Meta-science in neuroscience: examples and perspectives

In this talk, I will describe two meta-science projects and draw theoretical implications.

The Reproducibility Project-Psychology was a large-scale replication study in experimental psychology. As a follow-up to that project, we have modelled statistical power and publication bias, finding boundary intervals consistent with the observed results. We estimated that negative evidence was observed 55-98 times before one negative result was published. Another meta-scientific project concerns the diurnal variation of interleukin-6 in the blood. By reconstituting data sets from published diagrams, we were able to analyse 56 datasets from 43 studies, with a total of 1100 participants, and could show that interleukin-6 does not rise in the night, as previously thought. Based on these findings and others, I will propose some theoretical concepts for the data life-cycle as related to inference. In the talk I will also mention other meta-science related activities that I work with, most notably education in open science and reproducible research, for the purpose of exploring possible collaborations with Quest.