Consider, in this light, two of the theoretical virtues that Longino identifies as among those that may properly guide theory choice for feminists. One is "ontological heterogeneity." This is a preference for "splitting" over "lumping"-for emphasizing the qualitative diversity and individuality of subjects of study and the distinctions among properties commonly classified together. One purely cognitive motivation for this is to seek fine-grained descriptive accuracy. Barbara McClintock's revolutionary discovery of genetic transposition, which was based on close observation of the cytological differences among individual seeds on com cobs, demonstrates that such a focus can yield huge theoretical advances. But there are political reasons for emphasizing heterogeneity as well. Ideologies that purport to scientifically demonstrate the inevitability of male dominance often appeal to theories that assimilate disparate phenomena under vague, global classifications.--Elizabeth Anderson (1995) "Knowledge, Human Interests, and Objectivity in Feminist Epistemology."
Upon re-reading the quoted passage in Anderson's classic article, I had a mild panic attack: is my commitment to analytic egalitarianism an instance of an "ideology" that ends up demonstrating "the inevitability of male dominance"? After all, analytic egalitarianism is committed to the theoretical homogeneity of agents. Before I answer this pressing question, let's distinguish among three theoretical positions:
- Differentiated human nature; the differences also reflect axiological hierarchy. This seems to be Aristotle's position (say on natural slavery) in Book 1 of The Politics, and is revived regularly by modern Eugenicists and has a new instantiation by contemporary (smart/dumb obsessed) epistemocrats (e.g. Jason Brennan).
- Differentiated human nature, but without (axiological) hierarchy among the differences. This seems to be Socrates's position in the true city (or 'city of pigs') of Book 2 of the Republic. [Recall.]
- Analytic Egalitarianism (AE): homogeneous human nature such that we're equal for theoretical (including moral) purposes. Observed differences are due to cultural, educational, institutional factors.*
More positions are possible, of course, but these tend to be unstable. As I noted last week, Thrasymachus, for example, is an inconsistent analytic egalitarian because he does not include himself (and others 'in the know') inside his model (and so is unable to argue his case fully). Hobbes, by contrast, is consistent in this way. At first glance, Anderson's position is like Socrates' in Book 2 of the Republic (option 2). She insists there is human difference, but these differences (in one's theorizing) serve a kind of moral (and political) equality.
In fact, the original proponents of analytic egalitarianism, the economists David Levy and Sandra Peart, are strongly opposed to positing heterogeneous agents in part because after extensive engagement with the shared history of economics and philosophy, which is full of eugenicists and over confident experts, they are suspicious of models that posit hierarchy and thereby justify a hierarchical social organization. So, the difference between the second (Anderson's) and third theoretical option (AE) may be understood in two, not necessarily competing ways: (a) which approach allows us to do better modeling?; (b) In terms of inductive risk (see this important paper by Heather Douglas): which approach is more likely to generate (to use an euphemism) 'unpleasant' moral and political consequences? Both are empirical matters.
Note that my (a-b) have a resemblance to (but are not identical with) Anderson's "dual justification" approach to theoretical classification:
They must satisfy some standards of epistemic significance: there must be clear empirical criteria for determining when phenomena fall under a classification, some phenomena must actually meet these criteria, the classification must figure in some explanation or some causal or empirical regularity. Such classifications must also pass scrutiny from the standpoint of contextual interests and values. They must track the underlying contextual values accurately; that is, they must group phenomena together that share a common relation to these interests. And the contextual values themselves must be justified from an ethical point of view. (47)
For while I agree with Anderson's argument for this dual justification approach (although I note that the arguments from under-determination may become rhetorically less effective in this post-Quine-ean, 'knee-jerk' realist age), I also would suggest that we need to look at the inductive risks of a model as a further, third dimension of evaluation.
Here I cannot settle the (implicit) debate between Anderson and Levy/Peart. But it is worth noting that some of the most telling examples in her paper may also be thought to support AE, too. For, advocates of AE agree with the feminist criticism of "gender polarized classification schemes." (48) Analytic egalitarians do so on principled, not ad hoc, grounds because we reject human difference. If there are gender differences (a) they need to be discovered empirically and, if these are, found then (b) these pose the question what institutions, practices, and research methods generate such differences. The method is, then, self-critical.+ That is to say, AE method guards, in large part, against inductive risk.
Let me close with a final observation. At one point Anderson comments (on feminist criticisms of "theoretical classifications that presuppose the legitimacy of sexist and androcentric values"):
The changes in conceptual schemes for psychological research into gender that feminist normative criticism has recommended are not just window dressing. They open up opportunities for exploring human potentialities that were foreclosed under more rigid conceptual schemes. (Anderson, 47)
I welcome the feminist successes in advocating changes in conceptual schemes for psychlogical research. But I would claim that the problem here, then, is not rigidity as such (for AE is rigid, too), but the foreclosing of exploring human potentialities. AE is designed, precisely, to explore rigorously human potentiality under different institutional (cultural, educational, etc.) regimes. If it is an ideology, then it is a salutary one until, of course, the empirical evidence suggests otherwise.
* Two quick caveats: (I) gene expression is on the culture side in this approach; (II) the model is intentional not extensional--what matters is that people all people consider themselves to be (at least as) equal (Mandeville has also argued this).
+It can also be made compatible with stand-point theory because (a) the theorist is not superior to the agent modeled, and (b) the agent modeled may have information and experiences that are relevant to help explain purported empirical difference(s).