Neuroimaing Studies-Racial Data Less

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

NeuroImage 254 (2022) 119122

Contents lists available at ScienceDirect

NeuroImage
journal homepage: www.elsevier.com/locate/neuroimage

Diversifying participation: The rarity of reporting racial demographics in


neuroimaging research
Madeline G. Goldfarb, Darin R. Brown∗
Psychology Field Group, Pitzer College, Claremont, CA 91711, United States

a r t i c l e i n f o a b s t r a c t

Keywords: Functional neuroimaging has been instrumental to the field of cognitive neuroscience; however, its increasing
Neuroimaging prevalence has evoked conversations concerning limitations associated with reproducibility and bias. Prevailing
Diversity racial, cultural, and socioeconomic biases in scientific research perpetuate demographic homogeneity in partic-
Participant recruitment
ipation, contributing to failed replicability and generalizability and driving inaccurate representations of neuro-
Demographics
logical normalcy. The current report employs systematic and exploratory search methods to investigate ongoing
practices surrounding participant recruitment and documentation. The systematic search found that only 20 out
of the 536 articles collected reported the race and ethnicity demographics of their participants, exposing a dearth
of race and ethnicity demographics reporting in neuroimaging research. These results drive our recommendations
for increased transparency and diversity surrounding research participation.

1. Introduction (Lee, 2009; Loue, 2006), as race is a socially constructed method of ar-
bitrarily classifying humans. Concepts of race/ethnicity are historically
While Magnetic Resolution Imaging (MRI) and electroencephalo- prescribed to reflect economic, political, social, and cultural conditions,
gram (EEG) are widely regarded as progressive tools to study the neural and their boundaries are heavily influenced by colonialism and imperi-
correlates of behavior (Morita et al., 2016), issues of reproducibility alism (Lee, 2009; Yellow Bird, 1999). Regardless, race and ethnicity are
and bias have attracted much attention. In alignment with the 2015 not inconsequential identifiers, and contemporary dialogue should aim
Open Science Collaboration recommendations concerning reproducibil- to refute deterministic definitions of race and ethnicity.
ity (Open Science Collaboration, 2015), the current commentary aims to Racial categories have widespread effects, including findings of
obtain a sample the participant demographics of functional neuroimag- racial disparities in clinical outcomes (Brown et al., 2017; Díaz-
ing studies and explore the related methodological constituents that up- Venegas et al., 2016; Isamah et al., 2010). For example, Alzheimer’s
hold such biases in cognitive neuroscience. Dementia (AD) is nearly twice as prevalent among Black versus white
As neuroscientists increasingly recognize social and environmen- adults in the United States, and such rates cannot be explained by cere-
tal influences on neural mechanisms and their manifestations, statisti- brovascular factors and socioeconomic status (SES) alone (Manly et al.,
cally controlling for such participant idiosyncrasies has become essential 2008; Misiura et al., 2020). Instead, race effects on cognition likely
(Farah, 2017; Gilmore et al., 2017; Sauce and Matzel, 2013). Investiga- arise through the intersectional layering of unique sociocultural fac-
tions into social and epigenetic influences have largely refuted theories tors which are both sources for biological modification as well as ef-
of cognition rooted in biological reductionism, demanding the acknowl- fects of their biological embeddings (Lee, 2009; Manly et al., 2008;
edgment of relevant psychological, environmental, and social effects on Roberts, 2011). Further exploration of race differences in AD revealed
the brain (Champagne, 2010; Morita et al., 2016). The failure to ac- race-specific functional connectivity patterns and electrophysiological
knowledge individual variation can have profound effects, and many differences (Misiura et al., 2020). Similar neural signaling effects by race
hypothesize that such neglect contributes to the field’s contention with have also been found with respect to other clinical conditions (Chand
reproducibility and generalizability (Chatterjee, 2005; Isamah et al., et al., 2017; Hokett and Duarte, 2019). In an EEG study investigat-
2010). ing the influence of race on sleep, Black participants’ EEG revealed de-
The current commentary aims to explore demographic influences creased power in the delta and beta bands, and effects were independent
through the categories of race and ethnicity. Both constructs are widely from SES and education (Hall et al., 2020). Race has also been identi-
disputed and harbor culturally variable definitions and applications fied as a direct correlate of cognitive functioning among older adults,


Corresponding author.
E-mail address: darinb@pitzer.edu (D.R. Brown).

https://doi.org/10.1016/j.neuroimage.2022.119122.
Received 24 April 2021; Received in revised form 17 March 2022; Accepted 19 March 2022
Available online 23 March 2022.
1053-8119/© 2022 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/)
M.G. Goldfarb and D.R. Brown NeuroImage 254 (2022) 119122

Fig. 1. (a) Breakdown in demographic reporting by journal and technology. The complete search produced 20 articles that reported race or ethnicity demographics
and 516 papers that did not. 10 of the articles were published in NeuroImage and the other 10 were published in Cerebral Cortex. All 20 articles used MRI. (b)
Breakdown of EEG and MRI research identified from the search according to journal of publication. MRI research was published more frequently by both NeuroImage
and Cerebral Cortex.

contrary to expectations of an indirect influence through SES and ed- (Fig. 1). The reporting rates were less than 10% for both journals, sug-
ucation (Zsembik and Peek, 2001). These findings support the unique gesting neither requires such demographic information for publication
and confounding effects of race on cognition. (Fig. 1A). Moreover, all of the 20 articles reporting racial demographics
The present commentary aims to review current practices of docu- used MRI; however, the majority of the articles collected in general used
menting race in cognitive neuroscience research to determine the field’s MRI technology (Fig. 1B).
attitudes regarding race as a consequential demographic influencing the To investigate these discrepancies, a brief exploratory review of
replicability of findings. Moreover, the review aims to draw attention EEG literature was conducted (see supplement). Overall, the exploratory
to issues of diverse participant recruitment as it relates to the gener- analysis revealed a lack of consideration for participant race/ethnicity,
alizability of research conclusions and the perpetuation of inaccurate such that the dearth of reporting is not limited to research published
descriptions of neurological normalcy. in NeuroImage and Cerebral Cortex (Table S2). Nevertheless, certain
methodological practices in EEG research (see discussion) have led to
2. Methods documented difficulties in recruiting participants of color, suggesting a
degree of homogeneity across EEG research participation (Etienne et al.,
A systematic approach was employed to explore the racial demo- 2020).
graphics of participants in EEG and MRI research. We narrowed the
scope of the search to the top five journals in cognitive neuroscience 4. Discussion
according to their SCImago Journal Rank (SJR) Score in 2019. Because
three of these five selected journals predominantly publish review ar- The methods and techniques used in cognitive neuroscience are ad-
ticles, the search was restricted to EEG and MRI research published in vancing at an impressive rate; however, the most efficacious progress is
Cerebral Cortex (2019) and NeuroImage (2019), reflecting modern prac- often multifaceted, and the inability to recognize demographic variables
tices in cognitive neuroscience. These methods yielded 536 articles (Ta- in neuroscientific research has vast implications (Gilmore et al., 2017).
ble S1 of supplement), and papers that reported their participants’ racial The lack of consideration of these factors is problematic, resulting in
demographics were isolated (Table 1). inaccurate representations of neurological normalcy.
Further, racial demographic information was collected for the in- As shown in the results, only 20 of the 536 articles documented
stitutions and cities through which the research took place to provide their participants’ race/ethnicity, exposing a significant methodologi-
contextual information about the populations from which participants cal lapse in the field of cognitive neuroscience. Beyond the extensive
were recruited. Because race/ethnicity constructs vary by country, these research supporting the importance of racial considerations in neuro-
additional institutional and city-wide demographics were only supple- science research, the lack of transparency perpetuates division in the
mented for research conducted in the United States or regions with sim- field, hindering progress and discovery (Abiodun, 2019; Jones et al.,
ilar bureaucratically prescribed guidelines. These demographics were 2020). Thus, our first recommendation calls for increased transparency
collected from the 2019 US Census and official diversity reports. surrounding participation recruitment and demographic information.
Ideally, racial demographic documentation would supply a cohesive
3. Results narrative surrounding recruitment (i.e., the methods, geography, and
institutions) and expand upon the demographics widely collected by
The systematic search produced 536 articles (Table S1), and 20 pa- the field to date (i.e., sex, age, education). Importantly, these reporting
pers documented their participants’ race/ethnicity (Table 1). These 20 recommendations should be expanded to cover other social factors (e.g.,
articles represented less than 4% of the articles, revealing widespread SES) that have suffered from low reporting rates.
neglect of participant race/ethnicity demographics by the field. However, the complexities surrounding race/ethnicity hinder the
In an attempt to better understand the field’s reporting practices, idealistic portrayal of a given cohort (Konkel, 2015), likely perpetuat-
trends such as reporting rates by journal and technology were isolated ing the limited transparency and lack of uniformity in the demographics

2
M.G. Goldfarb and D.R. Brown
Table 1
Overview of articles documenting participant racial demographics in the systematic search.

Participant Demographics (%) Institutional Demographics (%) City Demographics (%)

Article Tech White Black Asian Indig. Latinx 2+ NR Source of Recruitment∗ White Black Asian Indig. Latinx 2+ NR City White Black Asian Indig. Latinx 2+ NR

Nemrodov et al. (2019) MRI 100 0 0 0 – – – Carnegie Mellon University 28.1 4.3 30.5 0 8.9 – 9.7 Pittsburgh, PA 66.9 23.2 5.7 0.2 3.1 3.5 –
Engelhardt et al. (2019) MRI 43.6 5.1 5.1 – 14.5 29.9 1.7 University of Texas Austin 39.9 4.9 19.5 0.1 21.7 2.7 1.2 Austin, TX 48.3 7.8 7.3 0.6 34.3 3.3 –
Spasov et al. (2019) MRI 99.98 – – – – – 0.02 ADNI, Harvard University 42.9 6.7 17.4 0.2 11.4 – 9.2 Cambridge, MA 67 11 16 0.2 9.2 4.1 –
Thomason et al. (2019) MRI 15.4 62.9 7.7 0 – 7.7 – Michigan State University 65.7 6.8 5.2 0.2 4.3 2.8 – Detroit, MI 14.6 78.6 1.6 0.3 7.6 1.9 —
Fan et al. (2019) MRI 69.5 8.3 19.4 0 2.7 – – Harvard University 42.9 6.7 17.4 0.2 11.4 – 9.2 Boston, MA 52.6 25.3 9.6 0.3 19.7 5.1 –
Boots et al. (2019) MRI 45.7 46.8 – 0 7.4 – – University of Illinois at Chicago 33.8 7.9 18.7 0.1 24.5 2.7 – Chicago, IL 49.4 30.1 6.4 0.3 29 2.7 –
Valenza et al. (2019) MRI 85.29 5.88 – 0 20.59 – – HCP, Harvard University 42.9 6.7 17.4 0.2 11.4 – 9.2 Cambridge, MA 67 11 16 0.2 9.2 4.1 –
Yu et al. (2019) MRI 0 0 100 0 0 – – Tangdu Hospital Shaanxi, China
Harnett et al. (2019) MRI 27.78 72.22 0 0 – – – Birmingham, AL area 59.9 21.2 5.7 0.04 3.2 3.6 – Birmingham, AL 25.3 70.5 0.9 0.2 3.7 1.6 –
Richmond et al. (2019) MRI 71.03 – – – – – 20.7 University of Melbourne Melbourne,
Australia
Jha et al. (2019) MRI 76 21.4 2.1 0.4 – – – University of North Carolina at 62 8 10.7 0.1 7.8 – – Chapel Hill, NC 72.9 9.9 12.5 0.2 6.6 3 –
3

Chapel Hill
Selvaggi et al. (2019) MRI 100 0 0 0 – – – Apulia, Italy Apulia, Italy
Herzmann et al. (2019) MRI – 57.5 – – – – 42.5 Washington University in 52.2 8.1 16.6 0.1 8.5 – 6.6 St. Louis, MO 46.2 46.9 3.2 0.2 4 2.3 –
St. Louis
Ferradal et al. (2019) MRI 45 – – – – – 55 Harvard University 42.9 6.7 17.4 0.2 11.4 – 9.2 Boston, MA 52.6 25.3 9.6 0.3 19.7 5.1 –
Marecková et al. (2019) MRI 100 0 0 0 – – – Masaryk University Brno, Czechia
Morton et al. (2019) MRI 62 17.4 6.5 – 15.2 – 14.1 Wake Forest University 67.5 8.4 3.8 0.8 6.4 2.3 – Boston, MA 52.6 25.3 9.6 0.3 19.7 5.1 –
Reineberg et al. (2019) MRI 82.7 11.3 4.5 – – – – University of Colorado Boulder 66.3 1.6 5.3 0.2 11 4.9 – Boulder, CO 87.2 1.2 5.6 0.3 9.8 3.9 –
Del Maschio et al. (2019) MRI 48 0 52 0 – – – Vita-Salute San Raffaele Hong Kong
University, University of Hong
Kong, Beijing Normal University
Xin et al. (2019) MRI 50 33 0 0 17 – – Allen Human Brain Atlas 42.9 6.7 17.4 0.2 11.4 – 9.2 Cambridge, MA 67 11 16 0.2 9.2 4.1 –
Xu et al. (2019) MRI 75 – – – – – 25 Parkland Hospital in Dallas Dallas, TX 62.5 24.3 3.4 0.3 41.7 2.5 –

Note. The demographics are as reported by the author of research, institution, or census. Demographic categories reflect those outlined by the National Institutes of Health (NIH) with adaptations suggested by scholars
and activists: white, Black or African American, Asian, Indigenous peoples of the Americas, and Latinx or Hispanic (National Institutes of Health. Updated June, 2018; Yellow Bird, 1999).
(∗ ) Source of participant reflects the sources from which the researchers found participants or the institution of the last author, if no source is documented.

NeuroImage 254 (2022) 119122


M.G. Goldfarb and D.R. Brown NeuroImage 254 (2022) 119122

presented in Table 1. These inadequate reporting methods are unique Supplementary materials
to race/ethnicity and likely avert scientists in the field from consider-
ing such demographics (Lee, 2009). Thus, our second recommendation Supplementary material associated with this article can be found, in
surrounds the development of standard reporting guidelines. the online version, at doi:10.1016/j.neuroimage.2022.119122.
Initiatives by the NIH (e.g., Human Subjects System) have required
References
clinical researchers to report their racial/ethnic demographics to pro-
mote diversity and inclusion; however, no such programs exist for basic Abiodun, S.J., 2019. “Seeing color,” a discussion of the implications and applications
science research (National Institutes of Health. Updated June, 2018). of race in the field of neuroscience. Front. Hum. Neurosci. 13. doi:10.3389/fn-
Without an analogous diversity standard, neuroimaging researchers hum.2019.00280.
Brown, T.I., Uncapher, M.R., Chow, T.E., Eberhardt, J.L., Wagner, A.D., 2017. Cogni-
could continue to recruit homogenous participant groups. An estab-
tive control, attention, and the other race effect in memory. PLoS One 12 (3), 1–21.
lished standard for reporting race/ethnicity demographics would sup- doi:10.1371/journal.pone.0173579.
port transparency in the field, clarifying the landscape of diversity and Champagne, F.A., 2010. Epigenetic influence of social experiences across the lifespan.
Dev. Psychobiol. 52 (4), 299–311. doi:10.1002/dev.20436.
promoting inclusion in neuroscience research. Such guidelines and re-
Chand, G.B., Wu, J., Qiu, D., Hajjar, I., 2017. Racial differences in insular connectivity and
quirements may be most effective if established by journals of publica- thickness and related cognitive impairment in hypertension. Front. Aging Neurosci.
tion. 9, 1–10. doi:10.3389/fnagi.2017.00177, MAY.
While the small sample size of articles with documented demograph- Chatterjee, A., 2005. A madness to the methods in cognitive neuroscience? J. Cogn. Neu-
rosci. 17 (6), 847–849. doi:10.1162/0898929054021085.
ics offers limited analysis of participant diversity, a prevailing bias in Díaz-Venegas, C., Downer, B., Langa, K.M., Wong, R., 2016. Racial and ethnic differences
research participation alludes to predominantly homogenous cohorts, in cognitive function among older adults in the USA. Int. J. Geratr. Psychiatry 31 (9),
driving the third recommendation of the current review to encourage 1004–1012. doi:10.1002/gps.4410.
Etienne, A., Laroia, T., Weigle, H., Afelin, A., Kelly, S.K., Krishnan, A., Grover, P., 2020.
diverse participant recruitment. Studies have revealed that approxi- Novel electrodes for reliable EEG recordings on coarse and curly hair. In: Proceedings
mately 80% of global research participants in neuroscience and psy- of the Annual International Conference of the IEEE Engineering in Medicine and Biol-
chology belong to wealthy, Western, and well-educated societies, and ogy Society doi:10.1109/EMBC44109.2020.9176067, EMBS, 2020-July, 6151–6154.
Farah, M.J., 2017. The neuroscience of socioeconomic status: correlates, causes, and con-
biases favoring white research participants have been well documented sequences. Neuron 96 (1), 56–71. doi:10.1016/j.neuron.2017.08.034.
(Henrich et al., 2010; Konkel, 2015). The conclusions made by cognitive Gilmore, R.O., Diaz, M.T., Wyble, B.A., Yarkoni, T., 2017. Progress toward openness, trans-
neuroscientists have vast implications, and thus, generalizations about parency, and reproducibility in cognitive neuroscience. Ann. N. Y. Acad. Sci. 1396 (1),
5–18. doi:10.1111/nyas.13325.
human behavior must be cautiously applied. Without strict disclosure
Hall, S., van den Heever, D., Vinding, M.C., Morris, L., 2020. Investigating motor prepara-
regarding the population to whom the results of a given study apply, it tory processes and conscious volition using machine learning. [Preprint]. Neuro-
is injurious to assume that findings can generalize beyond the typical science doi:10.1101/2020.09.07.286351.
Henrich, J., Heine, S.J., Norenzayan, A., 2010. The weirdest people in the world? Behav.
white, wealthy, and well-educated participant.
Brain Sci. 33 (2–3), 61–83. doi:10.1017/S0140525X0999152X.
To support our appeal for diverse recruitment, we lastly call for pro- Hokett, E., Duarte, A., 2019. Age and race-related differences in sleep discontinuity linked
gressive solutions to address currently prejudicial methodological prac- to associative memory performance and its neural underpinnings. Front. Hum. Neu-
tices. It is widely recognized that the mesh electrode cap used in EEG rosci. 13. doi:10.3389/fnhum.2019.00176.
Isamah, N., Faison, W., Payne, M.E., MacFall, J., Steffens, D.C., Beyer, J.L., Krishnan, K.R.,
research discriminates against those with textured or non-white hair and Taylor, W.D., 2010. Variability in frontotemporal brain structure: the importance of
inhibits diverse participant recruitment (Etienne et al., 2020). These recruitment of African Americans in neuroscience research. PLoS One 5 (10), 1–6.
trends were supported by our exploratory analysis (see supplement). doi:10.1371/journal.pone.0013642.
Jones, L.M., Ginier, E., Debbs, J., Eaton, J.L., Renner, C., Hawkins, J., Rios-Spicer, R.,
Such inequities are rivaled by ongoing work in the field to develop cre- Tang, E., Schertzing, C., Giordani, B., 2020. Exploring representation of diverse sam-
ative solutions surrounding the EEG cap (Etienne et al., 2020) and we ples in fMRI studies conducted in patients with cardiac-related chronic illness: a fo-
support these and similar efforts. cused systematic review. Front. Hum. Neurosci. 14. doi:10.3389/fnhum.2020.00108.
Konkel, L., 2015. Racial and ethnic disparities in research sutdies. Environ. Health Persp.
While functional neuroimaging offers progressive insights into the (12) 123. doi:10.1289/ehp.123-A297.
physiological mechanisms underlying behavior, demographics of race Lee, C., 2009. Race” and “ethnicity” in biomedical research: how do scientists con-
and ethnicity have been largely neglected by the field of cognitive neu- struct and explain differences in health? Soc. Sci. Med. 68 (6), 1183–1190.
doi:10.1016/j.socscimed.2008.12.036.
roscience. Race may or may not have a confounding role and contribute
Loue, S., 2006. Defining race, ethnicity, and related constructs. In: Assessing Race, Ethnic-
to issues of replicability and generalizability, but such conclusions can- ity and Gender in Health. Springer US, pp. 25–53. doi:10.1007/978-0-387-32462-3_3.
not be substantiated without the corresponding participant data. With- Manly, J.J., Tang, M.X., Schupf, N., Stern, Y., Vonsattel, J.P.G., Mayeux, R., 2008. Fre-
quency and course of mild cognitive impairment in a multiethnic community. Ann.
out clear demographic disclosure, cohorts cannot be determined to rep-
Neurol. 63 (4), 494–506. doi:10.1002/ana.21326.
resent the general population, and perceptions regarding the effects of Misiura, M.B., Howell, J.C., Wu, J., Qiu, D., Parker, M.W., Turner, J.A., Hu, W.T., 2020.
race on cognition cannot be ascertained. Given the histories of the field, Race modifies default mode connectivity in Alzheimer’s disease. Transl. Neurode-
participant race/ethnicity demographics are critical for contextualizing gener. 9 (1), 1–15. doi:10.1186/s40035-020-0186-4.
Morita, T., Asada, M., Naito, E., 2016. Contribution of neuroimaging studies to under-
contemporary research by exposing biased methodologies and support- standing development of human cognitive brain functions. Front. Hum. Neurosci. 10
ing investigations into their confounding effects on cognition. (SEP2016), 1–14. doi:10.3389/fnhum.2016.00464.
National Institutes of Health. (Updated June, 2018). Transition from Inclusion man-
Data and code availably statement agement system to new human subjects system (HSS) (NIH Notice Number:
NOT-OD-18-179). Retrieved from https://grants.nih.gov/grants/guide/notice-files/
The demographic data for the studies used to support the findings of NOT-OD-18-179.html.
Open Science Collaboration, 2015. Estimating the reproducibility of psychological science.
this commentary are available online on the NeuroImage [https://www. Science 349 (6251). doi:10.1126/science.aac4716, aac4716–aac4716.
journals.elsevier.com/neuroimage] and Cerebral Cortex [https:// Roberts, D., 2011. Fatal invention: how science, politics, and big business re-create race
academic.oup.com/cercor] webpages. Census data was obtained from in the twenty-first century. Press/ORIM, New.
Sauce, B., Matzel, L.D., 2013. The causes of variation in learning and behavior: why in-
the United States Census Bureau webpage [https://www.census.gov/ dividual differences matter. Frontl. Psychol. 4, 1–10. doi:10.3389/fpsyg.2013.00395,
data.html]. All webpages where data was obtained is open to the public. JUL.
Yellow Bird, M., 1999. What we want to be called: indigenous peoples ’ perspectives on
Funding racial and ethnic identity labels author ( s ): Michael Yellow Bird source. Am. Indian
Q. 23 No. 2 ( Spring , 1999 ), pp. 1-21 Published by: University of Nebraska. American
This research did not receive any specific grant from funding agen- Indian Quarterly, 23(2), 1–21. 10.2307/1185964.
Zsembik, B.A., Peek, M.K., 2001. Race differences in cognitive functioning among older
cies in the public, commercial, or not-for-profit sectors. adults. J. Gerontol. Ser. B 56 (5), 266–274. doi:10.1093/geronb/56.5.S266, Psycho-
logical Sciences and Social Sciences.
Declarations of Competing Interest (2019) NeuroImage, 184–203, All pages.
(2019) Cerebral Cortex, 26(1–12). All pages.
None.

You might also like