Expressivism and the Reliability Challenge


Suppose that there are objective normative facts and our beliefs about such facts are by-and-large true. How did this come to happen? This is the reliability challenge to normative realism. As has been recently noted, the challenge also applies to expressivist “quasi-realism”. I argue that expressivism is useful in the face of this challenge, in a way that has not been yet properly articulated. In dealing with epistemological issues, quasi-realists typically invoke the desire-like nature of normative judgments. However, this is not enough to prevent the reliability challenge from arising, given that quasi-realists also hold that normative judgments are truth-apt beliefs. To defuse this challenge, we need to isolate a deeper sense in which normative thought is not representational. I propose that we rely on the negative functional thesis of expressivism: normative thought does not have the function of tracking normative facts, or any other kind of facts. This thesis supports an argument to the effect that it is misguided to expect an explanation of our access to normative facts akin to the explanations available in regions of thought that have a tracking function. We should be content with explanations of our reliability that take for granted certain connections between our psychology and the normative truths.

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


  1. 1.

    I focus here on practical normativity. For brevity, I will use normative and normativity to mean practically normative and practical normativity, respectively, throughout the paper.

  2. 2.

    The reliability challenge differs in these respects from the evolutionary debunking arguments due to Joyce (2001, 2006) and Street (2006), which do not allow realists to take for granted the truth of our beliefs about objective normative facts.

  3. 3.

    Enoch (2011), Dreier (2012) and Joshua Schechter (“Does Expressivism Have an Epistemological Advantage over Realism?”, unpublished manuscript) offer similar statements of the reliability challenge, inspired by Field’s (1989) challenge to mathematical Platonism. How the failure to explain our reliability would undermine the realism is still a disputed question: it might provide a defeater for the justification we previously had for our realist beliefs, it might undermine the knowledge status of said beliefs, or it might just be that the absence of an explanation of our reliability is an important theoretical cost for realism. I will not attempt to settle this issue here.

  4. 4.

    This might be an advantage that expressivism has over non-naturalist realism, or one that it can provide to non-naturalist realism, depending on how the two views are related. More on this below.

  5. 5.

    Gibbard (2003) endorses certain claims associated with naturalist realism―e.g., that normative concepts are realized by natural properties, and that certain causal explanations involving normative facts are legitimate―but argues that there can be no naturalistic vindication of our normative judgments, remaining thus closer to non-naturalist realism. See also his claim in Gibbard (2011) that quasi-realism mimics the kind of realism that rejects a causal model for normative knowledge. It is an interesting question how exactly Gibbard’s epistemological non-naturalism can be reconciled with his naturalism about properties, but I will assume that there is no deep instability in his views on these matters.

  6. 6.

    This is, for instance, how Gibbard (2011) and Blackburn (“Sharon Street on The Independent Normative Truth as Such”, unpublished manuscript) respond to Street’s (2011) evolutionary challenge to quasi-realism.

  7. 7.

    Schechter (manuscript) makes a similar point about the irrelevance of a normative defense of our reliability in this context. Moreover, the strategy of defending our reliability with normative arguments is not a specifically expressivist resource to begin with. Realists with little sympathy for expressivism can adopt a similar stance.

  8. 8.

    This leaves open that genuine realism might consist in more than mere objectivism. The question of how to draw a divide between quasi-realism and realism has become known as the problem of creeping minimalism, due to Dreier (2004), who proposes that the divide between the two views should be understood in explanatory terms. In Golub (forthcoming), I argue that there need be no explanatory conflict between quasi-realism and anything recognizable as a general notion of realism. See Blackburn (1993), Rosen (1998), and Fine (2001) for earlier discussions of similar issues.

  9. 9.

    Schechter (manuscript) argues that the possibility of our attitudes being mistaken, which even old-school expressivists can allow for, is enough to give rise to a reliability challenge: how is it that we came to have by-and-large correct attitudes? I bracket this issue here, and focus on how the challenge applies to minimalist expressivism.

  10. 10.

    See Blackburn (1993), Essay 8; Blackburn (1998), Ch. 9; or Gibbard (2003), Ch. 9. As already mentioned, I will remain neutral on whether this combination of expressivism and objectivism―which I call quasi-realism, following established usage―can still be considered an anti-realist metaethical view, or has converged with realism.

  11. 11.

    Cuneo and Schechter focus on the reliability challenge. Street extends her evolutionary debunking argument to quasi-realism, but some of the issues she raises carry over to the present discussion.

  12. 12.

    Field (1989) and Fine (2001), among others.

  13. 13.

    Bedke (2013) argues that quasi-realism avoids the epistemological troubles of non-naturalist realism because it does not incur any robust metaphysical commitments. But I do not see how quasi-realists can make this argument. From a thoroughly minimalist standpoint, the quasi-realists’ commitments to truth and objectivity are as robust as they can intelligibly be, and give rise to the question of how it is that our normative beliefs are by-and-large true. (Moreover, even if minimalism provided some epistemological comfort in the face of this challenge, this would not have much to do with expressivism. Non-expressivist realists like Dworkin (1996) and Scanlon (2014) offer similar deflationary accounts of metaethical commitments, so the alleged benefits of minimalism would also be available to them.)

  14. 14.

    See Gibbard (2003), p. 261. Blackburn (1998) makes similar points about the epistemological upshot of expressivism. Note, however, that what is meant to do epistemological work here is not, strictly speaking, the expressivist semantics or any other thesis about language, but a psychological thesis about normative thought.

  15. 15.

    Thanks to Dan Waxman for helpful discussion on this issue. Cf. the arguments in Street (2010), pp. 377–9, and Schechter (manuscript) for similar theses.

  16. 16.

    See Dreier (2012), pp. 285–6.

  17. 17.

    See fn. 8.

  18. 18.

    More precisely, normative terms have the function of conveying and stabilizing our action-guiding attitudes, for the purposes of solving coordination problems in social contexts. See, e.g., the first few chapters of Gibbard (1990).

  19. 19.

    Gibbard (1990), Ch. 6. Gibbard used this idea of natural representation, defined in terms of evolutionary function, to articulate an inflationary notion of fact, and argue that normative thought is non-factual. But the functional claims of expressivism can be adopted by those who accept minimalism about factuality, as shown by Gibbard’s own evolution.

  20. 20.

    See, e.g., Blackburn (1984), p. 169, or Blackburn (1993), p. 7.

  21. 21.

    Blackburn (1984) attributes to realists the thesis that values “are themselves part of the genesis of our beliefs. It would be because values, etc. are distributed in some way around the world, and because we are capable of reacting to them (...) that we moralize as we do.” (pp. 181–2) Similarly, Blackburn (1993) defines realism as follows: “[T]he existence of facts explains the way in which our knowledge expands and progresses: here an explanatory role seems to carry with it an ontological commitment which (...) is surely problematic to the quasi-realist.” (p. 18)

  22. 22.

    Many realists might not assign a tracking function to normative concepts either: for instance, non-naturalist realists who believe that normative facts are not causally efficacious. I discuss this issue in Golub (forthcoming).

  23. 23.

    See Dretske (1981, 1988) and Millikan (1984), among others.

  24. 24.

    See Fodor (1990) for a discussion of challenges to etiological theories of content, e.g. the seeming possibility of mental items that have content but no history, or how to distinguish content-determining from non-content-determining causes.

  25. 25.

    Sinclair (2006) proposes a similar understanding of the negative functional thesis of expressivism, on which the relevant notion of function might involve evolutionary explanations or explanations of current practices. The idea that this functional thesis is not a mere add-on to expressivism but should be seen as one of its central tenets―indeed, as the best way to preserve its negative insights in a minimalist framework―is defended in O'Leary-Hawthorne and Price (1996).

  26. 26.

    To be clear, the Negative Functional Thesis is compatible with holding that our normative claims represent normative facts, in the sense captured by minimalism. On a minimalist account, to say that an utterance of “Genocide is wrong” represents the fact that genocide is wrong is not to say anything about why we use this linguistic item.

  27. 27.

    Such explanations are equally available to non-expressivist realists. See, in particular, Schafer (2010), Enoch (2011), Vavova (2015), and Clarke-Doane (2015) for explanations that appeal to our evolutionary heritage.

  28. 28.

    I am not suggesting that this explanation amounts to an evolutionary vindication of our normative beliefs of the sort available for, say, our beliefs about shapes or temperatures.

  29. 29.

    Several contributors to the debate on the reliability challenge hold that a good explanation of reliability should establish modal security thus understood. See Clarke-Doane (2015), Barkhausen (2016), and Warren (forthcoming).

  30. 30.

    Notably, Dreier (2012) and Schechter (manuscript) do not rely on this modal reading of the reliability challenge. Field (1989), in his reliability challenge to mathematical Platonism, only appeals to sensitivity as a relevant modal constraint.

  31. 31.

    To take a classic example from the philosophy of science: we can explain the occurrence of paresis by appealing to the fact that the patient had latent untreated syphilis, even though the chance that someone with latent untreated syphilis will develop paresis is about one fourth.

  32. 32.

    To be clear, this is not a specific virtue of the evolutionary explanation. The metaphysical necessity of normative facts ensures that the sensitivity constraint would be met by any explanation of our reliability.

  33. 33.

    Clarke-Doane (2015) makes both of these arguments. As he points out, the set of worlds relevant for assessing modal security cannot be too wide, lest we end up counting most of our ordinary beliefs as insecure.

  34. 34.

    An anonymous reviewer suggests that quasi-realists might have more to say in response: if we endorse Gibbard’s view that normative concepts are realized by natural properties and normative facts can play a role in causal explanations (see fn. 5), then the evolutionary explanation need not bottom out in brute connections between our psychology and the facts. We can claim that our evaluative tendencies are, at least in part, the result of interactions with normative facts―for instance, that the natural property that constitutes goodness causally influenced our tendency to value reciprocity. I cannot explore this suggestion in depth, but I suspect that neither Gibbard nor his opponents would think of these causal explanations as fully addressing any lingering concerns driving the reliability challenge. If normative thought does not have the function of tracking any facts, then the best explanation of why we have certain evaluative tendencies―one that would apply in a wide range of evolutionary pathways―does not involve normative facts. Even if our evaluative tendencies were causally influenced by normative facts, this does not give us sufficient insight into why we ended up on a truth-conducive path, given that other evaluative tendencies would have been equally compatible with the coordinating function of normative thought. Only a tracking account of normative thought could meet this explanatory demand.

  35. 35.

    Arguably, astrological beliefs are in fact more modally insecure than normative beliefs, but we can imagine a scenario in which they were just as stable, without being caused by astrological facts.

  36. 36.

    See also Nagel (1986), p. 141.


  1. Barkhausen M (2016) Reductionist moral realism and the contingency of moral evolution. Ethics 126(3):662–689

    Article  Google Scholar 

  2. Bedke M (2013) Ethics makes strange bedfellows: intuitions and quasi-realism. In: Haug M (ed) Philosophical methodology: the armchair or the laboratory? Routledge, New York

    Google Scholar 

  3. Blackburn S (1984) Spreading the word. Clarendon Press, Oxford

    Google Scholar 

  4. Blackburn S (1993) Essays in quasi-realism. Oxford University Press, New York

    Google Scholar 

  5. Blackburn S (1998) Ruling passions. Oxford University Press, New York

    Google Scholar 

  6. Clarke-Doane J (2015) Justification and explanation in mathematics and morality. In: Shafer-Landau R (ed) Oxford studies in metaethics, vol 10. Oxford University Press, Oxford

    Google Scholar 

  7. Cuneo T (2008) Moral realism, quasi-realism, and skepticism. In: Greco J (ed) The Oxford handbook of skepticism. Oxford University Press, Oxford

    Google Scholar 

  8. Dreier J (2004) Meta-ethics and the problem of creeping minimalism. Philos Perspect 18:23–44

    Article  Google Scholar 

  9. Dreier J (2012) Quasi-realism and the problem of unexplained coincidence. Anal Philos 53:269–287

    Article  Google Scholar 

  10. Dretske F (1981) Knowledge and the flow of information. MIT Press, Massachusetts

    Google Scholar 

  11. Dretske F (1988) Explaining behavior: reasons in a world of causes. MIT Press, Cambridge

    Google Scholar 

  12. Dworkin R (1996) Objectivity and truth: you'd better believe it. Philos Public Aff 25:87–139

    Article  Google Scholar 

  13. Enoch D (2011) Taking morality seriously: a defense of robust realism. Oxford University Press, Oxford

    Google Scholar 

  14. Field F (1989) Realism, mathematics and modality. Basil Blackwell, Oxford

    Google Scholar 

  15. Fine K (2001) The question of realism. Philosopers' Imprint 1:1–30

    Google Scholar 

  16. Fodor JA (1990) A theory of content and other essays. MIT Press, Cambridge

    Google Scholar 

  17. Gibbard A (1990) Wise choices, apt feelings: a theory of normative judgment. Harvard University Press, Cambridge

    Google Scholar 

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

    Google Scholar 

  19. Gibbard A (2011) How much realism? Evolved thinkers and normative concepts. In: Shafer-Landau R (ed) Oxford studies in metaethics, vol 6. Oxford University Press, Oxford, pp 33–51

    Google Scholar 

  20. Golub C (forthcoming) Expressivism and realist explanations. Philos Stud 1–25. doi:10.1007/s11098-016-0764-6

  21. Joyce R (2001) The myth of morality. Cambridge University Press, Cambridge

    Google Scholar 

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

    Google Scholar 

  23. Millikan RG (1984) Language, thought, and other biological categories. MIT Press, Cambridge

    Google Scholar 

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

    Google Scholar 

  25. O'Leary-Hawthorne J, Price H (1996) How to stand up for non-cognitivists. Australas J Philos 74:275–292

    Article  Google Scholar 

  26. Rosen G (1998) Blackburn's essays in quasi-realism. Noûs 32(3):386–405

    Article  Google Scholar 

  27. Scanlon TM (2014) Being realistic about reasons. Oxford University Press, Oxford

    Google Scholar 

  28. Schafer K (2010) Evolution and normative scepticism. Australas J Philos 88:471–488

    Article  Google Scholar 

  29. Sinclair N (2006) The moral belief problem. Ratio 19:249–260

    Article  Google Scholar 

  30. Street S (2006) A Darwinian dilemma for realist theories of value. Philos Stud 127:109–166

    Article  Google Scholar 

  31. Street S (2010) What is constructivism in ethics and metaethics? Philos Compass 5:363–384

    Article  Google Scholar 

  32. Street S (2011) Mind-independence without the mystery: why quasi-realists can't have it both ways. In: Shafer-Landau R (ed) Oxford studies in metaethics, vol 6. Oxford University Press, Oxford

    Google Scholar 

  33. Vavova K (2015) Evolutionary debunking of moral realism. Philos Compass 10:104–116

    Article  Google Scholar 

  34. Warren J (forthcoming) Epistemology versus non-causal realism. Synthese 98:271–295

Download references


I am grateful to Brian Ballard, Michelle Dyke, Tom Nagel, Daniel Skibra, David Velleman, several anonymous referees, the editors of this special issue, and especially Hartry Field, Josh Schechter, and Sharon Street, for their helpful comments on previous versions of this paper. Many thanks as well to audiences at New York University, Bilkent University, and the 2016 Conference on Objectivity in Ethics at the University of Utrecht for their useful feedback. I have also greatly benefited from discussions with Max Barkhausen, Harjit Bhogal, Paul Boghossian, Jared Warren, and Mike Zhao, and especially from many conversations with Dan Waxman.

Author information



Corresponding author

Correspondence to Camil Golub.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Golub, C. Expressivism and the Reliability Challenge. Ethic Theory Moral Prac 20, 797–811 (2017).

Download citation


  • Expressivism
  • Quasi-realism
  • Reliability challenge
  • Function
  • Tracking