"It seems like the brain’s ability to construct reality, or imagination, or our associations, our beliefs, our sense of self—these faculties had more dominance over sensory areas," Stoliker says, noting that while this interpretation remains a hypothesis, it offers a compelling framework for understanding the subjective state. "This could help explain why people have meaningful, complex imagery experiences, why they have mystical experiences with imagery that is personally relevant to them."
Simultaneously, the researchers recorded a drop in modularity across all four context settings (rest, meditation, music, and the cloud video). Modularity measures how cleanly groups of neurons remain divided into specialized functional units or "teams." Under psilocybin, internal communication within individual brain networks weakened, while functional connections between separate networks grew strikingly stronger.
This network integration led to a remarkable breakdown in how the brain distinguishes between internal and external environments. In a sober brain, neural activity looks vastly different when a person rests with their eyes closed versus when they watch a movie with their eyes open. Under the influence of psilocybin, however, this distinct boundary nearly vanished.
Within the brain’s visual network, the connectivity gap between the eyes-open state (watching clouds) and the eyes-closed state (resting) collapsed by an extraordinary 85 percent. Independent data from the EEG sessions confirmed this effect: alpha-band brainwave activity—which normally acts as a sensory gate to suppress visual processing when the eyes are shut—was reduced by nearly half.
"When somebody takes a psychedelic and they’re able to close their eyes and see complex imagery, there seems to be less boundary between the internal and external world than we ordinarily experience," Stoliker explains.
To standard analytic tools, these widespread shifts looked like classical brain chaos: modular systems breaking down, sensory boundaries dissolving, and brainwaves flattening out into unpatterned noise.
Hidden Order
Despite these apparent signatures of breakdown, the research team suspected that standard analytical methods were obscuring a deeper structure. Standard neuroimaging studies rely on two major mathematical averaging steps that inherently smooth over fleeting patterns:
- Averaging over time: An eight-minute fMRI scan generates hundreds of sequential 3D brain images. To calculate how two brain regions communicate, researchers typically average all these images into a single static correlation number representing the entire eight-minute window. This risks missing temporary, highly organized neural patterns that form for brief moments before dispersing.
- Averaging over participants: After generating a single network score for each participant, conventional methods pool all participants into a unified group average, treating individual variations in experience as statistical noise to be canceled out.
Because PsiConnect focused specifically on understanding individual, subjective experiences, Stoliker and his team eschewed these group-averaging shortcuts. Instead, they employed a custom artificial intelligence platform known as CEBRA—a machine-learning algorithm specifically engineered to reduce complex high-dimensional datasets down to their fundamental structure while preserving temporal chronological order.
The team funneled moment-by-moment neural activity from 332 distinct brain regions into CEBRA. For every participant, the AI compressed this massive data stream into a continuous trajectory—a mathematical path moving through a three-dimensional mapping space, where every single point represented a specific moment in time during the scan.
When the researchers analyzed these 3D trajectories, the apparent chaos transformed into clear, highly predictable geometry.
"When they’re not under psilocybin, the brain activity is less differentiated by context," Stoliker says. "But under psilocybin, we see that the activity becomes organized across time and more clearly differentiated by context."
Rather than floating randomly through 3D space, the psilocybin-induced brain trajectories organized themselves into four distinct, cleanly separated geometric clusters corresponding directly to the four experimental environments: rest, guided meditation, music, and the cloud movie. The organization was so distinct that a machine-learning classifier could analyze a single isolated snippet of a participant’s brain activity and accurately identify which of the four contexts the subject was experiencing at that exact second.
Crucially, the classifier’s accuracy scaled directly with the subjective intensity of the drug: the more profound a participant rated their psychedelic experience, the more clearly differentiated their brain trajectories were within the machine-learning model.
One with the World
To discover which specific brain structures were driving this unexpected context-specific organization, the researchers conducted virtual brain ablation tests. They took the machine-learning model and systematically replaced the psilocybin-state data of specific brain networks with their corresponding sober baseline data, observing how severely the classifier’s predictive accuracy degraded as a result.
