One of the humbling experiences in philosophy is to come across a paper which says what you (perhaps) inchoately thought, but in a much more knowledgeable way than you could have ever said it yourself. This at least was my experience when I read Amia Srinivasan’s “Normativity without Cartesian Privilege”. (Srinivasan also has related paper, co-authored with John Hawthorne, “Disagreement Without Transparency”, which is equally good.) Srinivasan’s argument is complex, but it’s worth trying to understand. Here’s a rough summary.
(1) Srinivasan starts from a claim in theoretical philosophy. She claims that we should be Anti-Cartesians—that is, we should think that there is no condition C such that we can always know whether we are in C. This is easy to see for the external world. We cannot always know whether my car is parked outside (rather than having been stolen, or having been moved without my knowledge). We cannot always know whether telling a joke will make others laugh—and so on. But it is tempting to think, following Descartes, that at least our own mental states can be reliably known: e.g., that we can know our own intentions, beliefs, feelings, and so on. But Srinivasan claims that this Cartesian hope is mistaken: our own mental states can fail to be luminous to ourselves. This is compatible with sometimes knowing these states, but denies that we always can.
(2) The exciting move Srinivasan makes is to connect anti-luminosity to ethics. Let’s distinguish two moral toy theories (my example). One objective moral theory demands that you ought to do A just in case C, e.g., you ought not to lie unless doing so is necessary to avoid catastrophic results. A subjective theory, on the other hand, relativizes this demand to your beliefs, or evidence, or some other mental states of the agent. Such a theory would say (e.g.) that you ought to do A just in case you have evidence that C. E.g., you ought not to lie unless you have evidence that doing so is necessary to avoid catastrophic results.
One of the main motivations for a subjective or “internalist” moral theory, Srinivasan claims, is the idea that an internalist theory promises transparency. On the objective toy theory, for example, it might be that you ought to lie, even though you couldn’t know that it was necessary to avoid moral catastrophe; or inversely, you might have had all the evidence that lying was morally necessary, but the evidence was misleading. Externalist theories like this one entail the possibility that we act wrongly even though there is no way for us to know, even in principle, that we do.
The internalist finds this possibility troubling. In particular, what internalists object to in externalist moral theories is the possibility that our moral fates are subject to cruel, world-imposed moral luck, such that the moral status of our own actions, beliefs, character, society (etc.) are beyond our control. (There’s also an interesting discussion of action-guidance in the paper which I’m glossing over.)
Insightfully, Srinivasan explains this worry as the conviction that the deontic facts (about what we ought to do) and the hypological facts (about what we should blame agents for) should align, or at least roughly align. Externalist moral theories allow that some people act wrongly even though we can identify no fault in their behaviour or character, nor in their reasoning; and badly irrational people might stumble into doing what is right.
(3) Srinivasan’s important result is now the following: if the anti-Cartesian critique of luminosity is correct, then the internalist hope to find “transparent” moral principles is entirely misguided. Subjective moral theories are subject to the very same fault (if it is one) as the supposedly intransparent objective moral theories. For example, you can fail to know what your own evidence is. In that case, our subjective moral pet theory also allows for cases where (i) you have evidence that lying is necessary to avoid moral disaster, and (ii) therefore you ought to lie, but (iii) you do not know that you have that evidence.
Thus, one of the main motivations for the internalist/subjectivist move falls away. This isn’t the deathblow for these theories, of course, but it suggests that they are less attractive than they initially seemed. Note that it doesn’t make a difference either if we try to salvage what we can by moving to a type of two-concept view, in which we distinguish the “objectively” and “subjectively” right, in the hopes that at least the latter reliably tracks homological facts. If Srinivasan is right, then this move also fails.
(4) What’s the upshot of all this? The internalist tries to capture, in some way, the intuition that our normative performance is under our control. That is, if we have carefully considered all the evidence available to us, shown the best of good wills, formed the firmest and nicest of intentions, then it seems strange that certain types of moral criticism should still apply to us—e.g., that we acted wrongly, or were responsible for some fault. But Srinivasan suggests this is an illusion:
Even in a sort of normative utopia—where all agents were competent, everyone knew the genuine normative truths, and everyone was maximally motivated or disposed to fulfil their normative obligations—we would still have deontic violations. Even in such a utopia, normative performance would not lie entirely within our control. (285)
In the last section of her paper, Srinivasan connects the issue back to Greek tragedy (and in turn captures much better what the tragedy of truth could be that I vaguely hinted at). She argues convincingly that the Greeks were much more tolerant of the possibility of bad moral luck—of a misalignment between the deontic and the homological. Oedipus, for example, violates moral, natural law when he sleeps with his mother. There isn’t some hidden moral flaw in Oedipus’ character that he could have remedied, or some piece of evidence that he culpably ignored: he acted wrongly, and no one with his knowledge could have avoided acting wrongly.
This makes his case a genuine tragedy. But that might just be the nature of morality. The attempt to avoid this tragedy, Srinivasan suggests, might even exhibit a type of category mistake:
Anti-Cartesianism invites us to return to a more tragic outlook of the normative. Another reason for such a return is itself normative. The impulse to eradicate luck from the normative sphere often issues from a laudable discomfort with the role that luck plays in other spheres of human life, especially its intimate connection with social and political inequality. Understandably, then, we hope to establish that one realm of human life is immune from luck – that is, the normative itself. But this philosophical longing might in the end be morally neutering: a revolt against injustice in theory, rather than in practice. In the place of such a longing, we might instead have a post-Cartesian normative theory that is more attentive to the fragility of our human estate.
(5) Let move on to some brief thoughts of my own. First, Srinivasan’s points are on a high level of abstraction. So it is useful to contrast her paper with a somewhat more concrete discussion. Fred Feldman, for example, has argued that expected-utility utilitarianism is not any more action-guiding than objective utilitarianism. Feldman’s main objection to the expected-utility proposal is on different, much less ambitious grounds: he thinks that even roughly adequate computations of expected utility are out of our reach.
Feldman’s objection rests on different foundations, and is directed towards a much more specific aim, than Srinivasan’s. But I think both results point to the roughly same outcome: the hopes to construct a fool-proof moral decision procedure are doomed to fail.
(6) Srinivasan focusses primarily on epistemic and moral theories. But it’s interesting to see whether any of these ideas transpose to politics. A toy externalist political theory might be, for example, that an educational system is just if it gives everyone equal opportunities for success. Assume that some bureaucrats think, on the evidence available to them, that their society’s educational system does provide everyone with equal opportunities for success, while in fact it doesn’t.
Interestingly, I think we are less inclined to make the internalist move in political theories—we are less inclined to think that evidence for justice-making properties makes something just. Either the educational system provides equal opportunities, or it doesn’t—it’s not clear to me why we would think that anyone’s evidence or beliefs matter in this respect.
(7) Another way to think through Srinivasan’s point is by highlighting the uneasiness we often feel with judging past societies. Assume that we discuss Roman, or medieval morality, or the morality of Southern slave owners. Many people insist that we shouldn’t judge the people of these times by “our standards”.
Part of the motivation not to judge far-away, long-gone societies might of course be a misunderstood moral relativism. But part of the motivation might also be the internalist worry that Srinivasan has identified: that blaming people for injustices they couldn’t know they committed is inappropriate. Just as we feel it wrong to indict Oedipus who couldn’t have done better, we feel it wrong to indict the slave-holders of the past (who, let’s problematically stipulate, couldn’t have epistemically done better). Once we free ourselves of this mistaken presupposition, judging people by our standards starts to look less troubling.
There are other reasons to abstain from making such judgments—for example, because habitual judging exhibits a kind of arrogant moralizing that is both epistemically and morally defective—and there might be other reasons why time-transcendent judgments are mistaken. But I think Srinivasan’s paper helps to identify at least one mistaken source of uneasiness regarding such judgments.
Edit. It’s Oedipus, not Odysseus, of course.