Nervous system — gross anatomy and systems

Intuition [Beginner]

The nervous system is the body's communication network. It gathers information from the environment and from inside the body, processes that information, and coordinates responses. Three functions define it: sensory input (detecting stimuli), integration (interpreting the input), and motor output (commanding a response).

The nervous system has two major divisions. The central nervous system (CNS) consists of the brain and spinal cord. It is the integration and command centre. The peripheral nervous system (PNS) consists of all the nerves that connect the CNS to the rest of the body. It carries sensory information in and motor commands out.

The PNS itself has two functional subdivisions. The somatic nervous system controls voluntary movements of skeletal muscles. The autonomic nervous system controls involuntary functions such as heart rate, digestion, and pupil dilation. The autonomic division further splits into the sympathetic branch (activates the body for "fight or flight") and the parasympathetic branch (promotes "rest and digest" activities).

Information travels through the nervous system as electrical signals (action potentials) along neurons, and as chemical signals (neurotransmitters) across synapses. A single neuron can receive inputs from thousands of other neurons, integrate those signals, and pass the result along. This massive parallel processing is what makes the nervous system so powerful.

Visual [Beginner]

The brain can be divided into four main regions, each with distinct functions. The cerebrum (the largest part, divided into left and right hemispheres) handles conscious thought, language, and voluntary movement. The cerebellum coordinates balance and fine motor control. The brainstem (midbrain, pons, medulla) controls vital functions such as breathing and heart rate. The diencephalon (thalamus and hypothalamus) relays sensory information and regulates homeostasis.

The spinal cord runs inside the vertebral column from the brainstem to the lower back. It carries sensory information up to the brain and motor commands down from the brain. It also contains local circuits called reflex arcs that can produce rapid responses without waiting for input from the brain.

Worked example [Beginner]

Trace the knee-jerk (patellar) reflex:

1. A tap on the patellar tendon stretches the quadriceps muscle.
2. Sensory neuron: Muscle stretch receptors (muscle spindles) detect the stretch and fire action potentials.
3. Spinal cord: The sensory neuron synapses directly onto a motor neuron in the spinal cord (a monosynaptic reflex).
4. Motor neuron: The motor neuron fires and sends action potentials back to the quadriceps muscle.
5. Response: The quadriceps contracts, extending the lower leg.

This entire loop takes about 30-50 milliseconds. It does not involve the brain — the decision is made in the spinal cord. This is why the knee jerk happens before you are consciously aware of the tap.

Check your understanding [Beginner]

Formal definition [Intermediate+]

The nervous system is organised hierarchically:

Central nervous system (CNS):

• Cerebrum: Paired hemispheres with an outer layer of grey matter (cerebral cortex, ~2-4 mm thick) and inner white matter. The cortex is divided into four lobes (frontal, parietal, temporal, occipital) and contains functional areas for motor control (primary motor cortex, precentral gyrus), sensory processing (primary somatosensory cortex, postcentral gyrus), vision (occipital cortex), audition (temporal cortex), and language (Broca's area, Wernicke's area).
• Basal ganglia: Deep nuclei (caudate, putamen, globus pallidus, subthalamic nucleus, substantia nigra) involved in movement selection and habit learning.
• Thalamus: Relay station for nearly all sensory information ascending to the cortex.
• Hypothalamus: Master regulator of homeostasis — controls body temperature, hunger, thirst, circadian rhythm, and the pituitary gland.
• Cerebellum: Coordinates motor timing, balance, and motor learning. Contains more neurons than the rest of the brain combined.
• Brainstem: Midbrain, pons, and medulla oblongata. Contains cranial nerve nuclei, reticular formation (arousal), and vital centres for breathing, heart rate, and blood pressure.

Spinal cord:

• Segmented into 31 pairs of spinal nerves (8 cervical, 12 thoracic, 5 lumbar, 5 sacral, 1 coccygeal).
• Grey matter (butterfly-shaped central region) contains neuronal cell bodies: dorsal horn (sensory), ventral horn (motor), lateral horn (autonomic, thoracolumbar only).
• White matter (surrounding region) contains ascending (sensory) and descending (motor) tracts of myelinated axons.

Peripheral nervous system (PNS):

• Somatic: Motor axons to skeletal muscle; sensory axons from skin, muscle, and joints.
• Autonomic: Two-neuron chains (preganglionic and postganglionic) innervating smooth muscle, cardiac muscle, and glands.
  • Sympathetic: Preganglionic fibres originate in thoracolumbar spinal cord (T1-L2). Short preganglionic, long postganglionic fibres. Primary neurotransmitter at target: norepinephrine. Function: mobilise body for action.
  • Parasympathetic: Preganglionic fibres originate in brainstem (cranial nerves III, VII, IX, X) and sacral spinal cord (S2-S4). Long preganglionic, short postganglionic fibres. Primary neurotransmitter at target: acetylcholine. Function: conserve and restore body resources.

Reflex arcs

A reflex arc is the simplest neural circuit producing a stereotyped response to a specific stimulus. Its components are:

1. Receptor — detects the stimulus.
2. Sensory (afferent) neuron — carries signal to the CNS.
3. Integration centre — one or more synapses in the CNS (monosynaptic or polysynaptic).
4. Motor (efferent) neuron — carries command to the effector.
5. Effector — the muscle or gland that responds.

Monosynaptic reflexes (stretch reflex) have the shortest latency. Polysynaptic reflexes (withdrawal reflex, cross-extensor reflex) involve interneurons and can produce more complex, coordinated responses.

Key theorem with proof [Intermediate+]

Theorem (Autonomic dual innervation and antagonism). Most internal organs receive dual innervation from both the sympathetic and parasympathetic divisions. These two inputs generally produce opposite effects on the target organ. The resting level of organ activity is determined by the balance (tonic firing rates) of the two autonomic branches.

Proof. Consider the heart as a paradigmatic example. The parasympathetic input arrives via the vagus nerve (cranial nerve X), releasing acetylcholine onto muscarinic receptors on the pacemaker cells. This slows the heart rate. The sympathetic input arrives via postganglionic fibres from the cervical and upper thoracic ganglia, releasing norepinephrine onto beta-1 adrenergic receptors. This increases heart rate and contractility.

At rest, the vagus fires tonically at a moderate rate, keeping the heart rate around 70-80 bpm. Cutting the vagus (vagotomy) raises heart rate to about 100 bpm, revealing the intrinsic pacemaker rate. Cutting sympathetic input lowers heart rate slightly. Cutting both leaves the heart beating at its intrinsic rate (~100 bpm). The resting heart rate is therefore below the intrinsic rate because vagal (parasympathetic) tone dominates at rest.

The same pattern of dual, antagonistic innervation applies to the gastrointestinal tract (sympathetic inhibits motility; parasympathetic promotes it), the pupils (sympathetic dilates; parasympathetic constricts), and the respiratory airways (sympathetic dilates bronchi; parasympathetic constricts them). $\square$

Bridge. This antagonistic organisation provides fine-grained control: rather than a single input that can only be turned up or down, the two divisions allow bidirectional adjustment around a set point. The dual-innervation principle builds toward the systems anatomy of autonomic outflows treated at Master tier (§ "PNS organisation"), where the cranial-sacral parasympathetic outflow and the thoracolumbar sympathetic outflow are localised to distinct CNS levels. The foundational reason for dual innervation is that homeostasis requires both fast-on and fast-off control, and this is exactly what two antagonistic tonically-active inputs deliver — the same control-theoretic motif appears again in 18.02.01 cardiovascular baroreflex regulation and in 18.07.01 hypothalamic-pituitary feedback. Putting these together with the somatic motor system's reciprocal inhibition (knee-jerk worked example), the central insight is that the nervous system generally implements bidirectional control by pairing antagonistic populations, not by adding inhibition to a single excitatory channel.

Exercises [Intermediate+]

CNS organisation: cortical maps, subcortical nuclei, and the topographic principle [Master]

The central nervous system is built on a single deep idea: topographic organisation. Nearby points on a sensory or motor surface of the body — adjacent fingertips, neighbouring patches of retina, adjacent points on the basilar membrane of the cochlea — project to nearby points on a cortical or subcortical map. The map is continuous in the topological sense (small displacements on the body produce small displacements on the cortex) but is not isometric: the area devoted to each part of the body depends on its computational demands rather than its physical size. This non-isometric continuity is the structural fact that organises sensory and motor cortex, and it recurs at multiple stations along every ascending pathway.

The cerebral cortex is a folded sheet roughly $0.16{\mathrm{m}}^2$ in unfolded surface area, $2-4\mathrm{mm}$ thick, with about $16$ billion neurons arranged in six laminar layers and an estimated $10^{14}$ synapses. Korbinian Brodmann's 1909 cytoarchitectonic survey ^{[Brodmann 1909]} divided this sheet into 52 areas distinguished by differences in cell density, lamination, and the relative thickness of the six layers. Brodmann areas remain the standard map of the cortex: area 4 is primary motor cortex; areas 3, 1, 2 (in posterior-to-anterior order) form primary somatosensory cortex on the postcentral gyrus; area 17 is primary visual cortex (V1, the calcarine sulcus); areas 41-42 are primary auditory cortex (Heschl's gyrus); areas 44-45 are Broca's region in the inferior frontal gyrus; area 22 contains Wernicke's region in the superior temporal gyrus.

Within each primary sensory or motor area, neurons are organised into cortical columns. Vernon Mountcastle's 1957 microelectrode recordings in cat somatosensory cortex ^{[Mountcastle 1957]} showed that a vertical penetration of the cortex (perpendicular to the pial surface) encountered neurons that all responded to the same modality (touch, vibration, pressure), the same skin region, and the same submodality, while a slightly displaced penetration moved to a column with a different submodality or a neighbouring skin patch. The column became the canonical microcircuit unit. Hubel and Wiesel's 1962 recordings from V1 ^{[Hubel-Wiesel 1962]} extended the picture to vision: V1 is organised into orientation columns (each column preferring stimuli at a particular bar orientation), ocular dominance columns (alternating left-eye and right-eye dominance), and the spatially repeating "hypercolumn" containing a full set of orientation and ocular preferences for one location in visual space.

The somatotopic principle is most starkly visible in the homunculi of Wilder Penfield. In the 1930s and 1940s Penfield carried out intra-operative electrical stimulation of the cortex of awake neurosurgical patients (epilepsy resections, under local anaesthesia), mapping the points whose stimulation produced movement of, or sensation in, a specific body part. The 1937 Penfield-Boldrey paper ^{[Penfield-Boldrey 1937]} gave the canonical sensory and motor homunculi: a distorted human figure stretched along the central sulcus, with enormous hands, lips, and tongue, modest trunk and limbs, feet on the medial surface in the interhemispheric fissure. The magnification factor — cortical millimetres per body-surface millimetre — is highest where tactile discrimination is finest (lips, tongue, fingertips, with two-point discrimination thresholds of $1-3\mathrm{mm}$) and lowest on the back, where two-point discrimination is $40-70\mathrm{mm}$. The same magnification pattern governs the motor homunculus: cortical area scales with the number of independently controllable motor units, not with the gross size of the represented body part.

The thalamus is the subcortical relay through which nearly all ascending sensory information passes on its way to the cortex. It comprises about $50-60$ nuclei, each projecting to a specific cortical region: the ventral posterolateral nucleus (VPL) relays body somatosensation to areas 3-1-2; the ventral posteromedial nucleus (VPM) relays face somatosensation through the trigeminal pathway; the lateral geniculate nucleus (LGN) relays retinal information to V1; the medial geniculate nucleus (MGN) relays auditory information from the inferior colliculus to primary auditory cortex; the ventral lateral and ventral anterior nuclei (VL, VA) relay basal-ganglia and cerebellar output to motor cortex; the pulvinar and intralaminar nuclei provide diffuse arousal-and-attention modulation. Thalamic relay is not a passive switchboard — each thalamic relay neuron is gated by a powerful reciprocal corticothalamic loop with cortical layer VI feedback, and by the inhibitory thalamic reticular nucleus, which sculpts thalamic activity in a state-dependent way (sleep spindle generation, arousal-dependent gain modulation, attentional selection).

The basal ganglia form a separate subcortical processing loop, anatomically distinct from the thalamic relay but functionally interleaved with it. Their major nuclei are the caudate nucleus, putamen (together: the dorsal striatum, the input nucleus), the globus pallidus (internal and external segments — GPi, GPe), the subthalamic nucleus (STN), and the substantia nigra (pars compacta — SNc, the dopaminergic input; pars reticulata — SNr, an output nucleus). The standard circuit model is the direct/indirect pathway organisation: cortical input arrives at striatum, where two parallel populations of medium spiny neurons project either directly to GPi/SNr (the direct pathway, dopamine-facilitated via D1 receptors, releasing the thalamus from inhibition and facilitating selected motor programs) or indirectly through GPe and STN to GPi/SNr (the indirect pathway, dopamine-inhibited via D2 receptors, increasing GPi inhibition of thalamus and suppressing competing motor programs). Loss of nigrostriatal dopaminergic neurons in Parkinson disease tilts the balance toward the indirect pathway and produces the bradykinesia and rigidity characteristic of the disease; excess striatal activity in Huntington disease (caudate atrophy) tilts in the opposite direction and produces choreiform movements.

The cerebellum is the third major subcortical structure, anatomically separate from the basal-ganglia loop but functionally complementary. It has a highly stereotyped microcircuit: the cerebellar cortex contains three layers (molecular, Purkinje, granular), with granule cells receiving mossy-fibre input from the pontine nuclei and projecting axons (parallel fibres) into the molecular layer where they synapse onto the dendrites of Purkinje cells. Each Purkinje cell also receives a single climbing fibre from the inferior olive, whose excitation produces a complex spike. Purkinje cells provide the sole output of cerebellar cortex, projecting (inhibitorily) to the deep cerebellar nuclei (fastigial, interpositus, dentate), which then project to the thalamus, the red nucleus, and the vestibular nuclei. The Marr-Albus-Ito hypothesis (Marr 1969, Albus 1971, Ito 1972) posits that climbing-fibre input carries an error signal that drives long-term depression at parallel-fibre-Purkinje synapses, implementing supervised motor learning. The cerebellum contains more neurons than the rest of the brain combined (about $69$ billion granule cells, roughly four times the cerebral cortex), reflecting its role as a high-dimensional adaptive controller and timing system.

The CNS is integrated by extensive white-matter fibre systems — association fibres connecting cortical areas within a hemisphere, commissural fibres (the corpus callosum) connecting the two hemispheres, and projection fibres connecting cortex with thalamus, basal ganglia, brainstem, and spinal cord. The topographic principle is preserved through these fibre systems: each thalamocortical, corticothalamic, and corticospinal tract carries its own internal somatotopic, retinotopic, or tonotopic map. A lesion at any point along these fibres produces a deficit whose body-localisation depends on which fibres of the bundle are interrupted, and clinical neurology is in large part the art of inferring lesion location from the pattern of preserved versus lost function.

PNS organisation: somatic vs autonomic divisions, dermatomes, and the cranial-spinal axis [Master]

The peripheral nervous system is the wiring that connects the CNS to every sensor and effector in the body. It comprises the 12 pairs of cranial nerves emerging from the brain and brainstem, the 31 pairs of spinal nerves emerging from the spinal cord, and the autonomic ganglia and plexuses that handle visceral and vascular control. The PNS implements both the somatic motor and sensory innervation of skin, skeletal muscle, joints, and special senses, and the autonomic motor innervation of smooth muscle, cardiac muscle, glands, and the enteric nervous system.

The cranial nerves carry sensory, motor, and parasympathetic fibres to and from the head, neck, and (for the vagus) much of the thoracic and abdominal viscera. Their Roman-numbered sequence — I olfactory, II optic, III oculomotor, IV trochlear, V trigeminal, VI abducens, VII facial, VIII vestibulocochlear, IX glossopharyngeal, X vagus, XI accessory, XII hypoglossal — runs rostrocaudally along the brainstem (with cranial nerves I and II emerging from the forebrain itself, not the brainstem proper). A classical mnemonic ("On Old Olympus' Towering Tops A Finn And German Viewed Some Hops") keeps the order. The functional classification — sensory only, motor only, mixed — is captured by another ("Some Say Marry Money But My Brother Says Big Brains Matter More") where the initials S/M/B mark each nerve. Sensory-only cranial nerves: I (olfaction), II (vision), VIII (hearing and balance). Motor-only: III, IV, VI (extra-ocular movement), XI (accessory, sternocleidomastoid and trapezius), XII (hypoglossal, tongue). Mixed: V (trigeminal — face sensation and muscles of mastication), VII (facial — facial muscles and taste from anterior tongue), IX (glossopharyngeal — pharyngeal sensation, parotid secretion, taste from posterior tongue), X (vagus — pharyngeal/laryngeal motor and sensory, plus the parasympathetic outflow to thoracic and abdominal viscera).

Of the twelve, four carry cranial parasympathetic outflow: III (oculomotor — pupillary constriction and lens accommodation via the Edinger-Westphal nucleus and ciliary ganglion), VII (facial — lacrimal and submandibular salivary glands via the superior salivatory nucleus, pterygopalatine and submandibular ganglia), IX (glossopharyngeal — parotid salivation via the inferior salivatory nucleus and otic ganglion), and X (vagus — thoracic and abdominal viscera, the dominant parasympathetic nerve of the body). The remainder of parasympathetic outflow comes from the sacral cord (S2-S4) and innervates the distal gastrointestinal tract, bladder, and genitalia via the pelvic splanchnic nerves. The sympathetic outflow is by contrast strictly thoracolumbar (T1-L2): all sympathetic preganglionic fibres exit the cord from these segments and synapse either in the paravertebral chain ganglia (sympathetic trunk) or in the prevertebral ganglia (coeliac, superior mesenteric, inferior mesenteric) before projecting via postganglionic fibres to their targets.

The autonomic division coined by John Langley in his 1903 paper ^{[Langley 1903]} is therefore organised on a two-neuron chain principle (preganglionic in CNS, postganglionic in PNS) with the synapse at the ganglion mediated by acetylcholine on nicotinic receptors in both sympathetic and parasympathetic branches. The branches differ at the target: sympathetic postganglionics release norepinephrine onto adrenergic receptors (with the adrenal medulla as a specialised "ganglion" whose chromaffin cells are modified postganglionic neurons that secrete epinephrine into the bloodstream); parasympathetic postganglionics release acetylcholine onto muscarinic receptors. The pharmacological selectivity that follows is the basis of modern autonomic-targeting drugs (beta-blockers, anticholinergics, alpha-agonists, muscarinic antagonists).

The enteric nervous system is now recognised as a third autonomic division — sometimes called the "second brain" — comprising about $500$ million neurons in two interconnected plexuses (myenteric/Auerbach for motility, submucosal/Meissner for secretion) embedded in the wall of the gastrointestinal tract from oesophagus to rectum. The enteric nervous system can regulate gut motility, secretion, and local blood flow autonomously without input from the CNS, although it is modulated by sympathetic (inhibitory) and parasympathetic (excitatory) inputs. Its size — comparable to the spinal cord in neuron count — and its functional autonomy give it a status of its own beyond the classical sympathetic/parasympathetic dichotomy.

Dermatomes are the skin regions innervated by a single dorsal root. The body surface is mapped onto the spinal cord in 31 dermatome strips (one per spinal nerve, except C1 which has no cutaneous innervation) running approximately horizontally on the trunk and obliquely on the limbs — the limb-bud development pattern stretches dermatomes longitudinally along the limbs (T1 reaches the ulnar side of the hand, C6/C7/C8 cover the upper limb in a stripe). Clinical localisation of spinal-cord and root lesions exploits the dermatome map: a band of altered sensation at a particular spinal level points to a lesion at that root or at the corresponding cord segment. The dermatome map has been refined repeatedly since Henry Head's 1893-1900 herpes zoster maps and Otfrid Foerster's 1933 root-section maps; modern textbook dermatome figures are composite averages that work clinically but mask substantial individual variability (overlap between adjacent dermatomes is the rule, not the exception).

Equivalent to dermatomes for the muscle side are myotomes (the skeletal muscle innervated by a single ventral root): C5 abducts the shoulder, C6 flexes the elbow, C7 extends the elbow, C8 flexes the fingers, T1 abducts and adducts the fingers; L2 flexes the hip, L3-L4 extend the knee, L5 dorsiflexes the foot, S1 plantarflexes the foot. The combination of dermatome and myotome testing localises lesions to specific spinal segments. The spinal cord itself is organised somatotopically in its grey matter — Rexed's ten laminae (Rexed 1952) classify dorsal horn cells by their cytoarchitecture and inputs (lamina I for nociception, lamina II for substantia gelatinosa, deeper laminae for proprioception), and ventral horn motor neurons are arranged in columns somatotopically by the muscle they innervate (medial-to-lateral: axial musculature, proximal limb, distal limb).

The PNS is highly plexus-organised at the limbs. The cervical plexus (C1-C4) supplies the diaphragm via the phrenic nerve (C3, C4, C5 "keeps the diaphragm alive"); the brachial plexus (C5-T1) gives rise to all the major nerves of the upper limb (musculocutaneous, axillary, radial, median, ulnar); the lumbosacral plexus (L1-S4) generates the femoral, obturator, and sciatic nerves of the lower limb. Each plexus implements a rewiring between segmental roots and named peripheral nerves: a single named peripheral nerve carries fibres from multiple roots, and a single root contributes to multiple peripheral nerves, so a peripheral nerve lesion produces a different deficit pattern than a root lesion at any of the contributing levels. Clinical localisation distinguishes "root lesion" (dermatomal sensory loss, single-myotome weakness) from "peripheral nerve lesion" (cutaneous distribution of the nerve, weakness of all muscles innervated by that nerve) by reading the deficit pattern against the plexus diagram.

White matter and connectomics: tracts, fascicles, and modern imaging [Master]

The CNS is not just a collection of grey-matter nuclei and cortical sheets; it is held together by a vast white-matter scaffold of myelinated axon bundles. Quantitatively, the human cerebrum contains roughly equal volumes of grey and white matter (about $700\mathrm{c}{\mathrm{m}}^3$ each), with the white matter carrying approximately $150,000$ km of myelinated fibres at the lower-bound estimate. The major fibre systems fall into three classes: projection fibres (cortex-to-subcortex), commissural fibres (hemisphere-to-hemisphere), and association fibres (within a hemisphere, region-to-region).

The major projection fibres form the internal capsule, a fan-shaped bundle through which essentially every fibre travelling between the cerebral cortex and brainstem-or-spinal-cord passes. Internal-capsule anatomy is somatotopically organised: the anterior limb carries frontopontine and thalamocortical fibres to/from frontal cortex; the genu carries corticobulbar fibres to brainstem cranial-nerve motor nuclei (face, jaw, tongue); the posterior limb carries the corticospinal tract (motor, with the leg medial, trunk middle, arm lateral, face most rostral) and somatosensory thalamocortical fibres in mirror-image arrangement. A small lesion of the internal capsule — most commonly a lacunar stroke in the small penetrating arteries from the middle cerebral artery — can produce a pure motor or pure sensory contralateral hemideficit, sparing cognition and the cortex itself; the anatomical compactness of the internal capsule makes it a clinically critical structure.

The corticospinal tract is the principal motor projection. Its cell bodies are the giant Betz pyramidal cells of cortical layer V in primary motor cortex (Brodmann area 4), with substantial contributions from premotor (area 6), supplementary motor area, and somatosensory cortex (areas 3-1-2). The axons descend through the corona radiata, the posterior limb of the internal capsule, the cerebral peduncle of the midbrain, and the basis pontis, before forming the medullary pyramids on the ventral surface of the medulla. At the pyramidal decussation (just rostral to the spinomedullary junction), about $85\%$ of corticospinal fibres cross the midline and descend in the lateral corticospinal tract in the contralateral lateral funiculus of the cord, synapsing on motor neurons (or, more commonly, on interneurons) in laminae VII-IX. The remaining $\sim 15\%$ descend uncrossed in the ventral corticospinal tract and cross at the segmental level. This crossed organisation is why a unilateral cortical or internal-capsule lesion produces contralateral motor deficits.

The major ascending sensory tracts mirror the descending motor system, with two principal pathways. The dorsal column-medial lemniscus pathway carries fine touch, vibration, and proprioception: peripheral afferents enter the cord via the dorsal root, ascend ipsilaterally in the dorsal columns (gracile fasciculus for legs, cuneate fasciculus for arms — Goll and Burdach in classical nomenclature), synapse in the gracile and cuneate nuclei of the medulla, decussate as the medial lemniscus, and ascend to VPL of the thalamus and then to somatosensory cortex (areas 3a/3b/1/2). The spinothalamic tract (anterolateral system) carries crude touch, pain, and temperature: peripheral afferents synapse in the dorsal horn within a few segments of entry, the second-order neurons decussate immediately across the anterior white commissure, and ascend in the contralateral anterolateral funiculus to VPL and then to somatosensory cortex. The two pathways' different decussation levels are the basis of the Brown-Séquard syndrome: a hemisection of the spinal cord produces ipsilateral loss of fine touch and proprioception below the lesion (dorsal column not yet crossed) but contralateral loss of pain and temperature below the lesion (spinothalamic already crossed).

The commissural fibres of the human brain are dominated by the corpus callosum, a sheet of about $200-250$ million myelinated axons connecting homologous regions of the two hemispheres. The callosum is topographically organised: the genu (anterior) connects prefrontal cortex; the body connects somatosensory and motor cortex (with somatotopy reading rostral-to-caudal: face anterior, arm middle, leg posterior); the splenium (posterior) connects parietal, temporal, and occipital cortex. Roger Sperry's 1960s split-brain studies in patients with callosal sectioning (for intractable epilepsy) demonstrated the integrative role of the callosum: with the callosum cut, the two hemispheres function independently, each unable to verbally report stimuli presented to the contralateral half of the visual field, and each capable of independent decisions when given separate sensory access. The anterior and posterior commissures and the hippocampal commissure add a small amount of additional interhemispheric connectivity.

The association fibres form long-range cortico-cortical bundles within each hemisphere. Major named association tracts include the superior longitudinal fasciculus (frontal-parietal-occipital, the dominant long association bundle); its dorsal branch the arcuate fasciculus (carrying the dorsal language stream from Wernicke's region to Broca's region); the inferior longitudinal fasciculus (occipital-temporal, ventral visual stream); the inferior fronto-occipital fasciculus (occipital-frontal); the uncinate fasciculus (orbitofrontal-temporal); the cingulum bundle (along the cingulate gyrus); the fornix (hippocampus to mammillary bodies). The early systematic description of these tracts is due to Joseph Jules Dejerine's 1895 anatomical atlas, supplemented through the twentieth century by post-mortem dissection (the Klingler freeze-thaw fibre-dissection method, Klingler 1935).

The modern non-invasive imaging revolution is diffusion-tensor imaging (DTI), introduced by Basser, Mattiello, and LeBihan in the early 1990s. DTI measures the anisotropy of water diffusion in tissue: along a myelinated axon bundle, water diffuses faster parallel to the fibres than perpendicular to them, and the principal eigenvector of the diffusion tensor at each voxel gives the local fibre orientation. DTI tractography algorithms then integrate the principal-eigenvector field across voxels to reconstruct three-dimensional fibre trajectories. The technique allows in vivo, non-invasive visualisation of white-matter tracts in individual subjects — the corticospinal tract, the arcuate fasciculus, the corpus callosum — and has revolutionised both clinical neurology (pre-surgical planning, white-matter disease assessment) and cognitive neuroscience (correlating tract integrity with behaviour).

DTI is the experimental foundation of connectomics, the research programme of mapping the entire structural-connection network of the brain. The term connectome was coined by Sporns, Tononi, and Kotter in 2005 ^{[Sporns-Tononi-Kotter 2005]} in deliberate analogy to the genome and proteome. The Human Connectome Project (Van Essen et al. 2013 ^{[Van Essen-HCP 2013]}) collected high-resolution DTI plus structural and functional MRI on more than $1,200$ subjects, producing a public dataset that has become the de facto reference for whole-brain connectomics. Parcellated connectome graphs typically have $\sim 100-400$ nodes (cortical and subcortical parcels) and tens of thousands of edges, each weighted by streamline count or fractional anisotropy.

The graph-theoretic analysis of connectomes has imported a substantial body of network-science machinery into neuroscience. The clustering coefficient $C_i$ at node $i$ is the fraction of pairs of $i$'s neighbours that are themselves connected; the network clustering coefficient $C$ is the average over nodes. The characteristic path length $L$ is the average length (number of edges) of the shortest path between all pairs of nodes. Watts and Strogatz's 1998 paper ^{[Watts-Strogatz 1998]} introduced the small-world coefficient

$$\sigma =\frac{C/C_{\mathrm{rand}}}{L/L_{\mathrm{rand}}},$$

where the subscript "rand" denotes the corresponding quantity for a random Erdős-Rényi graph with the same number of nodes and edges. Networks with $\sigma >1$ — high clustering yet short paths — are small-world networks. Human connectomes have $\sigma$ in the range $2-7$ at typical parcellations, comfortably above the small-world threshold and well above purely random or purely lattice values.

The connectome is modular: nodes partition into communities (the visual, somatomotor, default-mode, salience, fronto-parietal, attention, limbic networks) with denser within-community connections than between-community connections. Modularity $Q$ (Newman 2006) quantifies this partition quality; human structural connectomes have $Q\approx 0.4-0.5$. The connectome also contains hubs — nodes with high degree, high betweenness centrality, and disproportionate participation in the rich-club organisation (van den Heuvel & Sporns 2011). The rich-club nodes (precuneus, superior parietal, superior frontal, anterior insula, hippocampus, thalamus) form a densely interconnected "core" that integrates information across the modular networks; lesions in rich-club regions disproportionately impair cognition compared with lesions of equivalent size in non-hub regions.

Two cautionary notes are now mainstream in the connectomics literature. First, DTI tractography systematically fails at fibre-crossing voxels and at the cortical interface, so streamline-counted connectome edges have substantial false-positive and false-negative rates compared with tracer-based connectomes in non-human primates. Second, the graph-theoretic measures (clustering, path length, modularity) depend sensitively on the choice of parcellation, the edge-weight threshold, and the null-model graph used for normalisation; small-worldness coefficients can vary by a factor of two for the same dataset under different processing choices. The Bassett-Sporns 2017 review ^{[Bassett-Sporns 2017]} is the modern survey of methodological best practice and the empirical robustness of connectome findings.

Functional anatomy of the major systems with imaging correlates [Master]

The grey-matter map and the white-matter scaffold together define the anatomical substrate; the functional anatomy is the assignment of behaviours and computations to specific structures. Five major systems organise the integrated function of the nervous system, each anatomically distinct and each visible at characteristic imaging contrasts.

The motor system is hierarchical, with a long pathway from cortical intention to muscle contraction. The hierarchy reads: prefrontal cortex (decision and planning) projects to premotor and supplementary motor areas (motor sequencing and program selection) and to primary motor cortex (area 4, command of specific movements via the corticospinal tract). The corticospinal axons descend through the internal capsule, cerebral peduncle, basis pontis, medullary pyramids, decussate $85\%$ at the pyramidal decussation, and reach motor neurons in the ventral horn of the spinal cord — directly for distal limb muscles (the cortico-motoneuronal connections that allow dextrous finger control in primates) and indirectly via interneurons for proximal and axial muscles. Modulating this main pathway are two parallel side loops: the basal-ganglia loop (cortex to striatum to GPi/SNr to thalamus back to cortex) for action selection and habit; the cerebellar loop (cortex to pontine nuclei to cerebellar cortex via mossy fibres, climbing fibres from inferior olive, to deep cerebellar nuclei to thalamus back to cortex) for motor timing, coordination, and error-correction. The final common path is the lower motor neuron in the ventral horn or the brainstem cranial-nerve motor nucleus, whose axon innervates the muscle fibres directly. Damage to upper motor neurons (corticospinal pathway above the motor neuron) gives spastic paresis with hyperreflexia and the Babinski sign; damage to lower motor neurons (motor neuron itself or its axon) gives flaccid paresis with hyporeflexia, atrophy, and fasciculations. The upper-versus-lower motor neuron distinction is the foundational localisation principle of clinical motor neurology.

The somatosensory system is dual, comprising the dorsal column-medial lemniscus and the spinothalamic tract described above (§ "White matter"). The two pathways converge at the thalamus (VPL/VPM) and at primary somatosensory cortex (areas 3a, 3b, 1, 2 on the postcentral gyrus). Within S1, area 3a receives proprioceptive input from muscle spindles, 3b receives cutaneous input from fast-adapting and slow-adapting mechanoreceptors, area 1 integrates rapid and slow cutaneous information for texture analysis, area 2 integrates cutaneous and proprioceptive information for shape and stereognosis. Higher-order processing proceeds along two streams: a dorsal "where/how" stream toward the posterior parietal cortex (for tactile-guided action) and a ventral "what" stream into the secondary somatosensory cortex (S2) and the insula (for tactile object recognition). The somatosensory homunculus on the postcentral gyrus mirrors the motor homunculus on the precentral gyrus across the central sulcus; the magnification factors are similar, with fingertips, lips, and tongue receiving disproportionate cortical area.

The visual system is the most thoroughly mapped of the sensory systems. Retinal ganglion cells project via the optic nerve (CN II) to the optic chiasm, where nasal retinal fibres decussate while temporal fibres remain ipsilateral — the result is that each lateral geniculate nucleus receives the contralateral visual hemifield from both eyes. The lateral geniculate nucleus has six layers (parvocellular layers 3-6 for sustained colour-and-form, magnocellular layers 1-2 for transient motion), with each layer receiving monocular input from a single eye. LGN projects via the optic radiations through the temporal lobe (Meyer's loop, carrying superior visual quadrant) and parietal lobe (carrying inferior visual quadrant) to primary visual cortex (V1, area 17, the calcarine sulcus on the medial occipital lobe). The cortical map is precisely retinotopic: V1 contains an orderly representation of the contralateral visual hemifield, with the fovea hugely magnified (about half of V1 represents the central $10^{\circ }$ of vision). Beyond V1, higher visual areas form two streams: the ventral "what" stream through V2, V4, and inferotemporal cortex for object recognition and category-selective cortex (the fusiform face area, the parahippocampal place area, the visual word form area); and the dorsal "where/how" stream through V2, V3, MT/V5 (motion), and posterior parietal cortex for spatial vision and visually-guided action.

The autonomic system has its central control in the brainstem and hypothalamus. The medullary reticular formation contains the cardiovascular control nuclei (nucleus tractus solitarius for baroreceptor input, rostral and caudal ventrolateral medulla for sympathetic output, nucleus ambiguus for cardiac vagal output), the respiratory rhythm generators (pre-Bötzinger complex), and the central pattern generators for vomiting, swallowing, and coughing. The pontine micturition centre coordinates bladder emptying with urethral relaxation; the periaqueductal grey of the midbrain integrates pain modulation and defensive responses. Above the brainstem, the hypothalamus is the master integrator: its medial preoptic area regulates body temperature; the paraventricular nucleus drives autonomic and neuroendocrine outputs through the median eminence and the magnocellular projections to the posterior pituitary; the suprachiasmatic nucleus is the master circadian pacemaker; the arcuate, ventromedial, and lateral hypothalamic nuclei regulate feeding and energy balance. The hypothalamus projects autonomic outputs to the brainstem and to the intermediolateral cell column of the thoracolumbar cord (sympathetic preganglionic) and the sacral parasympathetic nucleus, integrating cortical and limbic inputs with autonomic outflow.

The limbic system — anatomically a ring of cortex (cingulate, parahippocampal) and subcortical structures (hippocampus, amygdala, mammillary bodies, anterior thalamic nuclei, septal nuclei) — coordinates emotion, motivation, and memory. The hippocampal formation implements declarative memory consolidation (the H.M. case from Scoville and Milner 1957 established the hippocampus as the critical structure for forming new episodic and semantic memories). The amygdala assigns emotional value to stimuli, particularly fear, with its central nucleus driving autonomic and behavioural fear responses via projections to the periaqueductal grey, the lateral hypothalamus, and the brainstem autonomic nuclei. The Papez circuit (hippocampus to mammillary bodies via fornix, to anterior thalamic nuclei via mammillothalamic tract, to cingulate cortex, back to hippocampus) is the classical anatomical loop of memory; the Yakovlev circuit (amygdala to medial dorsal thalamic nucleus to orbitofrontal cortex back to amygdala) is the parallel loop of emotional learning.

The imaging revolution provides two complementary functional windows. Positron emission tomography (PET) measures regional cerebral blood flow or metabolism via injected radiotracers (${}^{15}\mathrm{O}$-water or ${}^{18}\mathrm{F}$-FDG); spatial resolution is moderate ($\sim 5\mathrm{mm}$) and temporal resolution is slow ($\sim 30-60\mathrm{s}$ averaging). Functional MRI (fMRI) measures the blood-oxygenation-level-dependent (BOLD) signal (Ogawa et al. 1990), exploiting the magnetic susceptibility difference between oxygenated and deoxygenated haemoglobin; activated brain regions show increased local blood flow that over-compensates for oxygen consumption, producing a positive BOLD signal change at $\sim 1-5\mathrm{mm}$ resolution and $\sim 1-2\mathrm{s}$ temporal resolution. fMRI is now the dominant non-invasive imaging modality for human cognitive neuroscience, allowing experimenters to localise the brain regions whose activity changes with task demands, with stimuli, with cognitive states, and to map functional connectivity patterns from resting-state fluctuations.

The structural-functional interplay closes a loop. The DTI-derived connectome (§ "White matter") gives the anatomical scaffold of long-range connections; the fMRI-derived functional connectome — correlations among resting-state BOLD fluctuations — gives the patterns of co-activation those connections support; and lesion studies (stroke, surgery, traumatic injury) give the causal evidence that specific structures are necessary for specific behaviours. The convergence of these three modalities — anatomical mapping (DTI), functional mapping (fMRI/PET), and lesion-based causal inference — defines modern systems neuroanatomy. The structural map of nineteenth-century cytoarchitecture (Brodmann, Cajal, Sherrington) and the functional map of mid-twentieth-century stimulation studies (Penfield, Mountcastle, Hubel-Wiesel) are now integrated with non-invasive whole-brain imaging into a single unified framework — one that retains the topographic principle, the columnar organisation, the somatotopic and retinotopic maps, and the hierarchical-plus-parallel functional architecture that the founding generations established.

Synthesis. The nervous system's anatomy is the foundational reason its function is intelligible — the topographic principle, the cortical-columnar microcircuit, the thalamic relay architecture, and the basal-ganglia and cerebellar side-loops together build toward a single integrative picture. The central insight is that the nervous system implements computation at multiple scales simultaneously: at the synaptic and ion-channel scale described in 17.09.02 pending action potentials and 17.07.01 pending receptor signalling, at the circuit scale of cortical columns and reflex arcs, and at the system scale of distributed cortical-subcortical networks. Putting these together with the connectome, the small-world coefficient $\sigma >1$ is exactly the structural signature that supports the system-scale integration: high clustering preserves local specialisation while short path lengths permit rapid global coordination.

The somatotopic homunculus and the retinotopic V1 map identify cortical area with computational demand rather than physical body-part size, and the same magnification principle appears again in 18.04.02 pending cardiac and skeletal muscle specialisation as a general allocation rule of the nervous system. The bridge is between the structural connectome from DTI and the functional connectome from fMRI: the structural map constrains but does not determine the dynamic patterns of activity, and the discrepancy between the two is itself a major research programme in modern systems neuroscience. The hierarchical motor system from cortex through pyramidal decussation to lower motor neuron generalises to the sensory system in mirror image, and the upper-versus-lower motor neuron dichotomy is the same divide-and-localise principle that organises all clinical neurology. The pattern recurs from microcircuit to whole-brain network: nervous system anatomy is a structurally honest map of computational geography.

Connections [Master]

• Resting membrane potential and ion channels 17.09.01 is the cellular basis underlying every neural signal described here. The systems anatomy of this unit takes the cellular machinery — ion gradients, channels, and the resting potential — as the substrate from which action potentials and synaptic transmission emerge.

• Action potential ionic basis 17.09.02 pending treats the Hodgkin-Huxley cellular biophysics of the action potential. This unit stays at the systems level (tracts, nuclei, cortical maps) and cites 17.09.02 pending for the cellular mechanism by which signals propagate along the axons that make up every white-matter tract described above.

• Cell signalling and signal transduction 17.07.01 pending operates at synapses. Neurotransmitters binding to receptors trigger second-messenger cascades identical to those in non-neural cells. The pharmacological selectivity of autonomic neurotransmitters (norepinephrine vs acetylcholine at distinct receptor classes) depends on the receptor diversity treated in cell signalling.

• Skeletal muscle physiology 18.04.01 is the output side of the motor system. Motor neurons activate muscle fibres at the neuromuscular junction, and sensory feedback from muscles (muscle spindles, Golgi tendon organs) returns to the spinal cord through the dorsal-column pathway treated here.

• Muscle contraction — the actin-myosin cycle 18.04.02 pending provides the molecular machinery by which lower motor neuron commands translate into force. The motor system described in the master tier of this unit terminates at the neuromuscular junction described in 18.04.02 pending, and the pyramidal-decussation contralateral organisation closes only when paired with the muscle effector side.

• Endocrine hormones and regulation 18.07.01 is regulated by the hypothalamus, which controls the pituitary gland. The hypothalamic-pituitary axis is the anatomical bridge between the nervous and endocrine systems, and the master regulator at the apex of both is the hypothalamus treated above.

• Cardiovascular physiology — the heart 18.02.01 depends on autonomic control of heart rate, contractility, and vascular tone. The medullary cardiovascular centre integrates baroreceptor input via the nucleus tractus solitarius and adjusts autonomic output through the antagonistic dual-innervation mechanism whose key theorem is proved at intermediate tier.

• Respiratory physiology and gas exchange 18.03.01 pending is rhythmically driven by the pre-Bötzinger complex in the ventrolateral medulla, with chemoreceptor input from the carotid bodies (via CN IX) and the medullary central chemoreceptors. Respiratory rhythmogenesis is one of the brainstem central-pattern-generator circuits described above.

• Renal physiology and homeostasis 18.08.01 pending is integrated with autonomic and hypothalamic control: hypothalamic osmoreceptors drive ADH (vasopressin) release from the posterior pituitary; the sympathetic outflow to the kidney regulates renin secretion and afferent-arteriolar tone. The kidney is therefore one of the major target organs of the autonomic outflow described in this unit.

• Stochastic graph models and small-world networks [02.x.pending] — the Watts-Strogatz model and graph-theoretic measures (clustering coefficient, characteristic path length, modularity, betweenness centrality) — provide the mathematical machinery for the connectomics treated at master tier. The small-world coefficient $\sigma$ used here builds on the random-graph and complex-network material that lives in math section 02.

Historical & philosophical context [Master]

Santiago Ramón y Cajal, using the silver staining method developed by Camillo Golgi, established the neuron doctrine in the late 19th century ^{[Cajal 1904]}: the nervous system is composed of individual cells (neurons) that communicate at specialised junctions (synapses), rather than being a continuous reticulum. Cajal's Textura del Sistema Nervioso (Madrid, 1899-1904) remains one of the most influential works in neuroscience. Golgi, ironically, argued against the neuron doctrine despite providing the staining tool that proved it; the two shared the 1906 Nobel Prize for the discovery and the dispute.

Charles Sherrington's The Integrative Action of the Nervous System (1906) ^{[Sherrington 1906]} introduced the concepts of synaptic integration, excitation, inhibition, and the reflex arc. Sherrington coined the term "synapse" and demonstrated reciprocal inhibition experimentally on decerebrate cats. His work established the spinal cord as an integration centre, not merely a conduit, and shared the 1932 Nobel Prize.

The autonomic nervous system was characterised by John Langley in the 1890s and 1900s ^{[Langley 1903]}, who distinguished sympathetic from parasympathetic divisions based on anatomical origin, ganglion location, and pharmacological responses. Langley introduced the terms "autonomic", "preganglionic", "postganglionic", and developed the concept of receptor specificity that became the foundation of modern autonomic pharmacology.

The cortical map originates with Korbinian Brodmann's 1909 cytoarchitectonic atlas ^{[Brodmann 1909]}, dividing the human cortex into 52 numbered areas on the basis of laminar structure. The functional content of these areas was elucidated through the twentieth century by lesion-deficit correlations (Broca 1861, Wernicke 1874 on language), direct electrical stimulation in awake neurosurgical patients (Penfield-Boldrey 1937 ^{[Penfield-Boldrey 1937]} on the motor and sensory homunculi), and microelectrode recordings (Mountcastle 1957 ^{[Mountcastle 1957]} on cortical columns in somatosensory cortex; Hubel-Wiesel 1962 ^{[Hubel-Wiesel 1962]} on orientation and ocular-dominance columns in V1, recognised by the 1981 Nobel Prize). The Brodmann numbering remains in active clinical and research use over a century after publication.

The connectome research programme has a shorter lineage. Joseph Jules Dejerine's Anatomie des Centres Nerveux (1895) and Klingler's 1935 freeze-thaw fibre dissection method established post-mortem white-matter anatomy. Diffusion-tensor imaging (Basser, Mattiello, LeBihan 1994) made in vivo tract reconstruction possible. The term "connectome" was coined by Olaf Sporns, Giulio Tononi, and Rolf Kötter in 2005 ^{[Sporns-Tononi-Kotter 2005]}, and the Human Connectome Project ^{[Van Essen-HCP 2013]} provided the canonical large-scale dataset. The graph-theoretic apparatus imported from network science — small-worldness (Watts-Strogatz 1998 ^{[Watts-Strogatz 1998]}), modularity (Newman 2006), rich-club (van den Heuvel-Sporns 2011) — gave connectomics its quantitative language.

Two long-standing philosophical questions thread through this anatomy. The mind-body problem is sharpest at the nervous system: how does physical neural activity give rise to subjective experience? The brain regions described here — particularly cerebral cortex and thalamus — are the neural correlates of conscious awareness, but the explanatory gap between neural firing patterns and first-person experience remains open. The localisation-versus-distribution debate runs from the phrenologists (Gall, early 19th century) through the strict localisationists (Broca, Wernicke) to the holists (Lashley 1929, Hebb 1949) to the modern compromise: function is neither strictly localised to single areas nor uniformly distributed across the brain, but is implemented by overlapping distributed networks with specialised hubs and modules. The connectome is the contemporary mathematical formulation of this compromise.

Bibliography [Master]

@book{Cajal1904,
  author = {Ram{\'o}n y Cajal, Santiago},
  title = {Textura del Sistema Nervioso del Hombre y los Vertebrados},
  publisher = {Imprenta y Librer{\'i}a de Nicol{\'a}s Moya},
  address = {Madrid},
  year = {1899-1904},
}

@book{Sherrington1906,
  author = {Sherrington, Charles S.},
  title = {The Integrative Action of the Nervous System},
  publisher = {Yale University Press},
  year = {1906},
}

@article{Langley1903,
  author = {Langley, John Newport},
  title = {The autonomic nervous system},
  journal = {Brain},
  volume = {26},
  year = {1903},
  pages = {1--26},
}

@book{Brodmann1909,
  author = {Brodmann, Korbinian},
  title = {Vergleichende Lokalisationslehre der Grosshirnrinde in ihren Prinzipien dargestellt auf Grund des Zellenbaues},
  publisher = {Barth},
  address = {Leipzig},
  year = {1909},
}

@article{PenfieldBoldrey1937,
  author = {Penfield, Wilder and Boldrey, Edwin},
  title = {Somatic motor and sensory representation in the cerebral cortex of man as studied by electrical stimulation},
  journal = {Brain},
  volume = {60},
  year = {1937},
  pages = {389--443},
}

@article{Mountcastle1957,
  author = {Mountcastle, Vernon B.},
  title = {Modality and topographic properties of single neurons of cat's somatic sensory cortex},
  journal = {Journal of Neurophysiology},
  volume = {20},
  year = {1957},
  pages = {408--434},
}

@article{HubelWiesel1962,
  author = {Hubel, David H. and Wiesel, Torsten N.},
  title = {Receptive fields, binocular interaction and functional architecture in the cat's visual cortex},
  journal = {Journal of Physiology},
  volume = {160},
  year = {1962},
  pages = {106--154},
}

@article{WattsStrogatz1998,
  author = {Watts, Duncan J. and Strogatz, Steven H.},
  title = {Collective dynamics of small-world networks},
  journal = {Nature},
  volume = {393},
  year = {1998},
  pages = {440--442},
}

@article{SpornsTononiKotter2005,
  author = {Sporns, Olaf and Tononi, Giulio and K{\"o}tter, Rolf},
  title = {The human connectome: a structural description of the human brain},
  journal = {PLOS Computational Biology},
  volume = {1},
  year = {2005},
  pages = {e42},
}

@article{VanEssenHCP2013,
  author = {Van Essen, David C. and Smith, Stephen M. and Barch, Deanna M. and Behrens, Timothy E. J. and Yacoub, Essa and Ugurbil, Kamil},
  title = {The {WU-Minn} Human Connectome Project: an overview},
  journal = {NeuroImage},
  volume = {80},
  year = {2013},
  pages = {62--79},
}

@article{BassettSporns2017,
  author = {Bassett, Danielle S. and Sporns, Olaf},
  title = {Network neuroscience},
  journal = {Nature Neuroscience},
  volume = {20},
  year = {2017},
  pages = {353--364},
}

@book{Kandel2021,
  author = {Kandel, Eric R. and Koester, John D. and Mack, Sarah H. and Siegelbaum, Steven A.},
  title = {Principles of Neural Science},
  edition = {6th},
  publisher = {McGraw-Hill},
  year = {2021},
}

@book{BoronBoulpaep2017,
  author = {Boron, Walter F. and Boulpaep, Emile L.},
  title = {Medical Physiology},
  edition = {3rd},
  publisher = {Elsevier},
  year = {2017},
}

@book{Nieuwenhuys2008,
  author = {Nieuwenhuys, Rudolf and Voogd, Jan and van Huijzen, Christiaan},
  title = {The Human Central Nervous System},
  edition = {4th},
  publisher = {Springer},
  year = {2008},
}

@book{Sporns2010,
  author = {Sporns, Olaf},
  title = {Networks of the Brain},
  publisher = {MIT Press},
  year = {2010},
}