Pitfalls in the interpretation of genetic and pharmacological effects on anxiety-like behaviour in rodents.

Over the last 15 years, genetically modified mice have added important data to our knowledge on psychiatric diseases including anxiety. This has produced many behavioural publications, partially by non-behaviourists, in which differences between mutants and normal wild-type animals were described. The popularity of these novel tools allowing the study of new mechanisms also, however, led to observations that could not be confirmed. This review attempts to summarize various factors that can lead to difficult and partially incorrect interpretation of data collected in anxiety-related paradigms. These pitfalls are explained by using virtual data. Our analysis illustrates that determining anxiety in rodents is more complicated than measuring a single parameter in a particular paradigm. It is important to use proper controls such as additional measures in the same or other procedures, as well as a conservative estimation of the chance of finding an actual effect. In this way, it is possible to enhance confidence in the findings. Alternative explanations for findings, like side effects or main effects in a different domain, such as cognition, should always be taken into account. Finally, several examples from the literature are presented as illustrations of the theoretical issues discussed. We believe that considering the pitfalls presented here will help researchers to design optimized experiments that can be more readily interpreted and replicated across laboratories.