[This is one of the finalists in the 2022 book review contest. It’s not by me - it’s by an ACX reader who will remain anonymous until after voting is done, to prevent their identity from influencing your decisions. I’ll be posting about one of these a week for several months. When you’ve read them all, I’ll ask you to vote for a favorite, so remember which ones you liked - SA]

Introduction

I didn’t read The Righteous Mind for a long time after I knew about it. This was partly because I don’t get through much in the way of new reading material. A friend of mine told me yesterday that he’d read something like 130 new books this year. That was on February 20th. I’ve read one, and it was The Righteous Mind. Another friend releases Spotify playlists every Friday of the greatest hits from the many new albums he’s listened to that week. I’ve listened to one new album this year. It was Selling England by the Pound, which he recommended. It was my first foray into Genesis and I loved it. I now have to keep telling him that, no, I haven’t listened to any more Genesis or Peter Gabriel since then, but I’m sure I’ll get round to it within the year.

This is to make the point that I’m starting from a low base rate of reading things. I still think I put off reading The Righteous Mind for unusually long, though, given how interesting I find the subject matter. The reason, I think, is that I sort of felt like it wouldn’t be very interesting, because I’d kind of know and agree with all of it already. Given how slowly I absorb new books, I like them to either be challenging, or a new and informative look at things I just don’t know very much about yet. I don’t mean to come across as some sort of sage of intellectual piety and good habits of mind who scorns the comforting embrace of being validated. I read plenty of political bloggers that I mostly agree with! I just don’t tend to use books for that.

I had a general feeling that The Righteous Mind sits in the background of a lot of the political or meta-political content that I know and love. It had the aura of a sort of foundational text for the loose family of political views and affiliations I have. I don’t consider myself a centrist, which I think is how Haidt identifies himself, but I do share his disdain for tribal partisan politics and general sense that so much of what passes for political debate is just people yelling foundational definitional disagreements past each other mostly for the benefit of their own fans. I felt like I’d probably picked up most of its insights further downstream, and wouldn’t get much out of reading it.

I was completely wrong. I found The Righteous Mind to be a frustrating book, caught up in a disastrous confusion about what its central points were. I regularly found myself thinking I could make its arguments better than it was, leaving me with that awful feeling of listening to someone else make weak arguments for a position you hold and feeling the people around you being unconvinced. I think if I were among the people against whom it mainly seems to be arguing, I would find it unconvincing.

Most of all, though, it just feels horribly outdated only a decade after it was published, and that’s a real death blow for an attempt to get beyond the ephemera of partisanship and talk about political differences in a more fundamental way. At its core the book is an attempt to go beyond the surface partisanship of R vs D, and dive into the underlying moral psychology that Haidt thinks drives those differences. Unfortunately, just ten years from writing, his underlying structure looks almost as disposable and skin-deep as the latest scandal or wedge policy issue.

This means I enjoyed reading it a lot more than I expected! Reading it made me think harder about a whole bunch of topics. Working out exactly what I think is wrong with it was a really interesting and worthwhile experience. I spent a lot of the time I was “reading” it just staring out the window, working through things in my head. I ended up with a lot to say about it, which is what led to my writing this review.

Run through the three sections

For all I thought its argumentation was muddled and unclear, the way the book is structured is very clear and helpful. It’s divided into three sections, each of which has clearly stated main points it sets out to prove. Within those sections, each chapter ends with a brief summary that makes it clear what Haidt wants you to focus on and take away.

I’m a big fan of this. I read a lot of political or philosophical writing that seems almost designed to make it unclear what the author is trying to say, and I find it extremely wearing. I appreciate an author who will just state the claims and then try to back them up, rather than meandering about between anecdote, argument, autobiography, and rant and hoping you stitch together something out of the vague vibes. Haidt isn’t as unbelievably rigorous about this as Plantinga or Parfit, whose numbered key statements make piecing together their arguments and examining them in detail a pleasure, but the chapter summaries and clear three-part structure are great.

I’m going to run through the three sections fairly briefly, making what I think are the key points in outline format.

Intuitionism

The first part of the book lays out some history of moral psychology and then makes the case for Haidt’s intuitionism. At its core, this is the idea that human beings don’t naturally reason their way to moral decisions, but make them pretty much based on instinct, then cobble together whatever reasoning they need in order to explain their decisions to others and justify them internally.

This is the “rider and elephant” analogy that’s seeped well into rationalist thought, so a lot of it was familiar to me. It’s nicely written in order to gradually guide someone who might be fairly new to the idea that people might not be fully rational agents through the arguments, using Haidt’s own career working in moral psychology as a framework for doing so.

Moral Foundations

After making the case for intuitionism, the middle section of the book focuses on the core of Haidt’s research, his moral foundations model. By analogy to taste receptors, the idea is that we don’t have just one intuitive impulse producing our snap moral judgements that we then rationalise, but five (later this is expanded to six, but I’ll follow the book’s process of introducing that one later, as it was added later in Haidt’s own research). The five foundations are:

  • Care: Basic utilitarian idea that suffering is bad;

  • Fairness: Desire for equality and freedom, and the urge to punish free-riders and defectors (if this seems like a messy category at best, and a contradictory mess at worst, don’t worry, Haidt refines this one in time);

  • Loyalty: Basic in-group/out-group alignment;

  • Authority: Respect for those in positions of power/prestige, and the idea that they can determine what is right and wrong and instruct accordingly;

  • Sanctity: Religious-seeming (though not always explicitly so) ideas of purity vs profanity, not violating the sacred, and so on.

The idea is that these are five fundamental intuitions humans have that come together to create our instinctive moral judgements, but that they are weighted differently in different people. Haidt then goes on to claim, based on his research, a strong correlation between the extent to which people feel these intuitions and the US partisan groupings of conservatives and liberals (Haidt is aware that the term “liberal” means all sorts of things around the world, but is working in a US context so uses it that way throughout, as I will), with liberals mostly responding to care and fairness, and conservatives to all five.

Haidt then proposes that conservatives have an inbuilt advantage in moral persuasion, because their wider collection of moral intuitions allows them to tell a greater variety of stories and justify themselves in a way that generates broader appeal, whereas liberals are mostly stuck justifying everything through care and fairness.

Haidt clearly struggles with the fairness foundation and its somewhat grab-bag nature. He eventually splits it in two, leaving the free-rider punishment part (which he calls proportionality) in the foundation called fairness, and spinning off a new foundation called liberty which is based around freedom from oppression. The revised political division is then that liberals mostly respond to care, fairness, and liberty, libertarians to liberty above all, and conservatives to all six.

Group Selection

The third part of the book then tries to work out why we’d have these intuitions at all, rather than just being self-interested, as evolution would seem to suggest we should be. This turns into an impassioned defence of group selection. According to Haidt, group selection was a part of evolutionary theory from the start, but gradually got more and more abused by people reverse-engineering group selection just-so-stories for pretty much anything, until a backlash in the 60s and 70s consigned group selection to the academic graveyard and insisted that evolution worked entirely by individual selection.

Haidt believes this to be a mistake, and sees himself as part of a movement to cautiously return some degree of group selection to the academic mainstream. He draws heavily on E.O. Wilson, approving of his proposed synthesis between the biological and social sciences, and makes a strong case, based on studies of the emergence of hive behaviour in insects, among other case studies, that group selection is possible and has been observed in other contexts. He’s clearly aware of the dangers of loose applications of group selection and the interpretive freedom this would give to explain almost anything as in some way a product of evolution, but works hard to establish when we might expect group selection and when we might not, and to make sure that individual selection always stays in frame as the main driving force of evolution.

This is necessary for his central idea because the moral foundations he claims are innate (he does a good job of explaining how innateness does not require that something be present from birth) can be explained as group adaptations but not individual adaptations. Haidt is positioning himself against the idea that evolution would drive humans to be selfish (with exceptions for close genetic relatives), so any other behaviour must be the result of explicit reasoning on our part. He claims instead that that some degree of non-selfish behaviour arises naturally through group selection, and that this is expressed through our moral intuitions, with moral reasoning mostly serving as a post-hoc justification for this.

Aside on replication

Before diving in to my own thoughts on all these arguments, it’s worth mentioning that, as it positions itself well within the “people don’t actually behave rationally” line of psychological, sociological, and behavioural economic research, the spectre of the replication crisis in those fields looms over the book.

In the period Haidt was doing a lot of his work, there was an explosion of interest in the myriad ways that human behaviour might depart from the perfectly rational, and countless experiments purported to show all sorts of biases and irrationalities of varying degrees of plausibility. Sadly, much of this work was of low quality, and large swathes of it have failed to replicate or simply come to be viewed as unconvincing due to tiny and/or unrepresentative sample sizes, or poor practice of some other sort.

This makes reading this sort of book pretty difficult if you need convincing one way or the other on the underlying science. It really isn’t enough now to just wave at some behavioural studies showing people exhibiting some bias and move on. You’ve got to do quite a bit better than that these days.

This wasn’t a huge problem for me reading through, because I already believed the basic ideas of moral intuitionism: The way most people make most moral decisions most of the time is on instinct, and then retrospectively justified by argument if necessary. Had I not, I don’t think I’d have found the case made convincing.

I spotted a few things as I went through that I knew had failed to replicate since the book was written, and there is also a fair bit of leaning on the work of Dan Ariley, who turned out to be actively fraudulent. The rest of it, well, I don’t know. I haven’t gone and read all the papers describing the studies the book refers to in order to see if I think they’re of reasonable quality, or checked if they’ve replicated or if there’s been a meta-analysis that agrees or disagrees with them.

I didn’t need to be convinced about this aspect either way, but it was an uncomfortable experience to read about it and feel how unconvinced I’d be if I weren’t already on board, and I don’t really know what the solution is. It’s hard for popular science books to work with the kind of rigour that’s needed to play in these domains, but I don’t want popular science to die out completely. I could get very distracted going down this route, so I just want to say “the replication crisis is clearly an issue for this sort of thing; deal with that how you will” and move on.

When Haidt is presenting his own work on moral foundations in the second section of the book, it’s clear he’s working with huge sample sizes over a long period and he explains his methods in some detail, so the key issue of “might this just be a tiny, awful study that’s shouldn’t convince me of anything” that plagues some of the first section of the book isn’t present. I have my own concerns about Haidt’s methods, but those are of a different kind.

The fundamental MaB

In this rudimentary form, I think there’s nothing particularly objectionable going on in Hadit’s main theses, and it is in this form that I’d heard most of the ideas before. However, as I read through the book, I started feeling increasingly uncomfortable with a sort of implicit motte-and-bailey that was developing.

This was never stated outright, certainly not in the helpful summary sections, but I felt it, particularly in the first of the three sections. It was expressed more in the narrative sections – in the bits where, instead of firmly stating hypotheses and marshalling arguments, Haidt allowed himself a bit more of a classic pop-sci writing style, linking together the development of his views with autobiography, and expressing frustration with the shortcomings of his chosen foils, most notably Kant, Bentham, and Kohlberg.

I was starting to feel like this might be a very difficult thing to explain, that I might have to quote extensive parts of the book to make the point that I felt he was implying some much more expansive bailey than his nicely confined and well-justified motte statements in the summaries. I thought I’d have to then make some sort of appeal that it was consistently implied by his tone, and by what he disagreed with or expressed frustration with, rather than being explicitly stated. I thought it was going to be a drag. But, then, late on in the book, Haidt produced a single paragraph that encapsulated the issue so perfectly, that it can stand for the whole thing:

I find it ironic that liberals generally embrace Darwin and reject “intelligent design” as the explanation for design and adaptation in the natural world, but they don’t embrace Adam Smith as the explanation for design and adaptation in the economic world. They sometimes prefer the “intelligent design” of socialist economies, which often ends in disaster from a utilitarian point of view.

Darwin and Ken Ham aren’t making competing claims about whether evolution by natural selection is a morally upstanding or desirable way for species to have arisen, they’re arguing (not directly but you get the point) about whether it actually happened that way. Conversely, a modern socialist in the US is not “embracing intelligent design” in the sense of claiming the current US economy must have been intelligently designed based on some study of its history or current state, but arguing that it would be better if it were so – that a planned economy is morally desirable (and a libertarian is arguing that a free market is more desirable, not that they currently live in a free market).

One debate is about the facts of what has already happened, the other is about the morality or in some other sense the desirability of different plans for the future. There’s absolutely no analogy here and no contradiction or hypocrisy in someone favouring natural selection as the correct explanation for the origin of species, but favouring “intelligent design” via a planned economy as the correct prescription for a flourishing economic future. Similarly, there’s no contradiction in believing that biology is the product of intelligent design by God, but believing that the best prescription for economic policy is “natural selection” via market mechanisms.

There are just some similar sounding words in these two debates, one of which is normative (how should the economy be run) and one of which is descriptive (how did species evolve), but Haidt latches on to it seemingly because he cannot reliably distinguish the normative and the prescriptive (and because he can’t resist a dig at New Atheists, many of whom favoured somewhat left-leaning economic policy).

When Haidt is forcing himself to be explicit, he knows the normative/descriptive distinction and in fact belabours the point. On more than one occasion he’s at pains to point out that he is making descriptive, not normative claims. At one point he sits down and tries to lay out his own normative moral theory, “Durkheimian Utilitarianism” as a well-separated exercise from his descriptive work.

His “Durkheimian Utilitarianism” is utilitarianism but with an understanding of sociology and game theory. In its conception of what is fundamentally right and wrong, it’s just pure and simple care foundation, just like Bentham would have it, but Haidt then attempts to justify heuristics based around the other foundations by showing how following them, in aggregate, can often lead to better utilitarian outcomes than naive utilitarianism.

It’s nothing that wouldn’t be familiar to people involved in the rationalism or effective altruism communities and their discussions around morality, though it’s a little bare-bones, and lining it up with the other five of his foundations is a bit tortured for reasons I’ll come back to. Importantly, it’s treating the moral foundations as descriptive, and then taking a completely separate approach to morally normative questions.

But, in the very same book, Haidt then cautions us to be suspicious of normative moral systems based around just one moral foundation, seemingly oblivious to the fact that he has just laid out such a system.

When he slams Kant and Bentham in a really silly bit of the book where he goes on a tangent about how they were autistic (complete with a really laughable graphic that adds nothing to the text), he slams them for “reducing” morality to just one thing. When he belabours a metaphor about taste receptors, he very much seems to be analogising to normative morality, not psychological description. When he describes his time in India and the affect it had on his approach to morality, he’s seemingly talking about normative morality, not descriptions of how intuition works or where it comes from.

I think that he really wants to be more normatively pluralist than he actually is. When pushed, he’ll say his normative morals are based on the care foundation, and that all he’s doing is trying to get other liberals to understand that other people have other moral intuitions, and that these might have arisen by group selection.

But it seems to irk him that he doesn’t actually agree with any of the people he’s trying to get us to sympathise with. So, when he’s being metaphorical, autobiographical, or slapping down those he sees as insufficiently worldly to empathise with a wide enough universe of moral tastes, he absolutely talks as if he believes that these other five moral foundations are normatively important as ends in themselves, not just instrumental heuristics for care.

It’s like he wants the aesthetic of being non-judgemental, relativist, and well-read in the wisdom of the east or whatever, and cringes at systematising/autistic reductionists trying to make everything about a cold hard utils calculation, but he can’t actually sustain that aesthetic when being explicit, because at bottom he’s actually a normative utilitarian, and only really disagrees with his model of Bentham in terms of application. He’s only not a utilitarian in that he isn’t one in the most limited, naive fashion that I don’t think anyone, even Bentham himself, actually is.

This back-and-forth runs throughout the whole book, and made it a frustrating read. I would regularly find myself feeling mocked and panned for believing things that, a few pages earlier or later, Haidt himself would explicitly endorse. There’s quite an intense whiplash involved in being told at one point that only an autistic nerd would be cringe enough to think morality boiled down to one axis, only to then read an attempt to boil morality down to one axis involving a bunch of extremely nerdy argumentation about game theory, group selection, and sociology a few pages later.

His inability to separate the normative and descriptive goes hand-in-hand with a frustrating inability to be clear about instrumental versus terminal values. At some points Haidt is almost mocking his interview subjects for their attempts to justify one moral foundation as an instrumental good based on the care foundation as fundamental, for instance by straining to find victims in anecdotes about disrespecting the flag, but he doesn’t take this idea any further.

He doesn’t attempt to separate out belief in a moral foundation as terminal from an empirical belief that it functions as a decent instrumental guide to another foundation, or look at these as separate contributors. It’s therefore really hard to tell what parts of the differences he’s seeing are different empirical beliefs about what causes harm as opposed to different beliefs about what matters morally.

Poor steel-men of conservatism and religion

This confusion lies behind another frustrating aspect of the book: Haidt seems to think he’s offering a refined steel-man of conservative beliefs and, through this, is able to negotiate political disagreements in a less partisan way. Unfortunately, his picture of conservatives, particularly religious conservatives, is pretty weak, and not actually that different in substance from the naive partisan liberal view of conservatives he’s setting himself up against.

In my mind, there are two ways to go about steel-manning conservative positions. Obviously conservatism is vast and multifaceted, and this will be a horrendous simplification, but I think it’s actually reasonable as a starting point and would be recognisable to many conservatives, unlike Haidt’s model. One way is to meet the rank and file conservative where they are, taking their views at face value. The other is to dive into the world of conservative intellectualism, which tends to take extremely different routes to similar practical politics.

I grew up a religious conservative, surrounded by religious conservatives of two quite different traditions as my family life straddled the (often acrimonious) divide between Fundamentalist Evangelicalism in the Southern Baptist tradition and Charismatic Pentecostalism. I can assure you, I did not hear a word about multilevel selection, punishment of free riders in game theoretic equilibria, or Schmittian notions of that which it is advantageous for society to collectively believe.

Homosexuality was wrong; God said so in the Bible; end of discussion. I could rarely get firm answer as to whether it was intrinsically wrong and God was helpfully letting us know this through providing guidance we could absolutely trust, or that God himself made it wrong by decree. I couldn’t actually get people to understand that these were different concepts. I also created a great deal of acrimony by questioning whether God actually had said this in quite so many words (it did seem that you weren’t really meant to actually read the whole Bible yourself, which I went ahead and did, then annoyed everyone by quoting it all the time) but that’s somewhat beside the point of this discussion.

Point is, nobody talked about it being beneficial for society to have some collective set of beliefs about behaviour even if they might at the fundamental level be somewhat arbitrary. Nobody chastened my picking around the edges of their certainties with sociological concerns that such independent thought threatened a shared reality that allowed people to predict each other’s behaviour with low cognitive overhead and therefore better co-ordinate.

People genuinely thought that at a base, normative level, there could be no morality without explicit instruction from God, and that an atheist could not justify refraining from mass murder. Tim Lambesis of the metalcore band As I Lay Dying, one of my Christian childhood heroes (explicitly on the Pentecostal side; the Evangelicals considered him obviously satanic all along due to the metal and were quite smug about how it turned out) actually hired a hitman to kill his wife after becoming an atheist on exactly these grounds. In this view, fundamentalist normative religion wasn’t a co-ordination system to make society function, it was the literal truth.

Haidt mentions that, after spending a while in India eating with his right hand and wiping his ass with his left, he could start to see why you might consider the right hand sacred and the left hand profane, and want to handle communal objects of religious significance with the right hand. Then, given the chirality of hands, you want this to be a shared norm. We will leave unexamined, as Haidt does, why you can’t just wash your hands, and how much this sucks for left-handed people and just take it as read that this broadly works as a co-ordination norm.

I am confident, given my experience of religious conservatism, that Haidt understands the cause and effect exactly backwards here. The people he’s attempting to empathise won’t have derived it this way round. They will wipe their ass with their left hands and eat with their right hands because the left is profane and the right is sacred, not the other way round. This just seems to be out of Haidt’s conceptual framework, and I think that’s for the same reason that, even when seeming to believe that he has a pluralist moral system, that when he actually explains it, it’s just non-naive monist utilitarianism.

This is really damning for a book that’s meant to be about understanding each other better and reaching across partisan divides. Haidt is witheringly harsh about simple-minded liberal partisans, exemplified in his telling by the New Atheists (the fact that the New Atheists are the ultra-partisan liberals is a whole other kettle of fish I’ll explore later) and simple-minded utilitarians, but his actual understanding of conservatives is basically the same as that of those he’s criticising, just with individual level selection replaced with group level selection.

He’s mad at New Atheists for being simple-minded enough to think that religion is nothing more than the product of natural selection predisposing the individual brain to think in magical ways and that therefore we can discard it as no longer useful. No, Haidt claims, in fact religion is the product of group selection predisposing whole cultures to build communal rituals around magical claims. I’m pretty sure the average religious conservative feels like these two positions are both pretty offensive ways of just not engaging with their actual beliefs but talking around them like they’re non-agents. The difference between them is pretty technical, and crucially Haidt’s position leads just as surely to the possibility of “and therefore we can discard them as they’re no longer adaptive to the current environment” as does that of his New Atheist foils.

In practical terms, I think I’m much more of a partisan liberal (at least on the partisan divide he’s identifying back in 2012) than he is and very much the subject of his condescension for my lack of bipartisan empathy, but I think I actually have a much better model of my ideological opponents, and one they’d find a lot less offensive, than he does.

Of course, it can be true that conservatives (or anyone) are predisposed to think something for some psychological or sociological reason, and that the reasons they actually give for their beliefs have nothing to do with those drivers. The object-level debate is different from the meta-level psychologising. Haidt mixes up these levels a lot, and it makes it hard to see what he really thinks, but I think it’s pretty clear he doesn’t really engage with conservatives’ claims on face value.

Problem is, he doesn’t steel-man them at the next level up of “these are useful things for society to believe and people should just sort of get on as if they’re true and not question them too much lest our shared reality collapse and our social norms with it” either. This is, I understand, what a lot of conservative intellectuals end up thinking on some level, and is my second way of steel-manning the conservative position. We could, therefore, see Haidt as engaging only with this type of claim, and that would be broadly fine as an approach. He’s an academic and engaging only with the academic version of his opposition would be a reasonable path, though it would achieve less sweeping results than engaging directly with the objects of his curiosity themselves.

But he doesn’t really lay out this kind of position in detail or in a convincing way either. There’s a staggering sort of teleological evolution thing going on where he seems to assume that, if we can identify how some conservative morality has evolved through group selection, then that’s equivalent to some normative argument for its force in a moral realist sense, or at least some empirical argument for its validity as an instrumentalist rule-of-thumb in his “Durkheimean Utilitarian” framework.

But this is even more shaky for group selection than for individual selection. At least individual selection is optimising for the reproductive fitness of the individual, which has at least some correlation some of the time with the welfare of that individual. Even then, this is far from a great correlation (particularly for any individual that won’t reproduce any more, or at all), but for group level selection it’s much, much worse. Aspects of a social group that make it compete well with other groups for resources, win wars with other groups, resist internal takeover or subversion by individuals within the group, and therefore be more reproductively fit at the social level are often aspects that make that social group incredibly unpleasant for almost all or even all of its constituent individuals.

To get from identification of conservative forms of morality as the product of group selection to where Haidt seems to think he’s ended up (more sympathetic to these claims either as normative claims about morality or instrumental claims about what really actually produces the best utilitarian results – again it’s tough that he can’t seem to tell the difference), you need to talk about reduction in cognitive overhead and how the average person has some combination of neither the ability, inclination, or resources to reason everything out in utilitarian terms. You need to go into robustness to extremely bad outcomes. You need to chat game theory and how conservative norms might make a better approximation to optimal play in situations where naive causal decision theory leads to defect/defect equilibria or to two-boxing predictably in Newcomb problems and thus poor results.

Haidt just sort of seems to tacitly assume something like this must be true, occasionally paints an outline of it, but mostly just barely seems to even understand that these are the terms of the debate if you’re going to be arguing about whether conservative approaches to morality actually work out better even from a utilitarian perspective than raw individual utilitarianism plus causal decision theory. His steel-man seems to be “evolution is teleological, even group selection”, which is an absolutely awful steel-man of the intellectual conservative position.

As a result, he engages with neither rank-and-file God’n’guns religious conservatism nor the intellectual conservative tradition of “what is good for the masses to believe is not identical to what’s fundamentally true, please consult my 60,000 word essay on decision theory, game theory, computational load, and the Austro-Hungarian Empire for details, therefore Catholicism”. He doesn’t really seem to realise that either of these positions exists, and “steel-mans” conservatism into some sort of superposition across “group selection informs us as to normative ethics” and “group selection is teleological towards utilitarian human flourishing”, both of which are utterly insane positions that I think almost nobody actually holds.

Aside on Categories and Axes

I now want to dive off into the thorny world of the “realness” of categories and measurements. This is a ludicrously fraught area in which lots of people don’t really define their terms and/or are inconsistent in their standards. Almost all of the common issues I can think of along this line come up in my analysis, so I want to get a few statements about categories out of the way first in order to have a clear framework for what I mean when I start complaining about Haidt’s categories.

Some things are actually categorical. Things can be one thing or another and there isn’t really a sense of degree, or of in-between-ness as an option. Particles are either fermions or bosons. The distinction is discrete, and it’s not really clear what it would mean for something to be part fermion and part boson. An atomic nucleus is a hydrogen nucleus if it has one proton it’s a helium nucleus if it has two (and on to lithium and the rest). There’s no sense in which something could be “between” being a hydrogen and a helium nucleus by having “between” one and two protons, because a partial proton just wouldn’t be a proton at all.

You could get pedantic and say a nucleus can be in a superposition of having decayed and not decayed into another nucleus, and so that’s in some sense a violation of this. However you could never interact with that nucleus while keeping it in that superposition and see it behave partially like one nucleus and partially like another, because doing that would bring you into the superposition and you’d then be in the Everett branch with one nucleus or with the other (or “the wavefunction would collapse if you did that and then it would be one or the other”, whatever that means). So for the purposes of our interactions with them, nuclei are categorically of one element or another. Totally unrelated to the rest of the review but I’d be keen to hear from physicists if I’m wrong about this example, as I am a chemist by training so flying somewhat by the seat of my pants here.

Regardless of my grasp of the physics example, I hope the concept is clear. Other examples I could think of were all conceptual maths or logic stuff that didn’t have a clean real-world implementation and I wanted to show that such tight category distinctions can in principle exist in messy reality as well as in thought.

Then you get bimodal or other multimodal clustering about some measurement. Most people have one strongly dominant hand; normally it’s the right hand but a good 11% or so have it on the left hand. If you graphed “right hand dominance” on some normalised scale of -1 to +1 that you devised based on some standardised measurement of relative ability to do fine tasks with one hand or the other, you’d find most people clustered over near +1 with some variation between them, an order-of-magnitude or so fewer people clustered over near -1 in the same way, and maybe an order-of-magnitude fewer people again spread out in the middle. You’d call these right-handed (RH), left-handed (LH), and ambidextrous (AD).

This wouldn’t be completely clean. You might not be able to draw a firm line between LH and AD that didn’t seem a bit arbitrary, or between RH and AD, but you’d be able to confidently classify most people as either LH or RH without much argument. Perhaps my imagining of how this would work is a little off, and actually there would be a third cluster around 0, giving a bit more structure to the AD grouping and allowing it to be more rigorously and unambiguously divided from the other two. These two possibilities would be bimodal and trimodal distributions, respectively.

In these cases, most people can be unambiguously categorised without much controversy, but there will be edge cases: One “group” of edge cases in the bimodal case, which would be the whole AD category, which would have hard-to-define edges, and two in the trimodal case, which would be the areas between AD and LH/RH, which you might want to give names to and which would then work like AD in the bimodal case in that they wouldn’t have well-defined edges.

I’ve chosen this example precisely because I don’t actually know if AD is its own cluster or just the name we give to the bit between LH and RH that just gradually bleeds into each. In each case, the term “AD” refers to something, but in one case it refers to something quite woolly, and in the other case to something more like LH or RH. I want to make clear that AD is a useful category in both of these situations, but it would need to be treated quite differently in each.

Then there are just pure smooth distributions. Height is the classic example. Some people are taller than others but there’s only one cluster. There’s an average height; some people are tall, other people are short, but that doesn’t mean anything like what RH and LH means. You can define the difference between tall and short at the median, if you like, to get you two groups, or you can make three groups by using the 1-standard-deviation lines either side of the mean to give you an average-ish height group then tall and short groups. These are all reasonable things to do if you define your terms, and in these cases your choice of bins isn’t particularly dictated by the underlying distribution as it would be in the handedness cases, so your classification isn’t so unambiguously reality-driven, it’s more conventional, and your categories are more subject to debate and scrutiny – they’re a bit less “real”, but still useful.

Indeed the fact that we could talk about ambidexterity even in a world in which the handedness distribution is bimodal is instructive. In that case, RH/AD isn’t a non-arbitrary distinction, and neither is LH/AD, but RH/LH absolutely is. In the trimodal handedness scenario, all three are non-arbitrary distinctions. There are just varying degrees to which a categorisation system is driven by the underlying reality of the situation versus just being a convenient way to divide up the world and talk about it, and which distinctions are of which nature really matters when critiquing those concepts and connecting them to other concepts.

All of this has just been with one undisputed axis on which to graph a distribution, but in the world of psychology things are almost never that simple. Psychological concepts are normally bundles of more fundamental concepts that have been lumped together to create some higher-level abstraction that can then be measured and correlated with other things. There’s a similar distinction to the height vs handedness distinction at play in terms of how firmly implied such groupings are by underlying reality.

When two smooth distributions correlate neatly, like foot size and hand size, they can easily be combined into a single measurement and talked about like they’re a single thing when discussing them. They might have to be pulled back apart for causal purposes (a footballer’s large hands don’t causally impact their play, but large feet do, for instance) but there’s a very meaningful sense in which foot/hand size is a useful concept in a way that a bundle of uncorrelated things like foot size and degree of short/long-sightedness isn’t.

The degree to which the components of a composite measure correlate with each other compared with the degree to which they correlate with things outside the composite measure is generally taken as an indication of how useful the composite might be in a classification system (though this is, I’m keen to stress again, not a statement about internal causation or external co-causation). People go on at length about how different psychometrics systems are or are not “real” based on vague invocations of this as well as unimodal vs multimodal distributional along these composite axes, and often really don’t define their terms well, which is why I’m going on about it at such length here.

I’ve heard people complain about Myers-Briggs based on the unimodality of distributions along each of its four axes of classification, but all that shows is that Myers-Briggs properties are more like height than like handedness. Yes, some T-type people are more like some F-type people than they are like some other T-type people. Some people under 6 foot are closer in height to some people over 6 foot than they are to some other people under 6 foot. That doesn’t mean height is a myth or even that the particular threshold of 6 foot isn’t sometimes really important (for instance in determining whether you’ll fit through a certain doorway without having to duck).

The more thoughtful criticisms of Myers-Briggs I’ve heard claim that the individual measurements (generally questionnaire answers) that the process bungs into its calculations to produce its E/I, N/S, T/F, and P/J axes don’t naturally form those four clusters at all, that they’ve just been arbitrarily added together and thus aren’t really internally predictive of anything. These critics will say that you can’t know that just because someone has some N-type attributes, that they’re more likely than chance to have other N-type attributes, so the idea of an N-type is as pointless as category as large-footed-long-sighted.

To imagine this taken to the limit, imagine the facebook personality quizzes that ask you if you prefer parties or staying in, then ask you if you’re more interested in fashion design or motor racing, then based on your answers “reveal” that you’re the type of person that prefers staying in and watching motor racing to the other three combinations, or whatever. It’s just reflecting your inputs back at you and not saying much else. This can be saying something if you’re then correlating it with something else external, but still it’s probably then just better to study those two correlations separately to avoid confusion.

It’s this layer of useful vs useless bundling that I have a problem with in Haidt’s work. I don’t think his categories are particularly “real”, or “fundamental” or whatever, but that’s such a misused criticism across the board and against psychological research in general that I wanted to spend over 1,500 words making sure I was being exactly clear what my criticism was, and that it couldn’t be confused with the criticism where you just demand that every distinction be as pure as that between a fermion and a boson and declare everything else some flavour of nonsense.

Haidt doesn’t show his work in the book at all. He does comment on how much data he’s got from his moral foundations questionnaires (it’s a lot), and that removes any concern that he might just be mining noise from tiny samples, but there’s no explanation of how he got from his data to his moral foundations. That’s sort of to be expected in a pre-popular-knowledge-of-replication-crisis popular science book, but then there’s a lot of discussion that strongly suggests that the bundling is pretty arbitrary, and that makes me worries that Haidt just arranges the data in a convenient way that just reflects his own categorisation system back at him.

It’s important to remember that if you enforce a categorisation system on the data, then the fact those categories then act like they’re fundamental in your summary analysis, and can be said to go up or down or correlate with certain things, might tell you something, but it absolutely doesn’t tell you that your categorisation system itself was a good idea. It tells you it wasn’t such a startlingly bad idea that it destroyed all the information in your data, but it certainly doesn’t tell you that the underlying structure of the distributions suggests that categorisation in particular.

Unfortunately, Haidt very much acts like it does, like the very fact he can add up different bits of his dataset that he’s called “Purity” and it seems to be higher for conservatives than for liberals tells him that there’s some underlying causal thing driving that variation which has a structure similar to his Purity foundation, or even that captures it as a distinct part of the larger dataset in a robust manner. The way he messes around with the Fairness foundation makes me very suspicious that this is all pretty arbitrary.

The botched fairness foundation

In the initial five foundations (no liberty), the fairness foundation is a complete mess. Haidt defines it as being the urge to punish free riders, which has obvious game-theoretic benefits in policing prisoner’s dilemma situations. But then he also wants it to include equality. He then seems to realise these aren’t the same thing, and tries to chalk this up to the difference between equality of opportunity and equality of outcome.

I could go on at length about how I think “equality of opportunity” is almost never a useful term, is almost never used consistently, and doesn’t really make sense outside a completely scientifically illiterate blank slate view of human nature. Thankfully, I don’t have to, because this framing then gets abandoned as the fairness foundation is split. Explicit free-rider-punishment gets to keep the name “fairness” and it’s sort of suggested that this is sort of equality of opportunity in some way. A new foundation, liberty, is created, sort of to cover equality of outcome, but it’s operationalised as freedom from oppression, which is then taken to refer to things like racial justice, but also get-off-my-land libertarianism.

This re-organisation is left unjustified on anything but its own very shaky top-level idea of those categories being sort of natural political terms, which doesn’t make the slightest shred of sense to me. No attempt is made to show anything like whether the underlying data aggregates more helpfully in the new system. It’s just that “fairness” was clearly making no sense so he re-organised it into something else that sparks slightly fewer intense feelings of “hang on, that makes absolutely no sense; what are you doing and can I see the data, please?”

Liberty is the most obvious example of a related problem in that he sometimes talks about how each foundation gets expressed in conservatives vs in liberals in a way that makes it clear they’re very different things – in this case in conservatives as “get off my lawn” and in liberals as “the government must intervene on behalf of oppressed groups”. But what’s the justification for treating this as one foundation that both partisan groups care about in very different ways, rather than just splitting it again into two foundations, one of which republicans have and one of which democrats have?

I fear the answer is that too much splitting would make it obvious that Haidt’s conclusions about the differences between conservatives and liberals are baked into the structure of his analysis and not really telling us anything we didn’t already know. He’s reaching for a higher level, a more “foundational” layer of ethical reasoning he thinks is causally driving partisan differences under the surface, but the way that he’s constructing those in his analysis can clearly be fairly arbitrarily tweaked to get you anywhere from “conservatives and liberals have the same underlying moral foundations, but operationalise them completely differently” to “conservatives and liberals have very different underlying moral foundations”.

Fully going for the latter in the only way this analysis can support would just be bald-facedly reflecting the inputs back as outputs – he’d have essentially just described the differences, and obviously provided no explanatory power. He’d just have listed what things conservatives obviously care about (at that point in US political history) like freedom from big government, and things liberals care about (at that point in US political history) like harm reduction and racial equality, in a way that clearly added nothing and predicted nothing.

But if he’d gone fully the other way and found some way to call everything two sides of the same coin, so that each foundation was well-balanced between the tribes but they expressed it very differently (like his liberty foundation) then he wouldn’t have found any juicy differences at his foundational level, so it also wouldn’t have seemed so interesting.

By not being principled about how he draws his categories, he makes it all look a lot more complicated than it really is, and a lot more like it might have predictive value rather than just being a summary of how people answered political questionnaires. Is this extremely unfair? Maybe, but we never get to see the underlying calculations that are justifying his category distinctions and analysis choices. This means that, given the fairness/liberty jiggery-pokery, I’m inclined to just think it’s nothing.

Of course, that’s just scepticism on my part of what’s essentially an untested hypothesis. Whatever magic Haidt pulled behind the scenes, if it produces a model that predicts things accurately, then there’s something useful going on. The whole point of the system is that these foundations are more real and more persistent than mere partisan squabbles and the latest news cycle, that they actually causally drive how different political tribes interact with issues. If it’s predictive, it’s predictive, regardless of my methodological scepticism.

So how does the model hold up ten years after the book came out? Do conservatives and liberals still line up the way Haidt saw in 2012, with liberals focusing on harm, with some fairness and liberty thrown on top in a way that’s quite hard to explicitly lay out, while conservatives focus on a wider selection of moral foundations including authority, sanctity, and loyalty?

The short answer is that they absolutely do not. The longer answer is that it’s difficult to even answer the question, because Haidt’s theory doesn’t just fail to predict current political trends in terms of who scores where on his axes, but fails to even have axes that capture the majority of what’s going on these days. Fairness/Liberty may have been a bit of a mess as the time of writing, but it’s an absolute disaster now. There’s a whole major strand of political identification that just doesn’t fit anywhere in his foundations: Socialism.

Where are the socialists?

There are just no socialists in Haidt’s world. Anyone who cared about equality in the socialist sense would find themselves, in his first draft, being lined up as indistinguishable from those who cared a lot about punishing free-riders. Bernie Sanders would have looked the same as someone going on talk radio to complain about welfare queens, as they would have both prioritised “fairness”, along with people who wanted the government to get out of their ammo safe and out of their kink dungeon, and people who wanted massive government intervention to equalise racial differences in elite society.

Haidt’s revised six-foundation model isn’t quite that loopy, but socialism doesn’t find any more of a home in it. Where in his six foundations would fit the now very commonly expressed, and historically somewhat influential idea that it’s a good idea for everyone to roughly have the same level of wealth, or income, or welfare, or something, and that therefore it’s probably good for the government to redistribute wealth pretty aggressively along those lines, or stop anyone from getting too rich or powerful?

It either has to show up the same as welfare-queen-punishment, which is its polar opposite, in the fairness foundation like before, or find a home in the liberty foundation along with liberal social justice and, somehow, Ron Paul.

I realise that, at this point, I’m sounding extremely negative. It’s key to remember that there’s a lot of this book that’s well-argued and making an important case that I think a lot of people could do with hearing. The defence of group selection is one of the best I’ve read, as is the attempt to re-ignite Wilson’s “new synthesis” between biological and social sciences. The scientific would could do with a lot more of both of those things, and the message is as important now as in 2012, if not more so.

Similarly, when firmly in moral psychology territory, the first section on intuitionism is, if something that’s very much “in the water” in many communities by now, still lacks the mainstream acceptance it deserves. If you can ignore the odd attempts to cast it as an enemy of or even alternative to both Kantian deontology and utilitarianism, and skip entirely the bit about how Jeremy Bentham was autistic or something, and treat it just as psychology unmoored from normative morality, it’s great stuff.

But the political side of the book is just really bad. He makes such a song and dance of how pluralist and open-minded he is, and how we need to reach across partisan divides and understand each other better, but as well as sort of mangling his understanding of religious conservatism and treating social justice warriors as a type of libertarian somehow, he just doesn’t seem to have ever come across a socialist in his travels, or his reading, and that’s a monumental blind spot.

There’s one point in the book where he mentions contact with Marxism, and that’s when he was travelling to Brazil to get some data on the moral intuitions of people outside the US. He mentions that he went to a conference but people were all Marxist so he left and went somewhere else. This was in the sense of their approach to psychology, and to be honest I’ve heard what I think is the kind of “theory” he’s gesturing at hearing there and I’d leave as well, but it’s a real weakness that this is the only time that any sort of Marxism or socialism really turns up in a book that’s supposed to help liberals understand challenges to their ideology and empathise with other points of view, and supposed to provide a sort of “theory of everything” for political differences.

Indeed, the part of the book that’s explicitly written as a steel-man of challenges to liberal thinking takes a religious conservative and a libertarian as its opponents. I’m not claiming Haidt needed to cover every political ideology that’s ever been devised. I don’t mind that in the course of his field work he didn’t track down Curtis Yarvin, spend ten careers deciphering exactly what that man is trying to say, and then introduce a seventh moral foundation called “based and blackpilled” or something. Eventually you’re just cataloguing, not making enough generalisations to predict anything.

But socialism seems like a big thing to just completely ignore. Maybe in 2012 (and I need to remember that even if the book was published then, most of the work that went into it comes from the noughties or even earlier) there were so few socialists that it would have been like including neoreaction to bother to include them? Maybe those sentiments were so rare that an axis for them would add little to the analysis?

I was at university in 2012 and I was sort of a socialist, to the extent that I hung out with the socialist activists a lot, and there were, like, five of us, and we mostly focused on organising against censorship (our biggest event that year was a “Free Speech Day” – wow 2012 is ancient history) and against things like Sharia arbitration courts, state-funded faith schools, and female genital mutilation. I think everyone realised there was very little traction for the economic parts of the agenda, so focused on social stuff where there were live fights to be had. The nature of those probably seems insane from a US perspective but things are a little different in the UK. We have issues with things like even normal state schools running Religious Education lessons that teach Islam or Christianity as fact, and there are schools (though I’m not sure any of them are state funded, but the line isn’t clear and the systems intermingle) that advocate the death penalty for homosexuality and stuff. It’s wild, but that’s a different conversation.

Being as kind to Haidt as possible, maybe having to fit socialism into his system seemed as ridiculous as having to fit divine right monarchy into it. Altogether, it comes across as an attempt to dramatically reach across the ideological divide between Barack Obama and Mitt Romney and therefore unify all political thought, but then, well, that’s quite literally what the divide was in 2012, so maybe I’m being unfair by calling him out on this kind of limited scope.

But I’m not. The whole deal here is meant to be that partisan politics is fickle but the moral foundations are, well, foundational. If just one decade on from publication, large amounts of mainstream politics are dominated by an idea that doesn’t really even fit into these foundations, then that’s quite a strike against the theory. I suppose Haidt would argue that is does fit, but I can’t work out how to guess whether he’d try to jam into fairness (that sounds most plausible from the name, but it really doesn’t work at all with what that foundation is supposed to be about) or liberty (making it into even more of a catch-all for everything he couldn’t capture in round one). But if the theory can be stretched that far, it isn’t really predicting anything.

If this were the only problem, we could perhaps chalk it up to “nobody expected Bernie 2016” and conclude that politics is pretty hard. We need to have a look at the things he’s more explicitly dealing with – differences between conservatives and liberals on the six foundations he wants to work with – and see if they’re stable over time and therefore perhaps useful predictive tools.

They are absolutely not. I don’t think you’d get anything like the same distributions of moral foundations between political tribes nowadays, and therefore that they’re only slightly less ephemeral than the partisan issue of the day. Moral foundations were supposed to be the stable thing in the background driving political differences. If anything, I think the last ten years suggests the causation flows the other way.

Morality doesn’t drive tribalism; tribalism drives morality

Probably the biggest difference between conservatives and liberals in Haidt’s view is that the former value sanctity, loyalty, and authority, whereas liberals don’t really care about these at all. So, explain Trump, please!. Trump is the embodiment of the profane. He offends basic notions of the sacred, of dignity in political debate and in human conduct in general, and of respect and decorum, at such a rate that the outrage just can’t keep up.

And that outrage is not just the classic outrage of liberals towards conservatives that Haidt would predict: the outrage on behalf of those harmed (because harm is what liberals care about above all else in his model). Outrage against Trump was often outrage against the violation of the sacred. The very human being of Trump was cast as disgusting and dirty. He revelled in it and conservatives lapped it up. Conservatives luxuriated in violations of the sacred experienced as liberation, like it was a Sex Pistols concert in the 70s.

And as for how liberals treat sacredness now? I don’t need to go on at this point with a bunch of examples of how the exemplar crime in the eyes of the average liberal is now offence against some sacred taboo, like conservatives of the past when someone swore in a sitcom. This is probably the most commented-on aspect of the political realignment of the 2010s and needs no further elaboration.

And who took the side of national loyalty in 2016? Clinton’s nomination drowned in the US flag and other patriotic regalia, while Trump was cast as essentially a Russian spy, an enemy of the state, an infiltrator. And, as if to rub Haidt’s face in it, he was supposed to be in hock to the Russians (putting liberals on the side of loyalty) because of some alleged candid photos of him getting urinated on by prostitutes, or something (putting liberals firmly on the side of sanctity). Many conservatives ended up lionising Russia as superior to the USA (or at least many aspects of the USA), in a complete reversal of their Bush-era nationalism and in rejection of the whole cold war framing (though this has understandably got a little quieter in many quarters as of 2022).

And, though liberals maintain their anti-authority flavour in some ways, as relating to sticks-and-stones authority like that of the police, it’s now liberals who demand respect for the authority of all the great centres of prestige – universities, broadsheet newspapers, the judiciary, teachers, the bureaucracy, scientists. Liberal discourse is saturated with appeals to hierarchy and demands for deference, with detractors urged to stay in their lane, not question the experts, and respect those with the prestige to demand such respect.

Even the demands for justice for oppressed groups have over time taken on less and less of a “reduce material harm” flavour and begun to be expressed more and more in a “give prestige and deference” manner. Though the socialists talk about material conditions a lot, the liberal mainstream almost seems to view that as vulgar these days, and talks in terms of prestige and respect – hierarchical terms I associate more with Haidt’s authority foundation – because that’s the currency they work with now.

In short, the political world Haidt describes is unrecognisable in 2022, and for more reasons that just the unexpected return of socialism to the world of polite conversation and the routing of the New Atheists from their position as the bleeding edge of cultural liberalism. I wouldn’t say that conservatives and liberals have switched places, it’s far more complicated than that, but certainly the dynamics he describes, where liberals focus almost all their messaging on utilitarian appeals to the greater welfare, while conservatives deal with a more complex and nuanced approach to morality involving respect for authority, a focus on defending the sacred, and loyalty to the nation and flag, are shot to hell by now.

It is not good for a theory of what lies behind politics to look completely foolish after a decade of politics has passed! I can’t see how anyone, looking at events since The Righteous Mind was published, can put any stock in its theories of how differences in moral intuition drive political divides. Clearly the differences between people in different political tribes as to what moral intuitions they have are not foundational at all, but contingent on something else. Something was making conservatives all sanctity/loyalty/authority in the past, but is doing the same to liberals now.

If moral foundations are not foundational at all, then what is driving them? Can we just look one link back up the chain and find a foundation of moral foundations, deeper in the recesses of the mind? Sadly, I fear it’s worse than that for Haidt’s theory. Rather than just thinking he’s failed to find the source of the causal chain, I think he has it entirely backwards: I think that political differences are what drive our differences in moral intuition.

Think through all the examples I’ve given of how the conservative/liberal divide he describes has changed with respect to emphasis on different moral foundations. Every single shift is politically convenient.

Liberals were anti-authority freedom fighters until they controlled a whole bunch of prestigious institutions and sources of authority, and then suddenly experts were not to be questioned, all sorts of elite prestige factories like Harvard were suddenly to be treated with the utmost of reverence, and liberals started using phrases like “the adults in the room” to refer to themselves. But they never got the police, so be as anarchist and punk rock as you like around those pigs (whereas Conservatives still ostentatiously show deference to the police).

Conservatives clung to an image of resisting the corruption of the sacred by the profane, of wholesome family values and smiling rows of identikit suit-wearing children with biblical names like those from all the best homeschooling families I knew as a child, until their movement somehow ended up headed by a brash, shock-and-disgust-to-dominate real estate brawler with a disreputable sex life, and all that got quietly dropped.

None of this is particularly controversial or insightful stuff when it comes to the last decade of political evolution “everyone actually a total hypocrite; tribalism rules all” isn’t exactly new or interesting. My point here is that this clearly drives the moral intuitions Haidt thought he’d identified as the prime mover in the equation. People don’t become conservatives or liberals because they have different moral intuitions. People develop the moral intuitions they require in order to support their political tribe, and those intuitions can change quickly.

That’s because they’re all just downstream consequences of the most fundamental intuition, the ultimate unthinking flinch reaction upstream of all the post-hoc moral philosophising Haidt quite rightly views as secondary, the most basic elephant-in-the-brain: “My side is justified”.

Conclusions

Overall, I’d recommend The Righteous Mind. If that seems nuts given the preceding 10,000 words then that’s only because the ways in which it messes up are extremely interesting. This is not the kind of book you get mad at and want to throw at the wall. This is the kind of book you keep finding yourself “reading” with a thumb in the page, gazing out the window, thinking about politics or meta-ethics or the replication crisis.

Particularly given when it was written, it’s a solid contribution to the world of popular social science. It does deserve its place as part of the turn towards empiricism and rigour in that world, and part of the “perhaps think a bit more about politics than how everyone else is obviously evil” genre. Its three main problems are that it’s hard to know exactly what to think about it replication-crisis-wise, that Haidt can’t distinguish between normative and descriptive claims even though he tries really hard to be good about this, and that its attempt to find an underlying driver for partisan differences is completely and utterly backwards.

These might seem like damning criticisms but they’re really not. Nothing in this vein is free from the stain of the replication crisis. If you threw out every book that referenced even just the absolute bottom tier stuff like Dan Ariley and John Bargh, you wouldn’t have a discipline left. This whole discipline is crawling out of a huge hole of fraud, wishful thinking, and appalling lack of rigour. People like Haidt, even though I may be quite sceptical of his category formation, are broadly pulling it upward. It makes thinking about this area hard but it does not make it impossible or useless.

Similarly the normative/descriptive confusion is far from unique and it doesn’t matter to a lot of the book’s content. You can take the moral intuitionism as a descriptive theory, as Haidt in his more sober moments would insist you should, take his attempt at “Durkheimian Utilitarianism” as a good starting point for how to think about utilitarianism in a less naive way, and leave out the awkward bridge idea in which somehow the existence of a moral intuition implies it’s either normatively true or instrumentally useful. There’s a lot of good thinking in there but you just have to follow Haidt very carefully and know when to let go of his hand as he powers off into his pluralism aesthetic, ranting about how New Atheists suck while producing almost identical arguments at a slightly different level of natural selection.

The political stuff is the most obviously wrong on its face, but it’s wrong in a really enlightening and interesting way. It’s a good idea to understand that not everything is a simple disagreement over how to maximise one agreed value, but neither is the disagreement that the other side hates your value and wants it to be minimised just to spite you. A staggering amount of political discourse even now still takes place on one of these two absolutely ludicrous sets of terms. Haidt is a solid rebuttal to that, showing how different people have quite different moral intuitions, and how political debates often run aground on the problem of orthogonal values.

His attempt to then treat those intuitive values as he found them in the war-on-terror era USA as personality fundamentals specific to the people that formed the conservative and liberal political tribes is completely wrong, but it represents a good first step away from what he’s positioning it against. If anything, he just didn’t go far enough and see that so much of what he saw as fundamental was a contingent expression of tribal alignment, and could be jettisoned quite easily, or built up from scratch, as political realities demanded.

Perhaps the real lesson is that, viewed tribally rather than nationally, the loyalty foundation actually rules all? I hope not, and I’m not really that cynical. But I do think that sanctity, authority, and loyalty (in a nationalist sense) in particular are likely just contingent expressions of political convenience. If you have the power to determine what’s sacred, you’ll develop the sanctity foundation. If those who do so use it to keep you down, you’ll feel drawn to the profane. And so on.

One counter-argument to this recommendation is perhaps that the best bits, like the elephant-and-rider idea, has seeped onwards into the broader culture where you can find it without the baggage of having to think about Dan Ariley, wrestle with a confusing and contradictory approach to meta-ethics, and work out how anyone could bundle Ron Paul style libertarianism and Ibram Kendi style social justice into one supposedly foundational pole of morality and call it a day.

But, as I said at the start, I knew a lot of what I was already steeped in was downstream of The Righteous Mind , and most of what was unexpected on reading it was bad, but I still immensely enjoyed it. Few things I’ve read have been such good mental exercise.

I guess it depends what one would want out of a recommendation. I would never recommend The Righteous Mind to someone who wanted to know what I think is the state of the art in thought on moral psychology, ethics, or politics. It’s so flawed. But I absolutely recommend it if you’re interested in those areas and want to find out more things to consider, and be primed to think hard for yourself.

It’s by far the best largely wrong book I’ve read.