Victoria Bee Photography / Getty

Everything my teen self knew about gateway drugs, I learned from a frying pan and a college basketball player.

They were both tentpoles of the “Just say no” heyday in the 1980s and ’90s, beamed into my brain through after-school television or guest speakers at school assemblies. The frying pan played the part of a drug in an extended-metaphor public-service announcement in which my brain was an egg. The college basketball player, Len Bias, served as an oft-cited cautionary tale after his death from an overdose (of what, my assembly speakers were not specific).

Together they warned a generation of the pitfalls of even a single instance of substance use. Once the gateway had been opened, only ruin could follow.

Even though those images have endured in my mind well into my 30s, what constituted a gateway drug was never clear to me as a kid. Decades later, that definition is getting only more muddled. According to a new study from researchers at Columbia University, the teen entry point for psychoactive-substance use has shifted away from cigarettes and alcohol and toward cannabis. That stands in opposition to the long-held belief that kids discover weed after experiences with alcohol and tobacco, and that making the transition to marijuana then sends them down a path toward harder drugs.

These new findings don’t mean parents should be tempted toward reefer madness. According to the study, the shuffle in priority is the result of teens experimenting with fewer substances and doing so later in adolescence, not an increase in teens’ fondness for weed. With adolescent marijuana use stagnant and alcohol and cigarette use delayed and declining among minors, it might be time to leave the popular understanding of the gateway drug in the past.

To measure teen drug use, the Columbia researchers looked at surveys administered to nearly 250,000 high-school seniors across the United States from 1976 to 2016. At the beginning of the study period, a third of high-school seniors were smokers, a rate that declined to less than 5 percent in 2017. Teen alcohol use also dropped; a 2016 report from the National Institute on Drug Abuse found that 37.3 percent of high-school seniors reported having been drunk at least once, down from a high of 53.2 percent in 2001.

The gateway theory of drug use was developed by the Columbia researcher Denise Kandel and was introduced to the public in the 1970s. (Kandel and the researchers in the new study are colleagues at Columbia, but Kandel herself was not involved.) As first introduced, the theory held that the sequence of drug use isn’t random, but instead marked by a near-inviolable order in which trying certain substances makes users more likely to advance to more destructive drug use. At the time, it was thought that alcohol could prime an individual for marijuana use, and marijuana then primed that person to pick up harder drugs, making marijuana use a unique risk marker beyond just its direct impact on a person. The initial research was funded by the National Institute on Drug Abuse, which Kandel has noted was specifically interested in excavating the potential harms of cannabis.

The gateway theory quickly became codified into public policy, setting the stage for decades of public-health and criminalization initiatives and serving as a foundational principle in things such as the D.A.R.E. program and the “Just say no” campaign, in addition to the War on Drugs as a whole. For decades, marijuana’s supposed role as a crucial inflection point in substance-use disorder has been understood as fact and contributed to, among many other things, an epidemic of mass incarceration.

But now teens have inverted the sequence, calling into question the gateway theory’s value. Katherine Keyes, the Columbia researcher who led the new study, told me that the issue is less about the theory itself and more about how the science behind the notion of gateway drugs has been misunderstood over the years. “The traditional gateway sequence is really agnostic about the causal mechanism,” she says. “The one clear causal relationship is that kids who engage in substance abuse early get selected into different social groups where more drugs might be available.” So the traditional sequence could indeed be correct in a specific place and time, but the causative relationship is one of culture.

Keyes notes that Kandel’s more recent research in rodents does suggest the possibility of a biological gateway, but not for cannabis. “They’ve found that mice who are primed to be addicted to nicotine then find the effects of cocaine, for example, much more reinforcing,” she told me. “When you use these substances early, it primes you to enjoy substances more, and you also tend to have the opportunity to use those substances.” She also notes that the science is not yet in place to determine whether cannabis has a similar biological-priming effect.

Contrary to Keyes’s belief in the soundness of the gateway theory’s basis, Michael Vanyukov, a professor of pharmaceutical sciences at the University of Pittsburgh, interprets Keyes’s findings and their relationship to the gateway theory as further proof that the idea has been flawed all along. “Gateway theory only relates to initiation of drug use, not progression to abuse, and even in that regard it’s about as meaningful as the sequence of dishes served in a dinner,” Vanyukov says. “The order of drug use, in general, is simply opportunistic. What is accessible at the least personal cost is what comes first.” For example, in a 2012 study he led about substance abuse in Japan, more than 82 percent of hard-drug users had never used cannabis.

Instead of the gateway, Vanyukov is a proponent of an alternate theory of substance abuse known as the common-liability theory. It holds that substance use and eventual abuse is best predicted through a particular person’s biological mechanisms of reward. How your brain’s dopaminergic system responds to stimuli determines whether you’re likely to progress to abuse of hard drugs, not whether you smoked weed as a teen.

To Ethan Nadelmann, the founder and former executive director of the Drug Policy Alliance, the new data illustrate the gateway’s flawed logic. “With this dramatic drop in cigarette use and drinking, if those substances were actually a gateway to marijuana, you’d think we’d see similar decreases in marijuana use,” he says. “But it’s fairly stable among adolescents. You’re not seeing a significant causal relationship.” Nadelmann is right about the numbers: Adolescent cannabis use isn’t moving much, even in places where it’s been legalized recreationally, while usage rates for other psychoactive drugs in both lower and higher stages of the supposed gateway sequence continue to fluctuate in ways that have no obvious relationship to cannabis use.

A potentially important confounding factor that wasn’t discussed in the study is e-cigarettes. Their use among high schoolers rose 78 percent from 2017 to 2018, to more than 3 million high schoolers, but they wouldn’t yet have been a huge issue in adolescent health when the study’s data cut off in 2016. Keyes says that current use isn’t high enough to account for the entire simultaneous drop in youth cigarette consumption, though, and the Food and Drug Administration has begun to implement policy changes it hopes will stymie the burgeoning popularity of vape products like Juul among adolescents.

And despite the recent inroads the tobacco industry has made among minors via the e-cigarette market, there’s cause for optimism in how American institutions might fare in combatting their popularity, and in reducing teen cannabis use even as legalization spreads. “We’ve achieved a remarkable reduction in the harms caused by alcohol and cigarettes by using public-health principles and goals,” Keyes told me. Hopefully, we’ve also learned from our errors.

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