One Switch in the Brain, Pressed Both Ways

We tend to think of sleep as the brain switching off. The reality is closer to the opposite. Being awake does not keep itself going. Something in the brain has to keep holding it there.

There is a hand that holds it. In 1998, two teams almost simultaneously found that hand in a tiny cluster of neurons in the hypothalamus. One called it orexin; the other called it hypocretin — two names for the same substance.

What orexin does is not build the switch but press it so it stays on one side. When the signal is strong the switch is pinned to wake; when the signal goes, it tips over to sleep.

The slider on the left is the strength of that signal. Push it yourself and you can see how the hand moves.

This hand can fail in two directions. The first is a hand that won’t let go.

The model Saper and colleagues proposed in 2005 likened the way the wake and sleep circuits hold each other down to an electrical switch. One side has to win outright before it flips over with a click.

If the signal doesn’t switch off when it should, the switch spends the night pinned to wake. That is insomnia.

Pull the slider on the left toward SLEEP. It doesn’t move as far as you pull, and even all the way over it never reaches sleep. Let go and it springs straight back.

The second failure is the opposite. The hand itself disappears. In narcolepsy type 1, a large share of the neurons that make orexin are lost.

With nothing to press it, wakefulness doesn’t hold. The switch slips toward sleep again and again during the day, and with it come cataplexy — muscles losing their strength at moments of strong emotion — and sleep paralysis and hallucinations on falling asleep or waking.

Use the button on the left to wake them, and you can see what it means to have no hand to hold on.

If the hand won’t let go, block the gap between the hand and the switch: a drug that covers the spot where orexin binds before orexin can get there. With lemborexant approved in Korea in June 2026, it can now be prescribed there as well.

Older sleeping pills worked differently. BenzodiazepinesHalcion · Ativan · Valiumtriazolam · lorazepam · diazepamA family of sedatives and sleeping pills in use long before orexin was known. and zolpidemAmbien · StilnoxzolpidemOne of the most widely prescribed sleeping pills. Not a benzodiazepine, but it acts on the same inhibitory circuit in the brain.-type drugs act on the brain’s general inhibitory signal, turning down brain activity wholesale.

Rather than picking out the wake switch alone, it is like throwing the main breaker for the whole house — which is why grogginess into the next morning, and dependence, came with them.

Drugs that block orexin turn down only the wake circuit. But that raises a new problem: how long should the block last?

Hold on all night and you cannot let go in the morning. How fast the drug leaves the body becomes, directly, the difference the next morning.

The answer didn’t come at once. SuvorexantBelsomraMerck (MSD), USThe first drug to block orexin. FDA-approved in 2014. in 2014 and lemborexantDayvigoEisai, JapanFDA-approved in 2019; approved in Korea in 2026. in 2019 both had half-lives longer than a night, and the second actually lasted longer than the first. Only with daridorexantQuviviqIdorsia, SwitzerlandFDA-approved in 2022. in 2022 did the half-life fit inside a single night.

The fourth pushed that direction as far as it goes. VornorexantVorzzz (ボルズィ)Taisho Pharmaceutical, JapanApproved in Japan in August 2025 — the first approval anywhere. cut the half-life to around two hours and was approved first in Japan, in August 2025. In a phase 3 trial of 596 people, the time it took to fall asleep was about 10 minutes shorter than on placebo, and when the drug was stopped no rebound insomnia or withdrawal symptoms appeared.

Lost neurons cannot be brought back, so until now narcolepsy has been treated one symptom at a time: a stimulant to fight off sleepiness, another drug to reduce cataplexy.

Then came the idea of building a hand to press in their place. The first drug from Japan’s Takeda PharmaceuticalTakeda Pharmaceutical (武田薬品)JapanThe company behind oveporexton. did work. But the higher the dose, the more it damaged the liver, and development was halted.

What matters is that it was not abandoned for lack of effect. Filling in the signal worked; the problem was how much to put in.

The second drug was rebuilt with a different structure: oveporextonOrzeyfulTakeda Pharmaceutical, JapanFDA-approved in August 2026.. Designed to spare the liver and to be taken once a day, it was approved by the US FDA on 5 August 2026.

Orexin binds at two sites, and this drug stimulates only the one mainly involved in wakefulness. Press the button on the left and a hand appears where there was none.

How does it compare with existing drugs? Measured by how long people stayed awake when asked to, the best of today’s stimulants adds about 9 to 11 minutes. Oveporexton added 12 to 25.

The bigger difference is cataplexy. Existing treatment has handled sleepiness with stimulants and cataplexy with separate drugs; the FDA described this as the first drug to treat the full range of narcolepsy type 1 symptoms. In the same trial, at week 8, the placebo group had 8.8 cataplexy episodes a week; the treated groups had 2.5 to 5.9, depending on dose.

The most common side effect of the waking drug was insomnia. Nearly half the participants had it, and in most it cleared within a week. Urinary urgency and frequency followed.

The sleeping drug is the mirror image. What researchers specifically watched for in the vornorexant phase 3 trial was cataplexy, sleep paralysis, hypnagogic hallucinations and daytime sleepiness. Every one of them is a symptom of narcolepsy.

Fortunately, none appeared. But why that list was prepared in advance is what matters.

Each drug carries the other’s disease as a side effect. Push the switch too hard in either direction, and the opposite failure appears.