How neurons fire

Your brain is built from billions of cells called neurons, and they talk to each other with tiny jolts of electricity. Every thought, every wiggle of a finger, is really a storm of these little pulses.

Each neuron keeps itself a bit like a charged battery. There is a small voltage across its skin, ready to go. When signals arrive from other neurons, they give it a push. A weak push does nothing much: the neuron twitches and settles straight back down. But once the push is strong enough, past a certain line, something sudden happens. The neuron fires a full electrical pulse, always the same size, and sends it shooting down a long thin wire called the axon to the next cell.

That is the key trick. A neuron does not send a big pulse for a big push and a small pulse for a small push. It fires all the way or not at all, like a light switch that clicks rather than a dimmer that slides. Push too gently and nothing travels. Push past the line and you get the full spike, every time.

At the far end the wire does not quite touch the next neuron. There is a tiny gap. So the electrical pulse gets turned into a puff of chemicals that drift across and nudge the next cell, which may then fire its own pulse. Chain that up across billions of cells and you have a working brain.

A resting neuron is not electrically flat. It sits at a resting potential of about \(-70\) millivolts, meaning the inside is that much more negative than the outside. It holds this gap by carefully controlling which ions (charged atoms) can cross its membrane, and by running a tiny pump, the sodium-potassium pump, that pushes three sodium ions out for every two potassium ions it drags in. That pumping, plus the way potassium leaks back out, sets up the standing voltage.

Signals from other neurons nudge that voltage up or down. If the nudges together lift it to the threshold, around \(-55\) mV, the cell fires. Below threshold, nothing propagates: the little bump just leaks away and the membrane relaxes back to rest. This is the all-or-nothing rule. There is no half spike.

Once threshold is reached, a fast, self-driving sequence takes over. Gates in the membrane called voltage-gated sodium channels snap open, and sodium ions flood in. Because sodium is positive, the inside voltage rockets upward, overshooting all the way to about \(+40\) mV. This surge is depolarisation. A beat later the sodium gates slam shut and slower potassium channels open, letting potassium rush out. That drains the positive charge back out and the voltage plunges down again, a step called repolarisation, usually dipping briefly below rest before settling.

Straight after a spike the neuron cannot immediately fire again. This is the refractory period: the sodium gates need a moment to reset. It caps how fast a neuron can fire and, usefully, it stops the spike from running backwards. The pulse only goes one way, down the axon.

How does the spike travel? Each patch that fires depolarises the patch next door, tipping it past threshold, so the action potential regenerates itself all the way along without fading. Many axons are wrapped in a fatty sheath called myelin, with small bare gaps between the wrapped segments. The signal jumps from gap to gap rather than crawling along every millimetre, which is far faster. This leaping is called saltatory conduction.

At the axon's end sits a synapse, a narrow gap to the next cell. The arriving spike triggers the release of chemical messengers called neurotransmitters. They cross the gap and land on the next neuron, nudging its voltage up or down and feeding into that cell's own decision about whether to fire.

The resting potential is an equilibrium, not a switch-off. Each ion species has an equilibrium voltage where its concentration gradient and the electric field balance, given by the Nernst equation:

\[ E_{\text{ion}} = \frac{RT}{zF}\,\ln\!\frac{[\text{ion}]_{\text{out}}}{[\text{ion}]_{\text{in}}}. \]

Potassium sits near \(-90\) mV, sodium near \(+60\) mV. The resting membrane is far more permeable to potassium than to sodium, so it rests near potassium's value, dragged a little positive by the sodium leak. The Goldman equation blends all the permeabilities into the actual resting voltage. The sodium-potassium pump maintains the gradients over the long run; it does not set the fast voltage directly.

The spike is a regenerative feedback loop. Voltage-gated sodium channels have the property that opens the whole thing: their opening is itself driven by depolarisation. Depolarise a patch a little and some sodium channels open, sodium enters, which depolarises the patch further, which opens more channels. That positive feedback is why the upstroke is explosive and why the response is all-or-nothing once threshold is crossed. Hodgkin and Huxley captured it in 1952 with a set of equations in which the sodium conductance depends on gating variables \(m\) and \(h\), and the potassium conductance on \(n\):

\[ I = C_m\frac{dV}{dt} + \bar{g}_{\text{Na}}m^3h\,(V - E_{\text{Na}}) + \bar{g}_{\text{K}}n^4\,(V - E_{\text{K}}) + g_L(V - E_L). \]

The variable \(h\) is the key to timing: it inactivates the sodium channels a fraction of a millisecond after they open, which ends the upstroke. The slower \(n\) opens the potassium channels for the downstroke. Sodium inactivation is also what sets the refractory period: until \(h\) recovers, no new spike is possible, so the action potential propagates forward only and cannot double back.

Propagation is the loop repeated in space. Local current from a firing patch depolarises the adjacent membrane past threshold, which fires in turn, so the spike travels as a self-sustaining wave at constant amplitude, unlike a passive voltage that would decay with distance. Myelin raises the membrane resistance and lowers its capacitance between the nodes of Ranvier, so the signal spreads passively and fast under each sheath and only regenerates at the bare nodes. That saltatory conduction lifts conduction speeds from around a metre per second to over a hundred.

Meaning is carried in timing, not size. Because every spike is the same height, a neuron cannot encode a stronger stimulus with a bigger spike. It encodes it in the firing rate, more spikes per second, and in the precise timing of spikes relative to other neurons. At the receiving end, each synapse delivers a small graded change in voltage, and the neuron sums these inputs over space (many synapses) and time (inputs that arrive close together). This synaptic summation at the axon hillock is the actual decision: if the running total crosses threshold, the cell fires; if not, it stays quiet. A neuron is, in effect, a device that integrates thousands of small votes and answers with a single all-or-nothing spike.