This review applies some new experimental findings and theoretical ide
as about how reinforcers act on the neural mechanisms of learning and
memory to the problem of how addictive drugs affect behaviour. A basic
assumption of this analysis is that all changes in behaviour, includi
ng those involved in drug addiction and the initiation of drug self-ad
ministration, require the storage of new information in the nervous sy
stem. Animal studies suggest that such information is processed in sev
eral (this review deals with three) more or less independent learning
and memory systems in the mammalian brain. Reinforcers can interact wi
th these systems in three ways: they activate neural substrates of obs
ervable approach or escape responses, they produce unobservable intern
al states that can be perceived as rewarding or aversive, and they mod
ulate or enhance the information stored in each of the memory systems.
It is suggested that each addictive drug maintains its own self-admin
istration by mimicking some subset of these actions. Evidence supporti
ng the notion of multiple memory systems and data on the actions of se
veral drugs (amphetamine, cocaine, nicotine, alcohol and morphine) on
these systems are briefly reviewed. The utility of the concept of ''re
ward'' for understanding the effects of drugs on behaviour is discusse
d. Evidence demonstrating actions of drugs on multiple neural substrat
es of reinforcement suggests that no single factor is likely to explai
n either addictive behaviour in general or self-administration in part
icular. Some of the findings on the development and maintenance of sel
f-administration by animals of the five exemplar drugs are discussed i
n the context of these ideas.