Both human and animal studies support the relationship between depression and reward processing abnormalities, giving rise to the expectation that neural signals of these processes may serve as biomarkers or mechanistic treatment targets. Given the great promise of this research line, we scrutinize those findings and the theoretical claims that underlie them. To achieve this, we apply the framework provided by classical work on causality as well as contemporary approaches to prediction. We identify a number of conceptual, practical, and analytical challenges to this line of research, and use a pre-registered meta-analysis to quantify the longitudinal associations between reward processing aberrations and depression. We also investigate the impact of measurement error on reported data. We find that reward processing abnormalities do not reach levels that would be useful for clinical prediction, yet the evidence thus far does not exclude their possible causal role in depression. Footnotes * https://osf.io/whvam/​ * https://github.com/nimh-mbdu/great_expectations​ * https://mybinder.org/v2/gh/nimh-mbdu/great_expectations/de3c6434b364f76db5b7e4b31f218d4045...

10.1101/2020.03.04.975136