Another “This is Your Brain on Weed” Study Shows... Not Much Actually

A Critical Look at Cannabis Research

Stop me if you've heard this one before: Heavy cannabis use alters brain function.

At least, that's what breathless headlines are claiming about a new paper published in JAMA Open from a team led by Nora Volkow, director of the National Institute of Drug Abuse (NIDA)[1].The study concludes that "cannabis use is associated with short and long-term brain function outcomes, especially during working memory tasks." This tepid statement oversells what the authors actually found, but at least the authors managed to avoid using explicitly causal language. The media, predictably, showed no such restraint. CNN jumped straight to "more frequent cannabis use may damage an important memory skill."

Before I explain why this study—despite its impressive sample size and fancy brain scans—is about as revealing as an empty bowling lane, let me share something delicious from the authors’ own data: compared to both heavy users and non-users, moderate cannabis users showed optimal brain activation patterns during memory tasks. That's right folks, if you want your brain firing on all cylinders, maybe you should be smoking just enough weed. Not too much, not too little... just the right amount. You know, like Goldilocks, if she was into cognitive neuroscience and indica strains.

In the spirit of intellectual honesty, I should probably mention my own brief and spectacularly unsuccessful foray into cannabis activism. Back in 1993, when I was an undergraduate at McGill, I co-founded a cannabis legalization club called Green Leaf. Our crowning achievement? Never managing to hold a single meeting. I never held any placards, marched down streets, or even voted for the Marijuana Party of Canada—yes, that was and still is a real political party, bless their single-issue hearts. So, while I come to this topic with some historical baggage, what really gets me fired up isn't the anti-cannabis message of this research but bad science. And this study? It's a perfect example of what’s wrong with cannabis research, particularly the kind championed by NIDA under Volkow's seemingly endless tenure.

Let's dig into what the authors did. The researchers analyzed brain scans from the Human Connectome Project, examining over 1,000 young adults as they performed tests of working memory, emotional processing, reward sensitivity, language comprehension, motor function, relational reasoning, and social cognition (theory of mind) in an fMRI machine. They sliced their sample into three groups based on the number of discrete lifetime uses—yes, they literally counted each individual time someone lit up, ate an edible, or took a hit: heavy users (more than 1,000 uses, about 9% of the sample), moderate users (10-999 uses, about 18%), and non-users (10 or fewer lifetime uses; 73% of sample).

After all this scanning and categorizing, what did they find? Exactly one finding: heavy users showed slightly lower brain activation during a working memory task than non-users. Not worse memory performance, mind you, just lower brain activation patterns in regions like the anterior insula and prefrontal cortex during a memory task. Also, recall that moderate users had slightly higher brain activation patterns than non-users or heavy users. The other six cognitive tasks? Nothing.

Why am I skeptical? Because this study makes a lot of convenient choices that don't quite add up. Or, to put it less diplomatically: this looks like textbook p-hacking, where researchers massage their data until it tells the story they want to hear. And in this case, that story is that cannabis is dangerous and must be shown to be dangerous, evidence be damned.

First, let's talk about how the authors treated cannabis use. Participants checked boxes indicating their lifetime cannabis use: 0, 1-5, 6-10, 11-100, 101-999, or >1000 times. But instead of working with these categories, they bizarrely smooshed them into three crude groups: heavy users (>1000), moderate users (10-999), and non-users (<10). It's like taking a detailed height survey with six precise categories from “under 5 feet” to “over 6ft4in” and deciding to just lump everyone into 'short,' 'medium,' or 'tall.' That moderate use category is particularly absurd—someone who's smoked weed 11 times is categorized the same as someone who's used it 999 times. What?!

But wait, it gets worse. The authors actually preregistered their analysis plan—meaning they declared their game plan before seeing the data. Great! Except... they didn't stick to it. At all.

Their preregistration promised to analyze the data by comparing users (>100 lifetime uses or symptoms of cannabis dependence) with non-users (<10 uses, no symptoms). Instead, after seeing the data, they cooked up this three-category mess. Oh, they also never made predictions relating lifetime cannabis used and brain-related memory activity. When researchers abandon their preregistered plans this dramatically, you have to wonder what inconvenient truths they're trying to hide.

Hints to the answer, it turns out, are hiding in plain sight in their supplemental materials (kudos for transparency, I guess, even if it required some digital archeology). The authors assessed participants for cannabis use disorder—these are folks who genuinely struggle with their use, who can't cut back despite trying, who find cannabis interfering with their daily lives. In the old days, we might have called them addicted or dependent on cannabis.

Here's where it gets interesting: 43% of people with cannabis use disorder fall into that suspiciously broad moderate-use category, with the rest in heavy use. But here's the kicker—there was zero association between the clinically important variable of cannabis dependence and any outcome. Not emotions, not motor performance, not memory, not reward responsiveness, not logical reasoning, not language comprehension, and not a single brain activation pattern related to any of these. Nothing. Nada. Zilch. This complete absence of findings somehow didn't make it into the main paper. This aggression against good scientific reporting will not stand, man.

But let's play along and pretend we never saw that cannabis dependency data. Yes, there's a small association between heavy cannabis use and lower brain activation during working memory tasks. But let's put this in perspective: the authors examined multiple brain areas across seven cognitive domains, plus behavioral performance metrics for each domain. Do the math with me: let’s assume three brain areas per domain, one performance metric per domain, seven domains; we're looking at 28 individual tests. Their big discovery? Two brain areas related to working memory showed some association with cannabis use.

The authors, in other words, achieved a hit rate of 1 out of 14. That's a batting average of .071 for those keeping score at home. And no, we can't just focus on the memory-related brain areas that showed positive results; that would be like shooting bullets at a barn wall and then drawing bullseyes around where they landed and then claiming to be a sharpshooter. When you're running 28 statistical tests, you need to account for all of them or you're just playing a rigged game. But even if we take these results at face value—which would be like trusting a nihilist with your ransom money—does this sound like the kind of public health crisis that demands breathless headlines? And remember, their precious working memory-related brain activity peaked for moderate cannabis users.

Some crisis.

An honest reading of these results would be downright boring: there are no meaningful differences in brain function or performance between cannabis users and non-users, even among heavy users. That's actually fantastic news that the public deserves to hear! But as Columbia University Professor Carl Hart brilliantly argues in Drug Use for Grown Ups, NIDA under Volkow's leadership seems constitutionally incapable of seeing good or even benign news about recreational drugs.

Volkow has been running NIDA for 22 years now, longer than some of her participants have been alive. At 68, she appears permanently imprinted with the "Just Say No" messaging of the 1980s. Like a Malibu sheriff who's already decided whodunit before examining the crime scene, she sees only what confirms her pre-existing beliefs about the evils of drugs, even when her own data tells a different story.

Look, I want solid research on cannabis effects. If cannabis turns out to have genuine downsides for adult users, I want to know about them. For instance, here's compelling evidence that adolescent cannabis use isn't as harmless as I once thought. Adolescents who use heavily show unrecoverable losses in IQ of up to 8 points. This research changed my mind, and I would not recommend any adolescent use cannabis with even moderate frequency until the age of 19 or 20.

We need good science, not this methodological sleight of hand. We need longitudinal studies that track real changes over time, not these snapshot brain scans. We need research that explores both benefits and harms, controls for crucial variables, and—most importantly—approaches drug research without an institutional thumb on the scale.

Maybe it's time for a fresh approach to cannabis research, one that isn't stuck in Nancy Reagan-era propaganda or dependent on dubious methodological choices to stoke fear. Until then, we'll keep getting what we have now: researchers staring at brain scans like fortune tellers reading tea leaves, desperately searching for evidence of damage that isn't there. The brain isn't a bowling ball, and we can't just X-ray it to see where the finger holes should go.

And maybe, just maybe, it's time to admit that not finding a problem is sometimes the most important finding of all.

---

[1] As Dr. Nora Volkow is the senior (last) author on the paper, a position traditionally indicating project leadership in academic research, my characterization of her leading the study followed standard scientific publishing conventions.

Note: A previous version of this post incorrectly characterized the research team as "the NIDA team." In fact, while Dr. Nora Volkow is the Director of NIDA, most of the other authors are not affiliated with NIDA. I regret the error.