The epistemological challenge to metanormative realism: how best to understand it, and how to cope with it


Metaethical—or, more generally, metanormative—realism faces a serious epistemological challenge. Realists owe us—very roughly speaking—an account of how it is that we can have epistemic access to the normative truths about which they are realists. This much is, it seems, uncontroversial among metaethicists, myself included. But this is as far as the agreement goes, for it is not clear—nor uncontroversial—how best to understand the challenge, what the best realist way of coping with it is, and how successful this attempt is. In this paper I try, first, to present the challenge in its strongest version, and second, to show how realists—indeed, robust realists—can cope with it. The strongest version of the challenge is, I argue, that of explaining the correlation between our normative beliefs and the independent normative truths. And I suggest an evolutionary explanation (of a pre-established harmony kind) as a way of solving it.

This is a preview of subscription content, log in to check access.


  1. 1.

    For one locus classicus, see Mackie (1977, p. 38).

  2. 2.

    Enoch (2003, 2007).

  3. 3.

    With the possible exception of Oddie’s (2005) realism, which is committed—unlike my realism—to the causal efficacy of moral facts. I return to this point below.

  4. 4.

    See, for instance, Timmons (1990, p. 114).

  5. 5.

    See, for instance, Brink (1989, Chap. 5), Sayre-McCord (1995), Scanlon (1998, pp. 64–72) and Shafer-Landau (2003, Chaps. 10–12). For an antirealist presentation of the challenge in terms of justification (and also in terms of knowledge), see, for instance, Waldron (1992, p. 175). For a skepticism-friendly discussion of moral epistemology that focuses primarily on moral knowledge and justification see Sinnott-Armstrong (2006).

  6. 6.

    See, for instance, Brink (1989, Chap. 5) and Sayre-McCord (1995).

  7. 7.

    See Audi (1995) and Shafer-Landau (2003, p. 258).

  8. 8.

    At least with regard to justification, then, I agree with Sayre-McCord (1995, p. 138) when he says: “there is no distinctive epistemology of moral belief”.

  9. 9.

    It may be thought that the fact that normative beliefs are controversial in some fairly robust way could do the job here. But it is not at all clear that they are controversial in a way that many other beliefs aren’t, nor is it clear what follows from such disagreement that we do find here. For my own views on this, see Enoch (forthcoming).

  10. 10.

    Compare: If the causal theory of knowledge entails that we have no mathematical knowledge, this counts heavily against that theory, perhaps more heavily against it than against mathematical knowledge.

  11. 11.

    Compare: If the causal theory of knowledge entails that we have no mathematical knowledge, but otherwise scores sufficiently highly on the list of theoretical virtues, perhaps we would be justified in biting the bullet and acknowledging that we never have mathematical knowledge, indeed that such knowledge is impossible.

  12. 12.

    You may think that more is needed for reliability, and in particular that reliability contains some modal feature, so that a set of beliefs is reliable only if, roughly speaking, it is not an accident that sufficiently many of them are true. At least one natural way of understanding such a requirement leads to what I take to be the proper understanding of the epistemological challenge, in Sect. 3.

  13. 13.

    I have in mind error-theorists, who think that moral or normative discourse is systematically mistaken, and so that no normative judgment is (non-trivially) true, and expressivists who think (or at least used to think, before quasi-realism became the fashionable way of being an expressivist) that normative judgments are not truth-evaluable at all.

  14. 14.

    But see footnote 48 below.

  15. 15.

    This is especially true if the scope for which the relevant account of knowledge causes trouble is even wider than the normative—if it includes, for instance, the a priori in general. For this dialectical point, mostly in the context of discussing the causal theory of knowledge, see Huemer (2005, p. 123), Liggins (2006), and somewhat more generally, Lewis (1986, p. 109).

  16. 16.

    Much of Shafer-Landau’s (2003, e.g. p. 235) epistemological discussion in the context of a general defense of metaethical realism is really an attempt to deal with particular instances of general skeptical worries. And he repeatedly emphasizes this very point—that the challenge to realism is a particular instance of general skeptical worries—as a part of his defense of realism (see, e.g. p. 239). But I take this not as evidence that realism can cope with the epistemological challenge, but rather as evidence that this is not the best way to put the epistemological challenge.

  17. 17.

    There is, of course, a huge body of literature on the question whether normative—and, in particular, moral—truths explain anything worth explaining. I do not wish to enter this debate here. I take it that whatever your view on moral or normative explanations, the suggestion that normative truths cause our normative beliefs is at least a bit of a stretch. I return to this point below.

  18. 18.

    I think—but I am not sure—that Nagel (1997, pp. 76, 131) flirts with the claim that such correlations are to be taken as brute, and perhaps necessarily so.

  19. 19.

    Oddie is not a reductionist or a naturalist, but he thinks values are causally efficacious, and indeed his motivation for this (surprising, I think) claim is that this is his way of coping with some version of the epistemological challenge. See Oddie (2005, p. 181).

  20. 20.

    Ideal-response-dependence views may not be in a better shape regarding the epistemological challenge compared to robustly realist views. See Dancy (1986, pp. 178–179).

    How powerful is this epistemological challenge against expressivist, or noncognitivist metanormative views? I think that when push comes to shove, such views are response-dependence views, perhaps of a special kind. If so, the discussion in the text applies. But it is now common for expressivists to insist that their view is not committed to any kind of response-dependence. If such claims can be defended, then perhaps the challenge in the text forcefully applies to expressivist views as well.

  21. 21.

    This characterization of internalism comes from Pryor (2001, Sect. 3).

  22. 22.

    For a closely related point, see Joyce (2006, pp. 179–182).

  23. 23.

    Wedgwood (2006) mentions something very close to this challenge, but by the time he gets to his solution, it is no longer, I think, this challenge he is addressing, and in conversation he has made it clear that he is indeed interested in a different challenge (that of explaining how it is that we are justified in our moral beliefs). And while Huemer (2005, pp. 123–127) does seem to raise the challenge in the text, it is not clear to me how his sketch of an epistemology of the a priori—-offered as a response to the challenge—copes with the challenge, at least as I understand it. For anti- (or non-) realist hints at this challenge, see Wright (1988, p. 25, footnote 36), Timmons (1990) (though Timmons raises it as a worry only for the combination of (naturalist, new-wave) realism and a coherentist theory of justification), and Gibbard (2003, p. 258) (though Gibbard raises it as a worry for his quasi-realism rather than for realism).

  24. 24.

    For a similar accusation against (fellow-) realists, see Wedgwood (2006, p. 62, especially footnote 2).

  25. 25.

    Sinnott-Armstrong and Timmons (1995).

  26. 26.

    For a precursor, see Gibbard (2003, pp. 263–265). And for a related argument, see Joyce (2006, Chap. 6).

  27. 27.

    For Street’s own characterization using this term, see Street (2006, pp. 112–113).

  28. 28.

    Perhaps this is why Thomas Nagel—a robust realist (in some moods, at least)—suspects (1986, p. 145) that realism is incompatible with the availability of a purely naturalistic explanation of our normative judgments.

  29. 29.

    I think that it is in this spirit that Copp (forthcoming) replies to Street.

  30. 30.

    This is a point Shafer-Landau (2003, p. 234) emphasizes in our context.

  31. 31.

    Or, in Street’s version—the correlation between (what we take to be) the normative truths and normative judgments that are likely to have been selected for is not that strong. Consider the false but seemingly evolutionarily useful “The interests of others with whom we have no privileged genetic of reciprocal relations do not count at all.” and “The interests of non-human animals do not count at all”.

  32. 32.

    For sometimes “‘we’ means: ‘you and I, and I’m none too sure about you’.” (Lewis 1989, p. 84).

  33. 33.

    What I say in the concluding section below can be seen as a partial answer to this question.

  34. 34.

    Think of the relevant reasoning processes as the analogues of proper conditionalization, and of the normative truths as the analogues of probability-truths. Then the point in the text is analogous to the following well-known result: So long as our assignments of prior probabilities are not too far off, and so long as we are exposed to sufficient evidence, proper conditionalization will get us increasingly closer to the probability-truths.

  35. 35.

    A point acknowledged by Street (2006, p. 123), and emphasized also by Copp (forthcoming).

  36. 36.

    Wedgwood (2006, pp. 79–80) mentions the starting-point problem for his reflective-equilibrium-like theory of justification, noticing that his suggested solution of (his version of) the epistemological challenge does not solve it. But again, the challenge he is primarily interested in is different from the one I am primarily interested in.

  37. 37.

    For a general criticism of all response-dependence views that involve some idealization—of which Street’s seems to be one—see Enoch (2005). And for a general critique of constructivism, see Enoch (manuscript).

  38. 38.

    A line I pursue in Enoch (2003, 2007).

  39. 39.

    Street (2006, p. 134) is explicit about there being only two types of possible explanations of such correlations—the two mentioned earlier in the text. But in her discussion of pain (Sect. 9) she may be implicitly flirting with a third-factor kind of explanation. I briefly return to this part of her discussion below.

    Throughout Copp’s (forthcoming) discussion of Street’s Darwinian Dilemma, he speaks of “the tracking thesis” as something realists must endorse if they are to avoid the skeptical horn of the dilemma. But talk of tracking here is misleading. “Tracking” is naturally understood as a causal term. If so, the availability of third-factor explanations shows that the skeptical horn can be avoided without endorsing a tracking account. If “tracking” is understood as not necessarily causal (perhaps merely indicating some counterfactual relations), then Copp may be including under the same term causally-tracking and third-factor explanations. Indeed, at the end of the day it is not entirely clear to me whether Copp’s own way out—that relies on his society-centered version of moral realism—is an instance of the former or the latter. His talk of “a close relative of the tracking account” seems to suggest that perhaps his solution too can be understood as a third-factor explanation of sorts, though of course not the one I am about to offer.

  40. 40.

    These clarifications should make it clear that the assumption I will be relying on is significantly weaker than those considered by Gibbard (2003, p. 264) and Timmons (1990, pp. 107–108). If I am right in what follows, the weaker assumption suffices for my purposes.

  41. 41.

    Of course, evolution has neither a mind, nor an aim in mind. Talk of the evolutionary “aim” in the text is meant as shorthand for the usual respectable, non-teleological, evolutionary way of putting things.

  42. 42.

    Discussions with Pete Graham and Sigrún Svavarsdóttir helped me with this paragraph.

  43. 43.

    In particular, perhaps such reasoning mechanisms can explain—given reasonably good starting points—how “beings like us would be good judges of ultimate worth, even in those cases where ultimate worth comes apart from maximizing the long-run reproduction of one’s genes” (Gibbard 2003, p. 265), as indeed Gibbard himself suggests later on (2003, p. 266).

  44. 44.

    Notice that the use to which I put (speculative) evolutionary considerations is not that of (directly) justifying our normative beliefs. Rather, it is that of explaining the correlation between the normative truths and our beliefs, thereby possibly defeating a defeater of the justification of our normative beliefs. To an extent, then, I need not disagree with Nagel when he (1997, p. 136) says: “This means that the evolutionary hypothesis is acceptable only if reason does not need its support. At most it may show why the existence of reason need not be biologically mysterious”.

  45. 45.

    I am indebted here to comments from Ralph Wedgwood and Sigrún Svavarsdóttir.

  46. 46.

    Street (2006, p. 150) makes an analogous claim against the pain-explanation of the correlation she is interested in. And Copp (forthcoming) is explicit about his attempt to avoid a commitment to any such remaining miracles.

  47. 47.

    I thank Ralph Wedgwood for relevant discussion here. For the claim in a very close context that an explanatory challenge remains, see Huemer (2005, p. 123).

  48. 48.

    In discussing knowledge in Sect. 2.4, I said that Gettier cases do not pose a special problem for normative beliefs. But at this point the related following worry may arise (for which I thank Mark Van Roojen): One common way of stating the problem Gettier was on to is in terms of luck: What we need for knowledge is unluckily justified true belief. And it may be thought that the counterfactual robustness of my explanation of the correlation between our normative beliefs and the normative truths—though perhaps it suffices to defeat the possible defeater of our justification in holding those beliefs—nevertheless does not suffice for Gettier. Perhaps, in other words, our normative beliefs, though correlated with the normative truths, are luckily so correlated, and so do not count as knowledge.

    I am not sure how best to understand the anti-luck intuition, and so how its best version interacts with the considerations in the text here. But let me acknowledge that more discussion is needed here. And, of course, even if there are here problems for normative knowledge, they don’t extent to the justification of normative beliefs.

  49. 49.

    And there is, presumably, an evolutionary story backing up this causal story. See, for instance, Gibbard (2003, p. 255).

  50. 50.

    Schechter (2006) notices this point. McGinn (1997, p. 58) notices that the epistemological challenge to realism is a particular instance of the general problem of a priori knowledge (but he thinks it cannot be solved by creatures like ourselves).

  51. 51.

    In his PhD dissertation (2006), and in work in progress based on it, Joshua Schechter discusses (among other things) the epistemological challenge to (what he calls) the objectivity of logic, and he understands the challenge in a way precisely analogous to the way I understand the epistemological challenge to metanormative realism in this paper. Furthermore, he pursues an evolutionary way of dealing with it. Indeed, my thinking about these matters was heavily influenced by my exchanges with him, and I am grateful to him for that. But it may be helpful here to note an important difference: Schechter discusses an evolutionary explanation that utilizes the evolutionary advantages of our employing reliable deductive rules (and so of forming true logical beliefs). I do not argue that having true normative beliefs is evolutionarily advantageous (for a sketch of an argument along these lines, see Huemer (2005, pp. 218–219)). My argument rather is a pre-established-harmony kind of explanation. This difference between these two explanations is an instance of the point made in the text: The details of the appropriate ways of dealing with the epistemological challenge (properly understood) may very well differ with context.


  1. Audi, R. (1995). In W. Sinnott-Armstrong & M. Timmons (Eds.), Intuitionism, pluralism, and the foundations of ethics (pp. 101–136). Reprinted in his Moral knowledge and ethical character (pp. 32–65). Oxford: Oxford University Press (1997).

  2. Brink, D. O. (1989). Moral realism and the foundations of ethics. Cambridge: Cambridge University Press.

    Google Scholar 

  3. Copp, D. (forthcoming). Darwinian skepticism about moral realism. In S. Landau (Ed.), Oxford studies in metaethics (Vol. 4).

  4. Cuneo, T. (2007). The normative web: An argument for moral realism. Oxford: Oxford University Press.

    Google Scholar 

  5. Dancy, J. (1986). Two conception of moral realism. Proceedings of the Aristotelian Society, 60(Suppl), 167–188.

    Google Scholar 

  6. Enoch, D. (2003). An argument for robust metanormative realism. NYU Dissertation, available at:

  7. Enoch, D. (2005). Why idealize? Ethics, 115(4), 759–787.

    Article  Google Scholar 

  8. Enoch, D. (2007). An outline of an argument for robust metanormative realism. Oxford Studies in Metaethics, 2(2007), 21–50.

    Google Scholar 

  9. Enoch, D. (forthcoming). How is moral disagreement a problem for realism? The Journal of Ethics.

  10. Enoch, D. (manuscript). Can there be a global, interesting, coherent constructivism about practical reason?

  11. Field, H. (1989). Realism, mathematics and modality. New York: Blackwell.

    Google Scholar 

  12. Gibbard, A. (1990). Wise choices, apt feelings. Cambridge: Harvard University Press.

    Google Scholar 

  13. Gibbard, A. (2003). Thinking how to live. Cambridge: Harvard University Press.

    Google Scholar 

  14. Huemer, M. (2005). Ethical intuitionism. New York, NY: Palgrave MacMillan.

    Google Scholar 

  15. Joyce, R. (2006). The evolution of morality. Cambridge: MIT Press.

    Google Scholar 

  16. Lewis, D. (1986). On the plurality of worlds. Oxford: Blackwell.

    Google Scholar 

  17. Lewis, D. (1989). Dispositional theories of value. Proceedings of the Aristotelian Society, 63(Suppl), 113–137; reprinted in his Papers in ethics and social philosophy (pp. 68–94). Cambridge: Cambridge University Press (2000).

  18. Liggins, D. (2006). Is there a good epistemological argument against platonism? Analysis, 66(2), 135–141.

    Article  Google Scholar 

  19. Mackie, J. (1977). Ethics: Inventing right and wrong. London: Penguin Books.

    Google Scholar 

  20. McGinn, C. (1997). Ethics, evil, and fiction. Oxford: Clarendon Press.

    Google Scholar 

  21. Nagel, T. (1986). The view from nowhere. Oxford: Oxford University Press.

    Google Scholar 

  22. Nagel, T. (1997). The last word. Oxford: Oxford University Press.

    Google Scholar 

  23. Oddie, G. (2005). Value, reality and desire. Oxford: Oxford University Press.

    Google Scholar 

  24. Pryor, J. (2001). Highlights of recent epistemology. British Journal for the Philosophy of Science, 52, 1–30.

    Article  Google Scholar 

  25. Sayre-McCord, G. (1995). Coherentist epistemology and moral theory. In W. Sinnott-Armstrong & M. Timmons (Eds.), Moral knowledge? New readings in moral epistemology (pp. 137–189). Oxford: Oxford University Press.

    Google Scholar 

  26. Scanlon, T. M. (1998). What we owe to each other. Cambridge, MA: Harvard University Press.

    Google Scholar 

  27. Schechter, J. B. (2006). Two challenges to the objectivity of logic. Dissertation, New York University.

  28. Shafer-Landau, R. (2003). Moral realism: A defence. Oxford: Oxford University Press.

    Google Scholar 

  29. Sinnott-Armstrong, W. (2006). Moral skepticisms. Oxford: Oxford University Press.

    Google Scholar 

  30. Sinnott-Armstrong, W., & Timmons, M. (Eds.). (1995). Moral knowledge? New readings in moral epistemology. Oxford: Oxford University Press.

    Google Scholar 

  31. Street, S. (2006). A Darwinian Dilemma for realist theories of value. Philosophical Studies, 127, 109–166.

    Article  Google Scholar 

  32. Timmons, M. (1990). On the epistemic status of considered moral judgments. The Southern Journal of Philosophy, 29(Suppl), 97–131.

    Google Scholar 

  33. Waldron, J. (1992). The irrelevance of moral objectivity. In R. George (Ed.), Natural law theory: Contemporary essays (pp. 158–187). Oxford: Clarendon Press.

    Google Scholar 

  34. Wedgwood, R. (2006). How we know what ought to be. Proceedings of the Aristotelian Society, 106(1), 61–84.

    Article  Google Scholar 

  35. Wright, C. (1988). Moral values, projection, and secondary qualities. Proceedings of the Aristotelian Society, 62(Suppl), 1–26.

    Google Scholar 

Download references


I worked on this paper as a fellow of the Institute for Advanced Studies in Jerusalem. I am grateful for the Institute’s support. I would also like to gratefully ackowledge the support of the Israel Science Fund. I presented this paper at the IAS in Jerusalem, and at the research workshop at the Bar Han University philosophy department, and I thank the participants for the helpful discussions that followed. For helpful conversations and comments on previous versions I am grateful to Erez Aloni, Hagit Benbaji, Pete Graham, David Heyd, Yair Levy, Ofer Malcai, Josh Schechter, Mark Schroeder, Yonatan Shemer, Sigrún Svavarsdóttir, Mark van Roojen, Ralph Wedgwood, and Ruti Weintraub.

Author information



Corresponding author

Correspondence to David Enoch.

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Enoch, D. The epistemological challenge to metanormative realism: how best to understand it, and how to cope with it. Philos Stud 148, 413–438 (2010).

Download citation


  • Moral realism
  • Moral epistemology