Cajal had shown that neurons are separate cells. That raised a puzzle: how does a message cross the gap from one to the next? For decades scientists argued over "sparks" versus "soups," electrical jumps versus chemical messengers. The answer, it turned out, is mostly soup, and that fact is the key to how nearly every psychiatric medicine and recreational drug works.

A dream and two frog hearts

In 1921 the German pharmacologist Otto Loewi woke in the night with an idea for an experiment, scribbled it down, and could not read his own handwriting in the morning. The next night the idea returned and he went straight to his laboratory. He took two frog hearts, kept beating in salt solution. He stimulated the vagus nerve of the first heart, which slowed it down, then moved some of the fluid bathing it onto the second heart. The second heart slowed too, though its own nerve had not been touched. The conclusion was clear: the nerve had signaled to the heart by releasing a chemical. Loewi called it "vagus substance"; it was later identified as acetylcholine, and he shared a Nobel Prize in 1936 with the British scientist Henry Dale.

How a chemical synapse works

When an action potential arrives at an axon's tip, it opens channels that let calcium in. Calcium flowing into the axon terminal triggers the release of tiny bubbles, called vesicles, that spill neurotransmitter into the gap. The molecules drift across in less than a millisecond and fit into receptor proteins on the receiving cell, roughly like a key in a lock. Some receptors are themselves channels that open at once; others set off slower chains of chemical reactions inside the cell. Then the signal must be switched off. Neurotransmitter is pumped back into the sending cell, a process called reuptake, or broken apart by enzymes.

A smaller number of synapses are electrical: tiny pores connect two cells directly, and current flows straight through. These are faster but less flexible, and they are common where many cells must fire in perfect step.

Go and stop

Most neurotransmitters can be sorted by their usual effect. Glutamate is the brain's main excitatory transmitter, and GABA is its main inhibitory one. Excitatory input nudges the receiving cell toward its firing threshold; inhibitory input pulls it away. A single neuron in the cortex may receive thousands of synapses at once. It does not obey any one of them. Instead, it adds up its excitatory and inhibitory inputs and fires only if the total crosses threshold. Each neuron is a small decision maker, and the brain's balance between go and stop is delicate: too little inhibition and circuits can tip into the runaway firing of a seizure.

How drugs work

Because synapses depend on keys and locks, molecules from outside the body can interfere. An agonist is a drug that mimics a neurotransmitter or boosts its effect. An antagonist is a drug that blocks a receptor without activating it, like a key that fits the lock but will not turn.

Some of history's most famous poisons work at the synapse where nerves meet muscles, which uses acetylcholine. Curare, the arrow poison of South American hunters, blocks acetylcholine receptors on muscles, so the muscles never get the message and the prey is paralyzed. Nerve agents such as sarin do the opposite. They block the enzyme that breaks acetylcholine down, so the transmitter builds up and muscles spasm uncontrollably, including those needed to breathe. Botulinum toxin stops acetylcholine from being released at all; in tiny medical doses, as Botox, it relaxes muscles for months.

Drugs that act in the brain follow the same logic. Alcohol and anti-anxiety medicines called benzodiazepines strengthen the effect of GABA, which is why they calm, sedate and, in excess, can slow breathing dangerously. Cocaine blocks the reuptake of dopamine and related transmitters, leaving more of them in the gap. The common antidepressants called SSRIs, or selective serotonin reuptake inhibitors, work by slowing the reabsorption of serotonin, so it lingers longer at the synapse.

Worked example: why antidepressants take weeks

A doctor prescribes an SSRI for a patient's depression and warns that it may take four to six weeks to help. The patient is puzzled. If the pill blocks serotonin reuptake within hours of the first dose, why the wait?

The delay is itself an important clue. If depression were simply a shortage of serotonin in the gap, the drug ought to work almost at once. Instead, the synapse is only the starting point. Over weeks, receiving cells adjust the number and sensitivity of their receptors, genes switch on and off, and in animal studies new connections form in regions such as the hippocampus. Many researchers now think SSRIs help partly by making circuits more able to change, and that the lift in mood follows from those slower changes rather than from extra serotonin itself. The full answer is still debated, and the drugs help some patients a great deal and others very little. The case shows a wider lesson of this level: a drug's first action at the synapse is easy to describe, but its effect on a whole person unfolds through the living, adapting brain.