Kugler, Jost, and Noorbaloochi recently published a critique of Moral Foundations Theory in Social Justice Research. A couple of people have asked me what I think of this critique.
In 2011 I provided a signed review of a previous version of Kugler et al.’s paper for a different journal. I think this review is still an accurate representation of my thoughts on these issues. In a nutshell: I agree with the authors that at points in Haidt’s and my writings we have blurred the line between a. the descriptive moral psychology MFT is concerned with, and b. normative recommendations that it is not concerned with. They rightly point out places where we seem to have been trying to derive ought from is. However, I find the entire empirical project of this paper to suffer from this same blurring of normative and descriptive. The authors attempt to derive isn’t from oughtn’t.
I provide my review here for those interested in the often-blurred lines between normative and descriptive moral psychology. (There’s also some discussion of the normative implications of flatulence, if you’re into that sort of thing.) This is NOT intended to be a review of the published paper, which has changed since the version I reviewed; it’s merely provided to show my thoughts on descriptive vs. normative as applied to MFT, authoritarianism, and ideology. (For more on the need to make a clear distinction between the normative and the descriptive, see section 4.1.5 of this paper, or listen to this cringe-tastic interview.)
A scientific theory is only as good as its critics. Fortunately for Moral Foundations Theory (MFT), John Jost has been a constructive, supportive, yet nevertheless vigorous and tireless critic of both MFT’s theoretical model and its very approach to studying morality and ideology. His criticisms have led to new ideas and research by myself and other moral foundations researchers (e.g., current efforts to measure moral intuitions of outrage at bullying and domination not simply derived from Fairness-related considerations). So I am pleased to be able to offer my opinion about this manuscript by Kugler & Jost offering “an empirical critique of ‘moral foundations’ theory.”
Although much of my work is the target of the authors’ critique, I found some of their criticisms valid. Specifically, I agree with the authors that at points in Haidt’s and my writings we have blurred the line between a. the descriptive moral psychology MFT is concerned with, and b. normative recommendations that it is not concerned with. They rightly point out places where we seem to have been trying to derive ought from is. However, I find the entire empirical project of this paper to suffer from this same blurring of normative and descriptive. The authors attempt to derive isn’t from oughtn’t. Below I give my view of this manuscript’s confusions, and then – in the spirit of fairness – turn to my own confusions.
1. This manuscript’s confusions. The authors provide correlational evidence that ideological differences in Ingroup, Authority, and Purity concerns are mediated by authoritarianism, and ideological differences in Harm and Fairness are mediated by opposition to equality. From this they conclude, “Thus, it seems premature to abandon the notion that morality consists in fulfilling the requirements of “justice, rights, and welfare” and to argue instead for a “broadening” of the moral domain to incorporate ingroup, authority, and purity concerns” (p. 19). The leap seems to be in this “Thus” – because these foundations are related to authoritarianism, which is normatively bad, they cannot be part of a descriptive account of morality (and since Harm and Fairness are related to low opposition to equality, which is normatively good, they can stay in). This reflects a deep confusion about the aims of MFT (and the aims of any scientific understanding of human morality), which we have written about extensively, including in the papers cited. The confusion may be in the fact that the words “moral” and “morality” can be used normatively – to describe something that is morally good, that others should promote and protect – but it can also be used descriptively, as in scientific descriptions of the domains and processes of human morality, intended to better understand them.
The patterns of correlations described in this paper are entirely consistent with MFT’s descriptive account of the domains, concerns, and intuitions involved in human morality. It would be shocking to me if the moral concerns about respect for authorities and traditions didn’t relate to RWA. In fact, the Graham et al. (in press) paper cited in this manuscript makes clear that both RWA and SDO were used as external validation criteria in the very development of the Moral Foundations Questionnaire. Use of these scales as validity criteria is no more a normative defense of authoritarianism than our use of the Disgust Sensitivity Scale is a normative defense of disgust sensitivity. (Further, the findings by Schnall, Pizarro, and others that fart sprays can increase the severity of moral judgments – which I take to be descriptive evidence for including Purity concerns in accounts of human morality – do not necessitate any normative endorsements or condemnations of flatulence.) The MFQ and MFT are attempting to measure and describe the full range of moral concerns people have – the good, the bad, and the heinous – which I see as a scientific step beyond previous treatments of morality (e.g. Kohlberg, Turiel) that only considered moral concerns with which the scientists normatively agreed.
The authors seem to be saying that since some moral foundations relate to authoritarianism, they can’t be a descriptive part of morality, because authoritarianism is so UNrelated to morality. And yet the RWA scales have plenty statements of moral concern. To use one item the authors give as an example: “Our country will be destroyed someday if we do not smash the perversions eating away at our moral fiber and traditional beliefs.” Just because this is an item that the authors and I may normatively disagree with does not mean that people who agree with it are not expressing a moral concern or belief. Part of the scientific benefit of MFT is understanding how there can be deep intuitive moral passions associated with authoritarianism, system justification, prejudice, violence, and even genocide. Human morality isn’t always normatively good.
2. My confusions. Haidt and I (2009) wrote at length about how binding moral concerns (especially Authority) can be wrapped up in system justification, and we tried to make clear that this did NOT mean system justification was a normatively good thing (see passage quoted below). However, most of our writing about this last point (human morality isn’t always normatively good) is in a paper (Graham & Haidt, in press) that I should not expect the authors to have read. In this paper we make the point that ALL moral concerns (including Harm and Fairness, but especially Ingroup, Authority, and Purity) can be dangerous, and can lead to violence in the service of protecting some morally sacred object, person, or idea. Here it is clear that what is descriptively moral can be normatively bad, disastrous, heinous, or immoral. But I have to agree with the authors that this has not always been clear in our previous writings (or in the press). I think it perfectly defensible for Haidt and I (2007, 2009) to have argued that MFT’s descriptive account of the full range of human moral concerns can help academic liberals understand – descriptively – the moral bases of many conservative (or religious, or non-Western) opinions. But when we claimed that the binding foundations were “moral (instead of amoral, or immoral)” this confused the point, and allowed for the authors’ reading of this to say “these foundations are normatively good, not normatively neutral or normatively bad.” Kugler and Jost are right to criticize us for this blurring of the normative and the descriptive.
Most of the other quotes from our writings I would stand behind, because we are speaking descriptively (i.e., these concerns are part of human morality) and not normatively (i.e., these concerns are good and right and we should all adopt them), as the authors took them. I think the authors do not “get” conservative morality descriptively, but this does not mean I think they should adopt it normatively. But one passage quoted on page 17 had me rushing back to the original. Did we really claim, based on our data and descriptive theory of morality, that embracing conservative moral values could lead to “a more healthy, human, and satisfying place overall”? We did, in a way, and the context from which the quote was taken is illustrative:
- “Our goal in offering this descriptive account is not to claim that conservative morality is superior to liberal morality, normatively speaking. Our normative position is a kind of consequentialism—we think moral systems should be judged by the quality of the worlds they lead to. We believe the benefits of modernity have been enormous, and that it is neither possible nor desirable to reduce ethnic diversity, eliminate existing technologies, or otherwise return to the Gemeinschaft social systems that prevailed centuries ago. But our consequentialism leads us to ask whether there might be some hidden utility in the three traditional foundations. Even from a liberal perspective, in which all that matters is the welfare of individuals (particularly those who are least well off), might there be some paradoxical benefits to individuals of social policies that do not put the welfare of individuals first? The social policies favored by conservatives, shown in the right-hand column of Table 15.1, are, broadly speaking, Durkheimian policies. They increase the cohesion and stability of communities. They therefore also increase the social capital (Coleman, 1988) of communities, which includes the dense networks of obligation and trust, social information channels, and effective norms and sanctions for deviance. By extension, we might say they increase the symbolic capital, too—the culturally evolved network of shared symbols and meanings from which people construct their identities and make sense of their worlds. (See Appiah, 2005, on the challenges of identity construction for liberalism.) Given the many arguments from psychology and sociology about the costs of anomie and hyperindividualism (Bellah et al., 1985; Leary, 2004; Schwartz, 1986), and the benefits of close, enduring social bonds and shared meanings (Baumeister & Leary, 1995), we believe that a modern society that makes some use of the three traditional foundations might—at least in theory—be a more humane, healthy, and satisfying place overall than a society that builds its values and policies exclusively on the first two foundations. We are not confident that the traditional foundations offer such benefits in practice, but we believe that traditional and conservative ideas are frequently mischaracterized, prematurely dismissed, or simply ignored by many psychologists, philosophers, and other academics.”
As this passage makes clear, we expressed (in an uncharacteristically tentative and unconfident way) a normative view that is mostly in line with what the authors seem to profess (binding foundations are dangerous in modern society) but diverges in our allowing for the possibility that there might be some benefits, in theory, to making some use of the binding foundations. The authors could certainly disagree with this normative claim, and reply that modern society should make absolutely no use of these foundations. But the point here is that we were clearly distinguishing these normative wanderings from MFT’s descriptive account of human morality.
3. Recommendation. Despite its title, this manuscript does not offer an empirical critique of Moral Foundations Theory, because the empirical findings do not contradict any of the scientific claims or predictions of the theory. Because [journal] is an empirical journal, I thus do not think this manuscript is appropriate for it. However, as detailed above, I do find merit in some of the authors’ normative claims, and while I find some of their quotes unfair, I do think they are justified in criticizing points at which Haidt and I strayed into normative territory based on our descriptive findings. If this exchange serves to clarify for all of us (and especially for others) the distinction between normative and descriptive accounts of morality and ideology (much as the Tetlock-Sears debate did several years ago) then there is a value in this being a part of the public record. I think this would be more appropriate for the Jost-Haidt exchange planned for Perspectives on Psychological Science. Regardless of the outlet, however, I think that the use of this manuscript’s data to claim an empirical critique of MFT will only add confusions, rather than clear them up.