A rich and diversified discussion has been going on since many years about ignorance and its treatment by sciences. Sociologists and anthropologists have proposed different definitions to delimit this broad intellectual field and to demonstrate how the actors actively cope with uncertainty. The article focuses on a particular kind of ignorance, namely intentional concealing of data in the field of neuroscience. This contribution draws on first hand empirical evidence that researchers sometimes obscure technical problems as well as they have to deal with the difficulties to publish discordant or contradictory data. These practices illustrate that valorisation of spectacular discoveries and "positive" results clashes with the ideals of transparency and knowledge sharing conveyed in particular by infrastructures like databases. © 2017 S.A.C.. Tous droits réservés pour tous pays.