Implicit Bias and the Idealized Self

Published March 17, 2019.

Here are some worrying facts, for someone in the discipline, about academic philosophy.

  1. White males are vastly overrepresented at all levels (and much more so than in other disciplines).
  2. Non-white-males face more difficulties than otherwise being in, contributing to, and being recognized for their contributions.
  3. The disciplinal response to the above has been slow and inadequate.

These I take to be well documented and fairly noncontroversial. What I want to take up here is an interesting hypothesis proposed in Berenstain (2018) (who also does an admirable job fleshing out the above three jobs). In particular, she seeks to explain one aspect of this last point: namely that philosophy as a discipline has not properly enacted measures that, we have good evidence to believe, counteract implicit biases (e.g. triple-blind review processes). These measures are not implemented, Berenstain proposes, because those who are in a position to implement them routinely underappreciate the existence and impacts of implicit bias. Why is this underappreciated? Here is the interesting hypothesis. Because of a general commitment to a picture of the idealized rational self, which has the following four components.

  1. Consistency — the mind does not contain contradictory beliefs;
  2. Unity — the mind is simple; it does not have parts and is not subdivided;
  3. Rational transcendence — in principle, thinkers are capable of achieving an objective rationality that abstracts away from their contingent circumstances and transcends cultural, environmental, and bodily limitations on reasoning (e.g., exhaustion, hunger, and pain);
  4. Introspective transparency — the contents of one’s mental states are transparently available through introspection.

The claim in particular is that commitment to this picture has the following effects.

  1. It makes thinkers skeptical of the correct picture of implicit bias in general.
  2. It causes thinkers to be skeptical that they themselves are subject to implicit bias and to underestimate the extent to which their judgments are affected by it.

Here, I take it, are the ways adherence to this picture is supposed to have these effects.

  1. If the mind is consistent, then if a person endorses egalitarian attitudes, she cannot also endorse (or even tacitly hold) inegalitarian biases. Since most philosophers (even the white, male ones) explicitly endorse some egalitarian attitude, they are not implicitly biased. In particular, since I endorse these attitudes, I cannot be implicitly biased.
  2. Philosophers are, many (most) philosophers think, a highly rational lot. They form their (philosophical) beliefs, or at least I do, through rational methods—(pure) intuition, reflective equilibrium—that are safe from these sorts of biases. So even if other people are subject to bias, we are not.
  3. I know what is going on in my head, and there ain’t a racist or sexist thought or impulse there.


There is, I think, very likely something to this thought that philosophers are especially likely to be overconfident in their estimation of their own rationality. The question is whether this is explained by commitment to something like the picture above. I am skeptical of this claim for the following two reasons. First, I doubt that a commitment to this picture is in fact all that widespread. Second, to the extent that it is, I do not expect it to be a major cause of this overconfidence. Rather, I expect the explanation to have more to do with the training and reward structure of the discipline (philosophers are trained to and rewarded for what we might call public displays of rationality and it strikes me as not implausible that someone frequently praised for displaying rationality comes to overestimate and overemphasize their own rationality, particularly if there are not robust counterpressures in play). Of course all of these questions are empirical and it seems we do not yet have adequate evidence.

Nevertheless, here are the worries. It is unclear how widespread adherence to these tenets are. Certainly some are broadly held to as ideals, such as rational transcendence. It would be somewhat unusual to think the ideal rational agent would have racist or sexist attitudes. (This is an issue Berenstain takes up at the end of the paper; put a pin in it.) But, it seems to me, a widespread commitment to what we might call ordinary irrationality: in general, everyone fails to live up to the ideals much of the time. As such, philosophers have frequently been concerned with examining the sources of error. For example, Descartes, who Berenstain credits (?) with exemplifying “the most robust version of the idealized rational self conception,” begins the Meditations worrying how frequently he has been in error. Descartes’ explanation, moreover, for this error involves habitual inference (cf. Hatfield 1992). There is at least a major strain in the philosophy of mind that thinks of ordinary mental goings-on as subdoxastic, arational (implicit bias itself has come in for some philosophical attention, cf. Brownstein 2015). I think it is at best unclear the extent to which philosophers are committed to a picture of the mind incompatible with the existence of implicit bias.1

I expect it more likely that philosophers are by and large are committed to a picture of cognitive functioning that is at least compatible with the existence of implicit biases. It is almost certainly true that the broad picture of ideal, rational cognitive function is incompatible with the existence (or at least the efficaciousness) of implicit biases (unpin this point). The wise person, as Hume tells us, apportions their beliefs to their evidence. Correspondingly, they act on and only on what their reasons support. So the ideally rational person does not suffer from implicit biases, or, if they do, they never let them effect their beliefs or actions.

Is that not what we should want? After all, implicit biases are supposed to be bad things: distortions that cause us to act unjustly. If our picture of rationality tells us not to have or not to be effected by such biases, so much the better, it seems, for this picture. Here, though, is the worry. For an analogy, think a moment about virtue ethics. What I should do, a simple version of virtue ethics tells me, is to do what the virtuous person would do. Suppose I have wronged a friend in some way and start thinking about how to make it right with them. I ask myself what the virtuous person would do. What she would have done, a voice snaps back, is not get herself in this situation. What we wanted to know is what to do in the face of our inevitable deviation from virtue. Likewise, what we want to know is what to do in the face of our inevitable irrationality.

Fair enough, one wants to reply, these ideal theories do not always tell us what to do, do not provide direct guidance. But they do provide an indirect guidance: they tell us what to aim for. We are to try to be, to the greatest extent possible, the person who does not wrong her friends, or does not suffer from biases. We will have to depart from ideal theories, do some nonideal theorizing, to figure out exactly what that involves, but these projects are not incompatible, indeed, are, instead, complementary.

But Berenstain, following Mills (2005)’ related complaints against ideal political theory, thinks this picture is far too rosy. There are a cluster of complaints here, and I admit to not always having a great sense of all of them.

  1. Ideal theories are wastes of time. Nonideal theorizing is both independent and much more important.
  2. Ideal theories are actively misleading. They obscure, rather than illuminate, thinking about what subjects who suffer from cognitive biases, moral failings, and live in systematically unjust worlds.
  3. Ideal theories themselves are the result of, expressions of white, male supremacy. (“Like the non-racialized conception of the social contract, the idealized rational self is an idealizing abstraction that, while once explicitly racialized and gendered, now purports to be genderless and raceless.”)

As Berenstain elaborates it, this third point seems to reduce to the second.2 The problem there is a “conceptual sleight of hand that moves from a purportedly normative context to a descriptive one without acknowledging that such a shift has taken place.” This is, however, a user error rather than a real implication of the theory itself.

I think this last point poses some problems for Berenstain’s explanation and so is worth foregrounding. If the explanation for an underappreciation of implicit bias resulting from a commitment to a theory comes itself from a fallacious inference from the theory, then why think the commitment to this theory itself is what is doing the explanatory work? Surely white, male philosophers could be committed to a nonideal, arational picture of the self and make largely the same fallacious inferences. We could be, instead, committed Freudians, explaining everything in terms of the sex and death drives and thinking that, well, since I rejected that candidate out of a brute, incestuous desire, I could not have rejected her out of an implicit bias against women of color.

I am sympathetic to Mills’ problems with ideal political theory. It is plausible to me that the questions ideal political theory asks are misguided and uninteresting and distract from and perhaps distort a more grounded approach to injustice (cf. Gaus 2016 for a convincing treatment). It is less plausible to me that thinking about ideal rationality is similarly pointless or similarly harmful.


In the end, I agree with Berenstain’s recommendations. Even without having a robust understanding of the causes of these problems in the discipline, sensible steps can be taken towards their solution. Many of these recommendations (e.g. in the APA’s Good Practices Guide)—blind reviewing practices, anonymity in grading, being more deliberate and less intuitive in one’s professional evaluations, e.g. in hiring, and so on—are relatively low cost and low risk.3 It is certainly also worth investigating these causes. I doubt, however, that this has all that much to do with philosophers’ theoretical commitments, rather than, say, cultural aspects of the discipline.


Berenstain, Nora. 2018. “Implicit Bias and the Idealized Rational Self.”

Brownstein, Michael. 2015. “Implicit Bias.”

Gaus, Gerald. 2016. The Tyranny of the Ideal: Justice in a Diverse Society. Princeton University Press.

Hatfield, Gary. 1992. “Descartes’ Physiology and Its Relation to His Psychology.”

Mills, Charles W. 2005. “"Ideal Theory" as Ideology.” Hypatia 20 (3):165–84.

Smith, Justin. 2015. Nature, Human Nature, and Human Difference: Race in Early Modern Philosophy. Princeton University Press.

  1. The evidence Berenstain adduces here, for example the literature surrounding Kripke’s puzzle, seems far from telling.

  2. I want to note in passing that the historical claim that the idealized rational self was once explicitly racialized and gendered may be too quick. As Smith (2015) reveals, the moderns had a diversity of perspectives on race that, while overall troubling, included more egalitarian currents. Descartes, for instance, seems to have included black people under the rubric of this self.

  3. This is not a thoroughgoing endorsement of all recommendations given in the guide or elsewhere. For instance, the APA guide, admittedly qualifiedly, endorses Project Implicit, despite its measurements being controversial and, likely, largely meaningless.