Why we can't argue each other out of our worldviews

For years I have held a hunch I so much wanted to be true. It began as a curious nagging question I kept returning to whenever I watched two intelligent, well-meaning people — an environmentalist and an entrepreneur, say — try to talk each other out of their respective worldviews over a cup of tea. It never worked. It still doesn't. Why not? And the reason, I suspected, was not that either was stupid or stubborn. It was that the debate was happening, the words were being spoken, but they just didn't get through.

I want to be clear up front: I am not a psychologist or neuroscientist, and I have not studied any of this formally. My strength, for better or worse, is pattern-spotting across domains — noticing the same shape of thing turning up in different fields. What follows is a pattern I have been sitting with for years. When I recently asked Claude to check it against the actual literature, I was pleasantly surprised to see it surfaced real backing. I have read Dr Darren Stevens' PhD on Constructed Development Theory — that one particularly resonates — but the rest I am just happy to see AI summarise in a way I can match to my own thinking. So the shape of the argument is mine; the specific citations below were surfaced with AI help, and I offer them as signposts rather than as a claim to expertise.

Let me walk through it.

---

Why tea and good intentions don't rewrite a worldview

Start with the observation. Two adults who genuinely care about the world, both articulate, both in good faith, sit down to discuss climate, economics, parenting, or politics. They exchange reasons. They present evidence. They leave unchanged, and likely hot under their respective collars.

Both parties feel that the other is being irrational or ideological. I think this is framed almost entirely wrong. What is actually happening is that their positions are not mainly arguments held in the reasoning part of the brain at all. They are deep structural commitments — built into the same machinery that holds their sense of self.

My hunch had been that our deep beliefs about how the world works are literally wired into our brains through the brain's own plasticity.

The next three paragraphs are the science. Feel free to skim — I pick the thread up again on the other side.

The most striking direct evidence comes from Jonas Kaplan, Sarah Gimbel and Sam Harris at USC. In a 2016 fMRI study published in Scientific Reports, they presented committed liberals with counter-evidence to their strongly held political beliefs. Political beliefs turned out to be far more resistant to change than non-political ones. More importantly, where the resistance happened was revealing. Challenges to political beliefs lit up what researchers call the default mode network — the brain's self-referential system, which hums whenever you think about who you are — along with the amygdala and insula, the circuits that fire when you feel threatened. The people who did change their minds showed less activity in those threat circuits, not more. In other words, updating your worldview is not primarily a reasoning event. It is an event in which your threat system either stays quiet enough to let updating happen, or it doesn't. A 2026 pre-registered replication in a non-US sample (Kossowska and colleagues) reproduced the same pattern, which suggests this isn't a quirk of American partisanship.

Jay Van Bavel and Andrea Pereira at NYU formalised the broader picture in what they call the Partisan Brain model (Trends in Cognitive Sciences, 2018). Their argument is that beliefs acquire their weight in the brain not mainly from being true but from the identity they signal — what tribe they mark you as belonging to. When the identity-value of a belief outweighs its accuracy-value, the brain will bend perception, memory, even basic visual judgement to protect the group-congruent version. This is not a bug. It is the system working as designed.

The theoretical umbrella for all of this is what researchers call predictive processing, developed by Karl Friston, Andy Clark and others over the past two decades. In this account, the brain is essentially a layered prediction machine. Your deepest assumptions sit near the top and cascade expectations downward, pre-filtering what even counts as evidence. A worldview is a stack of very high-confidence assumptions. Because the brain grants those top-level assumptions an unusually heavy weight (the technical term is precision), strongly held worldviews are effectively self-protecting by design.

So when your environmentalist and your entrepreneur sit down with their tea, the machinery that would need to update is busy defending the self. The threat system is on alert. The deep assumptions have too much grip. The updating never happens, and both leave convinced the other is being unreasonable. Neither is. They are both being exactly as reasonable as brains that evolved to protect coherent selves can be. To survive we must have a bias toward self-belief — in my amateur terms, confirmation bias playing out in its most fundamental form.

---

How the imprinting happens — the Billy-and-the-fire example

Here is the example I use when I try to explain my thoughts to people. Imagine a small child — call him Billy — reaching toward a fire. His mother shouts, sharply and loudly: Billy, no! Don't touch that! Or picture an earlier version of the same scene: a mother gripping her toddler's hand in a wild landscape and shouting don't let go, you could be eaten — which, for most of human history, was literal rather than metaphorical.

What happens in Billy's brain in that moment is not just the formation of a memory. It is the formation of an emotionally tagged memory. Joseph LeDoux's work on the amygdala, developed over three decades, describes the mechanism: when you are emotionally aroused, stress-hormone signals from the amygdala instruct the memory-forming part of the brain to record the event more deeply, more durably, and with its emotional charge attached. There is also a faster, shortcut route from the senses straight to the threat centre that bypasses conscious thought, which is why future similar situations can trigger the feeling of rightness or wrongness before you consciously know why.

I can attest to this from my own life — and it may be where my interest in this whole question began. Forty years ago, as a student on a street corner near Manchester Piccadilly station, England, I heard one car hit another a few yards behind me. I had no warning. It was very loud. I remember almost no details of what I saw afterwards, but the shock tagged the memory, and decades later it has not faded.

Now compound that across thousands of such moments in a childhood. The accumulation is not a set of beliefs the child holds consciously. It is a set of deep filters, many laid down before rational evaluation was even possible, through which all later experience passes. By the time Billy is an adult — articulate, capable of reasoning about fire safety or political economy or environmental policy — the reasons he gives for his positions are largely after-the-fact justifications for verdicts his emotional system has already delivered. Jonathan Haidt's image of the elephant and the rider captures the same point: the gut verdict comes first, the rational explanation is reverse-engineered afterwards.

This is also why I came to suspect that stage-based developmental models — Clare Graves' original bio-psycho-social research being one — are describing something more biological than they are usually given credit for. If the progression from basic survival, through the young child's attachment to caregivers and magical sense of the world, through the toddler's discovery of power, maps roughly onto the infant-to-toddler arc — survival needs, then attachment to caregiver, then the power hierarchy that emerges at nursery — then these developmental stages are not arbitrary cultural constructs. They are tracking the actual sequence in which different kinds of filters get laid down in a developing brain during its most plastic years.

I want to be careful here. I am not claiming that any particular developmental model maps neatly onto discrete brain structures. The neuroscience does not support that. What I am claiming is that the content associated with each developmental step — the felt-sense of safety and tribe, then power, then rules, then achievement, then pluralism — gets absorbed into the same generic machinery described above: high-confidence assumptions, emotionally tagged, defended by threat circuitry, sustained by a social group whose membership is itself part of the self-model. Stage models describe real developmental sequences of content; the neuroscience describes the generic mechanism by which any such content becomes entrenched.

---

Why early imprinting is privileged — and not the only window

One of the things I found most useful in the AI-surfaced research was the clarification that there are at least two major developmental windows here, not one. I have always been fascinated by both — by how early events shape character, and by how teenagers test social boundaries.

The first is early childhood. Allan Schore's and Dan Siegel's work on interpersonal neurobiology, building on Bowlby and Ainsworth's attachment theory, shows that early attachment patterns produce lasting templates in the brain's emotional and stress-regulating circuits. Insecure early attachment is associated with measurable differences in amygdala size and stress reactivity well into adulthood. These patterns are modifiable later — the "earned secure" literature suggests roughly a quarter of adults who test as securely attached got there through corrective experience rather than a happy childhood — but modification typically takes years of long-term therapy, a secure intimate partnership, or sustained contemplative practice. The original patterns are not overwritten. They are gradually out-competed by a new, competing model.

The second window is adolescence. Sarah-Jayne Blakemore and Kathryn Mills argued in a landmark 2014 Annual Review of Psychology paper that adolescence is a sensitive period for sociocultural processing. The social-thinking circuits in the brain undergo major structural and functional reorganisation across the second decade, with heightened sensitivity to peer acceptance, social norms, and identity formation. A 2024 revisit by Cheng, Mills and Pfeifer refined and confirmed the view. If anything is the neural imprinting window for ideological commitments — as opposed to pre-ideological attachment patterns — this is almost certainly it. This is the developmental window for locking in rule-based, achievement-oriented, and pluralistic worldviews — the moral and political frames that tend to persist into adulthood. Political scientists have known for decades that party identifications, moral foundations, and ideological self-placement formed in late adolescence and early adulthood are unusually persistent. We now have a candidate mechanism for why.

This refined picture matters because it changes the policy implications. If early childhood shapes the pre-ideological emotional substrate, and adolescence shapes the ideological superstructure built on top of it, then adulthood is a period of consolidation in which change becomes progressively more costly. Frequently-used pathways literally thicken (the technical term is myelination), so well-travelled routes become faster and more efficient. The brain's natural habit of minimising surprise produces what we experience as confirmation bias. The brain's defence of identity recruits the threat circuits. None of this is character weakness. It is the physics of a biological prediction system that has already paid a lot of metabolic cost to arrive at a coherent self-model.

Another way of saying it is that our fast thinking is largely unconscious — and it needs to be, in order to function at all. We rarely apply slow thinking to our own unconscious biases.

---

When adults do change — and why it takes so long

None of this is to say change is impossible. It does happen, and the literature on how it happens is remarkably consistent across domains.

Jack Mezirow's transformative-learning theory, built from decades of studying adult learners, describes a characteristic sequence. A disorienting dilemma — a life event the existing view of the world cannot digest — starts the process. This is followed by hard self-reflection, exploration of new roles, provisional trying-on of a new perspective, and gradual reintegration of identity around a changed frame. The timeline is typically months to years, not minutes to hours. The process is emotional and embodied, not purely rational.

Robin Carhart-Harris and Karl Friston's REBUS model (Pharmacological Reviews, 2019) — the clunky acronym stands for Relaxed Beliefs Under pSychedelics — extends predictive processing to explain why psychedelics can sometimes achieve in a few hours what therapy struggles to achieve in years. Psychedelics appear to temporarily loosen the grip of those top-level assumptions, particularly those sitting in the default mode network, letting bottom-up information re-enter and revise the model. As the drug wears off, the hierarchy re-forms — but in a subtly different configuration. Durable shifts in attitude and worldview can result.

Three further conditions recur in the change literature, and they all point the same way. First, emotional and bodily safety: Kaplan's subjects changed their minds when threat-system activity was lower, not higher. Confrontation is counter-productive. Second, a new social environment: because the value of a belief depends on the group it marks you as belonging to, genuine worldview change almost always involves a shift in a person's reference group. Studies of religious deconversion and political realignment consistently find that people change relationships before they complete the change in belief. Third, time and repetition: rewriting deep assumptions requires many cycles of safely experiencing the world failing to match them — the neural equivalent of learning a new language, not swapping a fact.

This is why worldview change takes months or years. Not because people are obstinate. Because the substrate is literally, physically configured to resist rapid change, and the only routes to revision run through the whole embodied, social, emotional system rather than through argument alone. It is also why major life upheavals are so often the trigger: the unconscious processing starts to fail at predicting the world, and if the mismatch is sustained, the brain has no choice but to begin rewiring.

---

What about stages of thinking complexity?

I have one further speculation, and it reaches beyond what the research currently supports.

Patterns recur across reality. Electron energy levels cluster at discrete values. Atoms combine into molecules in constrained ways. The same carbon atoms, depending on how they arrange, settle into either graphite or diamond — two radically different materials built from identical components. Crystals take specific lattice forms. Social groups, solar systems, galaxies — complex systems settle into a relatively small number of stable configurations determined by their underlying constraints. I have wondered, for some years, whether the stages of human cognitive development might be a similar story. Not that brains become crystalline in any literal sense, but that the space of stable configurations able to sustain progressively deeper self-reflection and prediction may have a finite number of resting states.

Even neurons have to follow rules. They live in physical space, embedded in a network of many others, and the connections between them are not unlimited. New links have to grow rather than form instantly, and the geometry of where each neuron sits constrains who it can reach. It is a bit like trees: branches and leaves cannot just appear anywhere — they have to share light and space, and follow patterns set by the physics of the real world. Brains face the same kind of constraints — and that, I suspect, is part of why developmental stages exist at all.

This is the territory of dynamical systems neuroscience, and it is a live question. Kurt Fischer's dynamic skill theory, developed at Harvard until his death in 2020, showed that cognitive-developmental spurts happen in step with bursts in brain-wave power and cortical reorganisation. His work, along with Michael Commons' Model of Hierarchical Complexity, is the closest the field has come to an empirically grounded account of developmental stages as genuine neural discontinuities rather than arbitrary classifications.

Dr Darren Stevens' Constructed Development Theory offers another angle, describing fifty "Cognitive Intentions" as habituated thinking styles — heuristic shortcuts that become grooved through repetition. I have discussed this with Darren, and I suspect there is a bridge waiting to be built. Specific Cognitive Intention repertoires are likely prerequisites for stably holding specific developmental worldviews. You cannot stably inhabit a genuine pluralistic concern for systemic fairness, for instance, without the underlying habits of perspective-taking and tolerance for ambiguity. In that framing, stage models describe content; CDT describes cognitive-structural capacity; and together they approach what Fischer's work hints at — that stages are real neural phenomena, not just classifications.

The testable version of this is still waiting to be done. It would involve measuring the neural signatures of people at different developmental stages and asking whether there are genuine discontinuities in how their brain networks organise themselves. My intuition says yes. The evidence so far is circumstantial but suggestive.

---

Why this matters now — education and the pace of AI

Which brings me to why this matters so much now, and to a story I have carried for years.

Years ago a colleague in software told me, with deep frustration, that he could not respect politicians who argued like children. The line stuck with me. At the time I was a prospective local council candidate, quietly keen to understand politics' relation to wider society. Our local MP was Sir Andrew Stunell. He struck me, even then, as a rare politician — someone who seemed genuinely curious about questions rather than positioning himself in relation to them, and who went on to help negotiate a national coalition government. I put the question to him directly.

His answer has stayed with me ever since: it is better than war.

I have quoted him often. For me it is the short version of why democracy matters. Not because it settles all arguments — it manifestly does not — but because it lets different worldviews coexist, and because the switching of governments between those worldviews is how societies, in aggregate, evolve. Democracy is, at least in part, a civilisational workaround for a neural fact: it allows competing worldviews to alternate in power without either side ever having to be persuaded out of itself.

That workaround has always depended on one thing, though: time. Time for governments to switch, for generations to turn over, for culture to stretch at the margins and slowly absorb new patterns.

For most of human history, the unconscious training provided by parents, schools and culture did a reasonably good job. It reproduced generationally stable behaviour patterns optimised for environments that changed slowly enough that the patterns remained useful across multiple lifetimes. Teenage rebellion — the adolescent window I mentioned above — provided just enough generational flexibility to stretch culture at the margins without breaking it. Agricultural societies, early industrial societies, even the settled consumer societies of the twentieth century, could afford this model. The patterns laid down in childhood were broadly appropriate for the world the adult encountered.

This is no longer true.

The pace of change in human social, economic and cognitive environments has decoupled from the pace at which individual brains — or indeed whole generations — can adapt. AI in particular is now changing the substrate of work, communication, trust, and meaning on timescales of years, not generations. The patterns we imprint in children today are statistically unlikely to be fit for the world those children will inhabit as adults. Teenage rebellion is no longer enough. Generational turnover is no longer enough. The unconscious reproduction of cultural patterns, which served us for millennia, has started to break as an adaptive mechanism.

This is why I think our education systems need a fundamental rethink. The dominant paradigm — even at its best — is still optimised for the transmission of fixed patterns. Traditional university education at its best valued something different: intellectual agility, the capacity to hold multiple frames at once, to revise one's assumptions in light of evidence, to tolerate the discomfort of not-knowing. That sensibility was once considered a luxury of a small elite. I believe it is now the central competence education needs to cultivate in everyone, because it is the only competence that will remain useful across the transitions ahead.

In neural terms, this means designing educational experiences that deliberately keep minds open for longer, rather than locking in fixed patterns early — a shift that many parents and politicians will instinctively resist, for precisely the reasons described above. It means protecting cognitive flexibility as a trained capacity, not just an innate trait. There is a growing literature on neuroplasticity-informed education, cognitive-flexibility training, and adaptive learning systems, and it points in roughly the same direction. But the implications run deeper than pedagogy. They ask us to reconsider what education is for and where humanity is heading.

For most of history, education was for reproducing the culture. Increasingly, it has to be for equipping individuals to navigate cultures that will transform multiple times within their own lifetimes. That is a different task, and it requires a different brain — one that holds its assumptions with less grip, its identity with less rigidity, and its curiosity with more protection.

---

What Px is really about

This is why Potentialisation exists, and why I find myself saying, with increasing frequency, that Px is not just software. The assessment platform is a means to an end. The deeper work is helping people, as many people as possible, to see themselves clearly enough — their patterns, their assumptions, their felt-sense of rightness and wrongness — that they can hold those patterns more lightly. Not abandon them. Not be shamed for them. But hold them with enough awareness that they are no longer the unexamined substrate of everything else.

The narrative reports, synthesised from the data with AI assistance, exist for a very specific purpose: turning psychometric data into a mirror a person can actually use. The instruments — Big Five, HEXACO, Enneagram, Vital Signs (via 5deep), HSP, attachment, values — are not the point. The point is the reflective capacity they collectively enable: the ability to step back from your own operating system enough to see it running — and to see it as a whole, not one trait or one framework at a time.

If the neuroscience I have been describing is broadly right, then this kind of reflective capacity will not be a luxury we can live without in the future. It will be the core adaptive skill for a world whose pace of change has overtaken our inherited developmental machinery. Building tools that help people develop that capacity, at scale, in a world that needs it urgently, feels like work worth doing.

I am curious what psychologists and neuroscientists with deeper training than mine make of all this. I offer these thoughts not as settled conclusions but as a founder's working hypotheses — a pattern I have been watching for years, partly backed up by a literature I have not personally studied but which Claude helped me check against. If you are working in adult development, predictive processing, educational neuroscience, or the intersection of these — I would love to hear from you.

And for everyone else — if any of this resonates and you would like to see what your own patterns look like, there is now a way in: what-world-way, at whatworldway.com/with/jon. It is the kind of mirror this article has been arguing for — three short tests and a portrait at the end. The core is free, no sign-up to start.

---

Jon Twigge is the founder of Potentialisation (Px), a UK non-profit building a partly AI-supported psychometric platform for human development. The research citations in this article were surfaced with AI assistance; the framing and pattern are the author's own, refined over several years of working at the intersection of adult developmental theory, psychometrics and technology. Feedback from researchers in these fields is actively welcomed.