Psychedelics and Entity Contact: The Johns Hopkins Research

In 2019, researchers at Johns Hopkins University published a survey study of over 2,500 people who reported contact with “seemingly autonomous, non-human entities” while using DMT. The results were strange enough that they published them anyway. Sixty-nine percent of respondents described the entities as benevolent. Over 80% described the encounter as among the most meaningful experiences of their lives, alongside peak religious experiences and the birth of children. A significant subset reported that the experience permanently changed their views about the nature of reality, death, and consciousness. Most strikingly, a majority of respondents who would have described themselves as atheist or agnostic before the experience described the entities as “real” afterward, in some sense they found difficult to articulate but could not dismiss.

The study was not designed to validate claims about interdimensional beings. It was designed to characterize the phenomenology: what do people actually report, how intense is it, does it cluster, does it persist. The answer on all four counts was more robust than the researchers expected. The DMT entity encounter is not an occasional oddity. It is one of the most consistent and reproducible phenomena in all of psychedelic pharmacology. Consistent phenomenology across thousands of independent reporters, many with no prior knowledge of DMT or psychedelics, coming from radically different cultural backgrounds, is a data point that demands more than dismissal.

The Experiences Are Real. The Mechanism Is the Question.

Let’s be precise about what is not in dispute. The experiences are real. People actually have them. The psychological and neurological events are genuine events. The entities feel autonomous, feel intelligent, feel like they are communicating something. The emotional impact is real, persistent, and often described as more real than ordinary perception. A 2020 follow-up study at Johns Hopkins found that the majority of participants rated the encounters as more real than everyday waking reality. This “hyperreal” quality is consistent across reporters and is one of the features that distinguishes DMT entity encounters from ordinary hallucination.

What is in dispute is the interpretation: what mechanistic story best explains why the human brain, when flooded with N,N-dimethyltryptamine, consistently generates encounters with autonomous non-human intelligences.

The obvious candidate explanation is that the brain is producing entities from existing cognitive material: the social cognition hardware that we use to model other minds, running hot, decoupled from its usual inputs, generating “minds” the way it generates faces in clouds but at substantially higher resolution and with substantially more emotional force. We are deeply social animals. Our brains run constant simulations of other minds. Under normal conditions these simulations are constrained and calibrated by sensory input and social feedback. Psychedelics, particularly at breakthrough doses, can remove those constraints. The result might be social cognition running unconstrained on itself, generating minds that feel every bit as real as the minds of people we actually encounter, because they are being produced by the same machinery.

This is a reasonable hypothesis. It is not a complete explanation of the data.

The Cross-Cultural Convergence Is the Anomaly That Keeps Not Going Away

If DMT entities were simply brain-generated confabulations, we’d expect them to vary significantly with the cultural priors of the person having the experience. A devout Catholic would meet angels. A Hindu practitioner would meet devas. A secular technologist would meet aliens. A person with no relevant prior exposure would meet nothing coherent.

This is not what the data shows. The entities reported across cultures and centuries are remarkably consistent in their characteristics. They are described as small, non-human, often insectoid or jester-like (the “machine elf” archetype that Terence McKenna popularized is recognizable to people who have never heard of McKenna). They appear to be engaged in activity. They communicate, often wordlessly, through a kind of felt transmission of meaning that reporters consistently describe as more direct and more information-dense than language. They seem aware of the observer and frequently express what reporters describe as something between amusement and care.

Rick Strassman’s original DMT research at the University of New Mexico in the 1990s, the first human psychedelic research to be approved by the FDA after the long pause following the 1970 Controlled Substances Act, documented entity encounters in the majority of participants even before they had any social context for such experiences. Strassman, who began the research as a skeptical neuroscientist, found himself increasingly unable to account for the cross-subject consistency in purely mechanistic terms. He did not end up endorsing the literal existence of interdimensional entities (he’s cautious about this), but he documented the phenomenological convergence and took it seriously enough to spend twenty years thinking about it.

The phenomenological specificity is what makes the cultural convergence argument hard to dismiss with a wave. It is not just that people across cultures report “other beings.” People report beings of a specific rough type: small, active, apparently occupied with their own tasks, frequently communicating through non-verbal transmission rather than speech, often described as existing in a different relationship to time than the observer. The “machine elves” that Terence McKenna described, which became culturally influential mostly after his death, map onto entities described by Amazonian ayahuasca practitioners as plant spirits or doctores, and onto beings reported in medieval mystical traditions under circumstances that probably involved ergot or other psychoactive contamination of grain supplies. The convergence is not superficial. It is typological. If these were purely confabulatory products of existing cultural frameworks, the reports from secular American graduate students in a clinical setting in 1990 should not resemble reports from Peruvian shamans in ritual context or medieval Christian mystics quite as much as they do.

The convergence across cultures goes further back than modern psychedelic research. Ayahuasca, which contains DMT as its primary psychoactive component, has been used ceremonially throughout the Amazon for at least a thousand years, and the entity encounters documented in those traditions map onto modern DMT reports with uncomfortable specificity. The “doctores,” the spirits of the forest that ayahuasca shamans work with, the beings who teach healing songs and plant knowledge, are not generically supernatural. They are phenomenologically specific in ways that map onto what people in clinical settings report after receiving IV DMT for the first time.

The Active Inference Framework Has Something Useful to Say

Here’s where the neuroscience gets interesting rather than just descriptive. The REBUS model (Relaxed Beliefs Under Psychedelics) developed by Robin Carhart-Harris and Karl Friston in 2019 attempts to explain the overall psychedelic effect in terms of the free energy principle and active inference.

The brain, in the active inference framework, is primarily a prediction machine. It maintains a hierarchical model of the world, uses incoming sensory data to update that model, and generates perception as the intersection of top-down prediction and bottom-up data. Crucially, the brain weights sensory data against predictions based on precision: how reliable is this data source, how certain am I of this prediction. The whole system is constantly managing the tension between prior beliefs and new evidence.

Psychedelics, in the REBUS model, flatten this precision landscape. They reduce the weight given to high-level priors (the existing world model) and increase the relative weight of bottom-up sensory information. This produces a state in which the brain’s prediction machinery has more freedom to explore model-space: configurations of the world-model that are normally suppressed because they don’t fit the strong priors. The ego dissolution that characterizes high-dose experiences is what happens when the prior that there is a unified self at the center of experience gets flattened.

What might entity contact be, in this framework? The social cognition module, which normally generates models of specific people constrained by incoming data about those specific people, might, when the precision landscape is flattened, explore model-space freely. In those unconstrained regions of model-space, there may be stable attractors: configurations of the social model that are intrinsically coherent and self-sustaining, not because they correspond to specific real people but because they correspond to something like archetypes in the computational sense, stable patterns that the social model naturally settles into when it isn’t anchored to specific external inputs. The entities feel autonomous because, within the model, they are: they are running their own dynamics, not being driven by the reporter’s intentions.

This is speculative. Carhart-Harris and Friston are careful not to make the full leap to “entity contact = exploration of unconstrained model-space.” But the framework is coherent, it’s grounded in computational neuroscience rather than mystical claim, and it offers a mechanistic story for why the encounters would feel so consistent and so real while remaining entirely within the brain.

The Ontological Question Is Not Closed

The honest scientific position as of 2024 is that the entity encounter is real as a phenomenon, increasingly understood mechanistically through the REBUS framework and active inference, but that the ontological question (are the entities encountering anything, are they in any sense real as distinct from brain-generated) is genuinely open.

Strassman has proposed, more controversially, that DMT might function as a kind of receiver that tunes consciousness to a frequency it normally cannot access, a view that positions the brain as filter rather than generator. David Luke at the University of Greenwich has reviewed the parapsychological literature around DMT and concluded that the evidence for genuine information transfer in these states (perceiving things the subject couldn’t normally know) is mixed but not definitively null. These are not mainstream positions, but they are positions held by researchers publishing in peer-reviewed contexts.

What the active inference angle offers that pure “hallucination” explanations don’t is a mechanistic reason for why the entity encounters feel more real than waking reality. In ordinary perception, the brain generates experience as a weighted combination of prediction and incoming sensory data. The felt quality of “realness” in everyday experience comes partly from the tight coupling between model and sensory input: the world keeps confirming the model, which gives the model a kind of authority. When that coupling is loosened by psychedelics, you might expect experience to feel less real, more dreamlike. Instead, in many DMT reports, the opposite happens. The hyperreality quality suggests the brain is running predictions with very high confidence in a regime where the normal cross-check against sensory data isn’t operating. The entity is experienced as undeniably, pressingly real because the brain is generating it with the full confidence weight it normally reserves for things it has cross-checked extensively. Why this happens in that direction, rather than producing something vague and hallucinatory, is a question the REBUS model handles only partially. The explanation is on the table. It isn’t complete.

The version of this that rates as fringe: DMT entities are literally interdimensional aliens, they have an objective existence independent of any brain, contact with them is equivalent to contact with real beings. No solid evidence for this. That version is Tin Foil at worst, Copper at best.

The version that rates Silver: the experiences are real, consistent, reproducible, cross-culturally convergent in ways that purely confabulatory explanations struggle to fully account for. The mechanism is increasingly understood but not completely. The REBUS model explains a lot. The cross-cultural convergence and hyperreality quality are the remaining anomalies. The ontological status of what’s being encountered is genuinely open. Not evidence of interdimensional aliens. Evidence that the prediction machinery, when freed from its normal constraints, finds something consistent, and we don’t know what that something is.

One further anomaly in the data is worth naming directly. In the Johns Hopkins 2019 survey, participants who reported the entity encounters as among the most meaningful experiences of their lives were also statistically more likely to report positive life changes afterward: decreased death anxiety, increased sense of meaning, improved psychological wellbeing. This mirrors the broader psychedelic therapy literature, where psilocybin and LSD sessions producing “mystical experiences” correlate with lasting positive changes in people with depression, addiction, and end-of-life anxiety. The question here is not just what the entity encounters are but why they work. If they were simply vivid hallucinations, we’d expect the meaning-effect to be temporary, the way a vivid dream’s emotional impact fades. The persistence of the effect, documented at 12-month follow-ups in multiple studies, points toward something happening at a deeper level of psychological organization than “I had a strange experience.”

The therapeutic mechanism is beginning to be understood in terms of the REBUS model: flatlined high-level priors allow the cognitive-emotional system to reorganize in ways that aren’t accessible in normal states. Rigid maladaptive patterns (the depressive’s narrative of worthlessness, the addict’s behavioral loops, the dying person’s terror of annihilation) are high-level priors that are hard to update precisely because they’re entrenched at the top of the predictive hierarchy. Flatten that hierarchy and the patterns can be disrupted. Whether entity contact is necessary for this disruption, incidental to it, or central to it through some mechanism we don’t yet understand is a live research question. Several labs are investigating. The research is ongoing. The Johns Hopkins psychedelic program is now well-funded and well-staffed, which was unthinkable twenty years ago. The active inference angle will produce testable predictions. This is the right kind of weird: reproducible, quantifiable, with a genuine mechanistic framework and genuine remaining anomalies. That’s what Silver looks like.