You may have recently stumbled across one of the recent headlines in the media that look something like this:
The scientific paper they are all referring to is linked here (although you can’t read it without a subscription, which brings up a totally different problem in access to science):
- J.M. Gilman et al., Cannabis Use Is Quantitatively Associated with Nucleus Accumbens and Amygdala Abnormalities in Young Adult Recreational Users, Journal of Neuroscience (Neurobiology of Disease section), 34 (2014), 5529–5538.
When I first read these headlines, I was understandably intrigued. Had someone REALLY found good evidence in MRI data that there are differences in the brains of casual marijuana smokers? The idea is not totally far fetched. Alcohol is known to be neurotoxic and it can shrink the size of the brain through dehydration, but this is mostly corrected after you quit drinking. Carbon monoxide in cigarette smoke is also a known neurotoxin, but I’ve never read about any changes in the size of anatomical regions of the brain from smoking tobacco, unless you count brain tumors. However, no one has ever really found evidence to support major anatomical changes in the brain following marijuana use. So could this really be true?
Even before reading the paper, my intuition said the answer was no… brain imaging research is notoriously fraught with spurious findings linked to inappropriate use of statistics.
I gave the paper a casual read, and immediately, I noticed problems, MAJOR problems. First of all, sample size… only 20 people were included in the cross-sectional study. That is low. A cross-sectional study means that they had no within-subject comparisons. In other words, all of the data was collected at one time. A much stronger approach would have been to image subjects before they smoke and then through time as they begin to “casually smoke” marijuana. Of course, this is much more difficult, but with a study design containing so many potential confounds (see below), it’s pretty much required (imho) to say anything definitively.
Second of all, the confounds… the investigators did not control for various other aspects of these people’s lives that may cause changes in brain anatomy. How much did each subject drink? smoke tobacco? do other drugs? etc… These all could be equally correlated to the differences in brain anatomy which they discovered. Or it could something entirely different like genetics?
And lastly, the statistics… I came across this article by computational biologist Lior Pachter, and that was sort of the nail in the coffin. I suggest reading through it because Lior does a great job of highlighting the problems with multiple comparison statistics, causation vs. correlation, and many other mistakes.
He even calls it ”quite possibly the worst paper [he's] read all year.” The Journal of Neuroscience is a rather prestigious journal, so this is all the more upsetting.
I do not study the effects of marijuana use on the brain, so I can’t tell you how it may or may not cause harm. I am absolutely positive it has some effect. But it’s important to remember that pretty much everything you do creates changes in your brain. Reading a book, riding a bike, talking with your friends… these all create lasting memories that are encoded in your neurons. However, after reading this most recent article, there is ABSOLUTELY NO WAY you can assert that marijuana use is harmful, or creates anatomical changes, or anything really…
It’s just another case of sensationalist reporting of poorly conducted science.