Saturday, October 28, 2006

Boundedly rational dope fiends

The paper by Suren Basov & Svetlana Danilkina, 2006, 'A theory of boundedly rational addiction' is the latest output of our drugs project. In my view its a nice piece that relaxes the constraints of the rational addiction model intelligently.

Much more to come at this site soon.


conrad said...

I'm not into addition or drugs, but if something like that was submitted to a decent psychology journal, it would get caned in the review process. I don't see:

a)how the data has anything to say about _genetic_ disposition. To say this you really need to find much better data (twin studies etc.). The fact that some people exhibit more addictive behavior than others doesn't neccesarily tell us anything about genetics.

b) Why not get real evidence supporting the rationality of people to start with? That way you could test the model with real data. It should be easy to find individiual differences on a variable like that.

c) If age of first abuse has something to do with rationality, then that should be easily testable to. We should find differences depending on age of first addiction. Surely that data exists.

d) Who cares about commensense views of addictive behavior when there is a large scientific literature on it?

hc said...

Conrad, This is a theory paper probably intended vfor an economics journal.

* The genetic evidence could stand for anything that produces individual differences in behaviour. With the same initial consumptions different people will behave in different ways. In fact there is a ton of evidence suggesting a genetic basis for addiction and this is partly based on twin studies.

*One of the problems is that it is difficult to refute the rational addiction hypothesis. Its implications are weak. On the other hand the vast amount of evidence showing that people seek to pay bto eliminate their addictions suggests it isn't very strong.

*There iis strong evidence that young people have high discount rates.

* We have been studying the scientific evidence on addiction for about 2 years.

Thanks for your comments I will relay to the authors.

conrad said...

No doubt it is a theory Harry, but it is a theory paper where an additional parameter is added to a model. It henceforth becomes easy to test based on Occams Razor that the simpler model is better unless it explains more. If you got desperate and the model didn't, you could look at subgroups and so forth.

I don't know what its like in economics, but in my little branch of science (probably not dissimilar to economics in many respects), providing both theory and a reasonable empirical test is much better than simply solving a few equations that a fair chunk of people won't understand anyway (probably exponentially better).

There is gazillions of stuff of personality and addictivenss, and if your model happens to capture a lot of that data, then that would be a big deal, versus now, which just looks like a nice little bit of maths to me. One can imagine things like discount rates are strongly correlated with other variables, in which case you might be able to infer the discount rate from other variables, and hence find a single underlying factor that explains many others. You could, for instance, try and collapse age of first addiction and personality into discount rate.

I still don't buy the word "genetic". If I beat you a lot and you become depressed, that would cause an individual difference in behavior that has nothing to do with genetics. Perhaps the term is just used differently in different fields.