Small marijuana plants grow in a lab at Niagara College in Ontario, Canada.Carlos Osorio / Reuters

In 1971’s Fear and Loathing in Las Vegas, a mescaline-infused Hunter S. Thompson and his Benzedrine-addicted attorney infiltrate a police conference, where they are amused and appalled by the naïveté of the people charged with enforcing drug laws.

Thompson captures the tone with a fictionalized bulletin, “Know Your Dope Fiend,” that warns officers:

The Dope Fiend fears nothing. He will attack, for no reason, with every weapon at his command—including yours. Beware. Any officer apprehending a suspected marijuana addict should use all necessary force immediately. One stitch in time (on him) will usually save nine on you. Good luck.

At one point, Thompson convinces an oblivious district attorney from Georgia that when Dope Fiends attack, the only recourse is to chop off their heads. “Nobody’s safe. And sure as hell not in the South. They like the warm weather.”

“Jesus God Almighty,” the officer responds. “What the hell’s goin’ on in this country?”

Today, the idea of marijuana causing a murderous rampage still sounds absurd to many users of small amounts of the drug. Drug abuse and addiction have slowly moved from being treated with disdain and punishment, and toward rehabilitation, harm reduction, and prevention. The medical establishment has moved, if glacially, toward embracing potential therapeutic properties of cannabis. There is less fear, less loathing—though still most doctors do not prescribe the drug, and most researchers are prohibited from studying it.

Which is why it was jolting to read echoes of these old tropes this week in The New Yorker, where the journalist Malcolm Gladwell makes the case that marijuana is not as safe as “we think.” The primary source of his criticism is a new book by the novelist and former New York Times reporter Alex Berenson, Tell Your Children: The Truth About Marijuana, Mental Illness, and Violence.

The book is, by Berenson’s account in the prologue, “not balanced.” It is an argument—an accumulation of evidence to support a thesis and convince a reader of something. “Marijuana causes paranoia and psychosis. That fact is now beyond dispute,” Berenson writes. “Paranoia and psychosis cause violence. Overwhelming evidence links psychotic disorders and violence, especially murder.”

Berenson is forthcoming about his bias at the outset: “If you want to read about the way marijuana legalization will provide jobs, or anecdotes from people who believe that smoking cured their celiac disease, or discussions of the relative merits of indica and sativa strains, this book will disappoint you. Maybe I’m too cynical, but I believe most people smoke marijuana for the same reason they drink alcohol or use any other drug: because they like to get high.”

Gladwell distills the argument still further, and the result is paragraphs like this:

Berenson looks, too, at the early results from the state of Washington, which, in 2014, became the first U.S. jurisdiction to legalize recreational marijuana. Between 2013 and 2017, the state’s murder and aggravated-assault rates rose forty per cent—twice the national homicide increase and four times the national aggravated-assault increase. We don’t know that an increase in cannabis use was responsible for that surge in violence. Berenson, though, finds it strange that, at a time when Washington may have exposed its population to higher levels of what is widely assumed to be a calming substance, its citizens began turning on one another with increased aggression.

We don’t know—that’s true. Science can’t prove a negative. But this is innuendo, the public-intellectual equivalent of just sayin’. Innumerable factors go into homicide rates. One of the strongest predictors is income inequality, for example, which was increasing in Washington during that period; the state now has the 10th-biggest gap in the country. If marijuana legalization accounted for increased murder rates, Amsterdam would be a much deadlier place to live. As it is, the homicide rate in the Netherlands is one-fifth as high as it is in the United States.

The article sparked backlash against Gladwell for, among other issues, the implication of causation from correlation and the apparent inconsideration of the political context into which the argument is introduced—one of mass incarceration, where millions of American lives have been upended not by consuming marijuana but by having some in their pocket.

Berenson does acknowledge the role of racism in drug legislation and enforcement. He covered the pharmaceutical industry while at the Times, then left a decade ago to write fantasy novels. But he threw his hat back into the ring with this book, which he says is the result of crawling down a rabbit hole of evidence after a conversation with his wife, a psychiatrist, who told him accurately that it’s well known that marijuana can cause psychosis.

I read the book this week. It argues that we are in the midst of a dangerous cannabis-driven psychosis epidemic: “The epidemic isn’t coming. It’s here.” Berenson describes murders, in detail, committed by people with schizophrenia or acute psychosis. He went to Colorado and combed through homicide records looking for evidence of marijuana involvement. In one case, he visited Bent County Correctional Facility and spoke to a teary inmate named Richard Kirk, who’d murdered his wife and claims he’d done so because he’d just eaten “a nibble” of an edible called Karma Kandy Orange Ginger. After the nibble, Kirk drove home, opened his safe, took out his pistol, and shot her.

Berenson takes Kirk’s account pretty much at face value, concluding that the man “existed at the center of the Venn diagram of three great American maladies—opiate abuse, financial stress, and easy access to firearms. But he’d lived there for years and never been violent, not until he ate a bite of Kandy Karma Orange Ginger.”

Vivid as the anecdotes are, the strongest part of Berenson’s argument is a 2017 review by the National Academy of Medicine, “The Health Effects of Cannabis and Cannabinoids,” which Gladwell also references. These reviews tend to be the “gold standard” for getting everyone in the medical field on the same page about how research should translate into practice. They bring together groups of experts and go through all the evidence they can, and distill it for doctors who can’t possibly keep up. In this case, thousands of studies became a nearly 500-page report.

In the good-news category, Berenson notes, there was no clear association between marijuana and lung cancer. (He does not mention the “conclusive or substantial evidence that cannabis or cannabinoids are effective for the treatment of chronic pain, chemotherapy-induced nausea and vomiting, and multiple sclerosis spasticity symptoms” or the “moderate evidence” that they are effective for improving sleep in people with sleep-apnea syndrome, fibromyalgia, and chronic pain.)

Berenson focuses on the report’s finding that there is “substantial evidence of a statistical association between cannabis use and the development of schizophrenia or other psychoses.” Though, again, it also found “a statistical association between cannabis use and better cognitive performance among individuals with psychotic disorders,” and “moderate evidence of no statistical association between cannabis use and worsening of negative symptoms of schizophrenia.”

The report does not discuss violent behavior specifically, which requires another leap. Berenson argues that if marijuana can cause psychotic breaks from reality, and psychotic people are more inclined to violence, marijuana is a cause of violence.

This is where he and Gladwell lose and upset some experts. Berenson specifically takes issue with the National Alliance on Mental Illness’s famous assertion that people with mental illness are more likely to be the victims than the perpetrators of violence. “Those statements are deeply misleading,” he writes, claiming that the subset of mentally ill people with schizophrenia is much more likely than average to be violent.

Yasmin Hurd, the director of the Addiction Institute at Mount Sinai School of Medicine, works on figuring out who is prone to addiction and why. She is among the small number of U.S. scientists who have authorization to study marijuana. “There is nothing to support that marijuana legalization has increased murder rates,” she told me. “The association between schizophrenia and marijuana use is nothing new. Early use of THC [tetrahydrocannabinol, the main active ingredient of cannabis], especially in high concentrations, is associated with psychosis and schizophrenia. That has been studied a lot. But schizophrenic people are not the ones committing murders. Trying to put a mental-health disorder as the explanation for murder rates—that is incorrect and should not have a platform.”

Hurd also emphasizes, contrary to Berenson, that marijuana remains a promising alternative to minimize opioid use and dependence. “I’ve studied cannabidiol and found that it does have beneficial effects in reducing opioid use,” she says.

The consensus is that, yes, for most people there is such a thing as too much marijuana. In some cases, using too much can have severe consequences that many people could benefit from taking more seriously. There are documented instances in which people have been driven to violence by marijuana, though what Berenson describes mostly seem to be cases where marijuana use is heavy over the course of many years—which could itself be a result rather than a cause of a psychosocial problem.

In these and the other cases in which an occasional user ended up in a psychotic episode or violent fit, it seems mostly that the marijuana does the final unleashing. As with any drug that lowers inhibitions, marijuana is much less likely to bring forth an entirely new person than it is to expose the nature of a person formed over a lifetime of input and environmental exposures and genetic proclivities. Violence is always a multifactorial end point.

Gladwell makes the same basic point: “The experience of most users is relatively benign and predictable; the experience of a few, at the margins, is not. Products or behaviors that have that kind of muddled risk profile are confusing, because it is very difficult for those in the benign middle to appreciate the experiences of those at the statistical tails.” Those outliers, if you will, are where the headlines are made. They are the focus of both writers’ arguments. But they’re also the heart of the problem these arguments face in the first place: The question of exactly how muddled that risk profile actually is makes it impossible to say whether marijuana is “more dangerous than we believe.”

That’s also the straw man in the room: the assumption that most people believe marijuana just makes you laugh and gives you munchies. Berenson goes as far as writing, “No one disputes that occasional use of marijuana by people over 25 is generally safe.” In fact, the U.S. government still treats cannabis as extremely dangerous, among the most dangerous drugs. It is one of only a few Schedule I substances—the most forbidden category, along with heroin and LSD, “for which there is no accepted medical use.” Meanwhile, cocaine, Dilaudid, and methamphetamine are down in Schedule II.

The punishment for possession has long reflected this idea of danger, in which the criminal-justice system has treated marijuana similarly to carrying a bomb. Hundreds of thousands of Americans are still arrested every year for marijuana possession, and the penalty can mean a loss of livelihood, housing, and basic freedom. This is not yet a substance that society takes lightly, despite state-level moves toward decriminalization and legalization.

The feeling of being lashed back and forth by this book and the outcry over the Gladwell piece reminds me that science and medicine are rarely well served by writing in argument form. To do so well is compelling, and reading contrarianism is addictive. But an argument’s job is to undermine, downplay, or ignore contradicting evidence. Gladwell and Berenson offer no stories of anyone who has a positive relationship to the drug. By the end, I found myself questioning even my own experiences, in which I’ve mostly just laughed with friends about nothing. Though there was one night in Colorado when I ate a brownie and went back to my hotel room and became convinced that someone had followed me and was hovering just outside my door. I kept the lights out, and I sat on the floor next to the bed, and I ate an entire jar of almond butter.

I wasn’t deeply scared; some part of me knew it wasn’t real. If I had been another person, though—one who was more given to paranoia, who hadn’t been raised in a safe home by loving parents, and who was in possession of many firearms and had many sworn enemies, would I have opened fire through the door? I suppose it’s possible.

If there is anything on which there is unanimous agreement about marijuana, it is that we need to study it more. This did not happen for decades because it was regarded by regulatory agencies as an irrefutable evil, a dangerous vice that, if you were found possessing it, should ruin your career and rob you of your freedom. One stitch in time (on him) will usually save nine on you. This is the narrative that such articles and books feed.

The pendulum has been swinging away from that, and even Berenson believes decriminalization is the way to go. But whether it is legalized or decriminalized, if it’s going to be used medically, as a drug, it should go through the same process of clinical trials as other drugs: looking for side effects and attempting to discern proper dosages and delivery mechanisms, populations in which it is most likely to be effective and most likely to have drawbacks, and so forth. All this information is severely limited by the fact that studying marijuana has been illegal for most researchers and remains heavily restricted.

The fear-and-loathing narrative conflating marijuana and murder, Hurd worries, does nothing to stem abuse. Nor is it good for promoting further research. “It makes a huge difference,” she says. “Many people who are making the decisions about funding going to NIH and other organizations will now say we should have a moratorium on a drug that increases murder. Why would we want to do that and put people’s lives at risk?”

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.