Text by Marion Grimberg (Freie Universität Berlin)
© Pablo Dohms
"Behavioral scientists routinely publish broad claims about human psychology and behavior in the world’s top journals based on samples drawn entirely from Western, Educated, Industrialized, Rich, and Democratic (WEIRD) societies.” (Heinrich, Heine, and Norenzayan 2010, 1; emphasis mine).
With this sentence, the psychologists Heinrich, Heine, and Norenzayan pinpoint the critique I often heard about behavioural sciences during my anthropological studies. Ten years ago, the three psychologists Heinrich et al. wrote an inspiring but alarming article about sample composition of and generalisations in behavioural experiments. It triggered further thoughts[1] from a wide range of disciplines, such as economics, primatology, and anthropology. Taking an anthropological and constructivist standpoint, I will add to this debate by reflecting upon what has been taken for granted in behavioural sciences and ask where these blind spots could come from.
As Heinrich, Heine, and Norenzayan (2010) showed, researchers from behavioural sciences[2] mostly recruit test subjects on campus for their experiments, which leads to a very narrow sample diversity in the database. Most of the time results from studies conducted with such homogenous samples are then generalised to statements about “the human mind and human behavior” in general (ibid., 3), a great step which would deserve more questioning: Are the results really giving answers about human behaviour in general? How come there seems to be an assumption that results obtained by asking a limited amount of people in a defined setting can be generalised to broader claims? My anthropological scepticism awakes when I see this type of generalisations, and we will now see why.
First of all, how would a population group representative of human behaviour look like? At the moment, as we have seen in the opening quotation, samples are mostly gathered in so-called Western, Educated, Industrialized, Rich, and Democratic (WEIRD) settings (Heinrich et al 2010, 1). By comparing cross-cultural data, Heinrich et al. (2010) conclude that WEIRDs are not typical of “humanity at large” (ibid., 2), but “a truly unusual group” (ibid., 1). They argue that WEIRDs cannot be held as the norm for human behaviour, as their results in classical experiments often diverge strongly from that of other populations[3]. It is furthermore very difficult to estimate beforehand “what aspects will be reliably developing and robust across diverse slices of humanity” (ibid., 8)[4]. Today’s over-representation of WEIRD populations in samples poses a third problem: “[r]esearch methods and theoretical constructs are calibrated to the populations they have been selected and designed for” (Bennis and Medin 2010, 26; emphasis mine), in this case, WEIRD populations. Used in other contexts, the same methods might be “less well” or “ill fit” (ibid.). Researching non-WEIRD populations with these constructs could then be misleading. In short, even if there might be some instances where universality of human behaviour can be claimed, this universality claim needs to be thoroughly tested instead of assumed[5]. Let us now ask why these generalisations and biases exist.
Are Behavioural Sciences Western-Centric Sciences?
The underlying assumption for publishing universal claims based on narrow Western samples is that the investigated group is a good approximation of humans in general. In 2010, approximately 12% of the world’s population was WEIRD, but represented 96% of research participants in behavioural sciences, of which 68% came from the US alone (Heinrich, Heine, and Norenzayan 2010, 3). Not much has changed since then. Reasons for this can either be that researchers think that “one adult human sample is pretty much the same as the next” (ibid.) or that WEIRDs represent “standard subjects […] as representative of the species as any other population” (ibid., 1).
Next to the over-representation of so-called WEIRDs, I would like to criticise that entire countries qualify as WEIRD. Those who personify the WEIRD acronym, which are those who are at the same time Western, educated, industrialised, rich and democratic, represent only a subgroup within the so-called WEIRD countries. In Heinrich et al. (2010) the population of the US is implicitly categorised as wholly WEIRD[6]. This label is given nationally despite the strong disparities present within the US. However, the US are known for having an especially important wealth gap (ibid., 15), thus we can agree that not every US citizen is R for rich. I could continue with any other letter of the acronym to show how the label WEIRD is misrepresentative on a national level. I concede that categorizing can be helpful under certain circumstances, but we all have to be careful, as wrongly putting people into the same basket can lead to loss of meaning. Furthermore, as stated by Heinrich et al. (2010, 3), most participants of behavioural studies are young US-American undergraduates from psychology programs[7]. They thus form a subgroup of the US population I suggest to call YUSAPU (Young US-American Psychology Undergraduates)[8].
The origin of unreflectively generalising from (subgroups within) WEIRDs to the whole species may be of historical origin. Behavioural sciences have emerged in the so-called West. Still today, “[m]ost scientists are WEIRD, or were trained in WEIRD subcultures” (Heinrich, Heine, and Norenzayan 2010, 19). If behavioural sciences had “developed elsewhere”, there would likely be other “important theoretical foci and central lines of research” (ibid., 20). In a Western-centric and not fully globalized world, convinced by the ideas of enlightenment, it must have seemed logical to behavioural scientists of the past that results obtained in laboratory conditions with repeatable experiments were scientific truths valid for all humans. However, times and our knowledge of the world have changed. Intensified globalisation and anthropological observations have shown us that a myriad of human behaviours exists, the scientific community is hopefully on the path to abandoning Western-centrism and maybe, just maybe, claiming to have found universal truths could cease to be the ultimate goal for every science. Instead, I advocate for linking findings to the context in which they emerged.
Up to now, behavioural scientists have tried to create experiments that would uncover “underlying principles” (Ceci, Kahan, and Bramanc 2010, 27) of human behaviour, but it is misleading to take away all context and meaning from behavioural experiments. The reason why YUSAPU results look “particularly unusual” in cross-cultural comparisons (Heinrich, Heine, and Norenzayan 2010, 1) is not that YUSAPUs are atypical per se, but because the investigations focus on phenomena especially relevant to WEIRDs/YUSAPUs. These phenomena have emerged in a specific, historical and cultural context and might or might not be as relevant in other contexts. On top of that, Western-style experiments used on other populations might be understood differently, as their context and assumptions are often implicit (Baumard and Sperber 2010, 25)[9]. For participants, a different understanding of the experiment’s premises and tasks, in turn, changes their way of answering (ibid.)[10].
I believe the habit of removing context to be rooted in the understanding of knowledge prevalent in natural sciences. Distilling the core of behavioural phenomena through conducting basic experiments like the Ultimatum Game[11] mirrors the idea that truths are discovered or hypotheses proven in laboratory conditions. To anthropologists, working with people in their natural habitat seems more promising for knowledge generation than laboratory settings, as humans are social beings embedded in different contexts. I believe truths about human behaviour to be variable and malleable by context, meaning that truths are produced according to context. In short, I doubt that simply constructed experiments can produce universal proofs of how the human mind works because they ignore that it works in context.
Why the Generalisations? - The Quest for Easy Solutions and Universal Truths
To finish, I would like to dive deeper into the apparent quest for universal truths. What caught my attention in Heinrich et al.’s paper was that, even though the authors wish to see behavioural sciences change, they still pursue the goal to elicit universal truths about the human mind (Heinrich, Heine, and Norenzayan 2010, 19). To me, this search for universal truths reflects a wish to find easy solutions for complex issues.
When the media takes up results from psychological studies, they often present them in a simplified way: “it has been proven that our brain works like X.” Information on who participated, as well as on how clear the results were, is often lost. Most people I know take the time to question the validity of study results, but it takes training and a certain knowledge of the behind the scenes of science to do so. These people are usually either from disciplines where debates play a great role (for ex. anthropology or political sciences) or disciplines where statistics are widely used (for ex. business or medicine). But what about other readers? Unrepresentative study results are sometimes presented as new truths about humans, but to me, researchers first and journalists second have a responsibility to inform the public about the limitations and complexity of what they present as facts.
Non-fiction books claiming to give the key to self-motivation, happiness or other psychological puzzles also build on our hope for simple and universal truths. I concede that it would be relieving to find universal, simple solutions to behavioural enigmas. Books which promise to enlighten us and deeply change our lives, unfortunately, might end up disappointing readers. In the end, people and their behaviours are complex. Tailor-made solutions rather than universal recommendations are necessary, if imaginable at all. Maybe the wish for simple solutions to existential questions and internal conflicts is what has been holding us back for so long from accepting that there are no convincing reasons to expect a one fits all explanation for human behaviour? By us, I mean both the scientific community and individuals with questions about the human psyche.
It seems to me like scientists and thinkers have been looking for a universal key to the human mind, a key that could explain and predict human behaviour adequately. I suspect that there is no such key, only a toolbox to (partially) understanding the human mind. Sometimes, we might be lucky enough to use the right tool in the right situation. These moments might mislead us to think that we have found the key. Once we realise that there is none, we might develop more realistic expectations about behavioural research, adapt its methods and specify its limitations.
References
Astuti, Rita, and Maurice Bloch. 2010. "Why a theory of human nature cannot be based on the distinction between universality and variability: Lessons from anthropology." Behavioral and brain sciences. Cambridge University Press: 23-24.
Baumard, Nicolas, and Dan Sperber. 2010. "Weird people, yes, but also weird experiments."
Behavioral and brain sciences: 24-25.
Bennis, Will M., and Douglas L. Medin. 2010. "Weirdness is in the eye of the beholder."
Behavioral and brain sciences. Cambridge University Press: 25-26.
Ceci, Stephen J., Dan M. Kahan, and Donald Bramanc. 2010. "The WEIRD are even weirder
than you think: Diversifying contexts is as important as diversifying samples." Behavioral and brain sciences. Cambridge University Press: 27-28.
Heinrich, Joseph, Steven J. Heine, and Ara Norenzayan. 2010. "The weirdest people in the
world? ." Behavioral and brain sciences. Cambridge University: 1-23.
________________________________________________________________________________
[1] The paper The weirdest people in the world? by Heinrich, Heine, and Norenzayan (2010) is followed by a section of Open Peer Commentaries by various authors, and an answer of the authors to these commentaries. [2] I use Heinrich et al.’s understanding of behavioural sciences, which comprises experimental psychology, experimental behavioural economics and experimental cognitive sciences (Heinrich, Heine, and Norenzayan 2010, 1). [3] Among the cross-cultural behavioural studies reviewed by Heinrich et al. (2010) is the visual Müller-Lyer illusion: two lines of the same length are presented to participants, line a ending in arrows and line b ending in outward arrows. In a WEIRD context, many participants think that line b (with the outward arrow) is longer than line a. As Heinrich et al. (2010) state, “[m]any readers may suspect that tasks involving ‘low-level’ or ‘basic’ cognitive processes such as vision will not vary much across the human spectrum (Fodor 1983)” (Heinrich, Heine, and Norenzayan 2010, 4). What the study shows that this assumption doe not hold. The answers to the Müller-Lyer illusion given by 16 population groups differed considerably: American undergraduates were the most affected by the illusion. For them, line a had to be on average 1/5 longer than b before the two segments were perceived as equal. San and Kalahari participants, on the contrary, “were unaffected by the so-called illusion” (Heinrich, Heine, and Norenzayan, 2010, 4). Other illusions were tested in the same study and showed that thevariability in results between population groups was not unusual. Thus, American undergraduates are not necessarily adequate subjects to make claims about general human behaviour. [4] For further reasons on why generalising from WEIRD populations to humanity at large is problematic, see Heinrich et al. (2010) and the open peer commentaries to the article. [5] It is assumed because the great step of generalising is usually not made explicit or challenged. So-called WEIRD populations have been presented as being the norm for a long time. This assumption is often not tested for its validity (Heinrich, Heine, and Norenzayan 2010, 3). Publications “typically lack even a cautionary footnote about these inferential extensions” (ibid.). Demographic background information on participants is rare, “aside from their age and gender” (ibid.) and a “trend to qualify some findings with disclaimers such as ‘at least within Western culture’” (ibid.).
[6] Except for the last comparison where an inner-US-American difference is made between university-educated and non-university educated US-Americans in “Contrast 4: Typical contemporary American subjects versus other Americans” (Heinrich, Heine, and Norenzayan 2010, 16-18). For the other sections, the US citizen as a whole are qualified as WEIRD. [7] As stated above, 68% of all samples come from the US. Undergraduates from psychology programs represent 67% within these US-American samples and 80% in non-US-American samples (Heinrich, Heine, and Norenzayan 2010, 3). [8] Since anthropologists criticize this bias, it would be interesting to see if distribution in anthropological samplings is more representative of the human diversity. I suspect a reverse bias, a greater focus on so-called non-WEIRD populations, which might however be decreasing due to the current trend of conducting anthropology at home. [9] Giving background information and making sure that participants and researchers have the same understanding of an experiment is important, as Baumard and Sperber (2010, 24-25) exemplify. They use the example of economic games where participants are: “given a sum of money for free (which never happens in the real life) and have to share it with someone about whom they have no information (which also never happens in real life)” (ibid., 25). In a WEIRD context, this kind of experiment is widely known, the concept is familiar. Outside that context, participants could however ask themselves a number of valid questions: “Who owns the money? (…) Who is the other participant? Is he or she someone I know? Does he or she have rights over the money?” (ibid.). As the situation can be interpreted differently, different game decisions can either be due to diverging interpretations of the task, setting etc. or indeed show “deep psychological differences” in people (ibid.). Making sure everyone is on the same page is thus crucial to interpret a study’s results. [10] For other convincing examples on the importance of context, see for ex. Astuti and Bloch (2010, 24), Baumard and Sperber (2010, 24-25), as well as Ceci, Kahan, and Braman (2010, 27-28). [11]I name the Ultimatum Game as an example because it has a very simple design and is familiar to many. This famous economic experiment intends to test co-operation behaviours and motivations such as self-interest or fairness. For the game, participants are put in groups of two and are given a sum of real money. According to the study design, they do not now each other and interact only once. One person is the proposer and can offer the responder part of the money. The responder then decides to accept or reject the offer. Note that if the responder refuses the suggested amount, they both get nothing. If the responder accepts the amount, they get exactly that amount and the proposer gets the rest. (Heinrich, Heine, and Noranzayan 2010, 5)
Commentaires