not required.

• Loading the weak muscle drops previously strong muscles elsewhere.

• Evoked twitch and the afferent volleys stay flat while voluntary force changes.

The account is wrong if the weak muscle does not gain force during a real second nociceptive input, if both muscles move the same way, if light touch on a clean spot produces the same flip, or if twitch and voluntary force rise and fall together.

A missed spot is not a failed test. The account already says a wrong spot does nothing. A confirmed rise is still not a treatment. The source was outvoted while the input was there, not removed.