Following up on my post on some of the challenges for a program of “hereditarian leftism”, I want to note that there are at least four ways in which we can distinguish theories (or even pre-theoretical intuitions or stereotypes) about the differences among groups. Clarifying these distinctions can help us understand why “hereditarian leftism” will probably not soon be politically feasible, and why charges like “social justice advocates are the real racists!” fall on deaf ears. It may also help us understand why debates about the “reality” of “race”, “gender”, “sex”, and so on can be so confusing.

Here are the four dimensions, given as one extreme, a rough midpoint, and another extreme. It’s worth noting that not every dimension will be relevant to every theory.

  • internalist versus interactionist versus environmental theories;
  • essentialist versus polycausal versus statistical theories;
  • aretaic versus utilitarian versus dignitarian theories;
  • natural versus genealogical versus revolutionary theories.

The hereditarian leftists seem to be saying something like: “It’s okay to have a non-environmental theory as long as it’s non-essentialist and non-aretaic.” But what matters most to many on the left is that the theory be non-natural(-izing), and ideally that it be non-neutral (compare “colorblindness”). Or in other cases nothing is enough: it must be held that the differences don’t exist. (Compare Cordelia Fine, who simply cannot find it in her to accept that female college students are far less likely to accept offers of casual sex than male college students.)

It may seem at first like some of these dimensions are the same. But my argument is precisely that this is an illusion, and the seeming hypocrisy of all sorts of political actors bears me out. At the same time, the dimensions are not precisely orthogonal to each other. You’ll see what I mean.


Internalist, interactionist, and environmental theories. This dimension concerns the metaphysical ground or physical source or cause of a group difference. The clearest modern kind of internalist theory would rely on a single gene, a single hormone, a single physical process, etc., while the clearest modern kind of environmental theory would rule out any genetic component (including even propensities, epigenetic effects from trauma, and so on). It is important, however, to remember that not all theories of group differences were formulated in genetic terms; some precede genetic science, others ignore it entirely. Some racialists of several centuries ago seem to have believed that there were ethnic “characters” that distinguished, for instance, the British, the French, the Germans, and the Italians. An environmental theory of the day might have pointed to features of the British countryside or the Italian shore; an internalist theory might have located the differences in the “veins” or “humors” or so forth. An interactionist theory might suggest that there are both internal and external causes of some group difference, or it might suggest that external causes “trigger” internal propensities that themselves might or might not be distributed at different rates in different groups.

Essentialist, polycausal, and statistical theories. This dimension concerns the nature of the difference-concept itself, and especially its robustness and susceptibility to change (maybe in physical reality, maybe in metaphysical speculation). Even fully environmental theories can be essentialist. Consider the following formulation: “No man understands what it’s like to be socialized female into an oppressive, patriarchal system, and no woman fails to recognize that system or its purpose.” This is the sort of thing that will get you nods and snaps in certain situations. We might also call essentialist theories categorical: they suggest that there are very strict differences between groups that hold almost all the time. The fact that human groups never really seem to be natural kinds in the philosophical sense can be disregarded with an appeal to internalized [group]-ism: anybody from a marginalized group, for example, who fails to exhibit the right sorts of group characteristics or hold the right opinions can be accused of protecting, knowingly or not, the dominant ideology. For my part, I don’t particularly care if the essentialism is merely “strategic”; it’s all the same to me. Statistical theories, on the other hand, are just assemblages of correlations. Maybe seventy percent of British people like fried food, as opposed to only sixty percent of French people. One might shrug at such a difference and simply move on. Polycausal theories separate potentially essential categories into different parts and acknowledge that an individual can have some parts and not others. Multiple-gene models of intelligence fit here, for example.

Aretaic, utilitarian, and dignitarian theories. This is the ethical dimension of group differences. Under what I’m calling an aretaic theory, a different distribution of virtues – whether correlated with group membership or not – can lead us to assess that one individual is worth more than another individual. Under a dignitarian view, such an assessment is impossible. In the middle, utilitarians allow differential assessments of worth pragmatically insofar as different individuals can do less or more to increase total utility, but they don’t prefer any individuals over any others when it comes to the question of who should get what when that utility is finally distributed. It is important to make a historical note here, which is that old-timey dignitarians found a way around the linearity of this dimension: they found it amenable to assert that though all humans were equal in dignity, not all groups were groups of humans, or of full humans, or so on. I find such rationalizations pretty disgusting, and I imagine most of my readers do as well, but either way I am ex hypothesi talking about differences among groups of humans.

Natural, genealogical, and revolutionary theories. This is the political dimension of group theories. Note for sweet, naive scientist types: this is not the same as the above “ethical” dimensionNatural theories (for a while I was calling these apologist) begin with a certain set of social facts and attempt to explain them through understandable, universalizable processes that don’t themselves rely on a background of other group differences. Such processes might include evolution, ingroup/outgroup formation, fetal or childhood development, or even discrimination or colonialism in certain deployments. Genealogical theories sometimes take an ideology of group differences as a kind of unmoved mover. These ideologies are the “big bads” or “final bosses” of leftist politics that I talked about in the last post. It is noteworthy that there is little discussion, and probably less consensus, among genealogical theorists about why, exactly, these big bads came about in some places and not others. (Why is white supremacy such a historical force as opposed to black or brown supremacy? The fact that historical explanations actually only postpone but never answer questions like these, which are of course in a sense questions about group differences themselves, shows the weakness of the “genealogical” method.) Remember that the mission statement of poststructural social science is to show how what seems natural is actually historically contingent. Then, there exist revolutionary theories that assert something far stronger than historicizing theories: they can say that marginalized people are actually superior to their oppressors. Men might be cast as deficiently violent, or white people as inherently cruel or sociopathic. That this is politically permissible on the left, even encouraged, should never be overlooked. Even standpoint epistemology is a theory of this type (see above, the “no man understands . . ., no woman fails . . .” formulation).


These dimensions distinguish not just different theories about how the world is, but different conceptual analyses of the meanings of group-difference terms. Kwame Anthony Appiah, for instance, is known for the view that race, to exist, would have to be a biological (internalist) natural kind (essentialist). There are two sorts of people who disagree with him about this. First there are social constructionists, who take an environmental view of race. Then there are certain genetics researchers, who take a statistical view of race. (Personally I am relatively convinced that race is an internalist, probably biological term.

The following argument demonstrates why: We understand the phrase human race, and we know that robots would not be members of that race no matter their ethical status. And we understand the phrase race of the Atreides, and we know that even if Agamemnon adopted another child that child would not be a member of that race. So the term race is used both for biological groups smaller than modern ancestral taxa and for biological groups larger than modern ancestral taxa. So it is probably a biological term. Whether it refers to anything is a separate question, though it would take some grumbling about misapplications of semantic externalism to show why.)

Or consider gender. In my Quillette piece about the Hypatia scandal, I wrote that Beauvoir’s dictum that “gender is the social meaning of sex” could be taken to be trans-exclusionary. But this is only necessarily the case if we take the phrase to suggest that gender is environmental rather than internal. But if the phrase is taken to be a commitment to a genealogical theory, there may be room to make it consistent with trans-inclusiveness. (Personally I am relatively convinced that gender in Beauvoir’s sense is not real, but I think it might be useful in formulating theories of dysphoria and self-identification.)

Finally, consider sex. Some people think that facts about fetal development, or about intersex persons, should lead us to be “abolitionists” about the category of sex – to assert that it’s not real or not useful. But that requires a very extreme sort of essentialism: one that suggests that there are all humans are either physically male or physically female, and that they are all entirelyonly, and forever so. Sex, however, is a very useful category both for researchers and for laypeople. A more moderate essentialism would do just fine. Note, too, that a genealogical theory of sex would be insane.


One thing that leftist academics almost always got right was the danger of “reifying” our concepts. Often when it comes to group terms and group differences, it seems as though people argue over definitions as much as they argue over the facts on the ground – something I’ve complained about at length on this blog. I have urged, and here urge again, the adoption of new terms in situations where it’s feasible. For example, the success of genetic tests for ancestry suggests that there are, to some extent, biologically distinguishable subgroups of humans. Should we call those “races”? I think not, since I agree with Appiah that the word has an essentialist feel. (Note that this is different from saying that it has an essentialist history, something I care about not a jot.) I’ve put forth the older term “extraction” for this internal, statistical, natural category, but maybe it’s too old to work, or just too weird. In other situations terminology might not matter too much. Are there sex differences in cognition or are there gender differences in cognition? The answer will depend on what goes on in trans people’s brains, but our interest in the claim that there are important differences in cognition between cismales and cisfemales will not depend on that. My hope for this piece is that the distinctions above can help transform arguments about words into arguments about the world. That’s where all the good stuff is!

Advertisements