Interactive Dreaming: How Scientists Learned to Talk with People Inside Their Dreams

For most of history a dream was a sealed room — you could only describe it after you woke up. Then scientists found a way to get a deliberate signal out of a dream, and, decades later, to hold a real-time two-way conversation with a sleeping, dreaming person. Here is how interactive dreaming works, and what it has and hasn't shown.

Last scientific review · 2026-07-02

For as long as people have wondered about dreams, they have run into the same wall: a dream is private. You can describe it once you wake up, but by then it is a memory, reshaped and half-lost. There seemed to be no way to reach a dream while it was happening — to ask a question and get an answer from someone still asleep. That wall has now been breached. Beginning with a single deliberate flick of the eyes in a 1981 laboratory, and culminating in four independent labs holding real-time conversations with sleeping volunteers in 2021, scientists have shown that a dreaming brain can perceive the outside world and answer back. This is the science of interactive dreaming: not a fantasy of entering someone's dreams, but the careful, measurable business of exchanging signals with a mind that is asleep and dreaming.

What interactive dreaming means

Interactive dreaming: The experimental practice of exchanging information with a person while they are asleep and dreaming — sending a signal in, such as a spoken question, a flash of light or a touch, and receiving a deliberate signal back — instead of relying only on the dream report they give after waking. It depends on lucid dreaming: a dream in which the sleeper knows, as it unfolds, that they are dreaming, and can therefore act on a plan agreed in advance.

The key that unlocks all of this is lucidity. An ordinary dreamer is swept along by the dream and remembers little; a lucid dreamer knows they are dreaming and can hold on to an intention formed while awake — for instance, 'when I realise I'm dreaming, I'll signal.' That single capacity turns the dreamer from a passive subject into a collaborator who can send and receive messages across the boundary of sleep. Everything else in this field is built on it.

The breakthrough: a signal out of a dream (1981)

The first crack in the wall came from a simple physiological fact. During REM sleep — the stage of vivid dreaming — most of the body's muscles are paralysed, a safeguard that stops us acting out our dreams. But the muscles that move the eyes are spared. In 1981, Stephen LaBerge and colleagues realised this meant a lucid dreamer could send a message that the sleeping body was still able to make: a prearranged pattern of eye movements. A volunteer agreed, before sleep, that the moment they became lucid they would look sharply left, then right, then left, then right. Later, in the middle of laboratory-verified REM sleep, exactly that zig-zag appeared on the eye-movement recording. In a landmark demonstration, a person had knowingly, deliberately reached out of a dream in real time — an objective, time-stamped message from inside a world everyone had assumed was sealed.

That eye-signal method became the field's standard tool. Over the following decades researchers used it to timestamp dream events, to measure how long actions feel inside a dream, and to confirm again and again that lucidity is a genuine state a sleeper can enter and report from. But for forty years the traffic ran one way: the dreamer could signal out. Whether information could also travel in — whether a sleeper could receive a fresh question and answer it — was still open.

From one-way signal to two-way conversation (2021)

In 2021 that last barrier fell. Four laboratories — in the United States, France, Germany and the Netherlands — independently tried the same audacious thing: to ask sleeping, dreaming people questions and get answers back in real time. Working with trained lucid dreamers, experimenters posed simple problems from outside the dream — spoken aloud, or delivered as flashes of light or taps on the skin. The sleepers answered using prearranged eye movements or small facial-muscle twitches. They solved arithmetic ('eight minus six?' — two left-right eye movements), answered yes/no questions, and even distinguished the number of flashes they were shown. Crucially, polysomnography confirmed the volunteers were genuinely asleep in REM the whole time. The researchers called the phenomenon interactive dreaming: a real, if fragile, two-way channel into a dreaming mind, obtained not once but in four separate labs.

Preparation. A participant who can lucid-dream reliably learns a signalling code in advance — for example, moving the eyes left-right once for 'yes' and twice for 'no', or one eye movement per unit when counting.
Reaching the dream. The person falls asleep in the lab wired for polysomnography, which tracks brain waves, eye movements and muscle tone, so researchers can confirm the exact moment they enter REM sleep and, via an agreed lucidity signal, that they have become lucid.
The question goes in. From outside, the experimenter poses a simple problem — a spoken sentence, a pattern of lights, or taps — timed to the REM period.
The answer comes out. The dreamer, aware inside the dream, performs the prearranged eye or facial-muscle signal. The response shows up on the recording as a deliberate, decodable pattern.
Verification. Only answers made while the recording confirms unbroken REM sleep count — guarding against the obvious objection that the person had briefly woken up.

Why it works: the hybrid brain

How can someone reason and answer a question while fast asleep? The answer is that a lucid dream is not pure sleep. EEG studies of lucid REM sleep report a state that sits between waking and ordinary dreaming — most of the brain looks as it does in a normal dream, but activity over the front of the head has been found to speed up toward waking-like, fast-frequency (gamma) rhythms. Those frontal regions support the reflective, self-aware, goal-holding abilities that fade in an ordinary dream and return the instant you become lucid. In effect, part of the 'executive' brain appears to switch back on while the rest of the mind stays in the dream. That hybrid configuration is thought to be what allows a sleeper to notice a question, understand it, and choose to answer.

Dreamed actions leave a real fingerprint

There is a deeper reason to take these signals seriously: a dreamed action is not just imagined, it is physically enacted by the brain. In a striking experiment, a lucid dreamer inside a brain scanner agreed to clench their dream hand and signal when they did so. When the dreamed clench happened, the sensorimotor region that controls the real hand became active — the dreamed movement leaving the same neural footprint a waking movement would. The study was tiny, limited to the rare few who can lucid-dream on command inside a scanner, and it has not yet been widely replicated, so it should be read as a promising proof of concept rather than a settled result. But it points at something important: when a dreamer signals an action, the brain really is performing that action, in real time.

What we can and cannot say

What we know

A deliberate message can leave a dream. Prearranged eye-movement signals from lucid REM sleep are objective, repeatable, and time-stamped on the recording.
Information can also go in. Sleeping, dreaming people have perceived outside questions and answered simple ones — yes/no, counting, basic arithmetic — in real time.
The effect is not a fluke of one lab. Two-way communication was obtained independently across four laboratories in different countries.
Lucidity is a measurable brain state. The waking-like frontal activity of lucid REM sleep gives a physical basis for the reflective awareness these exchanges require.

What we don't know

How much a dreaming brain can really think. Simple yes/no answers and small sums are a long way from flexible reasoning; the true cognitive ceiling of interactive dreaming is unknown.
Whether it generalises. Nearly all of this comes from a small pool of highly trained or naturally frequent lucid dreamers, not ordinary sleepers.
How reliable it can become. Even trained participants respond on only a fraction of attempts, and answers can be missed, ambiguous, or forgotten on waking.
Whether lucidity can be switched on from outside. Claims that brain stimulation can induce lucidity are contested and not established.

The honest limits

It is easy to read 'scientists held a conversation with a dreamer' and picture a smooth, reliable dialogue. The reality is far more fragile. Everything depends on the participant first becoming lucid and then staying lucid long enough to notice and answer — and reliably inducing lucid dreams is still something science cannot do on demand. A systematic review of induction methods found that no technique dependably produces lucidity, and that the effects that exist are modest and variable. As a result, interactive-dreaming studies lean on small numbers of specially trained dreamers, and even they succeed on only a fraction of trials. The breakthrough is real, but it is a narrow, hard-won channel, not an open line.

Direction	One way: dreamer signals out	Two way: question in, answer out
What is exchanged	A prearranged 'I am lucid' eye-movement mark	Yes/no answers, counting, simple arithmetic
Landmark	LaBerge and colleagues, 1981	Four independent labs, 2021
What it proves	A sleeper can deliberately act on a plan in a dream	A sleeper can perceive and respond to the outside world in real time

Two milestones: one-way signalling versus two-way dialogue.

A contested frontier

Why it matters, and where it might go

Even in this early, hard-won form, interactive dreaming changes something fundamental: it reframes the dreaming mind as reachable rather than sealed off. Researchers have begun to imagine using the channel to study dreams as they happen instead of reconstructing them afterwards, to explore whether guidance from outside could ease recurring nightmares, or to probe learning and problem-solving during sleep. Those possibilities are genuinely exciting — and, for now, genuinely speculative; none should be read as a proven application. What is solid is the foundation: a dream is no longer entirely private, and a sleeping brain can, under the right conditions, answer when the waking world calls.

[ Caution · Safety notice ]

This article is a general educational account of laboratory research — how scientists communicate with dreamers and what the findings mean. It is not a how-to guide and not medical advice, and nothing here is a technique for you to attempt. The lucid dreaming these studies rely on is usually reached with methods that interrupt or shift sleep, such as waking during the night, which can reduce sleep quality and daytime alertness; do not disrupt your sleep on the basis of this article, especially if you drive, operate machinery, care for others, study, or do any safety-sensitive task on limited sleep. If you already practise lucid dreaming and it worsens insomnia, causes daytime sleepiness, or brings distressing experiences, ease off and, if it persists, speak to a clinician.

Where to go next

Interactive dreaming is one chapter in a larger story about turning the private world of dreams into something science can observe. To see how researchers first proved that lucid dreaming is a real, verifiable state, read how lucid dreaming was scientifically verified. To understand the experience these experiments depend on, start with what lucid dreams are. And to see how people learn to become lucid in the first place — the very bottleneck this research keeps running into — the MILD technique is a good next step. Each builds on the same surprising discovery: a dream can, sometimes, answer back.

Can scientists really talk to you while you are dreaming?

In a limited way, yes. With trained lucid dreamers, several labs have posed simple questions from outside — spoken, or as lights or taps — and received correct real-time answers given through prearranged eye or facial-muscle signals, all while recordings confirmed the person was asleep in REM. It is real two-way communication, but narrow and not fully reliable.

How does a dreaming person answer a question?

They cannot speak or move normally, because most muscles are paralysed during REM sleep. Instead they use the muscles that still work — chiefly the eyes — to make a prearranged pattern, such as looking left-right twice for a particular answer, or one eye movement per number when counting. That pattern is picked up on the eye-movement recording and decoded.

Does interactive dreaming work for everyone?

No. It depends on being able to lucid-dream on demand, which most people cannot do reliably, and inducing lucid dreams is itself still unreliable. The research so far relies on small groups of specially trained or naturally frequent lucid dreamers, and even they respond on only a fraction of attempts.

Is this the same as reading or controlling someone's dreams?

No. Scientists are not reading minds, implanting thoughts, or steering the dream's story. They are exchanging simple, agreed-upon signals with a willing, awake-enough part of the dreamer's mind — the dreamer chooses to answer. It is a conversation by prearranged code, not mind-reading or dream control.