5.22 The Motor Cortex and the Descending Motor Pathways

Halfway through a heavy pull, your training partner shouts — and the bar suddenly moves better. Your muscles didn't change in that half-second. The signal driving them did.

‌

That's the real story of strength. It isn't just muscle — it's a chain of command running from an idea to a contraction, and every link in the chain is trainable.

‌

Here's the chain in one breath. Your prefrontal cortex decides the lift is worth doing. Premotor and supplementary motor areas choose the movement and put its parts in order. Primary motor cortex sets the force and the direction.

‌

Then the corticospinal tract — roughly a million axons — carries the command down to spinal interneurons and motor neurons. And running alongside it are four brainstem pathways (reticulospinal, vestibulospinal, rubrospinal and tectospinal) that handle postural tone, anti-gravity drive, orienting, and a big share of the gross force in whole-body athletic actions.

‌

See who does what, and three gym mysteries dissolve at once: why rate of force development is trainable separately from peak force, why arousal changes strength within seconds, and why fine skill and gross power respond to different training.

‌

Key takeaways

1. One tract, a million wires. The corticospinal tract contains roughly one million axons per side in humans — and only about half of them start in primary motor cortex. The rest come from premotor, supplementary motor and parietal areas.
2. There's no muscle keyboard. Primary motor cortex neurons don't encode single muscles. They encode movement direction and force through population activity, and microstimulation produces coordinated multi-joint movements rather than isolated twitches.
3. Cued and self-started are different skills. Premotor cortex selects actions cued from outside; the supplementary motor area sequences actions generated from inside. That's why an athlete can look fluent in a cued drill and disorganised the moment they have to self-initiate.
4. The brainstem lifts with you. Brainstem pathways — especially the reticulospinal tract — drive postural tone and gross force, and they're highly sensitive to arousal. That's the most likely explanation for how a shout or a competitive setting adds force within seconds.
5. The first 100 ms belongs to the nervous system. Early force is dominated by neural drive — motor unit discharge rate, doublet firing and synchronisation — while peak force is dominated by muscle cross-sectional area. They're separately trainable.
6. Your wiring adapts today. Corticospinal excitability measured with transcranial magnetic stimulation changes within a single training session, well before any structural adaptation is possible.
7. The wires cross. The tract crosses at the medullary pyramids, which is why a one-sided cortical lesion causes weakness on the opposite side of the body.

‌

Beginner section: How a decision becomes force

‌

From decision to lift-off: the relay

Stand at a barbell and decide to pull. Between that decision and the bar leaving the floor, a specific, traceable sequence fires — and every stage of it is a real piece of tissue you could point at.

‌

‌

Stage one: your prefrontal cortex settles the argument. Pull now — not another grip adjustment, not a step away. Choosing an action also means suppressing its competitors.

‌

Stage two: the premotor and supplementary motor areas pick which version of the pull you're about to run, and order its components in time.

‌

Stage three: primary motor cortex — the strip of tissue running over the top of your brain — specifies how hard and in which direction.

‌

Stage four: a bundle of about a million nerve fibres carries that command down through the brainstem and into the spinal cord, where it's handed to the motor neurons that actually contract the muscle.

‌

That's the voluntary route — the one most people know about. Here's what most people don't know: it's not the only road down. For a big whole-body lift, it may not even be the dominant one.

‌

‌

The four routes you can't feel

Four more pathways descend from the brainstem rather than the cortex. You have no conscious access to any of them — and you'd fall over immediately without them.

‌

The reticulospinal tract supplies gross force and postural tone, and it's extremely sensitive to how aroused you are. That mid-lift shout? This is the pathway it most plausibly speaks to.

‌

The vestibulospinal tract reads the balance organs in your inner ear and drives your anti-gravity muscles — the ones holding you upright without ever being asked.

‌

The rubrospinal and tectospinal tracts round out the crew: a flexor bias for your limbs, and the reflex that turns you towards things you see or hear.

‌

Six terms that unlock everything else

• Primary motor cortex (M1) — the strip of cortex immediately in front of the central sulcus. The final cortical stage before the spinal cord.
• Corticospinal tract — the bundle of axons running from cortex to spinal cord. Also called the pyramidal tract, because it crosses at the medullary pyramids.
• Motor unit — one alpha motor neuron plus every muscle fibre it innervates. The smallest unit of force you can produce on purpose.
• Rate of force development — how quickly force rises once a contraction starts, usually measured over the first 50 to 250 milliseconds.
• Reticulospinal tract — a brainstem pathway supplying postural tone and gross force, strongly modulated by arousal and startle.
• Motor evoked potential — the muscle response recorded after a single magnetic pulse over motor cortex. The standard human measure of corticospinal excitability.

‌

Two systems, two training menus

Here's the distinction worth carrying into every session you ever write.

‌

Fine, precise, distal skill is corticospinal territory. It improves through attentive, specific, moderately loaded practice — crisp technique work, done fresh.

‌

Gross, fast, whole-body force leans heavily on the brainstem routes. It improves through intent, arousal, and heavy or explosive loading.

‌

You can watch the split play out across sport. A fighter's pad work with called combinations feeds cued selection; sparring feeds self-started sequencing — and one doesn't cover for the other. A thrower needs delicate release timing and savage trunk drive, built by different work.

‌

One warning on the arousal lever: it's acute, and it's a limited resource within a session. Save the noise for the attempt that counts.

‌

A programme that only feeds one system leaves the other underdeveloped. All quiet technique, or all heavy fury — either way, half your chain stalls.

‌

The graph that separates athletes with identical strength

Now for the most useful picture in this article.

‌

‌

Two athletes can reach the same peak force and be completely different competitors. One reaches 60 percent of that force in 100 ms. The other needs 250 ms. Same ceiling, different lift-off.

‌

That gap decides contests, because almost every sporting action that matters is over before peak force could have been reached. The early slope of the curve is often more important than its height.

‌

And the early slope is neural. The first 100 ms of a force-time curve is dominated by neural drive — motor unit discharge rate, doublet firing and synchronisation — while peak force is dominated by muscle cross-sectional area.

‌

Separate causes mean separate training. Better still, the wiring answers fast: corticospinal excitability shifts within a single session, long before any structure could have changed. You don't have to wait for bigger muscles to start building a steeper start.

‌

Want the whole pathway in two minutes — including where it crosses sides and why that matters clinically? Press play.

‌

‌

Advanced section: Coding, tract composition, and the neural determinants of force

What M1 neurons actually encode

The naive model of primary motor cortex is a keyboard: One cortical site per muscle. That model has been dead since the 1980s. Georgopoulos and colleagues recorded from M1 while monkeys reached in different directions and found that individual neurons had broad directional tuning, firing most for one preferred direction and progressively less as the reach deviated from it (Georgopoulos et al., 1986). The population as a whole codes direction accurately even though no single neuron does, an arrangement now called population vector coding.

‌

Later work sharpened the picture further. Graziano and colleagues showed that microstimulation of precentral cortex for behaviourally realistic durations evokes coordinated, multi-joint, goal-directed movements — a hand brought to the mouth, a defensive blink and shoulder shrug — rather than isolated muscle twitches (Graziano et al., 2002). Schieber demonstrated that the territory controlling individual fingers overlaps extensively rather than forming discrete zones (Schieber, 2001). Evarts had already shown that many M1 cells relate more closely to force and rate of change of force than to movement displacement (Evarts, 1968).

‌

‌

Taken together, the modern reading is that M1 sits at an intermediate level of abstraction. It is not commanding muscles, and it is not commanding goals. It is commanding movement fragments defined in terms of direction, force and useful synergies. For coaching, that has a direct implication: Instructions framed as coordinated actions with a direction and an intent map more naturally onto what the cortex represents than instructions framed as individual muscle contractions.

‌

Premotor cortex, SMA, and the difference between cued and self-initiated action

Premotor cortex divides into dorsal and ventral regions. Dorsal premotor cortex is heavily involved when an arbitrary external cue has to be mapped onto a response — a light means step left, a whistle means sprint. Ventral premotor cortex is closely tied to grasp: Its neurons code the shape of the hand required for an object, and it is the region where mirror properties were first described in macaques (Kandel et al., 2021).

‌

The supplementary motor area, by contrast, becomes most active for internally generated and sequenced movement. Classic imaging and lesion evidence shows SMA involvement scaling with the complexity of a learned sequence, and SMA damage impairing bimanual coordination and the ability to start movements without external prompting (Tanji, 2001). Pre-SMA and SMA proper differ, with pre-SMA more involved in learning a new sequence and SMA proper in running a well-learned one.

‌

This distinction is visible on any training field. An athlete who performs beautifully when a coach calls the cue and falls apart in open play has adequate premotor mapping and inadequate internally generated sequencing. The fix is not more cued drilling. It is removing the cue and forcing self-initiation, which is precisely what small-sided and constraint-led practice does.

‌

Tract composition, conduction velocity, and why the crossing point matters

The human corticospinal tract carries roughly one million axons on each side (Lemon, 2008). Only about 3 percent are the large, fast, thickly myelinated fibres arising from the giant Betz cells of layer V; the fastest of these conduct at around 70 metres per second (Lemon, 2008). The great majority are small and slow. Around 30 to 40 percent of the tract originates outside M1, in premotor, supplementary motor, cingulate and parietal cortex, which is why lesions restricted to M1 do not abolish movement (Lemon, 2008).

‌

Roughly 85 to 90 percent of the fibres cross the midline at the medullary pyramids to form the lateral corticospinal tract, with the remainder continuing uncrossed as the anterior corticospinal tract, which serves mainly axial and proximal musculature bilaterally (Nathan et al., 1990). This anatomy is why a stroke in one hemisphere weakens the opposite limbs while trunk function is comparatively spared.

‌

A distinctively human feature is the density of direct, monosynaptic connections from corticospinal axons onto alpha motor neurons supplying hand muscles (Lemon, 2008). Most descending control in other mammals goes through interneurons. That direct connection is the anatomical basis for independent finger movement, and it is also why hand dexterity is disproportionately vulnerable to corticospinal damage.

‌

The neural determinants of rate of force development

When a contraction begins, three neural variables dominate the first hundred milliseconds. The first is motor unit discharge rate: Trained explosive athletes achieve much higher initial firing frequencies, with reported instantaneous rates well above 100 Hz at contraction onset (Van Cutsem et al., 1998). The second is doublet firing, in which a motor neuron fires two action potentials separated by less than about 10 ms, producing force out of proportion to the mean rate (Van Cutsem et al., 1998). The third is the earliness of recruitment: In ballistic contractions the size principle is compressed in time, so high-threshold units are recruited almost immediately rather than progressively.

‌

Aagaard and colleagues showed that heavy resistance training increases the electromyography (EMG) rate of rise and V-wave responses, indicating enhanced efferent drive rather than only muscular change (Aagaard et al., 2002). Van Cutsem and colleagues showed that dynamic training at high velocity increased both discharge rate and doublet incidence (Van Cutsem et al., 1998). Duchateau and Hainaut had earlier demonstrated that training with ballistic intent produced different adaptations from training with the same load and slower intent, which is the experimental foundation for the coaching instruction to move the bar as fast as possible even when it is heavy (Duchateau & Hainaut, 1984).

‌

The reason this matters is that rate of force development (RFD) is not a by-product of getting stronger. It is a partly independent quality with its own training requirements: High intent, low fatigue, short exposures, and a bias towards ballistic or heavy-and-fast contractions rather than long grinding sets.

‌

Reticulospinal drive, arousal, and the StartReact phenomenon

The reticulospinal system is the least discussed and possibly the most relevant descending pathway for whole-body athleticism. It is diffuse, largely bilateral, biased towards proximal and axial muscle, and heavily modulated by arousal (Baker, 2011). Its output can be probed indirectly in humans using the StartReact effect: If a loud unexpected sound is delivered at the same time as a go cue, the prepared movement is released 40 to 80 ms earlier than normal and often with greater initial force, which is faster than any voluntary reaction could be (Valls-Solé et al., 1999; Carlsen et al., 2012). The prevailing interpretation is that the motor programme was already stored subcortically and the startling stimulus released it directly (Carlsen et al., 2012).

‌

That result reframes several familiar gym phenomena. A training partner shouting at the start of a lift, competition adrenaline adding kilos to a squat, and the extra height produced by a loud clap on a jump cue are all consistent with acute reticulospinal facilitation. None of these change muscle properties in the time available; they change descending drive.

‌

It also sets a limit. Arousal-driven facilitation is acute and non-cumulative. It can be spent on a competition attempt or a key training set, and it cannot be used as a substitute for the slow work of raising force capacity. Chronic over-arousal, in the sense discussed in 5.13, has costs of its own.

‌

Measuring the pathway in humans, and reading the literature critically

Four measurements dominate human research and each has a distinct interpretation. Motor evoked potentials from transcranial magnetic stimulation index overall corticospinal excitability but cannot separate cortical from spinal contributions. Short-interval intracortical inhibition and the cortical silent period index GABAergic cortical inhibition specifically (Roshan et al., 2003). Cervicomedullary stimulation isolates the subcortical portion of the pathway. The V-wave and H-reflex index spinal motor neuron excitability and reflex gain respectively (Aagaard et al., 2002).

‌

Because these measures are frequently conflated in popular summaries, claims that a supplement or a device improved neural drive should be read carefully. An increase in a motor evoked potential without a change in force is a physiological observation, not a performance finding. Conversely, an improvement in performance without any change in these measures does not disprove neural adaptation, because the relevant change may be in coordination rather than excitability.

‌

The defensible general statement is this: Corticospinal excitability changes measurably within a single session of skill practice or resistance training, inhibition decreases early in learning, and the magnitude of these changes correlates only loosely with performance (Carroll et al., 2001). Neural adaptation is real, fast and important, and it is not something you can read off a single number.

‌

Practical section: Training the drive, not just the muscle

Everything above converges on a small number of programming decisions. The list below is ordered by how much difference it usually makes.

‌

1. Train intent explicitly. For any set intended to develop rate of force development, the instruction is to accelerate maximally regardless of how slowly the bar actually moves. Duchateau and Hainaut showed that identical loads produce different neural adaptations depending on intent (Duchateau & Hainaut, 1984).
2. Keep explosive work early and short. Discharge rate and doublet firing are the first things fatigue degrades. Three to six sets of one to five fast repetitions with full recovery is a complete dose; adding a seventh set trains something else.
3. Separate skill practice from fatigue. Corticospinal learning happens best when the movement can still be performed correctly. Once technique drifts, you are consolidating the drift.
4. Use arousal deliberately rather than constantly. Save shouting, music and competitive framing for attempts that matter. If every set is maximal arousal, you lose the acute effect exactly when you need it.
5. Include heavy isometrics and heavy partial-range work when you want maximal drive without high mechanical strain. Maximal voluntary isometric efforts produce very high descending drive with minimal eccentric load.
6. Train both cued and self-initiated versions of key actions. Cued drills develop premotor mapping; self-initiated open play develops supplementary motor sequencing.
7. Respect the exercise specificity of neural adaptation. Corticospinal changes are largely task-specific, so transfer from a machine to a free-weight version of the same joint action is smaller than the muscular similarity suggests (Carroll et al., 2001; Nudo et al., 1996).

‌

One measurement worth adopting if you have access to it: Track a jump or a throw at the start of a session rather than only at the end of a block. Rate-of-force-development qualities are sensitive enough to show fatigue and readiness on a daily basis, which makes them a better daily monitor than one-repetition maximum, which is both slow to move and expensive to test.

‌

A final caution about electrical and magnetic devices marketed to athletes. Transcranial direct current stimulation and related tools do measurably alter cortical excitability in laboratory conditions. The performance literature in trained athletes is small, heterogeneous and frequently unblinded, and effect sizes are inconsistent. Treat the mechanism as real and the performance claim as unproven, and read article 5.19 before spending money.

‌

Sport applications

In sprinting the first two steps out of blocks are close to a pure test of early-phase force development, and the auditory gun start recruits exactly the StartReact mechanism described above. Blocks work should therefore be done fresh, in low volume, and with a sound cue rather than a visual one.

‌

In throwing and striking sports the corticospinal and reticulospinal contributions are both large: Fine distal timing at release or impact is corticospinal, while the proximal trunk and hip drive that generates the speed is heavily brainstem-mediated. This is the anatomical reason that throwers need both precise skill work and gross power work, and that neither substitutes for the other.

‌

In combat sports the SMA distinction is decisive. Pad work with called combinations trains cued selection. Sparring and open drills train internally generated sequencing. Fighters who only do pad work often look flat in the ring, and the deficit is not conditioning.

‌

In team sports the same logic applies to change of direction. Pre-planned cone patterns train the movement; reactive small-sided play trains the selection and initiation. Both are needed and the pre-planned version is the easier one to over-prescribe.

‌

In strength sports, the practical value of this article is mostly in the concept of intent. The heaviest single you will ever lift moves slowly, and it moves at all because you tried to move it fast.

‌

Common mistakes

• Training explosive qualities when already fatigued. Discharge rate and doublet firing collapse first. A fast set at the end of a hard session is a slow set with good intentions.
• Believing peak strength automatically produces rate of force development. They correlate, but imperfectly, and the early portion of the force-time curve has its own trainable determinants.
• Cueing individual muscles for complex actions. Motor cortex represents directions, forces and synergies. Ask for a direction and an intent.
• Using maximum arousal in every set. Acute reticulospinal facilitation is a limited resource within a session and a poor long-term strategy.
• Assuming a change in a motor evoked potential means better performance. Excitability and output are related but not interchangeable. Ask for a performance outcome as well.
• Only ever practising with an external cue. That develops cued selection while leaving self-initiated sequencing untrained, which is what open competition demands.
• Expecting neural adaptations to transfer broadly. They are largely task-specific. Similar muscles does not mean similar neural demand.

‌

Coaching cues

• Move it fast even when it is heavy. The intent is the stimulus.
• Fast work first, heavy work second, everything else after that.
• Ask for a direction and an outcome, not a muscle.
• Save the noise, the music and the crowd for the attempt that counts.
• If technique changes, the set that develops skill is over.
• Alternate cued drills with open, self-started versions of the same action.
• Test a jump at the start of the session to see what you actually have today.

‌

FAQs

Is it true that we only use a fraction of our muscle, and the brain holds back the rest?

The popular version of this claim is wrong, but there is a real phenomenon underneath it. Healthy motivated adults can usually activate close to full capacity in a maximal voluntary contraction, with a typical voluntary activation deficit of only a few percent in familiar movements, measured by superimposing an electrical or magnetic stimulus on a maximal effort. The deficit is larger in unfamiliar tasks, in the presence of pain, and in fatigue (Gandevia, 2001). So there is a small reserve, not a hidden 70 percent.

‌

Why do I lift more in competition than in training?

Several things stack. Acute arousal increases descending drive, particularly through the reticulospinal system. Reduced inhibition and pain tolerance play a part. You are also typically better rested and better prepared, and you are attempting once rather than repeatedly. The effect is real and usually worth a few percent, and it cannot be banked or repeated indefinitely.

‌

Does electrical muscle stimulation train the nervous system?

Not in the way the marketing implies. Peripheral electrical stimulation activates motor axons directly, largely bypassing the descending pathways that voluntary training targets. It has genuine uses in rehabilitation when voluntary drive is impaired, and it produces some cross-over into voluntary strength, but it is not a substitute for voluntary effort in a healthy athlete because the thing you most need to train is the drive itself.

‌

How fast do neural adaptations happen?

Measurable changes in corticospinal excitability and intracortical inhibition occur within a single session. Increases in voluntary activation and rate of force development typically appear across the first two to four weeks. Muscle cross-sectional area changes meaningfully from around three to six weeks onwards. That sequence is why early strength gains outpace visible muscle growth.

‌

Should I train explosively every session?

A small amount of high-intent work at the start of most sessions is well tolerated and helps maintain rate-of-force-development qualities. What does not work is high volumes of explosive work in a fatigued state, because the specific neural features being targeted are the ones fatigue degrades first. Quality and freshness matter more than total amount here than almost anywhere else in training.

‌

Does the brain control left and right sides completely separately?

Mostly, but not entirely. The lateral corticospinal tract is crossed and controls the opposite limbs, and this is why one-sided cortical damage produces one-sided weakness. However, about 10 to 15 percent of fibres remain uncrossed and serve axial and proximal muscles bilaterally, and brainstem pathways are largely bilateral. This is part of why trunk control is relatively preserved after a hemispheric stroke, and part of why unilateral training produces cross-education effects.

‌

What is the practical difference between the pyramidal and extrapyramidal systems?

The terms are old and somewhat unhelpful, but the underlying distinction is useful. Pyramidal refers to the corticospinal tract and its fine, fractionated voluntary control. Extrapyramidal was historically used for everything else, including brainstem pathways and basal ganglia influence, which handle postural tone, gross force and automatic adjustments. Modern textbooks avoid the term because the systems are not anatomically separable (Purves et al., 2018), but in coaching language the distinction between precise voluntary control and automatic gross control remains real.

‌

Recommended videos

Each video below was chosen because it covers a specific part of this article in more depth than text alone allows.

‌

‌

‌

‌

‌

‌

‌

‌

‌

‌

‌

Related reading on FitXplor

• 5.21 The Cerebral Cortex: Lobes, Layers, and Functional Organisation
• 5.5 The Athlete’s Brain: Functional Neuroanatomy for Performance
• 2.5 Power Development and Explosive Strength
• 1.5 The Nervous System and Athletic Performance
• 2.2 Maximum Strength Development

‌

References

Aagaard, P., Simonsen, E. B., Andersen, J. L., Magnusson, P., & Dyhre-Poulsen, P. (2002). Neural adaptation to resistance training: changes in evoked V-wave and H-reflex responses. Journal of Applied Physiology, 92(6), 2309–2318.

Aagaard, P., Simonsen, E. B., Andersen, J. L., Magnusson, P., & Dyhre-Poulsen, P. (2002). Increased rate of force development and neural drive of human skeletal muscle following resistance training. Journal of Applied Physiology, 93(4), 1318–1326.

Carroll, T. J., Riek, S., & Carson, R. G. (2001). Neural adaptations to resistance training: implications for movement control. Sports Medicine, 31(12), 829–840.

Duchateau, J., & Hainaut, K. (1984). Isometric or dynamic training: differential effects on mechanical properties of a human muscle. Journal of Applied Physiology, 56(2), 296–301.

Evarts, E. V. (1968). Relation of pyramidal tract activity to force exerted during voluntary movement. Journal of Neurophysiology, 31(1), 14–27.

Gandevia, S. C. (2001). Spinal and supraspinal factors in human muscle fatigue. Physiological Reviews, 81(4), 1725–1789.

Georgopoulos, A. P., Schwartz, A. B., & Kettner, R. E. (1986). Neuronal population coding of movement direction. Science, 233(4771), 1416–1419.

Graziano, M. S. A., Taylor, C. S. R., & Moore, T. (2002). Complex movements evoked by microstimulation of precentral cortex. Neuron, 34(5), 841–851.

Lemon, R. N. (2008). Descending pathways in motor control. Annual Review of Neuroscience, 31, 195–218.

Nathan, P. W., Smith, M. C., & Deacon, P. (1990). The corticospinal tracts in man: course and location of fibres at different segmental levels. Brain, 113(2), 303–324.

Nudo, R. J., Milliken, G. W., Jenkins, W. M., & Merzenich, M. M. (1996). Use-dependent alterations of movement representations in primary motor cortex of adult squirrel monkeys. Journal of Neuroscience, 16(2), 785–807.

Roshan, L., Paradiso, G. O., & Chen, R. (2003). Two phases of short-interval intracortical inhibition. Experimental Brain Research, 151(3), 330–337.

Schieber, M. H. (2001). Constraints on somatotopic organization in the primary motor cortex. Journal of Neurophysiology, 86(5), 2125–2143.

Tanji, J. (2001). Sequential organization of multiple movements: involvement of cortical motor areas. Annual Review of Neuroscience, 24, 631–651.

Valls-Solé, J., Rothwell, J. C., Goulart, F., Cossu, G., & Muñoz, E. (1999). Patterned ballistic movements triggered by a startle in healthy humans. Journal of Physiology, 516(3), 931–938.

Van Cutsem, M., Duchateau, J., & Hainaut, K. (1998). Changes in single motor unit behaviour contribute to the increase in contraction speed after dynamic training in humans. Journal of Physiology, 513(1), 295–305.

Carlsen, A. N., Maslovat, D., & Franks, I. M. (2012). Preparation for voluntary movement in healthy and clinical populations: evidence from startle. Clinical Neurophysiology, 123(1), 21–33.

Baker, S. N. (2011). The primate reticulospinal tract, hand function and functional recovery. Journal of Physiology, 589(23), 5603–5612.

Purves, D., Augustine, G. J., Fitzpatrick, D., Hall, W. C., LaMantia, A.-S., Mooney, R. D., Platt, M. L., & White, L. E. (2018). Neuroscience (6th ed.). Oxford University Press.

Kandel, E. R., Koester, J. D., Mack, S. H., & Siegelbaum, S. A. (Eds.). (2021). Principles of Neural Science (6th ed.). McGraw Hill.