Imposed boundaries and emergent realities: From drug trials and stocks to debates on sex/gender and race
It’s hard to define “group” without being circular: a group is a set of elements that are designated as a group. These designations can be based on characteristics of the individuals, mammals vs reptiles, or be arbitrary, teams created for a game of pickup basketball. The reason groups can tolerate even arbitrary designations is because the thing that’s important is NOT what the groups are, but rather what this designation permits: designating individuals as belonging to a group allows us to refer generally to the members of that group without having to call out each one individually. Hence, we usually designate groups when they are useful: with reference to certain interests, and when it is efficient to make group references.
Designating groups
There are 2 ways in which we designate groups:
Imposed designation: In imposed designations, the groups may not be discernable but for the criteria by which they are partitioned. For example, the members of the teams in pickup basketball, or an experiment where some subjects receive a drug while others do not. In the latter example, groups are defined by receipt of the drug; if the drug has no detectable effect, then you could not discern which individuals belonged to which group unless you knew who received the drug.
Emergent designation: Emergent designations emerge from characteristics that are common among some individuals but not others. For example, the traits that qualify an animal as a mammal or reptile, or using financial returns to sort stocks into positive earners and losers. Importantly, these characteristics are not necessarily inherent to the individuals; a successful company, for example, may be successful as much for good luck and poor competition as they are for their own merits.
In either case, we might designate according to a single criterion (effect of drug, financial returns) or several criteria—which may themselves be imposed or emergent:
In addition to receipt of a drug, if we were interested in differential effects between old versus young and/or females versus males, we might additionally partition individuals by age or sex.
In addition to financial returns, we might also value the ethical conduct and environmental impact of companies, and so discover groups across each combination of these interests.
Doubtful designations
There are a few things that can make it unclear which elements belong in which group—which goes on to threaten the validity/usefulness of the designation scheme used in forming those groups.
Arbitrary cutoffs
In our drug example, we could make a nice clear distinction between groups: those that took the drug versus those that did not. If we graphed this, it would be obvious which individual was in which group.
But what if we used a continuous measure to designate groups? Say we used height to group people as “short” and “tall”. Now we’re faced with the dilemma of declaring the cutoff between short and tall.
We’d surely agree that the extremes of the distribution are correctly labeled “short” and “tall”, but what about the individuals snug to either side of our cutoff? There is very little difference there, between the tallest short person and the shortest tall person. Further, the cutoff is arbitrary; had we chosen a different cutoff, some once-classified tall people would be newly classified as short and vice-versa.
Multiple criteria
When we designate groups along more than one criterion, then what might start as a clear designation along one criterion gets smeared along the multiple criteria. For example, if, in addition to financial returns, we additionally consider the ethics of a company, it’s possible that we can still identify separate groups; perhaps the most ethical companies tend to be less profitable/unprofitable.
We might thus stereotype companies like “Profitable companies tend to be less ethical”. But notice that, as we add additional criteria, not all members of a group exhibit all stereotypical features of that group. For example, the upper-left group is characterized as being more ethical but less profitable, while the lower-right group is characterized as being less ethical and more profitable. But there are some upper-left members that are more profitable or less ethical than the lower-right group; there are some lower-right members that are quite ethical or completely unprofitable. This doesn’t necessarily disqualify the stereotype as being useful, but it requires holding that stereotype more loosely.
Then there’s often cases that buck the trend such that the somewhat-clear groups must accept a few outliers.
In the extreme, the groups can disappear altogether.
In a sense, this is caused because, by adding additional criteria, we add more space in which data can spread rather than cluster. It follows, then, that this effect gets more and more pronounced as we incorporate more and more criteria.
“Real” groups
When we propose criteria to designate groups, we often want to know if the resulting groups are real. But “real” is a difficult concept to apply here because the designation of groups can be arbitrary.
But in many cases, we want to ask something similar: Is an imposed group also an emergent group? In other words, if we repealed knowledge of the designating criteria, could we still discern groups that map onto the imposed labels?
Determining designations
How might we go about determining if imposed groups are also emergent groups? We can follow a few steps:
Remove the imposed labels.
Map the data along all dimensions of interest.
Reapply the imposed labels.
Determine if the labels are sufficiently clustered to consider them emergent groups.
That’s it. Pretty basic.
There is a related technique that is often inappropriately applied to this problem: head-to-head comparisons along a single dimension. There’s a couple of problems here. The first is that head-to-head comparisons assume that the groups are real and then asks if they differ along some dimension of interest; they are not testing the validity of the designation. The second problem is, insofar as we are interested in multiple dimensions, focusing on a single dimension is folly; imposed groups can differ along one dimension and yet fail to cluster as emergent groups. We saw this in one of our previous examples (reproduced below): profitable and unprofitable companies may not cluster once we additionally concern ourselves with ethics.
Using a single dimension is not always a mistake, though. Sometimes the extent of our interest is exactly one dimensional. For instance, if we’re only concerned with financial returns, we might well accept the group designation.
Two diverging case studies
Given that groups are labeled along dimensions of interest, interests differ among people, and the designation of groups affects the treatment of members of those groups, it should be expected that disagreement and controversy can surround designation schemes—especially when applied to members of society. Here, I want to pee on the third rail and discuss two different juxtaposing cases: sex/gender identity and race. I’m not trying to make a case, ethical or otherwise, for any perspective here; I just find it interesting that these controversies, typically framed from an ethical perspective, can be reframed outside of an ethical framework and yet seem to explain much of the controversy.
I think both cases can be framed as disagreements over whether these designations are emergent or merely imposed.
Some believe that individuals should be allowed to participate in all aspects of society in correspondence with the sex they identify as, regardless of the underlying biology; I take this side to be arguing that gender identities are imposed (or adopted) rather than emergent. Conversely, others see male and female as critical categories that necessitate different treatment and considerations for members of these groups; I take this side to be arguing that these groups are emergent rather than imposed.
Some believe that races differ fundamentally (and therefore merit different considerations and treatment), the emergent point of view, while others believe this is false and that differences between races are largely false, exaggerated, or irrelevant, the imposed point of view.
Viewing it through this lens, I think, can explain how certain recent cultural disputes arise. If you believe that sex/gender is imposed but race is emergent, it seems that you would also believe that one can self-identify as whatever sex/gender but not cannot self-identify race. But if you believe the opposite in both cases, self-identifying gender seems invalid while a person’s skin color is but a single and uninteresting feature of that individual.
Bad arguments
Let’s first rid ourselves of the poor arguments that may come from the Imposed and Emergent points of view.
In both controversies, those in the Impositionist camp will rightfully point out that members of the conspecific groups overlap across any single dimension of interest. Some females are stronger than some males. This is true, but not the point I think we are interested in. Rather than concern over any single metric, group designations are typically concerned with several metrics.
The Impositionist camp may also point out that not all members from any of these groups will exhibit all of the characteristics associated with that group. This is also true, but our definition of group already allows for- and expects this.
As for the Emergentist, they could argue that, “Because most _(GROUP A)_ are more/less _(CHARACTERISTIC)_ than most _(GROUP B)_, that _(GROUP A)_ and _(GROUP B)_ represent different groups”. But insofar as we are interested in more than a single dimension, this is a weak argument for reasons already cited.
Common and competing values
If you buy our discussion so far, then the designation of groups is determined by a) the dimensions along which categorization is made and b) a soft threshold of clustering that must be met for the designation to be useful. Both are subjective. Hence, the extent to which designations are shared or disputed is the extent to which individuals wishing to participate in the relevant arena converge or diverge on these points.
Let us first focus on biological males, who identify as females, participating in female sports. In the realm of sport, the personal characteristics which contribute to a successful athlete are relatively few and well-defined; these characteristics differ by sport, but they include attributes like speed, strength, endurance, etc. When people object to biological males competing with females, they may well be running the thought experiment of mapping each individual along these attributes, looking for clusters, and recognizing that clusters exist and they largely abide by biological sex—thus making the labels “male” and “female” emergent groups. I have not conducted this analysis, and I doubt that many of us have, but these assumptions seem safe enough (to me, at least) to take disagreement to be misguided or dishonest. However, it seems that an Impositionist could argue that a) in addition to these athletic attributes, we should additionally be concerned with the feelings of the participants and b) the clusters are unlikely to be perfectly clear. These seems like reasonable points, and which may be used to argue that we should divide leagues along this expanded set of concerns rather than biological sex. It might be hard to argue with the validity of this, but it may be impractical to implement because it would require rather rigorous and frequent testing of athletic potential of all participants. Insofar as the biological-sex divide captures much of these athletic clusters, it is useful to continue using.
Now contrast this with the personal characteristics which contribute to a valuable person. A person may be valued along many, hard-to-define dimensions. There may be clusters in this vast ambiguous space, but it defies (at least, my) imagination that those clusters correspond in any way to race. This implies that race is an imposed- rather emergent designation. Of course, different values/priorities would change this analysis. For example, a focus on socioeconomic outcomes may produce clusters that are largely captured by race. In this case, race-based policies (e.g., affirmative action) may be appropriate. Notice that this is the mirror image of the sex/sport example: rather than focusing on race, one might argue that such policies should be based on individual socioeconomic status rather than race; the better solution will ultimately depend on what factors are considered and what is most practical.
Conclusion
Group designations can be imposed or emergent, depending on the criteria and the context. For an imposed group to be considered an emergent group, one must show that the group’s members naturally cluster in the relevant dimensions, even without the imposed label.
Many cultural disputes often boil down to disagreements about whether imposed labels reflect emergent clusters or are mostly imposed constructs. This extends into cultural disputes—say, over sex/gender and race.
Far from being merely ethical or moral debates, these controversies highlight a fundamental question about how—and why—we draw boundaries between sets of individuals in the first place. By clarifying the dimensions we value and the thresholds we use, we can make more informed decisions about when labeling is useful or detrimental.
