The Map: How the Nervous System Is Organized

Inject a protein tracer into a mouse’s blood and, in the cerebral cortex, it goes no farther than the single layer of cells lining the capillaries.

That result, from Thomas Reese and Morris Karnovsky in 1967, answered a question about where the brain’s barrier sits. This lesson is about where: a map of the nervous system, words for naming places on it, the fluid it floats in, and, hardest, what kind of experiment entitles you to say that a function lives in a particular place.

The map: two systems and a brain in four parts

The central nervous system (CNS) is the brain and spinal cord. The peripheral nervous system (PNS) is all the nervous tissue outside them: nerves, and ganglia, which are clusters of neuron cell bodies outside the CNS. Inside the CNS a cluster of cell bodies is a nucleus (not to be confused with the cell’s nucleus), and a bundle of axons running together is a tract.

Afferent fibres carry signals toward the CNS, as from a touch receptor. Efferent fibres carry signals away from it, to a muscle or a gland.

The PNS has three divisions. The somatic division serves the skin, joints and skeletal muscle. Somatic does not mean voluntary: the knee jerk is a somatic motor response that happens before you decide anything, and the stretch reflex gets its own lesson. The autonomic division (sympathetic and parasympathetic) serves smooth muscle, heart muscle and glands, largely without conscious control. The enteric division lies in the wall of the gut.

The adult brain has four parts:

• Cerebrum: the cerebral cortex, folded into ridges (gyri) and grooves (sulci) and divided into frontal, parietal, temporal and occipital lobes, plus the basal nuclei buried beneath it. The precentral gyrus contains primary motor cortex; the postcentral gyrus contains primary somatosensory cortex; primary visual cortex (Brodmann’s area 17) lies in the occipital lobe; primary auditory cortex lies in the temporal lobe. The basal nuclei take part in selecting and controlling movement, among other roles.
• Diencephalon: the thalamus, which relays and gates information passing to and from the cortex, and the hypothalamus, which contributes to homeostatic regulation of temperature, feeding, fluid balance and hormone release.
• Brainstem: midbrain, pons and medulla. They carry the long tracts between brain and cord, hold most cranial nerve nuclei, and take part in breathing and cardiovascular control.
• Cerebellum: compares intended movements with sensory feedback and helps correct them; the cerebellum lesson shows how.

Read every “contains” and “takes part in” literally: no region here has a single function, and no function lives in a single region. The cortex is a mosaic: one multimodal MRI atlas of healthy young adults divides each hemisphere into 180 areas (Glasser and colleagues, 2016), and borders vary between people. How the neural tube becomes these regions is a developmental story, and how cortex is divided into areas is another. The cells that make up the tissue come in the next lesson.

Counts need their source. One study using the isotropic fractionator on the brains of adult men estimated 86.1 ± 8.1 billion neurons and 84.6 ± 9.8 billion non-neuronal cells (Azevedo and colleagues, 2009), so the claim that glia outnumber neurons ten to one is not supported.

Many pathways cross the midline: the corticospinal tract mainly in the caudal medulla, the dorsal column pathway in the medulla, the spinothalamic pathway in the spinal cord, and at the optic chiasm only fibres from the nasal half of each retina. You will work lesions out from these in touch and pain and visual cortex.

Directions, planes and the bend in the neuraxis

The neuraxis is the long axis of the CNS. In a rat it runs nearly straight from nose to tail. In a human it bends by roughly a right angle where the midbrain meets the diencephalon. Think of a bent flashlight: “toward the beam” points forward at the head but up along the handle.

So the direction terms change meaning across that bend. In the forebrain, dorsal points toward the top of the head and ventral toward the base of the skull; rostral points toward the face and caudal toward the back of the head. In the spinal cord, dorsal points toward the back of the body, ventral toward the belly, rostral toward the head and caudal toward the feet. Always say which frame you mean. Medial means toward the midline and lateral away from it. Ipsilateral means on the same side of the body, and contralateral means on the opposite side.

Three section planes cut the brain:

• a sagittal plane separates left from right (the midsagittal plane runs down the midline);
• a coronal plane separates front from back;
• a horizontal plane separates top from bottom.

A section of the cord cut across its long axis is a transverse section. Because of the bend, a coronal section of the forebrain and a transverse section of the cord are both roughly perpendicular to the neuraxis at their level.

Ventricles, cerebrospinal fluid and two barriers

The brain contains four fluid-filled ventricles, the remains of the neural tube’s central canal: two lateral ventricles, one in each hemisphere, the third ventricle in the diencephalon, and the fourth between the pons and medulla in front and the cerebellum behind. Cerebrospinal fluid (CSF) is secreted mainly by the choroid plexus, a frond of capillaries covered by a secretory epithelium that projects into each ventricle. The choroid plexus is the predominant source in the adult mammalian brain, not the only one, and in isolated mouse choroid plexus water could be moved against an imposed osmotic gradient, so secretion is an active process (Steffensen and colleagues, 2018).

The net route runs one way, though the fluid also moves back and forth with each heartbeat and breath. Fluid passes from each lateral ventricle through an interventricular foramen into the third ventricle, through the narrow cerebral aqueduct of the midbrain into the fourth ventricle, and out through the fourth ventricle’s median and lateral apertures into the subarachnoid space around the brain and cord, where it cushions them. It leaves through arachnoid granulations into the venous sinuses and by other routes: in mice, lymphatic vessels at the skull base clear CSF macromolecules (Ahn and colleagues, 2019), and dural lymphatic vessels have been imaged in five healthy people and three marmosets (Absinta and colleagues, 2017), though that study could not show that they drain CSF.

How quickly is the fluid replaced? Take conventional adult human teaching values, not measurements from one person: a total CSF volume V of about 150 mL and a secretion rate Q of about 500 mL/day, with outflow equal to secretion so the volume is steady, and treat the CSF as one well-mixed compartment.

turnover rate = Q / V; volume-to-flow time τ = V / Q; fraction of an initial marker left = exp(−t / τ)

1. Turnover rate: 500 mL/day ÷ 150 mL = 3.33 per day.
2. Volume-to-flow time: τ = 150 ÷ 500 = 0.3 day, which is 0.3 × 24 = 7.2 h.
3. Check: 500 mL/day ÷ 1,440 min/day = 0.347 mL/min, and 150 ÷ 0.347 = 432 min = 7.2 h.
4. A marker present at time zero, with fresh fluid mixing in, falls as exp(−t/τ). At t = 7.2 h that is exp(−1) = 0.368, so 36.8 percent of the marker is still there after one full volume of fluid has been made. Replacing a volume does not empty it.
5. Now block the cerebral aqueduct while secretion continues. Fluid made in the lateral and third ventricles has no exit, so those ventricles enlarge; the fourth ventricle, downstream of the block, does not.

Secretion rate Q 500 mL/day CSF volume V 150 mL Time since the marker was present 7.2 h Block the cerebral aqueduct (secretion continues)

Lesson values (500 mL/day, 150 mL) Set t = τ Set t = 3τ Set t = 0

Turnover rate Q / V

—

Volume-to-flow time τ = V / Q

—

Check in mL/min

—

Fraction of the time-zero marker left

—

—

In this sketch: the lesson’s conventional adult teaching values (500 mL/day, 150 mL) are the starting point; the sliders vary them and are not measurements from one person; one well-mixed compartment; outflow equals secretion, so the volume is steady; the marker is present at time zero and none is added afterwards. The ventricles are drawn as a schematic, not to scale.

Turnover = Q / V; τ = V / Q; fraction of an initial marker left = exp(−t / τ): after one full volume of fluid has been made, 36.8 percent of the marker is still there. Block the aqueduct and fluid made upstream has no exit, so this one-compartment rule stops applying.

Two barriers separate the brain from the blood. The blood-brain barrier lies at the endothelial cells of brain capillaries, joined by tight junctions (the junction itself is covered in the epithelia lesson). Astrocyte endfeet wrap the vessels, but in Reese and Karnovsky’s mouse experiments the tracer never reached them. It is selective, not a wall: transporters carry glucose and amino acids across, and small lipid-soluble molecules diffuse through. A few small regions around the ventricles, the circumventricular organs, have leaky capillaries. The blood-CSF barrier is different: choroid plexus capillaries are leaky, and the seal is formed by the tight junctions of the epithelium over them.

Does sleep speed clearance? In 13 sleeping people, slow EEG rhythms were followed by pulses of CSF flow (Fultz and colleagues, 2019), a correlation, and mouse tracer studies disagree. Keep it open until sleep and rhythms.

Four kinds of evidence, four different licences

Evidence	What it can show	What it cannot show alone
Lesion	The damaged tissue contributes to a task	Which part of the damage mattered
Recording	Activity covaries with a task or stimulus	That the activity is used
Imaging (fMRI)	Where a haemodynamic signal goes with a task	Which neural signal, or necessity
Stimulation	Adding activity changes behaviour	That the region is required

A lesion tests necessity, but damage spreads. Scoville and Milner’s 1957 report on bilateral medial temporal resections in human patients is the classic case: the recent-memory loss was striking, but the resections removed several medial temporal structures at once, so the deficit could not be pinned to one structure. Memory systems tells that story.

A recording shows that firing goes with a stimulus or a choice, not that the brain uses it. Imaging has the same limit plus one more: BOLD fMRI measures blood oxygenation, and in anaesthetized macaque visual cortex it followed local field potentials, the summed synaptic and dendritic currents near an electrode, better than spiking (Logothetis and colleagues, 2001). An fMRI activation does not mark a region as the seat of anything.

Stimulation adds activity and asks whether behaviour follows. That shows a contribution, in that task, from the stimulated population, not from one identified cell type, because imaging in anaesthetized animals found that microstimulation sparsely activates neurons around the electrode, some as far as millimetres away. Tools for recording many neurons and for silencing or driving defined cells come in large-scale neuroscience and circuit experiment design.

Design problem

Rhesus monkeys judge the direction of noisy moving dots, and you suspect that area MT, where neurons are tuned to the direction of motion, contributes to the judgement. Your recordings already show that MT firing covaries with the monkeys’ choices. Design an experiment that tests whether MT activity contributes to the choice, say what result you predict, and say what your result would and would not license.

One way to do it

Place a microelectrode among MT neurons that share a preferred direction, mapped first. On randomly interleaved trials, deliver weak microstimulation while the dots are on, and compare choices on stimulated and unstimulated trials at the same motion strengths. Interleaving rules out drifts in attention or motivation.

Predict the sign first: if those signals are used, stimulation should push choices toward the preferred direction, not merely make the monkey worse. Salzman, Murasugi, Britten and Newsome (1992) did this in 139 experiments: stimulation had a significant effect in 89, and in 86 of those the bias was toward the preferred direction.

What it licenses: MT activity contributes causally to this choice in this task. Not that MT is the only place motion is processed, or that it is required. A lesion would test necessity but also damage neighbours and passing fibres. fMRI would show neither, since BOLD tracks synaptic input more than spiking in the anaesthetized macaque (Logothetis and colleagues, 2001).