Abstract
Recent work by Matthew Bedke and Max Hayward develops a new attack on metaethical non-naturalists: that they are committed to an immoral state of mind, because they must be willing to change their mind about the moral importance of certain actions given possible evidence about the layout of the non-natural realm. For example, they must be willing to decrease their credence that torturing babies is bad, if they ever get evidence that torturing babies is not in the extension of a non-natural property. The present paper aims to both develop the strongest version of this argument and propose a response to it. The first section argues that the strongest version of this attack does not depend on whether accepting non-naturalism commits one to believing a certain kind of conditional proposition. Rather, it depends on whether accepting non-naturalism commits one to having a certain kind of conditional credences. Moreover, the moral criticism at hand does not hinge on the causal inertness of non-natural properties. And in so far as the argument does not depend on causal inertness, the people threatened by immorality are not just non-naturalists, but metaethicists in general (Sect. 2). This is because the worry hinges instead on particular relationships of epistemic dependence between our metaethical and some of our ethical views. But it is not clear that those relationships obtain; we could adopt metaethical methodologies that preclude morally problematic dependence of relevant first order views on evidence about our metaethical theories (Sect. 3).
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
More precisely, Bedke’s argument is that the non-naturalists are committed either to an immoral state of mind, or an irrational state of mind. I will present a simplified version of the argument focusing on the immorality horn. When talking of “the non-naturalist” and their commitments, I will assume−unless otherwise noted−that the non-naturalist is rational.
Hayward puts this point quite nicely: “To understand the totality of a person’s moral outlook, it’s not enough just to know their directly world-directed moral attitudes and beliefs. Our moral outlooks are revisable. There are circumstances where we think we ought to change our views. So we have higher-order moral judgments about the conditions in which we ought to change our moral judgments. These higher-order judgments form a major part of everyday morality.” (p. 899).
(p.899).
See Melis Erdur (2016) for an argument of this kind.
This includes both what Golub calls “epistemic” and “methodological” versions of the argument.
Thanks to an anonymous reviewer for asking me to talk about this.
Here is how Hayward puts it: “CONDITIONAL PREMISE: Only if there are non-natural moral truths do some things matter. […] It is this conditional to which I object.” (p. 904).
For a response along the lines of the second direction, see Enoch (2021), to be discussed shortly.
The general chemist example, and the relevant theoretical and first order claims are borrowed from Bedke (2020). However, my example is tweaked in a crucial way: Bedke does not talk about versions of the case with particular explicit relationships of epistemic dependence. He just talks about a general approach akin to reflective equilibrium in balancing our first order claims about samples and theoretical claims in science in general. But this approach misses the importance of the subset of cases where we ought not make our credence in 1 conditional on our credence in 2 at all.
(Lewis, 1976) is one of the seminal papers discussing such a surprising trade-off.
Enoch (2021) develops a similar point. However, as we will soon see my answer diverges significantly from his, as I think this remark provides only the beginnings of an answer to the immorality charge.
This feature of one’s conditional credences may seem like a weak basis for charging them with immorality. However, it becomes more understandable once we take the immorality charge as a degreed charge; the more one is willing to lower their credence in the badness of torturing babies in light of metaethical evidence, the more the immorality charge is supposed to apply to them. As Bedke (2020) puts it: “Because the difference between a psychology poised to track the non-natural and one poised to abandon non-naturalism is a matter of degree, so too their difference in moral repugnance is a matter of degree.” (p. 1034).
Enoch’s example is of a physicalist who believes the conditional: If I don’t have a brain, I don’t have thoughts. Such a (rational) physicalist will not be willing to draw a Modus Ponens inference from such a conditional, upon being presented with evidence that she does not have a brain.
See Footnote 13 above. Bedke also clearly stresses that the immorality worry can come up for the non-naturalist’s credences as opposed to full-blown beliefs in (2022).
See Bedke (2022) p. 10: “[…] pick your favorite case where your confidence/justification concerning some normative matter is lower than your confidence/justification in believing non-naturalism. […] For you, these normative issues are rationally required to be left hostage to a non-natural realm. None of them are modus ponens junk, understood as an exceptional class of conditionals that cannot be used in those inferences.”.
(pp. 904–905).
(p. 1033).
In this discussion, by ‘our pretheoretic first order moral beliefs’ I shall mean our beliefs prior to (or, more accurately, bracketing) the influence and evidential force of specifically metaethical theorizing and related considerations.
Hayward (2019), p. 905.
The view I am describing here most closely matches the one in Boyd (1988).
An example of other such complex clusters of properties is the cluster of properties that constitutes healthfulness. While our use of “healthful” clusters around a number of natural properties, we do not have perfect a priori knowledge of what exactly constitutes healthfulness. Empirical discoveries are essential to learning what foods or habits are healthful.
Thanks to an anonymous reviewer for bringing this up.
If one finds the metaphysical oracles examples unhelpful, we can easily construct other troubling variants of the example: Imagine you are a non-naturalist who believes moral properties are irreducibly authoritative, but you find yourself in the following evidential situation: Last night you took a pill which benefits one’s philosophical thinking for a night, though it doesn’t make one infallible. Βut this comes at a cost: the morning after, anyone who takes this pill has no memory of the details of the arguments they uncovered the night before, only remembering key lemmas and conclusions. You took the pill, and reflected on the nature of reduction, authoritativeness, and moral properties. You wake up the morning after taking the pill, and you remember forming the conclusion that torturing babies has no irreducibly authoritative property after all, even though torture has all of the natural properties that you took to be morally relevant. How are you to change your beliefs?
And it seems like there should be plenty of worries. It is no accident that most metaethical theories do not provide such a specific reduction base. For one, it seems like to be able to provide one, we would have to solve both first order ethics and metaethics at the same time. And for another, it seems like we would be at a tough spot to avoid Moore’s (1903) open question argument (and some successors, like the Moral Twin Earth arguments by Horgan and Timmons (1991)).
It seems that people who originally supported this kind of view (e.g. Harman) later switched to non-cognitivist and expressivist views. The immorality worry is a little more complicated to levy against expressivism, since expressivism does not try to reduce morality to a set of properties that we have views about. Instead, it claims that what we do with our moral language is express preferences, or something of the sort. There is not enough space to go into detail about exactly how the immorality worry would arise here. However, most spelled-out versions of the view — as seen in e.g. Gibbard (2003) and Blackburn (1984)— are epistemically non-trivial in my sense, and build a logic of preferences which claims that agents are under rational pressure to update their moral views in light of certain kinds of evidence. In particular, agents are under rational pressure to adopt what they have evidence is a rationally idealized and maximally coherent version of their preferences. Because of this, expressivists will be under rational pressure to conditionalize their moral views on certain kinds of evidence which do not look like intuitively first order morally relevant evidence —particularly evidence about what one’s rationally idealized and maximally spelled out preferences would look like, and whether they would allow torturing babies. For this reason, I believe such expressivist views also face versions of the charge.
(p. 525).
Bedke (2020) pp. 1037-1038.
Enoch responds to this complaint by saying that, if a metaethicist is fully certain in their first order belief, e.g. that torturing babies for fun is wrong, then their unilaterally abandoning their metaethical theory in the face of new evidence is rationally unproblematic. This is because full certainty in a proposition entails not updating on it in light of any new evidence. If we are certain that torturing babies for fun is wrong, then the only thing we could budge on in light of the new evidence is the metaethical theory itself. While this might be true, it does not address the crux of the issue; the point that Hayward and Bedke are trying to stress is not that we (morally) should be certain that torturing babies for fun is wrong, but that we (morally) should only update our view on this in light of intuitively morally relevant evidence. It seems that if, for example, I got evidence that babies are automata with nothing even remotely resembling phenomenal or affective experiences, it would be fine to update my credence that torturing babies is wrong. If we want to uphold this intuition that even many of our first order moral views are in principle updateable, we must think that we are not fully certain in them. But we may still think that distinctively metaethical evidence is just not the right sort of evidence on which to update.
Golub (2021) proposes a move that may prima facie seem similar to the bottom-up-only approach, and which he argues could dissolve the immorality worry. According to Golub, the realist’s problems would be solved if they maintained that only evidence relevant to first order normative propositions can possibly be metaethically relevant evidence. However, this interesting line is crucially different from the bottom-up-only approach; the bottom-up-only methodology claims merely that evidence for metaethical views cannot undermine rational first order ethical views, but places no restrictions on what kinds of evidence can undermine metaethical views themselves. Golub’s methodology claims that only evidence relevant to ethical views can be evidence for/against (and thus can undermine) metaethical views. So how is Golub’s suggested methodology supposed to help the non-naturalist who gets evidence —say by an extremely reliable metaethical oracle, or a community of epistemic peers and superiors— that torturing babies for fun does not have a non-natural property? The claim might be that, when the non-naturalist gets evidence that appears to support that torture does not have a non-natural property, they do not get evidence that they should use to update their first order ethical views. And therefore, if abiding by Golub’s methodology, they do not get evidence that they should use to update their metaethical views either. But this is quite puzzling. How could we rationally justify ignoring this evidence (by reliable oracles or peers and superiors) that seems at the very least directly relevant to metaethical views? Aside from being counterintuitive, this would require a robust and so far unmotivated distinction between our methods for all kinds of philosophy and our methods for metaethics. If I get evidence that torture does not have a non-disjunctive, joint-carving, or reference-magnetic property, should I change my metaphysical views based on that, but leave my metaethical views intact due to this restriction on metaethical methodology? And what happens if the two are theoretically connected (if, for example, I believe that non-natural properties are non-disjunctive, joint-carving, or reference-magnetic)? At the end of the day, it seems difficult to justify why some cases of what looks like paradigmatically metaethically relevant evidence should not be considered when forming our metaethical views.
Considerations of this type are reminiscent of Quine’s (1951) general suspicion against the analytic-synthetic distinction, and the unrevisable data vs. revisable theory distinction. If one thinks that such distinctions are no good, it seems that they don’t have the bottom-up-only response to the immorality charge available to them.
(p. 1698).
This kind of view is suggested by old writings of Foot’s (1970, 1972). Foot thought that some imperatives in morality are “objective” in the sense that they can be known by means of mastery of the moral concepts (like some imperatives of etiquette are objective in the same sense). But this does not entail that all moral questions are settled by understanding the confines of the moral concepts (Foot mentions debates on the relative values of duties to not harm and duties to aid as examples of genuine questions that may not be settle-able by conceptual mastery with the moral concepts). On this view, conceptual mastery gets us genuine moral knowledge, but it can’t get us all the moral knowledge we would like. This is a short step away from taking conceptual moral knowledge as epistemically prior to metaethical knowledge, but acknowledging that there will be non-conceptual matters of moral facts that will not share this epistemic standing. A different version of this view is also proposed in Cuneo and Shafer-Landau (2014).
Potential costs to this methodological route may have to do with the general shortcomings of a picture on which we have a priori access to moral truths by virtue of conceptual competence. See Moore (1903) and more recently Horgan and Timmons (1991) for arguments putting pressure on genuine moral knowledge through mere competence with moral concepts, or at least against the naturalists’ picture of such moral knowledge.
One worry is how we can be sure about the generality of such a solution. If indeed there is no general story to be had about why our dearly held justified first order views are always epistemically prior to our metaethical views, how can we be sure that there will be no cases of the type that Hayward and Bedke are proposing where the relevant epistemic priority won’t hold? Perhaps the particular case involving evidence that torturing babies lacks a non-natural property is one where we think the first order view is epistemically prior. But why should we think no similar immorality example will arise at all? One surface-level and not fully satisfying remark is that the burden of proof is on the side that claims that there is an immorality problem: they are supposed to present a convincing example, while the examples currently in the literature seem to be plausibly solved by the epistemic priority claims presented here. Perhaps a more explanatory though more tentative remark is that our intuitions about epistemic priority in such cases seem to go hand in hand with our intuitions about what kinds of conditionalizations are morally acceptable. In other words, when we are presented with such cases and think “Of course the metaethicist shouldn’t change their mind about the moral importance of X action in light of Y evidence!” it seems that this thought naturally reflects both our intuitions about the immorality of changing one’s mind about X action in light of Y evidence, and the epistemic priority that we think the first order view about X bears in that case. Unless we have good reason to think that there will be cases of this structure where moral and epistemic shoulds are obviously at odds, it is reasonable to think that the cases where we have the intuition that we morally should not conditionalize our first order views on metaethical ones are precisely cases where there would also be something epistemically amiss with doing so. If indeed there are clear versions of the case where the moral and epistemic shoulds are clashing, this would at the very least be an interesting and surprising result; it would mean that following the epistemically right way to do metaethics sometimes renders metaethicists immoral.
Thanks to an anonymous reviewer for bringing this issue up.
Note however, that the methodologies are at least in principle compatible with metaethical error theories that claim to not have such revisionist first order moral consequences, a la Mackie -at least if such error theories are indeed coherent.
Indeed, in a recent paper (2021), Farbod Akhlaghi argues against a set of metaethical theories precisely on the grounds that accepting them requires us to preclude the epistemic possibility of error theory. It is beyond the scope of this paper to directly tackle that argument, but suffice it to say that, between Akhlaghi’s argument and the Bedke-Hayward argument, something needs to give. If it is indeed immoral to have credences that conditionalize one’s core moral convictions on evidence about metaethics, then it would be immoral to take metaethical theorizing to be the kind of enterprise able to reveal that nothing morally matters. If on the flipside both our theories and methodologies ought to allow for the epistemic possibility of error-theory, then our theories and methodologies ought to allow for possible metaethical evidence that torturing babies is not bad.
This includes both revisionist error theorists and those who do not in fact endorse error theory, but believe that the proper metaethical methodology allows for the epistemic possibility of a revisionist error theory. In other words, those who believe our metaethics could reveal to us that torturing babies is not morally wrong.
References
Akhlaghi, F. (2021). On the possibility of wholesale moral error. Ratio, 34(3), 236–247.
Bedke, M. S. (2020). A dilemma for non-naturalists: Irrationality or immorality? Philosophical Studies, 177, 1027–1042.
Bedke, M. S. (2022). Kowtowing to a non-natural realm. Journal of Moral Philosophy, 19, 1–18.
Blackburn, S. (1984). Spreading the word: Groundings in the philosophy of language. Oxford University Press.
Boyd, R. (1988). How to be a moral realist. In G. Sayre-McCord (Ed.), Essays on moral realism (pp. 181–228). Cornell University Press.
Cuneo, T., & Shafer-Landau, R. (2014). The moral fixed points: New directions for moral non-naturalism. Philosophical Studies, 171, 399–443.
Enoch, D. (2011). Taking morality seriously: A defense of robust realism. Oxford University Press.
Enoch, D. (2021). Thanks, We’re good: Why moral realism is not morally objectionable. Philosophical Studies, 178, 1689–1699.
Erdur, M. (2016). A moral argument against moral realism. Ethical Theory and Moral Practice, 19, 591–602.
Foot, P. (1970). Morality and art. Proceedings of the British Academy, 56, 131–144.
Foot, P. (1972). Morality as a system of hypothetical imperatives. The Philosophical Review, 81(3), 305–316.
Gibbard, A. (2003). Thinking how to live. Harvard University Press.
Golub, C. (2021). Is there a good moral argument against moral realism? Ethical Theory and Moral Practice, 24, 151–164.
Harman, G., & Thompson, J. (1996). Moral relativism and moral objectivity. Blackwell.
Hayward, M. (2019). Immoral realism. Philosophical Studies, 176(4), 897–914.
Horgan, T., & Timmons, M. (1991). New wave moral realism meets moral twin earth. Journal of Philosophical Research, 16, 447–465.
Lewis, D. (1976). Probabilities of conditionals and conditional probabilities. The Philosophical Review, 85(3), 297–315.
Moore, G. E. (1903). Principia ethica. Cambridge University Press.
Pryor, J. (2000). The skeptic and the dogmatist. Nous, 34(4), 517–549.
Quine, W. V. O. (1951). Two dogmas of empiricism. The Philosophical Review, 60(1), 20–43.
Schechter, J. (2018). Explanatory challenges in metaethics. In D. Plunkett & T. McPherson (Eds.), Routledge Handbook of metaethics (pp. 443–459). Routledge.
Williams, B. (1972). Morality: An introduction to ethics. Cambridge University Press.
Acknowledgements
Many thanks to David Christensen, James Dreier, Elizabeth Miller, the participants of Dissertation Workshop at the Brown Philosophy Department, and two anonymous referees for helpful conversations and valuable feedback on earlier drafts.
Funding
The author did not receive support from any organization for the submitted work.
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author has no relevant financial or non-financial interests to disclose.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Zormpalas, S.O. (Im)moral theorizing?. Philos Stud 180, 1881–1903 (2023). https://doi.org/10.1007/s11098-023-01939-1
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11098-023-01939-1