Here’s a phenomenon you may have observed: someone more or less on the political left moves right, but they don’t move just a little bit right, or adopt some of the positions associated with the right while retaining some associated with the left (though they sometimes say just that). Rather, they seem to move from left to quite far right, without appearing to move through the intermediate space that notionally separates them. They don’t adopt, or they don’t make much of, right-wing positions that might be reasonably debatable (e.g., the view that concerns about social solidarity mean we should slow immigration somewhat). Instead, they adopt right-wing views in their more rabid forms, including (or alongside) views that I will call – introducing here a technical term – bananas.Footnote 1
It’s hard to get a sense of how common this phenomenon is. Here, I won’t inquire into that question.Footnote 2 Instead, I aim to explain why it happens when it does.
Here are some examples. Nina Power, once a philosophy lecturer at the University of Roehampton, is the author of One Dimensional Woman (Reference Power2009), which aims to elaborate a genuinely anti-capitalist feminism. By 2018, she was messaging a friend wondering ‘what the real deal is with the Protocols [of the Elders of Zion]’, claiming that ‘the social disapproval for “racism” is a psyop to stop people realising that there *are* alien and fairy races’, and that ‘black and white were alien races and very different’ (Turner Reference Turner2025). Bret Weinstein, formerly a biology professor and self-identified liberal, is now an influential podcaster. He has used his platform to promote debunked Covid conspiracies, as well as the ‘Great Reset’ conspiracy theory, according to which elites aim to eliminate most people, and especially white people (Johnson Reference Johnson2024). Naomi Wolf, author of the best-selling The Beauty Myth (Reference Wolf1991) and later a consultant on both the 1996 Clinton and the 2000 Al Gore presidential campaigns, now seems to endorse pretty much every bananas conspiracy theory going, from the claim that Anthony Fauci was in the pay of Israel to the claim that ‘chemtrails’ are being used to engineer the climate.
What explains this trajectory? No doubt there are multiple factors, and individual dispositions and history play an important role. Naomi Wolf’s on-air humiliation over the meaning of a legal term that was central to the case she had made in a book may help explain her radicalisation (as Klein, Reference Klein2023, suggests), for example. I’m going to argue that there are common factors that are likely to play a role in many cases and that explain the trajectory from left to bananas right.Footnote 3
1. Pivot beliefs
One central factor in the radicalisation of thinkers, I suggest, is the existence of pivot beliefs. A pivot belief is a belief that belongs to a broader set of beliefs: the set of beliefs that are in some sense available to people on both the left and the right. A political outlook is constituted by, or perhaps brings with it, a set of commitments. The precise nature of these (shifting) commitments is difficult to ascertain, and highly controversial. It is controversial, in particular, which beliefs constitute the core of an outlook, which beliefs are entailed by the core in virtue of the particular facts that happen to obtain at a time, and which are optional, perhaps even inconsistent, extras. Nevertheless, that there are some such commitments seems undeniable.Footnote 4 The left emphasises equality (across multiple domains, but especially equality of resources) to a far greater degree than the right. The right emphasises individual liberty in a form on which the left places much less value (see Graham et al., Reference Graham, Haidt and Nosek2009, for an attempt to identify the moral ‘foundations’ characteristic of conservatives and liberals). Pivot beliefs are members of the set of beliefs that appear compatible, or not obviously incompatible, with the commitments of both or all mainstream outlooks (from now on, I will make the simplifying assumption that there are only two contending outlooks).
Being a member of this set of beliefs is necessary but not sufficient for being a pivot belief. Most of these beliefs are not politically charged at all. Most people can agree on them, because they have no obvious political entailments (‘cold water is quite refreshing when you’re hot’). Some members of this set of beliefs do have political implications, or just are political beliefs, but they’re not controversial (‘people should feel safe on the streets’).Footnote 5 Members of this set become pivot beliefs only when one or other side of politics undergoes an internal split over them.
Here are two recent examples. Neither left nor right was antecedently committed to a specific view about lockdowns as a response to the Covid-19 pandemic. The left has, typically at least, been concerned with public health and protecting the vulnerable, and those concerns seem to support lockdowns. On the other hand, the left has also typically been suspicious of the expansion of police powers, and also of responses that have vastly differential impacts on the better off (the laptop workers, for example, who could easily work from home) versus those who survive paycheck to paycheck, who are also often those in roles that required them to continue to be exposed to a risk of infection. The right has typically been suspicious of heavy-handed governmental interventions, and has instead emphasised individual responsibility. But it also has traditionally placed an emphasis on deference to duly constituted authorities.
In practice, of course, the left and right diverged in their support for lockdowns. The left strongly supported them, as well as associated measures (like mask mandates). The right opposed them. But because both positions were compatible with the commitments of the left and the right before they came to settle on a view, attitudes to lockdowns were apt to serve as pivot beliefs. Individuals who identified with one side or the other and who felt strongly about the issue, for one reason or another, might find themselves at odds with most people on their own side.
A second example continues to exert a powerful force over political debate: whether trans women are women on the same footing and with the same rights as cis women. Prior to this debate becoming an ideological flashpoint, there was significant opposition to treating trans women as women on the left. Sheila Jeffreys, for example, argued that trans women reinforce stereotyped norms of femininity (Jeffreys Reference Jeffreys1997; Reference Jeffreys2014), and Mary Daly argued that gender reassignment was an instance of patriarchal violence and that it enabled men to invade women’s spaces (Daly Reference Daly1991). These views were controversial from the first. Importantly, however, they were not grounds for ostracism or anything like it: for very many feminists, views like these were mistaken, but either reasonable or at any rate not grounds for excluding their proponents from the inner circle of allies (trans women, naturally, took a rather different view). Prior to the third wave of feminist thought, it wasn’t obvious which view would prevail. Both trans inclusive and trans exclusive views were available to mainstream feminists until fairly recently, when the camps diverged sharply. On the other hand, a small minority of those on the right – largely right libertarians – offered some degree of support to trans individuals. Again, this position was not seen as a reason to regard their advocates as beyond the pale.
To avoid misunderstanding, it’s important to note that I am not claiming that there wasn’t a fact of the matter on what left (or right) commitments entailed with regard to trans women or to lockdowns. My claim, rather, is that it wasn’t at all obvious what these commitments entailed, and as a consequence there were significant numbers of people on each side who saw a view incompatible with the one their side went on to take as available, or even required, by their political outlook. That this should occur is unsurprising: Working out the nature of our commitments is not at all easy, in part because these commitments are broad and somewhat vague (Eric Schwitzgebel, Reference Schwitzgebel2009: 57, calls them ‘only half-filled or quarter-filled with a real thought’). Whether or not the beliefs should have been available to both sides, the fact that enough people judged them to be was sufficient to make the beliefs (trans women are not women; lockdowns are wrong) pivot beliefs.
Pivot beliefs are pivot beliefs because on them a political outlook can turn. Someone on the left can leap to the right on the basis of having adopted such a belief, I claim.
2. Making the leap
How and why does the pivot happen? How and why does someone who dissents from their (former) side on some issue find themselves adopting a broad swathe of beliefs characteristic of the other side, including beliefs apparently unrelated to the pivot itself? I claim the pivot occurs due to a combination of psychological and epistemic pressures.
2.a Psychological pressures
It is of course possible to hold a political belief that is at odds with that of most people (or of those who are seen as thought leaders) on the side you identify with without psychological discomfort. In fact, that’s quite common. It’s certainly true of me. In most cases, though, these divergent beliefs occupy a relatively small place in political conversation. The more important the belief is to you and to others, the more psychological pressure the dissident leftist will likely feel.
By itself, one dimension of importance – importance to the dissident, or importance to others on their side – is usually insufficient to bring about significant psychological pressure. Say you’ve watched your sister struggle in her small business, and that’s made you sensitive to the concerns of small business owners. On that basis, you might find yourself thinking (rightly or wrongly) that taxes on such businesses should be lower than most people on your side think. You’re out of step with your side on an issue you feel strongly about, but so long as this issue isn’t one that exercises most people, you might find dissent psychologically sustainable. Even when others are exercised by an issue, your dissent may not much trouble you. You might, for example, think (rightly or wrongly) that cultural appropriation is a trivial issue, and find others’ deep preoccupation with it a bit silly. But because you think it’s trivial, being out of step on the question mightn’t particularly worry you.
It is, however, highly uncomfortable to find oneself out of step with your side on an issue that you and they both care deeply about. This occurs most commonly with those issues that are central to the culture wars; that is, that set of issues on which opinions are politically polarized and on which feelings run high. Because they’re central to the culture wars, others often express their opinion on them, or express opinions that obviously commit them to one side or the other. They make their opinion on these issues plain, and they may be quick to condemn those who diverge from them. In these circumstances, a dissenter can be expected to experience discomfort. This is especially likely if they have a public profile and for that reason (or for idiosyncratic reasons) feel compelled to take a public position on the issue, say by posting on social media.
This psychological discomfort constitutes pressure to change either one’s views or at least one’s public position. Some people react to this sort of pressure by withdrawing from the debate. If they feel strongly enough about it, though, or they value being heard, they may be unwilling to do that. They may try, for a while, to bring others around to their point of view. Of course, sometimes they succeed in doing so: arguments can change minds, and there are surely cases in which one side or the other begins to settle on one view but is converted to another by arguments. But of course such attempts often, perhaps usually, fail. If dissenters will not withdraw and fail to convince others that their views are worthy of respect, they must either resign themselves to facing not only disagreement but perhaps even condemnation and contempt, or look elsewhere for engagement. Unsurprisingly, many of them choose the latter option.
If a dissident takes the option of looking elsewhere for engagement and uptake, they will need to look beyond their side, to the non-aligned and to those on the other side of politics. In the polarised context of the culture wars, the non-aligned are hard to find, at any rate among those who are active in discussions. The vast majority of those who won’t condemn my judgment on some divisive issue are to be found on the other side. This is especially true when, as is commonly the case with culture wars issues, belief serves as a marker of identity: to adopt a position on lockdowns or on trans people is to identify oneself as a member of a side. Given these facts, dissenters can be expected to find themselves under pressure to seek validation, or any substantive engagement, from those on the other side. If they do, they will probably find themselves given an especially warm welcome and an especially high degree of credibility in virtue of their left-wing credentials, since their dissent from their side on this issue is taken to show the power of the right’s position (see Nyhan & Reifler Reference Nyhan and Reifler2013; Berinsky Reference Berinsky2017 for evidence that testimony is seen as especially persuasive when it comes from those who are seen to break ideological ranks).
These psychological pressures help to explain why a dissenter might find themselves talking more to those they see (or formerly saw) as on the other side than their own side. They thereby find themselves exposed to the claims and arguments of the other side and this very fact may bring them to become more sympathetic to these claims, perhaps by virtue of the mere exposure effect, whereby repeated exposure to something increases one’s liking for it (Bornstein & Craver-Lemley Reference Bornstein, Craver-Lemley and Pohl2022), or fluency effects, whereby things that are processed more fluently are seen as more plausible (Oppenheimer Reference Oppenheimer2008). The effect is especially likely to be powerful when we come to feel more positively about those who advance these arguments and with whom we engage civilly (see Fantl Reference Fantl2018 for a suggestion along these lines).
At the same time, the dissenter will hear relatively less from their own side – if only because time and attention are limited resources – and when they do their response to arguments from their own side will be coloured by the perception, perhaps the knowledge, that these people either have or would refuse to engage with them respectfully.
These psychological pressures might cause people to come to be more open to the other side and less open to their own. Alongside these mere causes, however, there are reasons for dissenters to pivot from their side to the other.
2.b Epistemic pressures
Disagreement – and agreement – are epistemic signals. On most views, that a peer disagrees with me is a reason for me to change my mind (though it may not be a sufficient reason). My epistemic peers are those agents who I have no reason to think are more (or less) unreliable in making judgments about an issue than I am. They have more or less the same capacities to judge in this domain as I do, and a roughly equal track record to mine. Moreover, there is no reason to think that in this particular case one of us is likely to have an edge. If my peer and I find ourselves disagreeing about the issue, we each have a reason to conciliate: to reduce our confidence in our judgement, perhaps even assigning equal weight to the probability that each of us is correct (Christensen Reference Christensen2007; Elga Reference Elga2007; see Matheson Reference Matheson2015 for discussion).
The peer disagreement literature rarely talks about either agreement or about the difference that the number of peers involved can make, but it has the resources to show that each makes a difference.Footnote 6 If the fact that a peer disagrees with me is evidence that I’ve made a mistake (one of us has, and I have no stronger reason to think it’s them than it’s me), then the fact that a peer agrees with me is evidence that I have not made a mistake.
To see this, consider the by now hackneyed restaurant case (Christensen Reference Christensen2007). After a meal, a group of friends decide to split the bill equally. Two of them, A and B, add up all the items on the bill, without worrying about who had what, add a 15% tip, and divide the total by the number of people in the group. A announces that each owes $43.70. B disagrees; saying it’s $45.30. Their disagreement is evidence for each that they have made a mistake. But now consider if A and B agree. Since the sum is moderately difficult – the kind of thing on which it is easy to make a mistake, for example by overlooking an item or glitching in one’s mental arithmetic – the fact that my peer agrees with me is evidence that I haven’t made a mistake. The likelihood that each of us made the same mistake (or at any rate a mistake of precisely the same magnitude) is smaller than the likelihood we’ve done our sums correctly.
The same sort of example can be extended to show the difference that numbers make. Suppose every member of the (let’s say) eight person group does the arithmetic and they all come to the same result. That’s strong evidence that that’s the correct answer: what are the chances that all eight made the same mistake, or a mistake of precisely the same magnitude? Suppose, on the other hand, A announces the answer is $43.70, while B through H all say $45.30. That’s good evidence that A is wrong: it’s far more likely that one person is wrong than seven. Agreement and disagreement among peers each provide evidence, and the larger the number of people on a side, the stronger the evidence provided.
So far, I’ve been assuming that the agents involved are (roughly) epistemic peers. But numbers can make another difference that the literature has largely neglected: they make peerhood less important. Suppose that A takes his eight-year-old nephew to the fast food restaurant, and they each add up the items on the bill afterwards to determine what A needs to pay. If A discovers that his nephew disagrees with his arithmetic, he has – at most – a weak reason to conciliate, since his nephew is no maths wiz (on the other hand, let’s suppose, A is good with numbers). Because his nephew is not a peer in this domain, A should think it’s far more likely that the mistake is on his side. But now suppose A finds himself in the same situation with a dozen eight-year-olds. If they all independently come to the same sum, dissenting from him, he does have strong evidence that the mistake is his. Even though none of the children are A’s peer when it comes to arithmetic, each is competent enough at the task for their identical judgments to have greater evidential weight than his lone dissent. Numbers can outweigh peerhood.Footnote 7
Numbers can render peerhood less relevant. At the same time, agreement and disagreement can provide evidence about who my peers are.
Agreement provides the strongest evidence about peerhood when it concerns difficult questions, whereas disagreement provides the strongest evidence about peerhood when it concerns easy questions.
Suppose that A is a mathematician, engaged in a task with a colleague who works in a different, but maths-heavy, field and about whom she knows nothing beyond that fact. Each independently constructs a proof of some conjecture. If A finds that her proof is identical to that of her colleague, she has good evidence that her colleague is a very good mathematician. Suppose, conversely, that the task is a relatively simple one for a mathematician, but her colleague produces the wrong answer and can’t see any mistake in their reasoning. A will conclude that whatever other qualities her colleague possesses, they’re not much of a mathematician. The simpler the task or the more absurd the answer her colleague produces, the more strongly she will relegate her colleague from peerhood. At the extreme, she will conclude that her colleague is either trolling her or suffering from some sort of glitch in their reasoning (Lackey Reference Lackey, Haddock, Millar and Pritchard2010).
With all this in mind, let’s return to the dissenter.
They find themselves in strong disagreement with their (former) side on some issue, and in agreement with those on the right. This disagreement entails evidence about the reliability of each side. As we just saw, agreement and disagreement provide different kinds of signals depending on how difficult the issue is perceived to be. This holds true in the dissenter’s case.
Commonly, the dissenter see the issue as one that is really quite simple: it’s just common sense (Thurlow Reference Thurlow2024), or an obvious implication of basic moral commitments. If they think the question is a simple one, then those who agree with them don’t send a strong epistemic signal thereby (anyone should see that). Conversely, disagreement sends a very powerful signal: those who disagree must be irrational, or morally corrupt, or fundamentally not serious. What else could explain their failing to see what is obvious?
Disagreement on an apparently simple question thereby sends a strong signal about the competence or the morality of those who disagree. Perhaps they’re trolls. Perhaps they’re fundamentally immoral. Perhaps they’re in the grip of some irrational ideology. Whatever interpretation the dissenter alights on, the signal constitutes a genuine reason (with ‘reason’ here understood internalistically, of course) to break decisively with their side. Their believing that casts doubt on the value of their testimony quite generally, by demonstrating that they are incompetent, irrational, or immoral.
More rarely, the dissenter might see the issue as a relatively difficult one. In that case, their side’s disagreement does not send a particularly strong signal. In this situation, though, the right’s agreement does send a strong signal: by getting this difficult issue right, it sends a signal of its competence (moral or rational). The signal doesn’t provide a strong push away from the left, but it does provide a strong pull toward the right. Again, the push and pull is constituted by genuine evidence; the dissenter has a reason to regard the right as more insightful than they had thought – indeed, as more insightful than the left.
Whether they see the issue as an easy one or a difficult to judge, then, the dissenter now finds themselves in a situation in which they should weigh testimony from the right more heavily than in the past, and testimony from the left less heavily. Agreement and disagreement provide evidence of competence and of the right values; both factors to which agents are very sensitive when it comes to assessing the credibility of testimony (Harris, Koenig, et al. Reference Harris, Koenig, Corriveau and Jaswal.2018; Sperber, Clément, et al. Reference Sperber, Clément, Heintz, Mascaro, Mercier, Origgi and Wilson2010). Our sensitivity to these factors are not the mere psychological quirks of animals like us, but responsiveness to epistemically relevant properties. Competent testifiers are obviously more likely to be accurate than less competent – we just saw how this works in the case of peer disagreement. Even our preference for the testimony of benevolent testifiers is epistemically rational. First, such testifiers are less likely to be deceptive or exploitative than malevolent testifiers. Second, many issues are morally inflected (including many that appear not to be), such that we ought to expect those with better moral perception to be more accurate in assessing them (Rini Reference Rini2017).Footnote 8
All these factors influence how the dissenter epistemically ought to respond to testimony from those who agree or disagree with them about the pivot issue, whether that testimony concerns that issue or another one. When testimony comes from those who agree with the dissenter about the pivot, it comes from someone who manifests moral competence (as assessed by their lights). Their agreement also exhibits empirical competence, insofar as coming to the right view requires assessing non-moral facts. Correlatively, testimony from those who disagree with the dissenter comes from those who – thereby – indicate a lack of moral and empirical competence.
The numbers of testifiers make a difference too, of course. The dissenter finds themselves in receipt of testimony not only from one or two people who agree with them on the focal issue, but hundreds, or even thousands. When this large group of people all appearing thereby to manifest competence and moral insight testify that p, for any p, they thereby provide strong evidence, by the dissenter’s own lights, in favor of p.
I’ll finish this section by considering two objections. The first objection focuses on the rationality of the dissenting agent placing sufficient weight on the pivot belief for it to serve the pivot function. The second focuses on the rationality of trusting the side that accepts the pivot belief, even if it is rational to accept that belief itself.
It may be doubted that it is rational for the agent to accept or to place significant weight on the pivot belief given their awareness that many, many people disagree with their judgment. Moreover, these are people who, until recently, they had seen as peers. Since peer disagreement carries information, should they not reduce their confidence in their judgment? If they do so, of course, the belief won’t serve as an epistemic pivot on which their outlook can turn. Conversely, if they maintain their belief in the face of all this higher-order evidence, they cannot be held to be acting rationally.
At least in most cases, I think the weight the agent places on the pivot belief can be explained rationally. If the issue is seen by the agent as difficult, placing this kind of weight on the pivot would be irrational. Given the difficulty of the issue, the agent cannot rationally disregard the higher-order evidence. Such cases are, however, rare: in most actual cases, the pivoting agent regards their belief as obviously correct. In such cases, they are rational in disregarding the higher-order evidence. Just as the person who – in Lackey’s (Reference Lackey, Haddock, Millar and Pritchard2010) example – entirely disregards dissent from the proposition that ‘2+2=4’ does so rationally, so the person who sees the pivot as commonsensical might rationally take disagreement to be grounds for dismissing their interlocutors from peerhood. Beliefs are apt to serve as pivots, therefore, when the dissenter is very confident in them, and that will usually be when they see the right view as obvious. In those cases, as we’ve seen, disagreement with the belief sends a powerful signal of incompetence or immorality.
The second objection holds that even if the agent is rational in accepting the pivot belief, they are not rational in pivoting on it. That is, they are not rational in taking the testimony and opinions of those people who also accept the pivot belief as strong evidence in favor of other questions. Pivoting like this is not rational because the agent possesses strong evidence that on other issues they are unreliable. I might be well disposed toward them in virtue of the fact that they agree with me enthusiastically on my pivot belief, but I also know that they are prone to get so many other questions, moral and empirical, wrong. I am therefore irrational in pivoting toward them.
It would be irrational for an agent immediately and completely to pivot. If I discover that agents who I regard as badly mistaken on most issues nevertheless agrees with me on one that is especially important to me and which is controversial, I would be irrational to adopt their opinions on these other issues all at once. I doubt that’s how pivoting occurs, though. Rather, the agent takes the testimony and opinions of their former foes more seriously and the testimony and opinions of their former allies less seriously, and gradually pivots. The process need not be all that slow: so far as I can tell, it can occur in a very few months. But it takes time. It takes time because it takes time for people to reassess evidence.
In sum, beliefs can become pivots on which outlooks can turn when they give rise to psychological discomfort or when they send epistemic signals concerning the relative reliability of testimony from each side of debate. I won’t try to assess the relative importance of each mechanism; it may be that either would be sufficient on its own, at least in some cases, but when they occur together – as they routinely do – they work powerfully to push and pull the dissenter away from one side and toward the other. The person who dissents on one belief can rapidly come to adopt a whole range of apparently unrelated beliefs, because these beliefs are espoused by those they now see as relatively reliable, and opposed by those they now see as relatively unreliable and because in adopting these new beliefs, they decrease the psychological discomfort that arises from disagreement with the beliefs of those they had formerly seen as their peers.
3. Going bananas
The mechanisms just described help us to understand how and why dissenters may pivot on their beliefs. Adopting a pivot belief can put agents under psychological pressure and bring about a host of changes in how they weigh testimony that together make it likely that they will relatively rapidly move from one side to the other. But nothing in the mechanisms described explains why agents might move to the far, even bananas, right.
It’s especially mysterious why this would happen given that the mechanisms we employ to assess the credibility of testimony are sensitive to plausibility, as well as the competence and benevolence of testifiers (Harris Reference Harris2012; Sperber, Clément, et al. Reference Sperber, Clément, Heintz, Mascaro, Mercier, Origgi and Wilson2010). The kinds of conspiracy theories that Power and Weinberg and Wolf each accept are highly implausible. It’s one thing to explain why they are now more sympathetic to the right than to the left, and quite another to explain why they’re endorsing bananas views.
Part of the explanation might cite contingent facts about the political landscape right now. In the United States as well as in some other countries the mainstream right – that is, the right with the most political power and the most supporters – is not any version of the reasonable right. It flirts with, or might just be, the bananas right. That powerfully shapes the content of the testimony the dissenter receives from those whose credibility is high in their eyes. They may assign sufficient credibility to bananas testifiers in light of the signals sent by agreement and disagreement to outweigh the implausibility of the claims testified to.
Another factor might be the political analogue of the so-called ‘zeal of the convert’; the supposed tendency of those who convert to a faith to be more zealous in adherence and promotion than those born into it. The phenomenon, if it is real, might in turn be explained by the perceived need of the convert to prove themselves to their co-religionists and to promote a sense of belonging for themselves. Evidence for the existence of the phenomenon in religion is, however, mixed (Beider Reference Beider2021; Liu Reference Liu2009). To the extent it is real, it would seem to predict the embrace of orthodoxy, as conceived in the tradition, and not fringe beliefs. While some extreme positions may signal that one is more right than the right, as it were, many of the more bananas belief don’t seem specifically right-wing at all (recall Nina Power’s fairies).
Perhaps, however, we might explain the adoption of bananas beliefs among some of those who pivot by a felt need to promote a sense of belonging akin to the need that is purported to explain the zeal of the convert. Researchers on radicalisation among young Muslim men have emphasised the sense of belonging promoted by a small group of people committed to a common cause (e.g., Atran Reference Atran2011). Something analogous may be occurring among those who adopt bananas beliefs. One might think of the way in which prominent members of the (so-called) intellectual dark web often appear on each other’s podcasts. They may thereby promote a sense of belonging, somewhat akin to the ‘band of brothers’ from which extremists often stem. The need to promote belonging might be especially strong in pivoters, who have recently lost the community to which they belonged.
There is, however, another factor at work.
Bananas beliefs are not equivalent to conspiratorial beliefs, both because they need not feature conspiracies (though they routinely do) and because not all conspiratorial beliefs are bananas. Some conspiracy theories are true, after all (Coady Reference Coady2006; Pigden Reference Pigden, Lippert-Rasmmussen, Brownlee and Coady2017). Nevertheless, the conspiracy theory literature can provide clues as to how generalized distrust can render agents susceptible to bananas claims.
Some theorists have argued that conspiracy theories are theories that are contrary to the official story – that is, the narrative that originates in or is endorsed by epistemic authorities (Coady Reference Coady2003; Harris Reference Harris2023; Levy Reference Levy2007). Whatever the merits of this view as an account of their nature or of the rationality of accepting them (see Hayward Reference Hayward2024 for critical discussion), it has the merit of drawing attention to the role of trust and mistrust in explaining why people accept conspiracy theories. Bananas beliefs are always contrary to the accounts promulgated by genuine epistemic authorities. Accepting – even seriously entertaining – such theories implies low trust in official sources. We should therefore expect that those dissenters who do not merely leap right but accept bananas claims to have had their trust shaken.
There are two reasons dissenters may find themselves less trusting of mainstream testimony.
First, in many cases mainstream institutions and sources endorse the view that the agent now decisively rejects concerning the pivot topic. Indeed, the view they reject may have its origin or have a major source of support in mainstream institutions (think of the CDC on both mask mandates and lockdowns, or the AMA on gender-affirming care). Since this view is false, by the agent’s lights, in promulgating or endorsing it, the mainstream institution sends a powerful epistemic signal to the dissenter, in the same way as the endorsement of that belief by their (former) side did. They thereby reveal themselves as untrustworthy, by the dissenter’s lights.
Second, the dissenter may now be more mistrustful generally, because they see themselves as having recovered with difficulty from having had the wool pulled over their eyes by their former comrades. For many years, they may think, I trusted these people. They seemed competent and benevolent; now I see them for what they are! Taking oneself to have been fooled on an important matter by people who seemed competent may promote suspicion. If they’re perfidious, who else isn’t to be trusted? The dissident may, in effect, come to adopt a ‘won’t be fooled again’ heuristic, being suspicious of all testimony that doesn’t come from those who explicitly share their pivot belief. That fact alone may make them more likely to accept bananas claims.
After all, an enormous amount of what we know we know on the basis of testimony (Roeber Reference Roeber2024). We would know almost nothing about history, or about the world outside our city – even our suburb – if we did not rely on it. Even facts about yourself (where you were born, who your parents are, and so on) you know only because you accept someone’s word for it. When it comes to facts that are contested by those who endorse bananas beliefs, our dependence on testimony is equally pervasive. Very few people are in a position to assess claims like the benefits of the Covid vaccine are significantly greater than the costs or 5G technology is safe for themselves: these claims fall within the domain of expert knowledge. If we’re not prepared to take anyone’s word for it except those who agree with us on the pivot belief, we cut ourselves off from the testimonial networks on which our knowledge depends. If we approach even our own side with heightened suspicion, we’re forced to fall back on ‘doing our own research’, and that’s a highly unreliable way of coming to beliefs about expert topics (Ballantyne, Celniker, et al. Reference Ballantyne, Celniker and Dunning2024; Levy Reference Levy2022).
Adopting a pivot belief can therefore open the way to going bananas. By promoting general mistrust, and mistrust, specifically, of epistemic authorities, they make the dissident vulnerable to all sorts of snake oil salespeople and charlatans. In their zeal to avoid being fooled, the dissident becomes liable to accepting some of the least credible claims going.
4. Conclusion
People we might have seen as our allies sometimes seem to go off the rails. Our former friends suddenly become culture warriors, first dissenting vociferously on some issue, and then adopting a raft of views that seem alien to the worldview they formerly espoused. Sometimes, these views are not merely wrong, by our lights, they are wrong by the lights of most people. They seem to have gone bananas.
In this paper, I’ve suggested that a combination of psychological pressures and genuine (internal) reasons explain the phenomenon whereby someone may quite suddenly switch allegiance. It’s uncomfortable to remain engaged with those with whom one disagrees deeply on an issue one really cares about. It’s uncomfortable even to see oneself as on the same side as those who are wrong about that. This fact alone makes dissenters more sympathetic to the other side, who agree with them on the pivot belief. At the same time, agreement and disagreement send strong epistemic signals, rationally changing the weight the dissenter places on testimony from each side.
This mechanism explains why people may pivot in their worldview on a belief. It doesn’t, by itself, explain why they pivot to bananas claims. I’ve suggested that a generalized suspicion of epistemic authorities may arise subsequent to a loss of trust in one’s side, and this generalized suspicion may render dissenters vulnerable to these claims. If something like this account is correct, we can explain how people may pivot from a set of beliefs that is reasonable and widely shared to a set of beliefs that includes some that are utterly bananas, all the while driven by a concern for evidence.
It’s important to note that nothing in the mechanisms described is partisan. It’s not because one starts on the left that one is subject to them. Pivots in the opposite direction, from right to bananas left, are at least possible in principle; in all likelihood such pivots have actually occurred. Perhaps contingent facts about the political landscape right now make them less common – for example, the strength of the political right in the United States and its growing power in western Europe and elsewhere might have the effect of pulling the space of available positions right, ensuring that the ‘loony left’ has decreased in visibility and importance. Perhaps those young people who support the populist right in increasing numbers today (Snowdon Reference Snowdon2024) would instead have found a home on the extreme left in a different social context Perhaps the apparent paucity of such cases might be explained by the tendency of people to shift rightward as they age.Footnote 9 It’s unlikely that such effects – the absolute magnitude of which is small (Peterson, Smith, et al. Reference Peterson, Smith and Hibbing2020) – rule out pivots from right to bananas left. In the right circumstances, perhaps any of us could go bananas.Footnote 10
Open access
