Reproducibility and replicability in zebrafish behavioral neuroscience research.

Reproducibility and replicability are fundamentally important aspects of the scientific method. From time to time the discussion about whether scientific findings are replicable enough flares up. In fact, some recent publications claim we are witnessing a replication crisis. This is a particularly important problem in laboratory organisms that are relatively new, i.e., for which only limited amount of information is available, and for which only a limited number of methods have been developed. The zebrafish is a relative newcomer in behavioral neuroscience. This review considers four distinct reasons as possibly underlying reproducibility issues in behavioral neuroscience studies using the zebrafish. One, publication bias for positive results. Two, statistical issues that surround the question of how to address type 1 and type 2 errors, and how to make statistical inference. Three, inappropriate control of factors that are known to potentially influence results. And four, methodological issues stemming from insufficient understanding of factors that may influence experimental results. The review will mainly focus on experimental issues and solutions, i.e. the latter two reasons listed above. It is not intended to be comprehensive, and its examples are drawn mainly from the author's own studies and experience with zebrafish. Nevertheless, most issues discussed are not unique to his laboratory, to the zebrafish, or even to behavioral neuroscience.