Batch effects, undesirable sources of variability across multiple experiments, present significant challenges for scientific and clinical discoveries. Batch effects can (i) produce spurious signals and/or (ii) obscure genuine signals, contributing to the ongoing reproducibility crisis. Because batch effects are typically modeled as classical statistical effects, they often cannot differentiate between sources of variability due to confounding biases, which may lead them to erroneously conclude batch effects are present (or not). We formalize batch effects as causal effects, and introduce algorithms leveraging causal machinery, to address these concerns. Simulations illustrate that when non-causal methods provide the wrong answer, our methods either produce more accurate answers or “no answer,” meaning they assert the data are inadequate to confidently conclude on the presence of a batch effect. Applying our causal methods to 27 neuroimaging datasets yields qualitatively similar results: in situations where it is unclear whether batch effects are present, non-causal methods confidently identify (or fail to identify) batch effects, whereas our causal methods assert that it is unclear whether there are batch effects or not. In instances where batch effects should be discernable, our techniques produce different results from prior art, each of which produce results more qualitatively similar to not applying any batch effect correction to the data at all. This work, therefore, provides a causal framework for understanding the potential capabilities and limitations of analysis of multi-site data.

Functional connectivity (FC) has shown promising utility in the field of precision psychiatry. However, to translate from research to clinical use, FC reliability and sensitivity to individual differences still require improvement. Movie watching as an acquisition state offers advantages at the whole-brain level that align with the requirements of FC for individualized measures. However, it is unclear whether these advantages hold in specific brain regions important for precision psychiatry. Here, we compared univariate and multivariate reliability-based measures of movie-watching and resting-state FC data in three psychiatrically relevant brain regions. We found that the reliability of movie-watching FC was comparable with resting-state FC in the dorsolateral prefrontal cortex and presupplementary motor area, and movie-watching FC was more discriminable than resting-state FC in the temporoparietal junction. Rest had higher reliabilities at lower data amounts (e.g., under 5 minutes of scan time). We then expanded this approach to all brain regions and showed that for image intraclass correlation coefficients (I2C2), no parcels were significantly different between movie and rest. For discriminability, 25% (94/379) of parcels were better for movie than for rest, and zero parcels were better for rest. For fingerprinting, 59 parcels were better for movie (mainly in visual and temporal regions, mean improvement in accuracy = 23%) and 4 parcels were better for rest. For researchers interested in cross-state differences in FC reliability, we provide an interactive visualization tool that displays the results for all measures and for all regions in both movie and rest. These findings suggest that movie watching as an acquisition state—even when using different movies across scans—may provide a useful alternative to resting state in research studies that require optimization of FC discriminability.

Diversity, equity, and inclusivity (DEI) are important for scientific innovation and progress. This widespread recognition has resulted in numerous initiatives for enhancing DEI in recent years. Although progress has been made to address gender and racial disparities, there remain biases that limit the opportunities for historically under-represented researchers to succeed in academia. As members of the Organization for Human Brain Mapping (OHBM) Diversity and Inclusivity Committee (DIC), we identified the most challenging and imminent obstacles toward improving DEI practices in the broader neuroimaging field. These obstacles include the lack of diversity in and accessibility to publicly available datasets, barriers in research dissemination, and/or barriers related to equitable career advancements. In order to increase diversity and promote equity and inclusivity in our scientific endeavors, we suggest potential solutions that are practical and actionable to overcome these barriers. We emphasize the importance of the enduring and unwavering commitment required to advance DEI initiatives consistently. By doing so, the OHBM and perhaps other neuroscience communities will strive toward a future that is not only marked by scientific excellence but also characterized by diverse, inclusive, and equitable opportunities for all, including historically under-represented individuals around the world.

Human neuroscience research remains largely preoccupied with mapping distinct brain areas to complex psychological processes and features of mental health disorders. While this reductionist and localizationist perspective has resulted in several substantive contributions to the field, it has long been viewed as only a piece of the puzzle. Emerging evidence now empirically demonstrates how a historical reliance on localizationist techniques may underlie recent challenges to reproducibility and translation in human neuroscience. To advance discovery, we must collectively better incorporate complex systems and machine-learning approaches that better capture the multidimensional, dynamic, and interacting nature of the brain. Moreover, we must begin to contend with how to best integrate complementary modalities beyond the brain to better understand complex mental processes.