Can We Enter the Dreams of Others?

Most technologies begin as prosthetics for the waking day. We build tools that extend our hands, our hearing, our memory. Then someone asks the more unsettling question: what about the third of our lives spent elsewhere, behind closed eyelids, in that private cinema where identity can split into many voices? In that inward territory we have long been explorers without instruments, returning with stories but no maps. The moment we attach sensors to the scalp and write software that listens for patterns in sleep, we begin to treat dreaming not as an absence, but as a state with structure, timing, and signals that can be measured.

A U.S. based sleep tech startup called REMspace has become one of the most visible names attached to that ambition. In October 2024 it issued a press release claiming a first: communication between two people “in dreams,” framed as a milestone achieved by inducing lucid dreams and exchanging a simple message. Local television coverage described an at home setup in which a server detected when a participant was lucid, delivered a randomly generated word through earbuds, and then recorded the participant’s response so it could be delivered later to a second sleeper.

The early fascination came bundled with an obvious scientific objection: independent verification. In late 2025, experts interviewed about REMspace’s “dream to dream” claim emphasized that lucid dreaming is real and that external cues can be processed during sleep, but that transmitting information between two separate sleeping people remained unproven at that time, and that the evidence had not been established through independent replication. The critique was not that people cannot be contacted in dreams at all. It was that the strongest interpretation, “information traveled from one brain to another,” demanded a stronger chain of evidence than press release narratives could provide.

In January 2026, the story acquired a more concrete artifact. An article titled “Server Human Communication in Lucid Dreams” appeared online first in the International Journal of Dream Research, authored by a team affiliated with REMspace. The framing is revealing. It does not claim brain to brain telepathy. It describes an engineered system: polysomnography to verify REM sleep and lucidity, a server side algorithm, and a compact electromyographical language, Remmyo, designed so that a dreamer can produce a constrained output that software can detect and interpret.

In that protocol, seven experienced lucid dreamers were trained to reproduce four simple Remmyo words, slept on coordinated schedules, and attempted to induce lucid dreams. Once a dream was verified through pre agreed eye signals and standard polysomnographic criteria, the system delivered one random word via audio; the dreamer attempted to reproduce it; and the server could store that result and later deliver the stored word to another participant once that participant entered a verified lucid dream. The system is therefore best understood as server mediated, and, in its most feasible form, delayed rather than simultaneous. Even the paper argues for delayed communication as a practical bridge because simultaneous lucid dreams are difficult to obtain on demand.

The results are not cinematic, and that is their value. Across two nights the paper reports 18 lucid dreams verified by polysomnography, 12 instances in which a word was delivered, and six communication events where the perception and reproduction of a word met the system’s criteria. It also reports false perceptions in which participants heard sounds or words that did not correspond to what the server sent, along with cases where dreams ended too quickly, words were mispronounced, or signals were misdetected. The paper treats these not as embarrassing anomalies but as the defining obstacle: the dreaming brain is a powerful simulator and can generate “signal like” experiences without any signal at all.

Seen in this light, REMspace’s “dream communication” is not mind reading and not a portal into someone’s private cinematic plot. It is an interface layer, narrow but testable, built on a path that is fundamentally ordinary: sound enters the sleeping brain, the dreamer responds using a trained code, and residual muscle activity is translated by software. The eerie part is not that the system is supernatural. The eerie part is that the sleeping mind can sometimes accept a prompt and sometimes respond intentionally, even while the dream continues to run its own story in parallel.

This sits on a longer arc of research. In 1981, a key study verified lucid dreaming by showing that dreamers could signal awareness from within ongoing REM sleep using pre arranged actions with observable correlates. That work turned lucidity from anecdote into evidence and implied an even larger possibility: if a dreamer can deliberately send a signal without waking, then a dialogue, however constrained, is possible.

In 2021, a multi lab study in Current Biology reported multiple demonstrations of two way communication during polysomnographically verified REM sleep, with lucid dreamers receiving cues and responding with eye movements or facial muscle signals across independent research teams. This work established an anchor point for the field: interactive dreaming is real, but it is fragile, dependent on agreed protocols, and typically limited to narrow response channels.

REMspace’s distinctive bet is that the output modality can be expanded by “unmuting” the dreamer, not by decoding the full richness of dream speech, but by reducing speech to a minimal alphabet of detectable facial muscle activations. Outside coverage has described Remmyo as a sleep language built to be detectable through facial EMG. REMspace’s own descriptions emphasize that Remmyo was developed for “phase states” such as lucid dreams and sleep paralysis and has been under development since 2021. The 2026 journal article explicitly frames this approach as a practical workaround for REM sleep paralysis, since eye movements and other signals are too limited for full language, and simultaneous lucid dreams are hard to time.

A sober summary of the current landscape reads like this. Lucid dreams exist and can be studied physiologically. Limited interaction with dreamers during verified REM sleep has been demonstrated in peer reviewed experiments. REMspace publicly claimed dream to dream communication beginning in 2024. In 2026, an affiliated team published a proof of concept article describing server mediated word delivery, word reproduction by trained signals, and delayed transfer between participants under polysomnographic verification, while also documenting substantial rates of misperception and technical failure.

Long before anyone wrote software that listens for REM sleep, people treated the dream world as a place you could travel to, sometimes deliberately, sometimes under instruction, sometimes with companions. The modern phrase “shared dream space” sounds like a feature announcement. The older versions sound like religion, folklore, and communal survival strategy.

In the ancient Mediterranean, dream incubation was a structured attempt to seek healing or guidance through ritual sleep in sacred settings. Modern scholarship on Greek incubation describes it as “ritual sleep in a sanctuary in order to obtain a dream, mostly for healing,” especially associated with sanctuaries of Asklepios and related sites. In incubation, the dream functioned as a medium through which instruction could arrive, framed in religious terms but enacted through ritual, expectation, and the interpretive power of narrative.

Across many Indigenous North American traditions, documented accounts describe dreams as carrying social obligations rather than merely personal meaning. Reference summaries of Haudenosaunee life, for example, describe dreams as supernatural messages and highlight dream guessing rituals during the Midwinter Festival, in which the community helps interpret and fulfill needs expressed in dreams. Scholarship on Jesuit era encounters similarly emphasizes that dream practices could be central to group identity and welfare as understood within the community, a fact that shaped conflict and misunderstanding in missionary contexts.

Australian Aboriginal traditions offer a different and vital caution: English words can flatten rather than reveal. Encyclopedic accounts emphasize that “The Dreaming,” as a coordinated system of belief and action, links people, land, and ancestral beings, giving individuals identity and obligations that extend from origins into the present and future. When outsiders translate this as “Dreamtime,” the temptation is to imagine only nocturnal dreams. But the concept, as presented in authoritative summaries, is a total framework for life, place, and moral relation, not merely a record of personal sleep experiences.

In Tibetan Buddhist traditions, dream yoga offers another bridge between lucidity and discipline. Contemporary descriptions present it as a technique cultivated to induce and refine lucid dreaming so that contemplative practice can extend into sleep and shape waking life. Here the dream is not randomness. It is training, a controlled encounter with illusion, used to cultivate awareness that can persist even when waking consciousness returns.

Astral projection and out of body travel occupy another shelf in the library of night beliefs. Popular explanations trace astral projection’s conceptual roots to 19th century theosophy, a mystical system that proposed multiple “bodies” or layers of self and imagined travel in a subtler form. Scientific and skeptical accounts generally treat astral projection as a subjective experience rather than literal travel through physical space, but the persistence of the motif matters for history: it shows how readily humans imagine the self as portable, detachable, and able to cross boundaries that the physical body cannot.

The twentieth century attempted to force dream based contact through a more experimental and more controversial doorway. The Maimonides dream telepathy program, as summarized by the Society for Psychical Research’s Psi Encyclopedia, used EEG tracking, repeated awakenings for dream reports, randomized target images “sent” by an agent, and blind judging procedures designed to test correspondences between targets and dream content. The same summary notes longstanding debates over replication, protocol variability, and criticism from skeptics, which have remained part of the scientific controversy around dream based ESP claims.

Even widely repeated “dream sharing” stories can become folklore in their own right, especially when they promise a technology of the soul. The famous “Senoi dream theory” inspired Western dreamwork movements with the idea that morning dream telling and dream control could produce a healthier society. Yet scholarly critique has been blunt. Anthropologist Robert Knox Dentan wrote that anthropologists working with the relevant peoples considered the influential account of dream practices to be rather idealized, attributing inaccuracies to systematic methodological bias. Domhoff’s extended critique similarly argues that central claims about communal dream control and sharing were largely the theorist’s own, rather than accurate ethnographic description.

When you line these traditions up, a pattern emerges that feels almost biological. Humans keep asking the same question in different dialects: can the private theater of sleep become a place of contact? Folklore answers yes, mediated by gods, spirits, and subtle bodies. The laboratory answers maybe, mediated by sensors, protocols, and constrained codes. And even when the answer is “no,” we keep asking, as if the question itself is part of the species’ design.

What a dream network could Mean

Suppose dream communication becomes reliable. Not one word, once or twice, under conditions optimized for expert lucid dreamers. Suppose it becomes robust: a protocol independently replicated, a system usable by non experts, a channel broad enough for meaningful exchange rather than symbolic pebbles. The future would not look like magic. It would look like infrastructure, like early networked mail looked to the first users: sparse messages hinting at a larger network.

Dream research still relies heavily on reports collected after waking, a method vulnerable to memory distortion and narrative smoothing. Interactive dreaming research has already demonstrated the value of real time engagement: dreamers can perceive cues and respond while in verified REM sleep, allowing experimenters to probe cognition inside the dream as it unfolds. Even the narrow channel described in the REMspace affiliated 2026 paper, if replicated, would broaden what can be tested in sleep research by linking verified lucidity to controlled stimulus delivery and measurable response.

We can immediately see applications which would likely be practical and therapeutic, but more modest than the boldest marketing implies. Reviews of lucid dreaming emphasize both possible applications and incomplete knowledge of mechanisms, especially given variability in people’s ability to induce lucid dreams. The plausible near future is not an “infinite dream internet.” It is limited, clinical, specialized: new ways to deliver cues that shape nightmares, new ways to practice skills or coping strategies in controlled dream states, and better experimental methods to understand how sleep interacts with memory and emotion.

When a message arrives during a dream, it arrives as an event inside a story. A voice becomes a character. A random word becomes a symbol. The 2026 REMspace paper’s emphasis on misperceptions is therefore more than a technical footnote. It is a warning that dreams are powerful simulators: they can generate the feeling of an external cue without any cue, and they can misattribute dream generated content to an external sender. Any “shared dream” technology that wants to move beyond novelty would have to solve not only detection and transmission, but also attribution and trust inside a state of consciousness built to blur boundaries.

Even narrow dream communication systems gather intimate biosignals. Brain computer interface research and commercialization raise concerns about privacy, consent, and the possibility that neural signals might someday reveal sensitive information about intentions or mental states. The dream adds a special twist because it has historically been a refuge. Dreams have been sacred partly because they were unshareable, or shareable only through imperfect waking language. If a reliable channel across sleep becomes normal, the sleeping mind becomes a new site of data collection and potential intrusion, even if the initial use cases are benevolent.

For now, the state of play is modest and strangely instructive. A small set of trained lucid dreamers. A handful of words. A measurable fraction of nights when the cue arrives. A measurable fraction when the dreamer’s own world invents a false signal. That is how new continents look from the shoreline. Not a city of gold, but a foggy line of trees, and the knowledge that something is there, waiting for better instruments.