Rational addiction
Overview
The hypothesis that addictions (to heroin, tobacco, television, etc.) are nothing more than a specific kind of rational, forward-looking, optimal consumption plans. Kevin M. Murphy and Nobel Prize Winner Gary S. Becker published the "Theory of Rational Addiction" in the Journal of Political Economy in 1988 (Volume96: 675-700). A theory of addictions in the broad sense - whether to heroin, tobacco, television, exercise or food - the article tried to reconcile addictions with the standard rational choice framework of modern economics. Though controversial, the theoretical approach has become the standard approach to understanding addiction in economics, and a variety of extensions and modifications have been developed and published by other authors over the years - several in leading economics journals.
1988 article
The original theory models addictions as the gradual implementation of a forward-looking consumption plan made under full certainty and perfect information. The addict knows exactly how the good will affect him, and the reason he consumes more and more ("gets hooked") is that this is the pattern of consumption that maximizes his discounted utility. He knows that consuming the addictive good will change his "utility technology," altering both his future baseline level of utility and the marginal effect of consuming the addictive good in the future. Put differently, you'll wake up feeling worse, but the drug will work even better. This means that your present consumption is influenced both by how attractive the addictive good is today, and partly by its use as "investment" in changing your body and mind for future consumption. As a result, if cigarette taxes are credibly announced to double in one year's time, cigarette smokers will cut their smoking today according to the theory.
A sizeable econometric literature has grown up around the theory - and according to some the theory is consistent with the main empirical evidence that has been used.
Later extensions
Later extensions by other authors allow for uncertainty and a form of regret (you make a "rational bet" with positive expected return, but you're unlucky), cyclical consumption (interpreted as bingeing), chaotic consumption (fully deterministic but highly irregular consumption patterns strongly influenced by initial conditions), endogenous time preferences (where you know and rationally take into account how the drugs make you care less about your future), and quasi-hyperbolic discounting (the way the consumer takes the future into account makes him inconsistent over time - laying plans that he later deviates from). This latter theory is an example of an extension trying to shift the theory away from its laissez-faire and drug-liberal implications for policy. The fully rational "rational addiction models" are widely held to have the policy implication that drug users - even if they are unhappy on drugs - would be even more unhappy and worse off if they were forced off drugs. The government's policy - in this view - should be restricted to ensuring that people are well-informed and don't impose externalities on others (e.g., second-hand smoke).