Our research centre is already planning to adopt the excellent guidelines suggested by Story Landis and colleagues to improve reporting standards in preclinical research (S. C. Landis et al. Nature 490, 187–191; 2012). However, I am concerned about the wholesale adoption of recommendations that could, paradoxically, have a restrictive effect on the early stages of basic research.

Curiosity and observation are critical to science, and although blinding is good experimental practice in most research scenarios, it should follow an exploratory period. In several of the behavioural studies conducted in our lab over many years, we have done the experiment at least twice — once unblinded and then blinded — followed by a check that we get the same results. Usually we do, but not always. Sometimes we deliberately include a procedure that causes performance to fall to chance (see, for example, D. Tse et al. Science 316, 76–82; 2007). If it does not, there might be uncontrolled variables.

Researchers thrive on noticing something subtle and pursuing it, but this is most effective when they do not have one hand tied behind their backs. Stringent guidelines should then follow.