Under-Reported Task Parameters Can Impact Conclusions Drawn From EEG Research

• In Event: Towards a More Open, Reproducible, and Representative Field of Developmental EEG Research

Abstract

Building on advances in the field of developmental cognitive neuroscience, EEG findings are increasing being pursued as potential brain-based targets in translational research. As one example, the Error-Related Negativity (ERN)—an EEG-based neural marker of the ability to self-detect one’s own mistakes—has gained interest as a potential biomarker for anxiety symptoms in both youth and adults. Prior meta-analytic studies confirm the link between anxiety symptoms and an enlarged ERN. However, more work is needed to characterize how variation in study parameters—especially for methodological details that may be underreported in the literature—might systematically impact the observed link between ERN and anxiety symptoms. In the current study, we leveraged an open-source database of reported ERN-anxiety effect sizes (> 50 studies) and tested, via meta-analytic regression, whether several common task parameters exhibit systematic influences on observed ERN-anxiety associations. Exemplar task parameters of interest where: paradigm (Flanker vs. Go-Nogo), response format (one vs. two-hands), and block-level feedback (present, absent). Assessing rates of methodological reporting, 100% of studies reported the task paradigm used to elicit the ERN, whereas 30% failed to report response format, and 40% did not report presence/absence of block-level feedback. Meta-analytic regression found that task paradigm did not moderate the ERN-anxiety association across studies (p = .36). In contrast, both response format (p = .04) and presence of block-level feedback (p = .03) exhibited significant, systematic effects on the ERN-anxiety association. There are several implications of these findings. First, meta-analytic evidence for the effects of response format and feedback on ERN-anxiety associations corroborate findings from individual studies, with theoretical implications for why an enlarged ERN is associated with anxiety. As such, they can guide the choice of task parameters most appropriate for eliciting an ERN in the context of translational research on anxiety. Second, overall rates of reporting methodological details for two fundamental task parameters (response format and presence of block-level feedback) were surprisingly low, which impairs replication/reproducibility efforts generally. Third, the underreporting of some task parameters is particularly problematic, given that these parameters were shown to exhibit systematic influences on the observed link between anxiety symptoms and the ERN. Collectively, these findings not only advance our theoretical understanding of why an enlarged ERN is associated with anxiety symptoms, but highlight the necessity of detailed, exhaustive reporting of methodological details in EEG research. The importance of appropriate methodological reporting is further magnified by the fact that EEG-based measures are increasingly being employed in translational contexts, with real-world implications for the effectiveness of assessment and intervention efforts.

Authors

• George A Buzzell, Florida International University
  Presenting Author

• Yanbin Niu, Vanderbilt University

• Emily Machado

• Renata Dickinson