Brain Systems

The brain has never touched the world

Start with an anatomical fact that is easy to state and surprisingly hard to absorb. The brain is enclosed in a rigid box of bone, cushioned in cerebrospinal fluid, and shielded chemically by the blood-brain barrier. No light reaches it. No sound of any consequence reaches it. It is not in contact with the objects it thinks about, and it never has been. Whatever the brain knows about the world, it knows at second hand, through wires.

Now look at what travels down those wires. A sensory nerve fibre carries action potentials: brief, stereotyped voltage spikes, each roughly a millisecond long and each essentially the same size as the last. That last clause is the crux. Action potentials are all-or-none events. A brighter light does not produce a bigger spike; a louder sound does not produce a taller one. It produces more of them, or a different pattern of them in time, but the individual units are interchangeable. A spike in the optic nerve and a spike in the auditory nerve are, considered purely as electrical events, the same kind of thing.

Put those two facts together and the problem becomes acute. The brain is cut off from the world, and the only messenger it receives is a spike that carries no intrinsic label. There is nothing about a spike from the retina that says "I am about light." So how does the brain avoid total confusion? How does it not hear the sunset and see the door slam?

Labelled lines: why pressing your eye makes light

The answer, worked out in the nineteenth century and confirmed by everything since, is that the meaning of a signal is given by where it arrives. This is the principle of labelled lines, sometimes still called by its old name, the law of specific nerve energies. The brain does not decode the impulse. It decodes the address.

Signals arriving at the primary visual cortex are experienced as vision, whatever produced them. Signals arriving at the primary auditory cortex are experienced as sound, whatever produced them. The line is labelled by its destination, and the destination is fixed by the wiring, which was laid down in development long before any stimulus arrived.

This is not an abstract claim. It makes a hard, falsifiable prediction: if you could excite the visual pathway by some means other than light, you should see something rather than feel something. And you can test that prediction right now, in about three seconds, with no equipment.

The phosphene is not a quirk. It is the principle in the clearest possible form, and it generalises. Electrical stimulation of the visual cortex during neurosurgery produces phosphenes, not tingles. A blow to the head produces "seeing stars." Pressure on the ulnar nerve at the elbow, the funny bone, produces tingling in the little finger rather than pain at the elbow, because the fibres you compressed are labelled for the skin of the hand and the brain reads them accordingly. In every case, the sensation follows the wire, not the cause.

The consequence is worth stating carefully, because it reframes everything that follows. Sensory qualities, the redness of red, the pitch of a note, the sting of a burn, are not properties of the incoming signal. They are properties of the pathway and the machinery at the end of it. The world supplies energy; the brain supplies meaning. Every remaining page in this section is, in one way or another, an account of how a particular line does that job.

The architecture every sense obeys

Labelled lines tell you how the brain keeps the senses apart. They do not yet tell you what happens inside one. And here the striking thing is how little the senses differ. Vision, hearing, touch, and taste solve wildly different physical problems, photons, air pressure, skin deformation, dissolved molecules, and yet all four are built to the same plan. Learn the plan once and you have most of a sensory system for free.

1. Transduction: everything becomes voltage

   Each sense begins with a receptor cell whose job is conversion. A photoreceptor turns absorbed light into a change in membrane voltage. A hair cell in the cochlea turns a mechanical deflection of a few nanometres into a voltage change. A mechanoreceptor in the skin does the same for indentation. A taste cell does it for a dissolved molecule. The physics differ completely; the output is identical in kind. Every sense converts its own physical quantity into the single common currency of the nervous system, a change in the potential across a membrane, using ion channels that open or close in response to the stimulus. From that moment on, the brain is no longer dealing with light or sound. It is dealing with voltage.

2. Receptive fields: each cell watches one patch

   No sensory neuron listens to everything. Each responds only to stimulation within a limited region, its receptive field: a small area of retina, a patch of skin, a narrow band of sound frequencies. The size of the field sets the resolution. Fingertips have tiny receptive fields and can resolve two points a couple of millimetres apart; the skin of the back has fields many times larger and cannot. Acuity is not a property of the sense as a whole. It is a property of how finely the sensory surface is divided up.

3. Topographic maps: the sensory surface is redrawn in cortex

   Neighbouring points on the receptor sheet project to neighbouring points in the brain. The retina maps onto visual cortex (retinotopy), the cochlea's frequency axis maps onto auditory cortex (tonotopy), the body surface maps onto somatosensory cortex (somatotopy). These maps are inherited from the thalamus and preserved up the hierarchy. Crucially they are distorted: cortical territory is allotted by importance and receptor density, not by physical size, which is why the hands and lips occupy an enormous share of the somatosensory map and the trunk almost none.

4. Hierarchy: simple features are assembled into complex ones

   Early stages report primitives, spots of light, edges, tones, points of pressure. Later stages combine these into progressively more abstract descriptions: orientations, motion, textures, shapes, objects, faces, words. Nothing at the first stage knows what an object is; nothing at the last stage sees a raw pixel. The construction is gradual and it is the general strategy of the cortex, not a quirk of one sense.

5. Adaptation: only news gets through

   Hold a stimulus constant and the response to it dies away. You stop feeling your clothes minutes after dressing. You stop smelling your own house within moments of walking in. Stare at a waterfall for a minute and the rocks beside it appear to drift upward when you look away. This is not fatigue and it is not a defect. It is a deliberate allocation of a scarce resource: a nervous system that reported everything all the time would spend most of its bandwidth on facts that have not changed since the last report. Adaptation spends the bandwidth on news.

These five features are not a list to memorise. They are a set of consequences, and each of them follows from a constraint the brain actually faces. Transduction follows from the fact that neurons speak only one language. Receptive fields follow from the fact that a finite number of cells must divide up an infinite world. Maps follow from the fact that wiring costs space and energy, so connected things should be placed near each other. Hierarchy follows from the fact that no single cell could recognise a face from raw light. Adaptation follows from the fact that information is carried by change.

The thalamic gateway, and the one sense that skips it

Once a sense has transduced its stimulus and passed it up through its early stations, it faces a bottleneck that all of them share. To reach the cerebral cortex, sensory information must first pass through the thalamus, the pair of egg-shaped grey masses at the geometric centre of the brain. Vision arrives at the lateral geniculate nucleus. Hearing arrives at the medial geniculate. Touch, temperature, and pain from the body arrive at the ventral posterolateral nucleus, and the face and taste at the ventral posteromedial. Each nucleus then projects to its own patch of cortex.

The word usually attached to this is "relay," and it is misleading. A thalamic relay nucleus receives only a minority of its synapses from the sense organ it serves; most of its input comes back down from the cortex and in from the brainstem. The thalamus is not a wire. It is a gate, and how far that gate opens depends on attention, arousal, and what the cortex expects. When you fall asleep, the gate largely closes, and that is a measurable change in how thalamic neurons fire, not a metaphor.

And then there is smell, which does none of this.

Why should one sense be built differently? Partly history: olfaction is evolutionarily ancient, and much of the forebrain of early vertebrates was olfactory. The other senses were, in a sense, fitted around a system that already existed. But the exception is more than a fossil, because it changes what smell can do. A signal that reaches the limbic system without a cortical gatekeeper is a signal that can trigger emotion and retrieve memory faster than you can name what you are smelling, and that is exactly the phenomenology people report.

Perception is for action, and action is part of perception

It is tempting to treat the senses as an end in themselves, as though the brain's purpose were to build a picture and then admire it. It is not. Perception is expensive, and evolution does not fund expensive things that do nothing. Every sensory system in the brain exists because it improves behaviour, and the output side of that arrangement is the motor system: the cortical, subcortical, and spinal machinery that turns decisions into muscle contractions.

The two halves are not merely connected. They are interleaved, and two everyday phenomena show it.

Notice what these two examples imply together. If the brain must predict its own actions in order to perceive correctly, then perception is not a read-out at all. It is an ongoing comparison between what arrives and what was expected, and the same comparison logic runs through vision, hearing, and touch. That idea, and the anatomical evidence for it, including the awkward fact that far more nerve fibres run backwards from cortex to thalamus than forwards, is the subject of sensory processing, which sets out the principles all the individual systems share.

Explore the systems

Each page below takes one system and answers one question about it. The principles page is the one to read if you want the logic that all the others assume.

Related topics