No Pill Can Cure Addiction | Psychology Today | Hacker News Reader

No Pill Can Cure Addiction | Psychology Today

psychologytoday.com

5 pointsmuon2 comments

I upvoted this, but I think the headline is slightly misleading - I think they are saying that no one drug can cure all different types of addiction.

This seems, in a sense, almost self-evident; it is a bit like saying that it is impossible for one patch file to fix all possible bugs.

At the time of writing, one of the Wikipedia definitions of addiction was "a continued involvement with a substance or activity despite the negative consequences associated with it". Just like it takes some reasoning and domain specific knowledge to decide if something is a feature or a bug, it is not trivial to decide if a consequence is negative or positive, and requires a lot more logic than can be achieved with drugs binding to receptors in the brain.

Pleasure evolved because it makes animals do things which cause the genes for pleasure to be passed on (for example, eating food, having sex, or even more indirect things like solving a problem or acquiring knowledge to help achieve goals in the future). Addictions are really conditions where things which aren't useful to an individual invoke the pleasure pathways, because they are too recent for evolution to have got the pathways right, causing a distortion of priorities.

So it seems self-evident that any workaround to the distortion of priorities will need to specifically lower the pleasure / bump the pain of harmful things, relative to the beneficial things, rather than simply making everything unpleasurable by the same amount.

It's interesting that this would even be surprising: do people really expect every high-level, qualitative human mental state and behavior to be caused by a really simple element of brain wiring or chemistry, like a particular neurotransmitter being imbalanced, or presence/absence of a particular structure? Much of pop-neuroscience seems to, believing that for every/almost-every interesting high-level X, there's a "gene for X" or "neurotransmitter that causes X", and therefore a cure (in the case where X is bad) that consists of a fairly simple blunt instrument, like a pill that changes levels of one neurotransmitter.

But even in intro math classes studying relatively simple dynamical systems, it quickly becomes nontrivial to try to trace back high-level, qualitative behavior of a system to specific elements of the system that cause that behavior--- much less to point to which parts you should tweak to cause a particular desired high-level change. Craig Reynolds sometimes calls it the "reverse emergence problem" in computer graphics and agent simulations: our system is doing high-level-thing A, we want it to do B, which of our low-level rules are causing that and how do we fix it? There often isn't a simple answer or clean change to make, because the answer is "lots of rules interacting causes this, and any change you make will change lots of other things too". It would be pretty surprising if the common case in neuroscience, dealing with a hugely complex dynamical system, really turned out to be much cleaner and simpler than that.

Reply on news.ycombinator.com