“If there is one quick truism about psychedelic drugs it is that anyone who tries to write about them without first-hand experience is a fool and a fraud.” — Hunter S. Thompson

People have been doing psychedelics for at least three thousand years, and yet our understanding of how these substances interact with our nervous systems is still in its infancy. Until recently, illegality hampered clinical research. Even now, as some psychedelics pass the threshold to decriminalization, large-scale trials remain few and far between—a problem for research into such a variable experience. That’s not to mention the inherent contradiction of conducting clinical trials on any social (or at least highly contextual) drug. People usually trip in scenic environments with their friends. That’s hard to replicate under scientific observation.

The temperamental nature of psychedelics is essential to their appeal. Drug nerds—the true bearers of this occult knowledge—love to share their unique experiences with each other. To the researcher, this same variability represents a bottleneck, and a costly one. Venture capitalists are pouring millions of dollars into new psychedelic-based antidepressants that move beyond the classic SSRI model, but without a precise index of the neural processes at work, pharmaceutical patents are unlikely to be approved. 

Last year, however, a group of interdisciplinary researchers announced a simple but powerful work-around: using AI and brain imaging, they found a way to draw directly from the experiences of some of the internet’s chattiest psychonauts—potentially paving the way for a new class of hybrid drugs.

*

Samuel Friedman, a machine learning researcher at MIT, came to the psychedelic question via an enigmatic region of the central nervous system called the “default mode network.” Formerly known as the “task-negative network,” the DMN is thought to govern some of our most introspective behaviors. Neuroimaging shows increased blood flow towards this area during daydreaming, reminiscing, “wakeful rest,” and the quiet contemplation of our lives and relationships. An overactive DMN is also associated with a range of mental health issues—particularly, the obsessive pondering of one’s own unhappiness, known as “rumination,” a symptom of major depression. Psychedelic treatments have shown great promise at interrupting an overactive DMN, apparently correlating with subsequent improvements to a patient’s mental health. 

Hoping to illuminate these connections, Friedman organized a talk in 2019 with Danilo Bzdok of McGill University, who was working on a comprehensive mapping of the DMN, and his childhood friend Dr. Galen Ballentine, a SUNY psychiatrist pursuing research into the therapeutic potential of hallucinogenic drugs. Over dinner, afterwards, the three men lamented the miniscule sample sizes of most studies into psychedelic-assisted therapy. Drug trials of this sort have particularly stringent exclusion criteria, resulting in volunteer disqualification ranging from 90% to 96.3%. Notably, people already taking antidepressants aren’t eligible. Bzdok wondered if there wasn’t some preexisting data set out there that could be analyzed. That’s when Ballentine remembered the Erowid Center—a labyrinthine online drug forum, where thousands of people have been sharing their experiences with psychoactive substances for over two decades.

This opening yielded an experiment combining all their specializations. In the following months, the team downloaded nearly seven thousand testimonials for 27 different psychedelics from Erowid’s archives. They used AI to boil down commonly recounted experiences with a particular drug into their most essential, expressive keywords. Bzdok’s brain imaging maps subsequently connected these keywords to a particular psychedelic as well as a specific region of the nervous system.

The fruit of this labor is “Trips and Neurotransmitters: Discovering Principled Patterns Across 6850 Hallucinogenic Experiences,” the largest ever study of psychedelic influence on the brain. Published in March 2022, it’s been hailed as a “rosetta stone” that translates symptoms to molecules. Along eight distinct axes of psychedelic experience, it offers an expansive yet precise map of neurotransmitter-receptor combinations that need to be stimulated to induce a specific state of conscious experience. It also clarifies how the DMN ordinarily works to maintain a stable sense of self capable of withstanding the constant stimulation of being in the world—and how psychedelics might shake things up.

*

Obtaining this volume of actionable results through a classic clinical trial would have required a nearly prohibitive amount of time and money. “One of the mysteries about these [machine learning] models is that they need so much more data than humans to learn things,” Friedman tells me over Zoom. “But if you have the data, then they are able to learn things that humans can’t.”

On a shared screen, he talks me through a presentation of their findings. The slide in front of me has two columns of words, one in red and one in blue, flanked by scans of the human brain, with clouds of color indicating neurotransmitter activation. The column in red has words like “love,” “dancing,” “magic,” “happy,” “crying,” and “time,” while the blue column contains “distortion,” “deepening,” “nausea,” and “voices.” Even reduced by the AI model, the remaining kernels of language retain the vividness typical of Erowid entries.

Erowid was founded in the midst of Bill Clinton’s presidency, when DARE officers were still giving lectures nationwide on the “Three R's”: Recognize, Resist, Report. In opposition to this abstinence-focused approach, the founders of Erowid—a Bay Area couple known only by the sobriquets “Earth Erowid” and “Fire Erowid”—created an open-access educational resource. Their stated goal was a "world where people treat psychoactives with respect and awareness; where people work together to collect and share knowledge in ways that strengthen their understanding of themselves, and provide insight into the complex choices faced by individuals and societies alike.” It remains ad-free and looks much the same as most websites from the earlier and more hopeful days of the internet. 

Each drug on Erowid has its own “experience vault,” full of first-hand testimonials that have been vetted for authenticity and accuracy. Beyond that they’re pretty free-form. “At about 8 PM, I chose to vomit (as I was fairly nauseous and thought it might expel some of the toxins). During my toilet-filling episode, I thought I was going to die and I asked God to help me,” goes one testimonial on psilocybin. “With each spew, I felt better and I got the message that Jesus had saved me.” Another recounts the pleasure of “rubbing my hands and body against the bark of some large evergreens.”

While some entries can be bleak—particularly for harder drugs like meth or heroin—the vast majority are written in a companionable, curious voice that will be familiar to anyone with an older sibling or cousin who likes to test the limits of consciousness from their own backyard. The testimonials include highly specific descriptions not just of the chosen amount and imbibing method, but also the subtle shadings of each experience; sometimes with humor, but always with rigor, vibrancy, and clarity, often down to the passing minutes. These are good faith arbiters, truly interested in exploring the variance of human perception and making sure others can do so safely. There are none of Hunter S. Thompson’s “fools or frauds” here, though any one writer tends to give the distinct impression of being a bit of a weirdo.

They would probably be intensely curious about the new substances their testimonials might be used to formulate; how their words, written in an online forum as much as 20 years ago, were fed into an artificial intelligence program and then translated into molecules.

*

Researchers have been quick to pick up on the potential of “Trips and Neurotransmitters” for clinical application.“The problem is not the drug—drugs are just molecular tools—but rather, not pairing the right tool with the right patient,” wrote two psychiatrists in Science on the implications of the study. “The task facing the companies and researchers developing psychedelic compounds is to identify which specific compounds alleviate which specific types of human suffering.” Bzodoc echoed this sentiment in the press release for “Trips and Neurotransmitters,” along with the hope that their work would help generate a new class of drugs: “The U.S. is going through a profound mental health crisis that has been exacerbated by the Covid-19 pandemic, yet there have been no truly new psychiatric drug treatments since Prozac [...] These insights may lead to new ways to combine existing or yet to be discovered compounds to produce desired treatment effects for a range of psychiatric conditions.”

If an entire nation is experiencing a mental health crisis, it's worth reconsidering whether this is something that can be individually remedied by the right combination of molecules. Depression diagnoses increased by more than 300% after the advent of SSRIs. A new psychedelic-style Prozac would be a gold mine for any company lucky enough to patent it and then prescribe it to a population in distress. A study from 2018, however, attributed lack of access to mental health services as a major factor in the mental health crisis in America—not imperfect medication—citing high costs, insufficient insurance, and long waits.

The particular pressures of American work life also play a measurable role in the development of mental illness in the U.S. Contrary to popular belief, for example, the rate of suicide in the United States actually dropped during the Covid-19 pandemic, perhaps thanks to the federal relief funds that allowed many people to stay home. According to suicidologist Craig Bryan, the lockdown reduced stressors for a large portion of Americans: while experts worried about the impact of unemployment, Bryan notes, “People who hated their jobs suddenly had a reason to stop working. Even if it made for challenging financial situations, that may have been preferable to many [...] Unable to go to work, people started spending more time at home, typically with other family members, friends, and their support networks.”

Alleviating human suffering is complicated, but the world of for-profit pharmaceutical research is adept at streamlining it into a fruitful cycle of consumption. In recent years, for-profit companies like atai Life Sciences, MindMed, and Compass Pathways have flooded universities with millions of dollars in support of psychedelic research. MindMed, for instance, provided enough funding to NYU to support four research positions over five years. Their efforts have been buoyed by loosening regulations, including the FDA’s approval of ketamine-derived nasal spray as a treatment for major depression. Canada set a precedent last August when four terminally ill patients were granted access to psilocybin as part of end-of-life care. In the United States, meanwhile, voters have passed measures both legalizing shrooms for controlled therapeutic use in Oregon and decriminalizing plant-based psychedelic substances in D.C. Legalization, however, does not seem to be a major concern for these companies, beyond how it affects their ability to conduct research.

One of the giants of this emergent field, atai founder Christian Angermayer, said recently that he supports decriminalization but thinks legalizing psychedelics could create a backlash for the industry. "Biotech is all about having monopoly,” he noted, in an interview with Insider. “That's the whole of biotech. You do something novel, you finance it, and you own it."

*

Getting on the right medication can be life changing, but for many it’s an arduous process. Psychiatrists—or, for that matter, neurologists—can’t reliably predict who will improve on antidepressants and who will not. Until recently, clinical professionals operated on the assumption that depressed people had a serotonin deficiency and that SSRIs worked by boosting it; however, a paper published last July presented evidence that people with depression have the same amount of serotonin as people who are not depressed. The new consensus is that antidepressants can help the brain form healthy connections between cells that have previously withered, possibly due to stress, and that serotonin might simply be acting as a growth factor.

Part of the excitement around psychedelics is that they demonstrate the ability to operate in a similar fashion, improving mental health by rapidly inducing neural plasticity. A recent study at Yale University showed that mice given one dose of psilocybin grow longer and denser dendrites—the tapering branches that extend from the body of a neuron—yielding a full 10% increase in neural connections.

This kind of increased neural plasticity is associated with a concurrent loosening of the DMN’s top-down control over what constitutes the inner self, leading to a leaky filtering of sensations, thoughts, and feelings—essentially, a disintegration of normal self-awareness. Regions of the brain that aren’t usually in conversation end up talking to each other. Psychedelics seem to have the special aptitude to slacken certain connections while strengthening others. Crucially, the more powerful this dissolution of ordinary processing and ego-maintenance, the better the clinical result in the long term.

For people with treatment-resistant depression—that is, patients who do not improve with medication and psychotherapy—the acuteness of this oft-described “mystical” experience turns out to be particularly consequential. Compass Pathways, a London-based startup, found that three weeks after psilocybin-assisted therapy, 29% of people with treatment-resistant depression who received the highest doses were in remission. In trial scenarios, such high-dose applications lead to “spiritual,” “mystical,” or “religious” experiences. A 2018 study on hallucinogens notes that, “regardless of the terms chosen to define them, evidence suggests that profound psychological experiences can be predictive of subsequent psychological health,” whether induced by psychedelics or not.

Despite its obvious import, research scientists remain wary of the mystical experience. For one, it’s challenging to quantify. The same 2018 paper notes that “mystical” suggests “an association with the supernatural that may be obstructive or even antithetical to scientific method and progress,” and that they are not keen to “endorse any associations between it and supernatural or metaphysical ideas” by describing the phenomenon. Researchers have done their best to get around this by developing various questionnaires for people undertaking a psychedelics trial. The Mystical Experience Questionnaire (MEC), for instance, features questions designed to quantifiably measure a subject’s “transcendence of time and space”; “sense of awesomeness, reverence, and wonder”; and feelings of “ineffability and paradoxicality.”

One of the most important facets of “Trips and Neurotransmitters” is that it highlights this key aspect of psychedelic-based therapy on a molecular level. “Earlier research on ego dissolution has placed a focus on binding to the 5-HT2A receptor,” a single kind of serotonin receptor, but “together, a wider group of receptors may individually or collectively underpin the process of self-disintegration that is thought to be critical for the success of hallucinogen-assisted psychotherapy.”

There’s another major obstacle facing hopeful pharmacologists. Unless the hallucinatory effects are eliminated, prescription medication derived from psilocybin, MDMA, and LSD can’t be used by patients without supervision. This would make it a lot harder to market them as a replacement for the classic SSRI. For that reason, some researchers are attempting to develop a psychedelic analog to CBD, a derivative of cannabis that lacks its psychoactive properties—and so evades stringent legal oversight.

The evidence presented in “Trips and Neurotransmitters” suggests that this will be a challenge—the mystical experience and the hallucinatory effects are deeply entangled. When I ask Friedman how his team considered the problems facing drug developers, he emphasizes the emergent nature of a data-driven approach: “We don’t really get to say, ‘Okay I want to segregate the visual hallucinations from the mystical experience, and I want it to show me how to break down the experience upon those axes,’” Friedman says. “What did come out [of the study] is there seems to be this very hallucinatory thematic side, with words like ‘nausea’ and ‘sleep’ and ‘stomach,’ and also ‘hallucinations,’ ‘euphoria,’ ‘visuals,’ ‘headache.’ Then on the other side was this very mystical, intense emphasis on ‘reality,’ ‘breakthrough,’ and ‘consciousness.’ You can imagine pushing along that axis like a knob that you could turn.” He smiles. “That would be cool.”

How close are we then to actually seeing new drugs show up on prescriptions? Friedman considers for a second, then replies that we might need to start thinking differently about how we conceive of medication. “​​It’s such a different model than antidepressants, where you take it chronically,” he explains. “These are one or two interventions with really long-lasting effects. And it seems that the setting matters so much, and the person administering it plays such a critical role…”

Friedman is the first to say that we’re still in the “very early days” of viable psychedelic medication, even if it becomes possible to parse out the visual hallucinations molecularly. The mystical-leaning axis of molecule-to-experience in “Trips and Neurotransmitters'' is as characterized by intense fear and dysphoria—“terror,” “horrible,” “death,” “fear,” “panic,” “pressure,” “shit,” “insane,” “surrender”—as it is by a parallel experience of transcendence: “forever,” “lightbulb,” “bliss,” “void,” and “life.”

For now, there’s no separating the agony from the ecstasy. ♦