A registered report of error-related negativity and reward positivity as biomarkers of depression: P-Curving the evidence

Abstract

Performance-monitoring event-related brain potentials (ERPs), such as the error-related negativity (ERN) and reward positivity (RewP), are advocated as biomarkers of depression symptoms and risk. However, a recent meta-analysis indicated effect size heterogeneity in the ERN and RewP literatures. Hence, advocating these ERPs as biomarkers of depression might be premature or possibly misguided due to the selective reporting of significant analyses on the part of researchers (e.g., p-hacking or omission of non-significant findings). The present study quantified the degree of selective reporting and the evidential value for a true relationship between depression and ERN and RewP using a p-curve analysis. We predicted that the ERN and RewP literatures would fail to show evidential value for a relationship between each ERP and depression. Contrary to expectations, both literatures showed evidential value, albeit weak.The statistical power of the included ERN studies was between 20% and 25%, and the statistical power of the RewP was around 27%. Taken together, these findings provide support for a relationship between these ERPs and depression, which strengthens claims that these ERPs represent candidate biomarkers of depression symptoms and risk. In light of the evidence for these relationships being weak, some recommendations moving forward include conducting a priori power analyses, increasing sample sizes to improve statistical power, assessing the internal consistency of ERP scores, and carefully planning statistical approaches to maximize power.Keywords: performance monitoring, depression, p curve, error-related negativity (ERN), reward positivity (RewP) 3 Performance-monitoring event-related brain potentials (ERPs), such as the error-related negativity (ERN) and reward positivity (RewP), are advocated as biomarkers or endophenotypes of depression symptoms and risk (e.g., Meyer, Bress, Hajcak, & Gibb, 2018;Nelson, Perlman, Klein, Kotov, & Hajcak, 2016). However, recent meta-analytic work indicates that both ERN and RewP literatures are replete with heterogenous effect sizes (Moran, Schroder, Kneip, & Moser, 2017). Although heterogeneous effect sizes might be due to known or unknown moderating factors, it is possible that heterogeneity is due to the selective reporting of significant analyses or undisclosed use of other questionable research practices (QRPs), and this possibility was not examined in the meta-analysis. If effect size heterogeneity is due to selective reporting/omission of non-significant findings or other QRPs, then advocating performancemonitoring ERPs as biomarkers of depression would be misguided or at least premature. The aim of the present registered report was to the quantify the degree of selective reporting and the evidential value for a true relationship between depression and ERN and RewP using a priori registered hypotheses and methods.

Dimensions of anxiety and depression and neurophysiological indicators of error‐monitoring: Relationship with delta and theta oscillatory power and error‐related negativity amplitude

“…When looking at individuals diagnosed with MDD, although there is some evidence that individuals with MDD have an enhanced ERN when compared to controls (Aarts et al., 2013; Chiu & Deldin, 2007; Holmes & Pizzagalli, 2008, 2010), other evidence suggests that either ERN amplitude is blunted (Olvet, Klein, & Hajcak, 2010; Weinberg, Meyer, et al, 2016) or that there is no difference in ERN amplitude between those with MDD and those without (Gorka & Phan, 2017; Moran, Schroder, Kneip, & Moser, 2017; Weinberg et al., 2012). A recent registered‐report meta‐analytic p ‐curve analysis suggested that the studies of the ERN in individuals with depression are underpowered to detect small‐to‐medium effects (average 20% power), making it currently difficult to understand the direction and strength of the relationship between ERN and depression symptoms (Clayson, Carbine, & Larson, 2020). Findings from the p ‐curve suggested that there is evidential value for the relationship between ERN and depression symptoms, but that the relationship is heterogeneous and small (Clayson et al., 2020).…”

Moderators of the internal consistency of error‐related negativity scores: A meta‐analysis of internal consistency estimates

“…The analysis on psychopathology groups as a whole, however, sheds light on the common misconception that the internal consistency of ERN scores is similar across psychopathology and healthy control groups. For example, a recent meta-analysis on the relationship between depression and ERN emphasized that not a single of the 23 examined studies evaluated the internal consistency of ERN scores (Clayson, Carbine, & Larson, 2020). Given the potential for mistaken statistical inferences in research with poor score reliability highlighted above, the adoption of new standard operating procedures that include routine evaluation of ERN score internal consistency in ERN psychopathology research seems warranted.…”

The Association Between Experimentally Induced Stress, Performance Monitoring, and Response Inhibition: An Event-Related Potential (ERP) Analysis

“…The ERN can be localized to the ACC (Gehring et al, 1993;Dehaene et al, 1994;Herrmann et al, 2004;Debener et al, 2005) and a convergence of information suggests the ACC plays a considerable role in self-monitoring and emotion regulation (Perrone-McGovern et al, 2017). Notably, some studies suggest a small-to-modest association between ERN amplitude and various forms of psychopathology (e.g., Pasion and Barbosa, 2019;Clayson et al, 2020).…”