Research findings on addiction are contradictory. According to biographical records and widely used diagnostic manuals, addicts use drugs compulsively, meaning that drug use is out of control and independent of its aversive consequences. This account is supported by studies that show significant heritabilities for alcoholism and other addictions and by laboratory experiments in which repeated administration of addictive drugs caused changes in neural substrates associated with reward. Epidemiological and experimental data, however, show that the consequences of drug consumption can significantly modify drug intake in addicts. The disease model can account for the compulsive features of addiction, but not occasions in which price and punishment reduced drug consumption in addicts. Conversely, learning models of addiction can account for the influence of price and punishment, but not compulsive drug taking. The occasion for this target article is that recent developments in behavioral choice theory resolve the apparent contradictions in the addiction literature. The basic argument includes the following four statements: First, repeated consumption of an addictive drug decreases its future value and the future value of competing activities. Second, the frequency of an activity is a function of its relative (not absolute) value. This implies that an activity that reduces the values of competing behaviors can increase in frequency even if its own value also declines. Third, a recent experiment (Heyman & Tanz 1995) shows that the effective reinforcement contingencies are relative to a frame of reference, and this frame of reference can change so as to favor optimal or suboptimal choice. Fourth, if the frame of reference is local, reinforcement contingencies will favor excessive drug use, but if the frame of reference is global, the reinforcement contingencies will favor controlled drug use. The transition from a global to a local frame of reference explains relapse and other compulsive features of addiction.

The matching law accounts for both addictive behavior and the usefulness of a person's evaluating choices in overall categories. To explain why overall bookkeeping, once learned, does not easily win out over local bookkeeping, another implication of matching is needed: intertemporal bargaining. The role of melioration, though probably important for new addiction is separate.

It is difficult to accommodate compulsive drug use that destroys regulation by consequences in a unitary framework of melioration or maximization, especially as both are strategies of behavioral regulation. The “contradiction” of addictive behavior despite aversive consequences is an illusory issue because consequences that do not decrease addictive behavior cannot be considered punishing.

Heyman explains useful ways to bring addictive drug use under environmental control. We doubt that relapse is explained by drug features such as immediate reinforcement, clouding of judgment, and so forth. Relapse may require explanation in terms of enduring sensitization of incentive neural substrates, but even if its causal assumptions are wrong, Heyman's model makes useful predictions for behavioral control.

Heyman's view that addiction can be viewed as a natural outcome predictable by melioration and the matching law is provocative. Remaining to be explained more fully, however, are exactly how his view is an improvement on other reinforcement-based accounts. Included in these elaborations should be an account of how different “bookkeeping schemes” are developed and controlled and what new approaches to treatment and prevention of drug addiction are indicated.

Heyman's target article makes a strong case for a behavioral approach to addiction, yet some important assumptions require justification, and promising behavioral alternatives to the author's melioration approach should be considered. In particular, the behavioral economic approach to addiction appears well developed and comprehensive. How does the melioration approach complement or improve on a behavioral economic account?

Heyman's two crucial theses are that people over-value immediate rewards, and that addictive substances “subvert the value of competing commodities.” These perennial ideas were discussed by Plato. Whereas Heyman provides scientific clarification and support for the first, the second remains problematic. I outline how this deficiency might be remedied via evolutionary considerations.

I support the major theme of Heyman's target article that behavioral choice theory can enhance our understanding of drug dependence, but I raise concerns about the critique of the operant model of drug dependence, the underscoring of melioration to the exclusion of other theories of choice, and assertions about the unique effects of drug reinforcement.

Heyman's analysis of addiction suggests that drug taking is irrational. The irrationality of drug taking, however, may depend on the acceptance of mainstream society's view of what is valuable. Consideration of the addict's viewpoint and cultural aspects of drug taking may be useful in trying to fathom the “rationality” of drug taking.

I discuss various theoretical issues concerning maximizing, matching, and melioration. The model of addiction based on melioration has the key feature that a reduction in drag use increases both the value of drag-taking and the value of conventional activities. I found Heyman's target article stimulating. I don't feel competent to comment on drags but I do have some thoughts, both general and specific, on the theoretical issues.

Heyman's target article is an analytical tour de force, but it makes too hard a distinction between voluntary and driven behavior. It is more fruitful to think about brain and behavior as shifting, interacting “agents,” represented by multiple neural circuits. This has the virtue of better connecting behavioral analysis with wet neuroscience.

For use in applied settings, the melioration model will need to incorporate changes in the shapes of local value functions over time, treat current value as a continuous function of time since previous choices, and take into account discounting of the effects of current behavior on future value. The policy implications of the model for regulating drugs are limited.

Economic models of addiction are choice-based. These models push the choice paradigm too far by modeling addiction as rational, normative behavior. Heyman's target article provides a sensible alternative to this economic approach by emphasizing that addiction is characterized by ambivalence and a perceived loss of self control. However, matching may not be a satisfactory platform on which to build this alternative model. Matching experiments do not provide evidence of ambivalence or perceived loss of self control.

Choosing the smaller, less delayed reinforcers of drug abuse (including cessation of withdrawal) over the larger, more delayed reinforcers of a healthy life can be described as impulsiveness (the opposite of self-control). Examination of drug abuse using a self-control/impulsiveness framework can help explain many of the puzzles about this behavior, while tying drug abuse into a large research database.

The conflict between drug and conventional rewards leads to a paradox: Sanctions against drug use decrease access to conventional rewards and push drug users toward drug abuse, whereas increased access to the rewards of family, friends, and work may help reduce drug abuse. Lack of control is not specific to drug addiction and is unlikely to yield to a shift in bookkeeping.

Heyman makes a convincing case that a melioration choice strategy is sufficient to produce addictive behavior. But given a plethora of addiction theories, the question is whether melioration theory is superior to rivals more sophisticated than a simple disease model or operant conditioning account. Heyman offers little direct evidence that melioration actually causes the addictions we observe.

Numerous individual differences determine the effectiveness of the multiple external reinforcement contingencies that control drug seeking behavior. Differences in the capacity to frame reinforcement and also the toxic effects of drugs modulate susceptibility to addiction. As a result, not all individuals are at the same degree of risk for addiction.

Addictive behavior has never seemed rational because it persists in spite of drastic aversive consequences. This is a particular problem for models of addiction such as operant psychology which hold that behavior is controlled by its consequences. Inspite of claims to the contrary, Heymans target article illustrates how operant psychology resolves this contradiction. By using the matching law, Heyman suggests a mechanism that explains why delayed aversive events may not control behavior, and a conceptual framework in which we can understand successful therapies.

Knowledge of specific instances of melioration does not imply knowledge of the variables controlling behavior. Drug self-administration has been successfully analyzed in terms of standard variables such as the immediacy and magnitude of the drug reinforcer. An appeal to melioration may not be necessary.

Heyman's model, paradoxically, predicts that whereas a maximizing approach to drug choice will prevent escalation of drug use it will never yield complete abstinence. We suggest an alternative model that overcomes this difficulty by focusing on changes in drug tolerance. A small modification allows maximization to predict either abstinence or moderation (e.g., social drinking).