They seemed unconscious, yet a third of them steered a tone by thought

A patient lies in intensive care a few days after a severe brain injury. He does not squeeze the doctor’s hand, does not track with his eyes, does not answer. He wears a cap of electrodes, and a steady beep plays through headphones. A recorded voice asks him to try opening and closing his hand. The hand stays still. The beep climbs in pitch.

That is the setup of a Columbia University experiment published in the journal Brain on October 22, 2026. Neurologist Jan Claassen and his team tested 41 people with acute brain injury who showed no sign of consciousness on a standard bedside exam. After brief training, roughly one in three learned to raise and lower the tone using brain activity alone. And every patient in whom a separate test had already found hidden awareness succeeded. Two different methods pointed at the same people, so the beep is not just catching random noise. The body stayed silent while the brain followed instructions.

A beep steered by an attempted movement

At rest, the brain’s motor cortex hums with a steady rhythm, like an engine idling. Try to clench your fist, or even just imagine it, and the hum breaks up. Ordinary electroencephalography (EEG) picks up that change through the scalp, with no surgery and no implanted chip. Software at the bedside compares the signal with a resting baseline and nudges the tone up or down in real time.

Here is the detail that matters: the patient hears the sound too. That closes the loop. The person makes an effort, the beep changes, and they can notice that their effort moved something in the outside world. Clinicians call this kind of hidden awareness cognitive motor dissociation. The tests for it used in intensive care run in one direction only: doctors look into the patient’s brain, and the patient never knows. For someone locked inside a motionless body, the beep in the headphones may be the first feedback since arriving on the ward.

Twenty years of tennis in the head

The idea that a “vegetative” patient might hear and think was first taken seriously in 2006. British neuroscientist Adrian Owen reported in Science on a 23-year-old woman who, five months after a road accident, met the criteria for a vegetative state. Inside a scanner she was asked to imagine playing tennis, then to imagine walking through her home. Her brain lit up in the same regions as those of twelve healthy volunteers. Skeptics pushed back: perhaps her brain was simply reacting to the word “tennis,” with no intention behind it.

Six years later Owen went further. Scott Routley, a Canadian, had shown no response on any bedside test for twelve years after a car crash. In the scanner he was offered a code: tennis meant “no,” walking through the house meant “yes.” Asked whether he was in pain, Routley’s brain answered with tennis . Headlines said a patient had told doctors he was not in pain. Strictly speaking, his brain picked one of two options and the headlines supplied the words. But the answer itself was new.

The method has an obvious flaw. An MRI machine is huge and expensive, and you cannot wheel a critically ill patient out of intensive care to reach it. So Claassen’s group went looking for the same signal with EEG, which fits on a cart next to the bed. In 2019 they showed in NEJM that machine learning detected a brain response to hand-movement commands in 15% of unresponsive patients. A year later, nearly half of them could stay home alone during the day, compared with just one in seven of the rest. In 2024 a large international study found hidden awareness in one in four patients who could not follow a single command.

Why the clock is measured in days

What sets the new study apart is timing. Patients were tested a few days after admission, in the ICU, rather than months later in a rehabilitation center. That matters because the hardest decisions are made in those first days.

In a Canadian study covering six trauma centers, about 70% of deaths after severe traumatic brain injury followed a decision to withdraw life-sustaining therapy. Half of those deaths came within the first three days. Mortality differed fourfold between hospitals. For a family, that means the outcome depended partly on which hospital the patient was taken to and on the conversation doctors started with relatives in those first days. Decisions rest on a prognosis, and the prognosis rests on the bedside exam: does the person open their eyes, squeeze a hand. By definition, that exam cannot see hidden awareness.

Claassen does not claim that every unresponsive patient will recover. In 2019 he himself called for his results to be replicated in a larger group. Still, in 2022 his group reported that hidden awareness predicted recovery better than any established factor they examined. Age, cause of injury and the Glasgow Coma Scale score on admission all came second. A beeping test at the bedside could become one more piece of evidence in the conversation between doctors and a family.

What the beep cannot do yet

Controlling the pitch of a tone is not the same as having a conversation. Claassen himself calls the system only a first step , one that shows that in principle some of these patients might one day be given a voice. The team’s next goal is a system that can convey simple things such as pain or hunger.

The sample is small: 41 people. Two thirds could not control the sound, and for now there is no way to tell which of them lacked awareness and which were tired, asleep or simply did not hear. Other groups still need to repeat the result with other patients.

Some questions no EEG will answer. It is unclear what to ask first of a person who can only reply “higher” or “lower.” It gets harder when the question concerns their own treatment. Whether such an answer counts as consent is something doctors and ethicists have debated for years, and there is still no common rule.

What has changed is the question itself. Twenty years ago scientists argued over whether a person in a “vegetative state” could be aware of anything at all. Now they argue over how to talk with them, and the first word in that conversation turned out to be an ordinary beep.