Cambridge Journal of Human Behaviour

Matthew Thomas Hilton

Methodological Challenges Faced by Researchers Studying Early Neurodevelopmental Outcomes in Majority Settings

ABSTRACT

In 2010, Henrich and colleagues highlighted that the majority of psychological research comes from a minor proportion of the global population (Minority countries). These countries are often Western, Educated, Industrialised, Rich, and Democratic. Over the past ten years, there has been a rise in research projects based in Majority settings (i.e., where most of the global population resides), such as the Brain Imaging for Global Health (BRIGHT) project (Katus et al., 2019). Many of these projects have used new technologies, such as functional near-infrared spectroscopy (fNIRS), to focus on cross-cultural differences in early neurodevelopment. This article reviews several studies and examines the issues researchers conducting studies currently face in Majority settings. The issues discussed relate to problems adapting task paradigms, technological issues, study design problems, cultural confounds, difficulties obtaining funding, and publishing papers. While much progress has been made over the past ten years, these limitations still hinder research on neurodevelopmental outcomes. However, at the same time, there is still a pressing need for research in Majority settings. Potential solutions are suggested, which, if implemented, may help to further our understanding of neurodevelopment in these areas. 

Keywords: early development, cross-cultural research, methodological challenges, WEIRD, Majority setting

Introduction

Methodological Issues Posed by Task Paradigms and Indices

Technological Issues

Issues Related to Study Design

Issues Related to Culture

Issues Related to Publication and Funding for Studies in Majority Settings

Conclusion

Article References

Commentary References

INTERDISCIPLINARY COMMENTARIES:

Methodology & Archaeology: Problems of scale and specificty: A commentary on methodological challenges faced by research studying early developmental outcomes in non-WEIRD settings (Betteridge, M. A.)

Natural Sciences: Culture across the animal kingdom: Insights into the study of human behaviour

PSYCHOLOGICAL & BEHAVIOURAL SCIENCES

Matthew Thomas Hilton
Emmanuel College, University of Cambridge
Volume 1, Issue 2, pp. 106–114

Received: December 5, 2022
Revision recieved: December 30, 2022
Accepted: December 31, 2022
Published: January 20, 2023

Preferred citation:
Hilton, T. M. (2023). Methodological challenges faced by researchers studying early neurodevelopmental outcomes in non-WEIRD settings. Cambridge Journal of Human Behaviour, 1(2), 106–114. https://www.cjhumanbehaviour.com/pbs0017

© Matthew Thomas Hilton. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial 4.0 License.

INTRODUCTION

At the beginning of the last decade, Henrich and colleagues (2010) argued that psychological research had been disproportionately conducted in Western, Educated, Industrialised, Rich and Democratic (WEIRD) settings. It was found that 96% of studies were performed in WEIRD settings, even though these areas only account for around 12% of the world’s population. Similar rates were found when analysing published articles in child development journals (less than 3% of articles included participants from Central America, South America, Asia, Africa or the Middle East; Nielsen et al., 2017). The last decade saw a rise in studies conducted in non-WEIRD settings, helping us to understand how cultures across the world differ socially, cognitively, and neurobiologically (Katus et al., 2019, 2020; Wijeakumar et al., 2019; Xie et al., 2019). This progress includes the Brain Imaging Global Health (BRIGHT) project, which studies early neurodevelopment in Gambian and British cohorts (Katus et al., 2019, 2020). Cross-cultural studies have become more feasible in part due to emerging technologies, such as functional near-infrared spectroscopy (fNIRS), which is more portable than similar measures such as functional Magnetic Resonance Imaging (fMRI; Blasi et al., 2019).

Recently, there have been calls to abandon the WEIRD/non-WEIRD dichotomy in favour of a view which emphasises differences between non-WEIRD settings (Ghai, 2021). Furthermore, there have been criticisms of how some of the terms in the WEIRD label, such as ‘democracy’, can be operationalised (Syed, 2021). Consequently, several researchers have begun using the terms ‘Majority World’ and ‘Minority World’. Majority World refers to the understudied regions where most of the global population live, and Minority World to heavily studied, typically Western, regions (Alam, 2008; Draper et al., 2022; Khan et al., 2022). Conducting research in Majority settings involves various methodological challenges. In this article, I discuss these issues and how we might be able to overcome them. For clarity, I group the challenges into five themes: (1) problems adapting task measures, (2) technological issues, (3) cultural confounds, (4) challenges related to study design, and (5) difficulties obtaining funding and getting papers published. However, many of these issues are interrelated, and a big-picture view should always be taken, since the best interventions will target multiple problems at once. 

METHODOLOGICAL ISSUES POSED BY TASK PARADIGMS AND INDICES

Many studies on early neurodevelopment in Majority settings have involved measuring behavioural responses to particular task paradigms that were created and initially applied in Minority settings (Katus et al., 2020; Wijeakumar et al., 2019). These task paradigms are important for understanding neurodevelopment in children aged between one and five years, since measures such as IQ are not accessible to young children (Milosavljevic et al., 2019). In order to ensure that these paradigms are accessible to those in Minority countries, these paradigms often have to be adapted (ibid.). Task adaptation often consists of translation into the relevant language for the culture being studied. Furthermore, it is often necessary to ensure that the task is comprehensible and relevant to the culture being studied. If the task is not appropriate to the culture, floor effects may be seen, since the task becomes harder for participants (Goodwin, 2016). Thus, it becomes harder to distinguish between individual performance on these tasks. Problems may also arise when comparing Majority samples results to Minority samples and to other Majority samples. For example, if a Majority sample performs poorly on a badly adapted measure, then it is likely that they will also perform worse than Minority samples, or Majority samples for whom the adaptations are sufficient. In this case, a difference in performance is unlikely to be caused by underlying psychological differences and may instead be due to a lack of cross-cultural validity, making it hard to infer any conclusions from such findings. 

Performing linguistic adaptations of task paradigms can often be achieved through collaboration with local researchers (Katus et al., 2019; Milosavljevic et al., 2019)—guidelines for performing translations have been published (Flaherty et al., 1988; Peña, 2007). Researchers can aid with pure translation, and ensure “functional, cultural and metric equivalence of the translated items” (Milosavljevic et al., 2019). Here, translation can prove to be a protracted process. In line with World Health Organisation guidelines, Milosavljevic and colleagues translated the Mullen Scales of Early Learning (MSEL; Mullen, 1995) in three stages (Sousa & Rojjanasrirat, 2011). First, they performed a consensus translation with a panel of three researchers. Then a local assistant reviewed the consensus translation and provided comments. These comments were sent back to the panel, who updated their translation appropriately. Another local assistant then provided a translation back into English (blind-back translation), which was compared to the original English version of the MSEL (Mullen, 1995). These protracted processes can limit the variety of ethnic groups researchers can work with. For example, the BRIGHT project studied a sample from The Gambia, where there is a multitude of ethnic groups (Katus et al., 2020). However, the available resources only allowed collaboration with a single ethnic group, the Mandinkas. Only collaborating with a small subset of the Gambian population constrains the generalisability of the results to other ethnic groups within The Gambia. 

Task adaptations also need to ensure that the paradigms used are appropriate to the Majority culture being studied. For example, Nimkar and colleagues (2020) attempted to adapt the MSEL for use in the southern and western regions of India. Many objects used in the MSEL—such as telephones, chimneys, and lamps—were not culturally appropriate for children from these areas. These objects had to be altered to ensure the paradigms were culturally relevant. For example, a line drawing of a lamp was altered to one of a bulb. This adaptation was considered suitable because both drawings represented objects from a similar category (a light source) in each culture. Other adaptations included replacing a chimney with a flag, as these are both common household fixtures in each culture. However, questions still linger about the cross-cultural validity, since the bulbs and flags were still presented as line-drawings, a phenomena of Western art, which are less common in other cultures. To ensure that these drawings were intelligible to children from Southern and Western India, these adaptations were reviewed by local psychologists, paediatricians, researchers, and healthy children, who all provided feedback. Thus, ensuring the cultural relevance of task paradigms can involve a multitude of local participants. 

Once a task has been adapted, various methods can be used to try and assess cross-cultural construct validity through measures of measurement invariance, such as confirmatory factor analysis and Rasch analysis (Prince, 2008). However, problems can still be found with indices that show adequate model fit based on measures of measurement invariance. For example, McCoy and colleagues (2016) used data collected using the Early Childhood Development Index (ECDI) to assess the prevalence of developmental deficits in infants globally. The ECDI showed stable confirmatory factor analysis results. However, the ECDI was also designed to be brief (10 questions long) such that it would be valid across a diverse range of contexts (McCoy et al., 2016). Consequently, they could not obtain estimates on the global prevalence of deficits in some cognitive subdomains, such as executive function and language. Studies may benefit from implementing multiple measures/indices: some which are more general and satisfy cross-cultural validity measures (such as the ECDI), as well as others which are more detailed and have undergone thorough adaptation (such as the MSEL). This could be rewarding, as it may help us to see how specific cognitive domains vary globally, allowing us to understand these domains better and perform better targeted interventions. 

Obtaining population data in Minority settings—particularly low- and middle-income countries (LMICs)—can also be difficult due to a lack of resources in these settings (McCoy et al., 2016). Thus, when McCoy and colleagues (2016) used cluster sampling from censuses in 35 different LMICs to estimate the global prevalence of impaired development, their results were only based on data from roughly 21% of the combined child population of all LMICs. While this sufficed to provide estimates of global deficits in neurodevelopment, few inferences could be made at country and regional levels. Even those studies targeting single countries fail to produce data that facilitates inferences at these levels. Take the Gambian example, discussed previously. The protracted processes involved in translating and adapting measures can limit the variety of ethnic groups one can work within a particular Majority country. For example, the available resources for the BRIGHT project only allowed for collaboration with a single ethnic group, the Mandinkas (Katus et al., 2020). By only collaborating with a small subset of the Gambian population, the generalisability of the results to other ethnic groups within The Gambia was constrained. In order to obtain country-level data, adequate funding must be provided for research at this level. One step towards this goal would be to inform members of funding bodies of the methodological challenges posed when “collecting heterogenous data” in Minority settings (Nielsen et al., 2017). 

TECHNOLOGICAL ISSUES

Projects around the world have now begun to implement techniques such as eye tracking (Katus et al., 2019), fNIRS (ibid.; Perdue et al., 2019), and electroencephalography (EEG; Jensen et al., 2019; Katus et al., 2019; Shephard et al., 2019). These techniques have found differences in brain function and structure in cohorts as young as four to six months old. Behavioural measures, instead, only show altered developmental pathways from roughly two years of age (Lloyd-Fox et al., 2016). However, these methods also present challenges to researchers studying in Majority settings (Blasi et al., 2019; Katus et al., 2019). These constraints are less the result of cultural difference; instead, they arise from poorer infrastructure in non-WEIRD settings. This highlights that researchers have to face different kinds of challenges when studying in these regions. For example, Katus and colleagues (2019) used an eye-tracking paradigm in The Gambia. However, due to poor lighting conditions in the laboratory (that were beyond researchers’ control), the sampling rate had to be lowered from 120Hz to 60Hz. The authors assumed that, as the sampling rate and shutter speeds of the eye tracker is lowered, less visual noise is acquired in the eye image. In line with their experience, poor lighting conditions appeared to compound the visual noise acquired with higher sampling rates. However, while empirical research on optimal sampling rates covers interference due to head movement (e.g., Hessels et al., 2015), there is little empirical evidence to indicate which sampling rates are best when interference is the result of poor lighting conditions. Ultimately, this solution did allow the necessary data to be collected by Katus and colleagues; however, should higher-quality data be required, it would be recommended that projects attempt to implement stable lighting conditions in their laboratories. At the same time, it would be helpful for groups such as Katus and colleagues to make their data on the optimal sampling rates in different lighting conditions available so that other researchers do not have to perform similar preliminary research.  

One particularly common brain recording measure that is being implemented in Minority settings is fNIRS, which allows us to measure correlates of brain activation (Blasi et al., 2019). fNIRS resembles more conventional techniques for recording correlates of brain activation such as fMRI, with “good” correlations being found between both measures (Alderliesten et al., 2014). However, fMRI scans have been very hard to implement globally, due to the scanners’ size and power requirements. Furthermore, fMRI scans are hard to perform on infants, due to the loud noise they make and the requirements to keep still. Thus, they lack utility for studying early neurodevelopment (Ivry et al., 2018). Contrary to fMRI, fNIRS allows for slight movement and is much more portable than fMRI scanners as they can be dismantled and reassembled in under four hours (Katus et al., 2019). Thus, this technique is useful for studying infants in Majority settings, and highlights how emerging technologies can greatly aid global neuroimaging projects. Similarly to fNIRS, electroencephalography (EEG) allows for both movement and portability (Katus et al., 2020). However, unlike fNIRS, EEG has high temporal specificity and low spatial specificity, meaning that the conclusions that one can draw using this technique are of a different nature (Ward, 2019). 

However, there are still problems with using fNIRS and EEG. Firstly, the atlases which are used to map activation onto specific brain regions have been created using Western children, and so may not account for global morphological differences in brain structure (Wijeakumar et al., 2019). Consequently, the accuracy of any conclusions drawn is vastly impaired. A future direction for research is to create different atlases for different ethnic groups. This undertaking would require vast amounts of structural brain imaging to be collected, and so it is likely that this problem will remain for some time (Ivry et al., 2018). 

Secondly, the climates of some Majority settings can be hotter and more humid than in typical Minority settings (Katus et al., 2019, 2020). These climates can lead to participants sweating, which in turn can lead to misalignment of headgear. Consequently, recordings must take place in air-conditioned rooms, a rarity in some Majority countries. Furthermore, air conditioning can create electrical noise which interferes with recording methods. Most studies account for these two opposing problems by only turning on air conditioning before any measurements are taken. However, this can still lead to sweating during later stages of recording. This problem thus highlights how researchers in Majority settings may have to tackle conflicting issues to collect sufficiently high-quality data. Sweating artefacts can also be removed using filters (Katus et al., 2020); however, such a technique could filter out relevant data. 

A further issue is that recordings can sometimes be distorted by braided hair, which can be common in some Majority countries, such as The Gambia (Katus et al., 2019). While parents can be asked to untie their children’s braids, this is sometimes not possible, requiring data to be collected from neighbouring regions to those being studied. Furthermore, dark skin and hair may also interfere with fNIRS readings, as darker skin and hair absorb more light (ibid.). However, if relative changes in blood flow are measured (i.e., changes recorded from the same participant), then the data is not affected because the interference from darker hair and skin is constant.  

These issues highlight that, although useful, brain recording techniques still face problems when used in Minority settings. However, these issues should not put researchers off, as such studies have advanced the field of early development. For example, Katus and colleagues (2020) measured auditory event related potentials (ERPs) in children from both the UK and The Gambia. Both cohorts showed an association between an auditory evoked potential and neurodevelopmental outcome, despite the vast cultural differences between both cohorts. Thus, they concluded that such ERPs could potentially be used as an objective indicator of neurodevelopmental outcome in the future in a wider range of settings than previously thought. Furthermore, these markers provide insight into a child’s development from five months of age, which is earlier than behavioural assessments, and would allow for earlier intervention. Such studies highlight the importance of cross-cultural research. 

ISSUES RELATED TO STUDY DESIGN

The design of many studies in Majority settings can lead to confounds in the data. Wang (2016) highlights how many cross-cultural studies are designed to study group differences between two different cultures. Consequently, many neurodevelopmental studies in Majority settings pool the data from individual participants within each ethnic group, to create a group average (Jensen et al., 2017). As Jensen and colleagues highlight, this approach can “hide the true effect of risks or interventions” (ibid., p. 233) because it does not take into account the heterogeneity of the individual participants within the sample. They give the example of an intervention performed by Morgan and colleagues (2017), which did account for such heterogeneity, by dividing their South African sample into groups with different polymorphisms of the SLC6A4 gene. The effects of their intervention differed vastly between groups, revealing an interaction effect that would not have been found had they averaged the whole sample’s data. Thus, it is important for researchers in Majority settings to not only study different ethnic groups to assess cross-cultural differences, but also to study those in Majority settings in their own right and investigate individual differences (Wang, 2016).

Furthermore, the design of studies performed in Majority settings needs to take into account interactions between confounding factors. Many early neurodevelopmental studies in Majority settings have focused on risk factors leading to lower neurocognitive outcomes (e.g., Caspi et al., 2016; Hackman & Farah, 2009; McEwen & Gianaros, 2010). However, many of these risk factors and outcomes are highly interrelated. For example, HPA-axis activation is correlated with impaired executive functioning, language, and memory in children (Gunnar & Nelson, 1994; Sameroff et al., 1993). Furthermore, HPA-axis activation has been shown to mediate the associations between risks such as low income and cognitive abilities (Blair et al., 2011). These correlations lead to multicollinearity, where factors are so highly correlated that it is hard to discern the individual input of each factor towards an outcome (Jensen et al., 2017). In the broader psychological literature, ways to tackle problems of multicollinearity can include randomised control trials (RCTs), where the variation of one independent variable is used to measure the effect on the dependent variable (Goodwin, 2016). Such trials help us to better infer causal relations. RCTs can be used in certain cases in Majority settings, by comparing the effects of an intervention to a control group (e.g., Stewart et al., 2022). However, even intervention studies may not be able to control well for potential multicollinearity. For example, Walker and colleagues (2005) studied the effect of increased cognitive stimulation on children’s cognition in Kingston, Jamaica, through visits by healthcare workers who facilitated play between child and mother. A significant effect of stimulation on child cognition was found. However, it is uncertain if the intervention had a direct causal effect on cognition or an indirect one. The intervention also had a significant effect on lowering maternal depression, and it may have been this factor that contributed to the effect on child cognition. Ultimately, it is likely that both direct and indirect pathways contributed to this effect (Jensen et al., 2017). Consequently, researchers in Majority settings should not only try to perform intervention-based studies where possible, but they should also attempt to measure a range of factors to best understand causality in these settings. 

As Wang (2016) argues, one of the important benefits of cultural psychology is that it allows us to discover trends and phenomena that cannot be uncovered by only performing research in Minority settings. However, the assumption that cultural psychology should be comparative in nature is still common (ibid.). This assumption limits the factors studied to those that inherently transcend cultures and overlook traditions, unique to a specific culture. For example, a study by Stewart and colleagues (2022) performed an intervention to lower maternal depression in the Kanyelang group in The Gambia. Maternal depression has been linked to negative outcomes in offspring, such as internalising problems (Betts et al., 2013) and elevated cortisol levels (Barry et al., 2015). The Kanyelang group has a rich cultural tradition of performing participatory music and dance in order to help prevent issues related to fertility (McConnell, 2020). Stewart and colleagues (2022) performed ethnographic research to understand how this tradition could be implemented to reduce perinatal stress and found a significant decrease in stress symptoms. The authors note that, due to the high specificity of this tradition, the exact intervention is unlikely to generalise to other ethnic groups (although similar musical interventions may work). Consequently, as Wang (2016) argues, it appears important that we move away from the assumption that cultural psychology should be comparative in nature, in order to “uncover mechanisms that are unique to non-Western populations” (p. 591).  

ISSUES RELATED TO CULTURE

Cultural issues can also pose challenges to researchers studying in Majority settings. For example, the indices employed in research may measure development based on Western norms. For example, studies suggest that some tests, such as IQ tests, are culturally specific (Sharma, 1971), and that cultural background potentially contributes to variation in results. Kleinman (1987) argues that qualitative, anthropological studies can provide the necessary insight into a culture to ensure that Western measures correlate with how other cultures interpret specific phenomena. Kleinman gives the example of a study performed by Manson (1995), who translated the Diagnostic Interview Schedule (DIS) into Hopi, a Native American language. The DIS combines several concepts (guilt, shame, and sinfulness) which are understood to be wholly separate within the Hopi ethnic group, and to combine them would have led to categorical fallacies within the data, according to Hopi healthcare professionals. While this study was performed on adults, similar problems may arise when applying concepts from research in early neurodevelopment to different cultures. Qualitative research can help elucidate how Majority cultures view certain concepts in relation to the predominant Minority classifications, to ensure that such fallacies are not made. An even better solution is to promote research design within Majority settings, rather than in the Minority settings where most research is published (Barrett, 2020). When research is designed in Minority settings, it often imposes a Minority worldview onto Majority cultures (Sternberg, 2004). However, such an endeavour seems difficult, due to the low number of psychologists performing research in certain Majority settings and their poor connectivity with other researchers, constraining their ability to collaborate (Quayle & Greer, 2014). 

Other factors that need to be considered when performing research in Majority settings, especially when performing so-called “helicopter studies” (where researchers perform projects in foreign countries; Haelewaters et al., 2021) include observance of religious holidays and how participants will react to foreign researchers. For the former, Katus and colleagues (2019) had to ensure that, during Ramadan, recordings and tasks were performed just after sunrise when fasting participants were most alert. For the latter, Katus and colleagues (ibid.) also had to ensure that, when working with infants who were 12 months old or older, tasks were completed by local researchers since the infants were typically fearful of the Caucasian staff. 

ISSUES RELATED TO PUBLICATION AND FUNDING FOR STUDIES IN MAJORITY SETTINGS

Obtaining funding can be difficult even when researchers have designed a study with the relevant task adaptations, and have accounted for all of the potential technological, cultural, and study design issues (Draper et al., 2022). Furthermore, even if researchers obtain funding, and have written a manuscript, it can still be hard to get an article on cultural psychology published, as Wang describes:

Not long ago, I submitted an article to a journal specialized in my area of research. … [The action editor] informed me that she had decided not to send the manuscript out for review because cross-cultural research did not fit well enough with the goals of the journal. Really? I could not believe my eyes. I took another look at the journal’s Aims and Scope, which clearly states that the journal covers “human” conceptual processes, memory, learning, problem solving, and more. Ironically, all these aspects of cognition have been shown to be susceptible to cultural influences. If the journal indeed tries to exclude cross-cultural research, then “WEIRD human” … would be a more accurate description in its Aims and Scope. I wrote back and copied my e-mail to the editor-in-chief, hoping to engage them in a constructive discussion, but never heard back from either. (Wang, 2016, p. 583)

One particular problem is that many Majority countries lack funding for this kind of research, with financial resources being allocated to other pressing needs, such as infrastructure (Draper et al., 2022). Consequently, many psychological research projects in Majority countries require funding from international sources and/or collaborations with researchers from Minority countries (ibid.). Funding from Minority countries may be hard to obtain. Many of the hurdles discussed in this paper make research in Majority settings expensive, particularly because it involves cross-disciplinary research, such as anthropological assessments. Tancock (2017) has noted the impact of unconscious biases within editorial teams, and it may be that similar biases exist within funding bodies, although empirical research on this subject is yet to be conducted. Consequently, obtaining funding from Minority countries can often be difficult. 

Even if funding from a Minority country is obtained, the resultant power dynamic affects the outcome of the study. For example, local researchers involved in projects often struggle to get authorship for any resultant papers (Morton et al., 2022). As many have described, psychological academia suffers from a ‘publish or perish’ culture (Goodwin, 2016). Consequently, a lack of authorship for local researchers can limit their career trajectories, and this issue may further hinder the ability to obtain funding in Majority countries. Haelewaters and colleagues (2021) discuss ways to initiate more equal collaborations between researchers from Majority and Minority countries. Examples include designing studies with local researchers from the outset, to ensure more equal power dynamics within the research group. Furthermore, such collaborations may help Minority researchers to anticipate and navigate administrative and political issues, since local researchers are more likely to understand these issues better. 

Finally, even if the study is completed, there can still be issues getting papers published. Draper and colleagues (2022) note how journals appear to hold research from Majority countries to much higher standards, expecting researchers to explain the context of their studies in much more detail than research from Minority countries. They also discuss ways to combat publication bias, suggesting measures such as hiring editors from Majority countries, and publishing statistics annually on where their articles come from.

CONCLUSION

In conclusion, researchers in Majority settings face various methodological challenges including, but not limited to, problems with adapting measures, technological issues, cultural confounds, study design issues, and funding and publication difficulties. Thus, researchers may face significant hurdles at every step of the research process. First, funding must be obtained, which often has to come from Minority countries. Second, they must engage with the culture they choose to research, to ensure that the constructs they intend to research are valid. Third, the tasks they implement must be sufficiently adapted to the relevant culture. If researchers implement methods such as fNIRS and EEG, they must ensure that the labs have the necessary conditions (such as good lighting and air conditioning). Finally, once the research is conducted, a journal has to be found to publish the article. While some steps have been taken to mitigate some of these issues, progress is still necessary in other domains. However, these issues should not deter researchers from conducting research in non-WEIRD settings, as highlighted by the vast wealth of findings already generated by such cross-cultural research. 

INTERDISCIPLINARY COMMENTARY

METHODOLOGY & ARCHAEOLOGY

Problems of scale and specificity: A commentary on methodological challenges faced by researchers studying early developmental outcomes in non-WEIRD settings
Michael Allen Betteridge
Jesus College, University of Cambridge

Hilton’s article engages with some challenges for research projects that span multiple cultures with admirable nuance and coherence. Despite this, some issues with generalisability and cultural specificity remain unaddressed. In this commentary, I will note where these issues emerge and how they reveal limitations in the ideas Hilton presents. Unfortunately, I am not able to resolve these issues, which reflect some of the most pernicious and unresolved issues in modern anthropology. Instead, this commentary functions to highlight some places where the author’s argument could be extended to incorporate a more complete review of cross-cultural research.

Introduction
Hilton’s article adeptly confronts many of the most challenging aspects of cross-cultural comparison. It highlights how many Western-developed practices fail to consider the immense cultural variation between Western, Educated, Industrialised, Rich and Democratic (WEIRD; Henrich et al., 2010, as cited in Hilton’s article) and non-WEIRD societies, and between different non-WEIRD societies. It is important to elaborate on some areas considered by Hilton, and I will suggest some further considerations in this commentary. Specifically, when discussing the practices of a study conducted by the Brain Imaging for Global Health (BRIGHT) project in The Gambia, Hilton explores how only focusing on one group (the Mandinka) limited the degree to which the study could be generalised to the entire Gambian population. This raises questions about the degree to which research should aspire to be culturally generalisable, and while Hilton returns to this point in their final section, some important questions remain undiscussed. Secondly, this article, especially in its final section, touches on challenging notions regarding the usefulness of cross-cultural value scales, which, when explored further, may reframe much of the article’s core debate. In this brief commentary, I intend to build on Hilton’s discussion of these two deeply interlinked issues and provide some directions from different fields to initiate a wider discussion.

Generalisability
It is important to note, as Hilton does, that research conducted exclusively on Mandinka individuals cannot be generalised to the entire Gambian population. But this is not the end of the discussion. For example, social anthropologists have noted that in many communities, Western-initiated studies are more likely to focus on individuals with greater social power, and disproportionately focus on men (Milton, 1979). This is evidenced by some anthropological studies, in which wealthy older men were observed to interact substantially more frequently with researchers than young women. This undermined the usefulness of the anthropological research by generating conclusions that significantly favoured one section of a community over another (Reita, 1975). The most pernicious thing about this “double male bias” (ibid., p. 13), as it has sometimes been called, is that the degree to which individuals are representative of the greater population of interest is often not clear to researchers from outside a community. While some power structures might be easier to identify even in unfamiliar cultural contexts, such as patriarchy and class hierarchy, others, such as moiety and caste systems, are often extremely difficult to identify unless one is immersed in a cultural system (Evans, 2003).

Archaeologists have noted the degree to which it is extremely difficult to identify these structures in unfamiliar cultures, and this habitual oversight is a trap that anthropological, psychological, and medical researchers often fall into (Moore, 1995). If we imagine that the example used in this article took place in the indigenous Tlingit community of Northwestern Canada, the question would not only be whether Tlingit results can be generalised to apply to all of Canada, but also whether results from the Raven or Eagle moiety can be generalised to apply to all of Tlingit society (De los Santos et al., 2019). Cultural sub-groupings like these are common and can often be extremely difficult for researchers to understand, even if local informants are consulted (Berreman, 2012). This limited ability to decipher the nuances of societal power structures and systematic variation within a culture further complicates any attempts to generalise results and explain the tendencies of a society as a whole. Hilton suggests working closely with local researchers to consider how materials can be adapted to specific cultural contexts, but this is not a solution if those researchers come from backgrounds that, unbeknownst to Western researchers, overwhelmingly favour one sub-community grouping. An alternative way to resolve this challenge may be to accept that research like this is difficult to generalise at all, but that it can be combined with a wider body of research to better explain phenomena on a larger scale. In practice, this means that while the study which exclusively worked with Mandinka individuals cannot be used to produce generalised results about Gambian society, when combined with other studies, including those highlighting the disparity between Mandinka and non-Mandinka data, it can play a part in establishing a broader understanding of Gambian society.

Cultural specificity
In its final section, Hilton’s article questions whether it is appropriate to attempt cross-cultural studies in fields like neurodevelopment at all, or whether the results are so heavily swayed by cultural background that comparing cultures is impossible. This directly addresses a challenge that looms over the entire concept of cross-cultural research. While I ultimately agree that cultures should be studied “in their own right”—or, perhaps, “on their own terms”—some further challenges should be considered when this point of view is adopted. Ultimately, Hilton’s argument is that rather than trying to rank results from enormously different cultural contexts—Thailand and Germany, for example—on the same scale, it may be more beneficial to analyse Thai or German data in isolation. The first issue this raises is essentially the issue of scale I discussed above. It is totally arbitrary to base an intellectual boundary, the point where comparison ceases to be useful, on national boundaries. One could use data gathered in Bangladesh, with a Bangladeshi scale and Bangladeshi parameters, and still unfairly compare members of indigenous Adivasi communities to those from highly globalised communities in Dhaka. I will not belabour this point any further, but it is important to consider that if the aim is to study cultures “in their own right”, we must have a boundary of where one culture ends, and another begins. Either that boundary is generated in the mind of Westerners, in which case it undermines any attempt to study cultures in their own right, or it is defined by members of a community who may intentionally deny or downplay the existence of sub-cultural divisions. Any notion that cultures are firmly defined, discrete entities falls into one of these two traps, and, therefore, advising that cultures are studied “in their own right” is, at best, highly optimistic. 

A second, perhaps more hopeful, question that the article raises is that different cultures have different parameters for concepts like intelligence, disability, or even success. As noted in the article, conditions like deafness are not always understood as disabling within cultures. Similarly, markers that are considered to be signs of neurodevelopmental success or failure in WEIRD countries may not be understood in the same way globally. With this in mind, communities and individuals—particularly those who are neurodiverse—should be empowered to define development and success on their own terms instead of being forced into one standard Western-led model of development by administering Western tests that prioritise Western ideals of development. This does not rob researchers of the ability to identify neurological differences between individuals and communities, but it does temper the Western cultural tendency to ascribe moral judgements to different trajectories of neurodevelopment. In other words, the fact that individuals from some cultures score ‘lower’ in Western-designed tests does not mean that they have performed ‘worse’ than peers in their own cultural context. Even though the study in Bangladesh did not compare its results to Western cohorts, it did rank participants based on ideas of “developmental potential” (Xie et al., 2019, p. 1), which is a culturally-specific concept. Culturally-specific concepts like this may not have been equally meaningful for all participants involved for a range of cultural reasons, and to truly study cultures “in their own right”, researchers should allow metrics such as success and “impairment” to be defined by individuals in their own terms, rather than imposing Western parameters.

Conclusion
Overall, Hilton should be commended for constructing a survey of the main challenges of cross-cultural research projects. The idea that it is often inappropriate to compare test results gathered from vastly different cultural contexts is useful, but requires elaboration to consider the degree to which cultures are not discrete, clearly delineated entities separated by international borders. Since specific cultures are hard to identify, isolate, and understand without oversimplifying, the idea of studying cultures “in their own right” is, while a valiant objective, extremely hard to actually achieve. To tackle these issues more completely, it is important to consider how different research projects can require vastly different levels of cultural specificity depending on their scale, from individual to global.

NATURAL SCIENCES

Culture across the animal kingdom: Insights into the study of human behaviour
Georgina Scott
Christ’s College, University of Cambridge

Culture influences almost every aspect of human behaviour, and yet most psychological studies are performed on only a very small subset of the population, causing only one culture—that of so-called WEIRD populations—to be meaningfully represented. Here, I discuss the meaning of culture in an ecological context, and present examples of culture seen throughout the animal kingdom, in an effort to understand why culture is so important in the study of human behaviour, as well as the challenges it presents therein.

Introduction
Hilton’s article discusses the material difficulties faced by psychological researchers attempting to perform studies outside the typical, WEIRD (Western, Educated, Industrialised, Rich, Democratic)  settings, from which most psychological data is collected (Henrich et al., 2010 as cited in Hilton’s article). This is a problem, as pointed out by Henrich and colleagues, not only because general psychological conclusions are often drawn from this small pool of data, but also because this specific group of people, far from representing the wider human population (as is often assumed by these studies), has actually been shown to be an outlier in many behavioural metrics when compared to people from a range of different cultures—for example, their being more susceptible to “perspective drawing” optical illusions (Segall et al., 1966). It is therefore imperative that more studies are performed outside of this narrow testing range to obtain data that is more representative of the human population throughout the world.

This issue highlights a common problem in the study of human behaviour, particularly in a biological context: how can innate human behaviours be separated from the cultural context? In other animals, certain behaviours have been found to follow simple Mendelian inheritance—for example, the shape of the burrows dug by deer mice (Weber & Hoekstra, 2009). However, in humans, the situation is much more complicated. While culture certainly exists in other animals to varying degrees, nowhere is it as prevalent as in humans: almost all of our actions and behaviours are influenced by culture, and attempting to remove these contextual elements to find the underlying biological bases of different behaviours has proved extremely challenging. Here, I will discuss the nature of culture in the animal kingdom and discuss why culture presents such a challenge to the study of human behaviour.

What is culture?
While the term culture can have many different meanings in many different contexts, in evolutionary biology we define culture as the inheritance of behaviours via social learning (Whiten, 2021). The most commonly discussed form of inheritance is genetic—or “hard”—inheritance, which can and has persisted for millions of years. However, it is not the only form of inheritance. There also exist forms of “soft” inheritance. These are not as long-lasting as genes, but they can still have a significant impact on evolution, and indeed can allow rapid adaptation to new or changing conditions. Forms of soft inheritance include property (Strassmann & Queller, 2014), beneficial symbionts (e.g., Ledón-Rettig et al., 2018), and, of course, culture. This form of culture often occurs as the acquisition of behaviours in younger generations by the mimicking of older generations (such as parents). In this way, behaviours can persist within a population without any genetic basis and, crucially, without the need for natural selection. This means that, while highly detrimental behaviours are unlikely to persist, cultural behaviours do not have to increase survival and reproduction in order to be maintained – they are maintained within the social group simply by continued imitation.

In his book The Selfish Gene (Dawkins, 2016), well-known evolutionary biologist Richard Dawkins coined the term “meme” as a unit of cultural inheritance. These memes are perpetuated— “inherited”—from person to person in their minds and thoughts, and there is a form of selection for those ideas which are most able to propagate themselves, be that by being especially interesting, profound, funny, ridiculous, captivating, or having any other reason to be remembered and transmitted above other ideas. While natural selection on genes means that these have to improve the survival and reproduction of the resident organism, culture does not have such limitations. While it certainly can be, and indeed is, passed down generations from parent to child, culture can also be “inherited” laterally, i.e., between any two individuals of the species. The only requirements for transmission are the physical and cognitive capacities to copy and learn the behaviour (which will generally be present in all members of the species) as well as an inclination to do so.

This means that culture can interact in various ways with genetic evolution: it can work alongside it, to increase the rate of survival and reproduction of organisms; it can have a neutral effect, neither beneficial nor harmful to the organism; or it can have a detrimental effect. Indeed, Dawkins points to an intriguing example of the latter in his book—that of celibacy in monks. This behaviour, by definition, could not be passed down by genetic means, as it would be erased from the gene pool within a single generation; however, it can pass via cultural means between non-genetically related members.

Culture throughout the animal kingdom
While humans have developed extremely complex cultures to an extent that has not yet been reflected in the rest of the animal kingdom, culture itself is not a unique human trait. If we define culture as behaviour passed between individuals via social learning rather than genetic means (Whiten, 2021), then we can find many examples of this phenomenon in the wild.

Cetaceans, in particular, appear to show a great propensity for creating cultures. For example, Allen and colleagues (2013) analysed the spread of lobtail feeding throughout whale populations. This form of hunting was an innovation first documented in 1980, and involves the striking of the tail on the water surface several times, before continuing with the more usual bubble feeding behaviour which whales had previously been known to engage in. The prevalence of this behaviour rose sharply throughout the following decade. In the study, a network-based diffusion analysis provided strong support for a social learning component driving the rapid spread of this behaviour, suggesting a strong influence of culture. They also found that spatial and temporal distribution of sand lance was a significant predictor of whales acquiring this hunting technique, suggesting that it aided in the adaptation to this particular prey type, with those individuals who learn the behaviour having a fitness advantage over those who do not. Hence, we can see how culture can be adaptive in nature, allowing new behavioural innovations to spread through a population without the need for genetic evolution (which would likely be much slower). 

We can even find evidence of cumulative culture outside of the human sphere—that is, socially learnt behaviours which are continually improved upon in each generation, leading to an overall increase in knowledge and adaptiveness of the behaviour over time. This has been found, for example, in the tracking of ungulate migrations across North America, particularly in bighorn sheep and in moose (Jesmer et al., 2018). By comparing the migratory propensities of historical populations who had inhabited certain areas (prior to local extinction) and those of reintroduced populations, the researchers found that the propensity of the newer populations was initially much lower than that of the historical populations. However, over time, this propensity increased as the groups learnt to “surf” green waves, i.e., tracking plant phenology across the landscape to maintain high-quality foraging throughout the year. The increase in collective knowledge over time is indicative of cumulative culture and highlights that there is more to these animals than just their genetic make-up.

Not only is culture found in mammals, but also across the animal kingdom. Culture has been found in the form of variations in birdsong (Marler & Tamura, 1964), predator recognition in fish (Mathis et al., 1996), and even mate choice in fruitflies (Danchin et al., 2018). However, no animal has shown more evidence of culture than our closest relative, the chimpanzee, as shown by Whiten and colleagues (1999). This meta-analysis of several studies found evidence of 39 cultural behaviours (i.e., practices which are specific to certain populations, with differences not attributable to ecological conditions), more than any other recorded species. Interestingly, this list includes tool use, an arguably important feature of human evolutionary history. Evidence for seemingly non-adaptive cultural behaviours has also been suggested, such as “grass-in-ear” behaviour (van Leeuwen et al., 2014), which entails individuals inserting a stiff blade of grass into one of their ears that they leave in as they proceed with other activities. This highlights the fact that, while culture certainly can be adaptive or enable adaptation, cultural behaviours do not have to be adaptive; there may not be an evolutionary explanation for every behaviour.

Studying human behaviour
While culture does play a significant role in behavioural traditions throughout the animal kingdom, it is particularly influential in humans. Indeed, it seems that our culture has grown to the point that it has more influence on our evolution than genetics, and this influence is only growing over time (Waring & Wood, 2021).

Both within the field of psychology and in popular culture, a common tool to explain human behaviour has been evolutionary psychology. This is a school of thought which suggests that our behaviours have been shaped by evolution in response to the environments and circumstances faced by our ancestors (Smith et al., 2008). However, this approach often fails to take cultural influences into account, either by overlooking them or by use of methodologies which do not capture the variation between cultures. For example, in the 1960–1970s, psychologist Paul Ekman declared that he had found six universal human emotions through the study of people from different cultures, asking them to assign the appropriate facial expression to each emotion (Ekman et al., 1969). This appeared to support Darwin’s (1872) idea of discrete, universal facial expressions. However, more recent studies have not corroborated the universality of these emotions across different cultures, finding Ekman’s study to place constraints on participants’ answers by only allowing them to select emotions from his chosen categories (Gendron et al., 2018). Other studies have since found that participants from non-Western cultures, when asked to freely describe the emotion shown in a photo, would rarely come up with Ekman’s supposedly “universal” emotions of their own accord (e.g., Crivelli et al., 2017).

Hence we can see that methodology is very important when studying human behaviour, as researchers will have their own cultural biases; they may unconsciously assume that what they perceive as “normal” must be universal, something which may then colour their research.

Conclusion
What is clear, in the end, is that culture has played a huge role in the evolutionary history of modern humans and our relatives. With its presence even in Protostomata, from whom humans diverged around 670 million years ago, to its startling abundance in the mammalian clade, particularly within our own primate lineage, the importance of culture is undeniable.

While animals are often thought of as being something of a monolith in their behaviour, with no conscious thought or decision-making, we can see that in some cases there is a cultural influence on individual behaviours. The extent of this influence is increased in humans, naturally following the complexity of culture. It is for this reason that the study of innate human behaviours—including early neurodevelopment—is so difficult, as removing the cultural context is almost impossible, and the methodology can be biased by the researcher’s own cultural perceptions of ‘normality’. This is why it is important to perform psychological studies on people from many different cultures all around the world with careful consideration of methodology, and why solving the challenges discussed in the paper which this commentary accompanies is so important.

ARTICLE REFERENCES

COMMENTARY REFERENCES

Henrich, J., Heine, S. J., & Norenzayan, A. (2010). Most people are not WEIRD. Nature, 466, 29. https://doi.org/10.1038/466029a  

Milton, K. (1979). Male bias in anthropology. Man, 14(1), 40–54. https://doi.org/10.2307/2801639 

Moore, J. (1995). The archaeology of dual organization in Andean South America: A theoretical review and case study. Latin American Antiquity, 6(2), 165–181. https://doi.org/10.2307/972150 

Xie, W., Kumar, S., Kakon, S. H., Haque, R., Petri, W. A., & Nelson, C. A. (2019). Chronic inflammation is associated with neural responses to faces in Bangladeshi children. NeuroImage, 202, Article 116110. https://doi.org/10.1016/j.neuroimage.2019.116110