This work goes beyond the normal logical confusion of "correlation implies causation." It's just a really poorly designed experiment. A better experiment would have compared blood flow in brains of people who have never used marijuana and them had them use marijuana and then again measured blood flow. (This also lacks a control but at least it measures something useful. And I spent all of five seconds coming up with it and I'm not in the field. So.....)
So what's wrong? Maybe people who have abnormally low brain blood flow are prone to using marijuana, perhaps even to _increase_ blood flow, but less than perfectly. And instead of _causing_ psychosis, maybe people who are psychotic are prone to using marijuana. I know people who have told me that they need to use marijuana to feel normal, and maybe normal for them means non-psychotic.