RESUMO
The use of animal models continues to be essential for carrying out research into clinical phenomena, including addiction. However, the complexity of the clinical condition inevitably means that even the best animal models are inadequate, and this may go some way to account for the apparent failures of discoveries from animal models, including the identification of potential novel therapies, to translate to the clinic. We argue here that it is overambitious and misguided in the first place to attempt to model complex, multifacetted human disorders such as addiction in animals, and especially in rodents, and that all too frequently "validity" of such models is limited to superficial similarities, referred to as "face validity", that reflect quite different underlying phenomena and biological processes from the clinical situation. Instead, a more profitable approach is to identify (a) well-defined intermediate human behavioural phenotypes that reflect defined, limited aspects of, or contributors to, the human clinical disorder, and (b) to develop animal models that are homologous with those discrete human behavioural phenotypes in terms of psychological processes, and underlying neurobiological mechanisms. Examples of past and continuing weaknesses and suggestions for more limited approaches that may allow better homology between the test animal and human condition are made.