An integrated homeostatic reinforcement learning theory of motivation explains the transition to cocaine addiction
Drugs of abuse implicate both reward learning and homeostatic regulation mechanisms of the brain. Theories of addiction, thus, have mostly depicted this phenomenon as pathology in either habit-based learning system or homeostatic mechanisms. Showing the limits of those accounts, we hypothesize that compulsive drug seeking arises from drugs hijacking a system that integrates homeostatic regulation mechanism with goal-directed action/behavior. Building upon a recently developed homeostatic reinforcement learning theory, we present a computational theory proposing that cocaine reinforces goal-directed drug-seeking due to its rapid homeostatic corrective effect, whereas its chronic use induces slow and long-lasting changes in homeostatic setpoint. Our theory accounts for key behavioral and neurobiological features of addiction, most notably, escalation of cocaine use, drug-primed craving and relapse, and individual differences underlying susceptibility to addiction. The theory also generates unique predictions about the mechanisms of cocaine-intake regulation and about cocaine-primed craving and relapse that are confirmed by new experiments.\n\nSignificanceChronic use of addictive drugs renders increased motivation in planning to obtain and consume the drugs, despite their adverse social, occupational, and health consequences. It is as if addicts gradually develop a strong need for the drug and use their cognitive abilities and the knowledge of their environment in order to fulfil that need. In this paper, we build a mathematical model of this conception of addiction and show through quantitative simulations that such a model actually behaves in the same way that human addicts or laboratory animals that are exposed to cocaine behave. For example, the model shows gradually increasing motivation for drugs, relapse after long periods of abstinence, and individual differences in susceptibility to addiction.