The credibility of scientific activity has recently been under scrutiny with reports questioning the reproducibility of results. In response to this "reproducibility crisis", the National Institute of Health (NIH) designed a plan for more reproducible science in 2015, but progress to curb the issue seem to be slow at best. It is possible that the solutions proposed by the NIH are targeting symptoms rather than causes. In this talk, Dr. Poline will first consider the field of neuroscience and human neuroimaging and analyze the main causes of irreproducibility, considering the statistical and computational aspects of neuroimaging or imaging genetics. He will then discuss the social components that are likely to contribute to irreproducibility. In a second part, Dr. Poline will review solutions to foster a more reproducible research at the level of the tools and the statistical methods used - for example in high dimensions. He will also consider the academic ecosystem and propose community actions that are both possible and could be effective to reshape the way we practice research.
Dr. Jean-Baptiste (JB) Poline is an Associate Professor in the Department of Neurology and Neurosurgery at McGill; the co-Chair of the NeuroHub and Chair of the Technical Steering Committee for the Canadian Open Neuroscience Platform (CONP) at the Montreal Neurological Institute & Hospital (the NEURO); and a Primary Investigator at the Ludmer Centre for Neuroinformatics & Mental Health.