When you're out to prove a point that the data doesn't support, what can you do? Change definitions, of course.

According to New Scientist, new experiments show that caged monkeys will repeatedly inject themselves with THC, as many as thirty times an hour. Well, duh. Caged animals (including humans) will do most anything to releive the stress of imprisonment. That has nothing to do with addiction.

It wasn't all that long ago that addition was a clearly defined syndrome: if a drug created neurological changes that lead to a tolerance/withdrawl cycle, and a user failed several attempts to quit and continued use in the face of negative health consequences, then the user was addicted. Note the dual physiological and behavioral aspects of this definition - addition lies both withing the pharmacological features of the drug, and in a user's behavior.

But for prohibtionists, self-righteous moralists, and the recovery industry, this definition is much too narrow. Not only does it not include drugs like cannabis, but non-drug problem behaviors like gambling and sex are excluded. So we started to hear of "psychological addiction", a deliberately fuzzy concept which essentially boileds to "you're doing something that we don't like more often than we think you should."

Now the distinction between physiological and psychological addiction is no longer made, and dependance is judged based on the behavior of animals in unnatural and stressful situations. It's a poor substitute for science.