How Brain Cells Signal

The two languages of the brain

A nervous system faces a design problem. It must move information over long distances quickly, which favours an electrical signal, and it must be able to adjust the strength of each connection with fine grain, which electricity does badly. Evolution's answer was to use both, and to split the job by scale. Inside a single cell, where speed over distance is the constraint, the signal is electrical. At the junction between cells, where flexibility is the constraint, it is chemical.

The conversion happens at the axon terminal. An electrical impulse arrives, opens calcium channels, and the calcium influx causes vesicles of neurotransmitter to fuse with the membrane and empty into the cleft. On the far side, the transmitter binds receptors that produce a new electrical change in the receiving cell. Signal converted, delivered, converted back. It happens in under a millisecond and it happens, in the human brain, on the order of a hundred trillion times over across all its synapses.

The resting potential

A neuron at rest is not idle. It is holding a charge, and holding it costs energy continuously. Across its membrane there is a voltage difference of roughly 70 millivolts, with the inside negative relative to the outside. Small as that sounds, the membrane is only a few nanometres thick, so the electric field across it is enormous. This stored potential is the battery the neuron discharges when it fires.

The voltage arises from two things working together. First, the concentrations of ions are unequal across the membrane: potassium is concentrated inside the cell, sodium and chloride outside. Second, the membrane at rest is far more permeable to potassium than to sodium, thanks to leak channels that are always open. Potassium therefore tends to drift out down its concentration gradient, leaving behind unbalanced negative charge, until the growing electrical pull inward exactly cancels the chemical push outward. The resting potential is that equilibrium, sitting close to but not exactly at the potassium equilibrium potential.

The ion gradients themselves do not maintain themselves. The sodium-potassium pump, an ATP-consuming protein embedded in the membrane, grinds away continuously, ejecting three sodium ions for every two potassium ions it brings in. This pump is a major part of why the brain is so metabolically expensive: a substantial fraction of the organ's very large energy budget goes on nothing more glamorous than keeping the batteries charged.

The action potential

When the neuron's membrane voltage is pushed up past a critical threshold, roughly minus 55 millivolts, something abrupt happens. Voltage-gated sodium channels snap open, sodium rushes in down its steep gradient, and the inside of the cell swings positive within a fraction of a millisecond. That depolarisation opens the sodium channels a little further along the axon, which depolarises the next patch, and so on. The action potential regenerates itself as it goes, which is why it can travel a metre without losing amplitude.

Two properties define it. It is all-or-nothing: below threshold there is no spike at all, and at or above threshold the spike is full-sized, regardless of how far past threshold the stimulus went. And it is self-regenerating: unlike a passive electrical signal, which decays with distance, the action potential is actively rebuilt at every point along the axon. The consequence is that a neuron cannot signal intensity by firing a bigger spike. It can only signal intensity by firing more often. Information in an axon is carried in the timing and rate of identical pulses.

1. Threshold

   Summed input depolarises the membrane past about minus 55 millivolts at the axon hillock, the trigger zone where voltage-gated sodium channels are densest.

2. Rising phase

   Sodium channels open in a rapid positive feedback loop. Sodium floods in and the membrane potential swings sharply positive, overshooting zero.

3. Falling phase

   Sodium channels inactivate and slower voltage-gated potassium channels open. Potassium leaves, driving the voltage back down towards rest.

4. Refractory period

   Inactivated sodium channels cannot reopen immediately. For a brief window the neuron cannot fire again, which caps the maximum firing rate and forces the impulse to travel forwards rather than backwards.

Speed depends heavily on myelin. Where an axon is insulated by glial wrapping, the voltage-gated sodium channels are concentrated at the bare gaps in the sheath, the nodes of Ranvier, and are largely absent beneath it. The insulated stretch between two nodes cannot fire, but it carries current well: the wrapping raises the resistance of the axon wall and lowers its capacitance, so charge leaks away far more slowly and spreads far further. The impulse is therefore rebuilt only at the nodes, having crossed the gap between them passively and fast. This is called saltatory conduction, from the Latin for leaping, and the name is a description of how it looks rather than of what happens: nothing actually jumps. Myelinated fibres conduct at over 100 metres per second; unmyelinated ones manage roughly one. See the action potential for the full mechanism.

Crossing the synapse

At the end of the axon the electrical signal reaches a dead end. The synapse is a gap, typically about twenty nanometres across, and voltage cannot simply jump it. What happens instead is a small chemical event, exquisitely timed.

The arriving depolarisation opens voltage-gated calcium channels in the terminal membrane. Calcium concentration inside the terminal is normally kept extremely low, so the influx is a dramatic proportional change, and calcium acts as the trigger. It binds sensor proteins on the synaptic vesicles, small membrane sacs pre-loaded with neurotransmitter and docked ready at the release site. The vesicles fuse with the terminal membrane and empty their contents into the cleft.

Transmitter diffuses across in microseconds, binds receptors on the postsynaptic membrane, and the receiving cell's voltage changes. Then the signal must be shut off, or the synapse would blur into a continuous smear. Clearance happens by reuptake, where transporter proteins pull the transmitter back into the presynaptic terminal or into surrounding astrocytes, or by enzymatic destruction, as when acetylcholinesterase splits acetylcholine in the cleft within milliseconds.

Fast and slow receptors

A neurotransmitter does nothing until it meets a receptor, and it is the receptor, not the molecule, that determines the effect. The same transmitter can excite one cell and inhibit another, purely because they express different receptors. Receptors divide into two families that differ fundamentally in speed and mechanism.

The division is not a detail of pharmacology; it is the reason the brain has two operating timescales. Ionotropic signalling handles the moment-to-moment computation, the millisecond-precise transfer of specific information. Metabotropic signalling handles the context: how excitable this circuit is right now, how much attention is being paid, whether this is a moment for exploration or for exploitation. Most transmitters act on both families, which is how one molecule can both carry a message and adjust the conditions under which messages are read.

Excitation and inhibition

Fast synaptic traffic in the brain is dominated by two amino acids. Glutamate is the principal excitatory transmitter: when it opens an ionotropic receptor, positive ions flow in and the postsynaptic cell moves towards threshold. GABA is the principal inhibitory one: opening a GABA-A receptor lets chloride in, pushing the cell away from threshold. Between them these two molecules account for the overwhelming majority of synapses in the cortex.

It is tempting to think of inhibition as merely a brake, something to stop runaway activity. It is far more than that. Inhibitory interneurons sculpt the timing of firing, sharpen the selectivity of responses, generate the rhythmic oscillations visible in an EEG, and enforce the competition that lets one representation win out over its neighbours. A circuit without inhibition would not simply be over-excited; it would be incapable of making distinctions.

The health of a circuit depends on the balance between the two. Tip it too far towards excitation and activity runs away: this is, in essence, what a seizure is. Tip it too far towards inhibition and the circuit falls silent. Most drugs that alter consciousness act somewhere on this axis. Benzodiazepines and alcohol, for example, both enhance GABA-A receptor function, which is why both are sedating.

Neuromodulation: the third mode

Fast excitation and fast inhibition are not the whole story. Layered over them is a third mode of signalling that works on a completely different scale of time and space. A small nucleus deep in the brainstem or midbrain, containing perhaps tens of thousands of neurons, sends its axons everywhere: across the whole cortex, the hippocampus, the thalamus, the striatum. When those neurons fire, they release transmitter not into a tight cleft aimed at one target but into the extracellular space more broadly, where it diffuses and reaches many cells at once. This is called volume transmission.

Because these transmitters act mainly on metabotropic receptors, the effect is slow to arrive and slow to fade, lasting seconds or minutes rather than milliseconds. And because it reaches so many cells simultaneously, it does not carry a message. It changes the state of the circuit: how excitable its neurons are, how easily its synapses change strength, how much signal it passes relative to noise. Neuromodulation sets the gain rather than the content.

The four classical systems are dopamine, noradrenaline, serotonin, and acetylcholine, each originating in a small cluster of cells and projecting extraordinarily widely. Between them they determine whether you are alert or drowsy, focused or diffuse, motivated or indifferent, ready to learn or set in your ways. They are the reason the same cortex, wired the same way, behaves so differently at three in the afternoon and three in the morning.

Integration and the decision to fire

Put all of this together at a single neuron and you get the brain's fundamental computation. At any instant a cortical pyramidal neuron may be receiving input at thousands of synapses. Some are excitatory, nudging it towards threshold; some are inhibitory, pulling it away. Each arriving signal produces a small, graded, decaying change in local membrane voltage, and these changes spread passively through the dendrites towards the cell body, growing weaker as they go.

The neuron sums them. It sums them in space, because inputs arriving simultaneously at different dendrites add together, and it sums them in time, because inputs arriving in quick succession at the same synapse pile up before the first has decayed. If the running total at the axon hillock crosses threshold, the neuron fires. If it does not, nothing happens.

And the threshold itself is not fixed. Neuromodulators shift it, inhibitory tone shifts it, recent firing history shifts it. So the decision to fire is made against a background that is itself continuously changing. That, in one sentence, is why brains are hard: the computation and the conditions of the computation are being adjusted at the same time, by the same tissue.

Explore signalling

Related topics