What this covers

Gordon Pennycook makes a case for a fundamental reframing of why people believe false claims. Rather than attributing misinformation and conspiracy acceptance to motivated reasoning—the defense of political identity and group belonging—he argues the primary culprit is unthinkingness), a failure to engage reflective thought. The episode moves through the psychological evidence: people who rely on intuition over reflection are worse at distinguishing true from false claims regardless of ideology; overconfidence predicts conspiracy belief independent of political identity; and conspiracy believers systematically overestimate how many others agree with them through a massive false consensus effect. Pennycook draws on Harry Frankfurt's philosophical distinction between bullshit and lies, and examines pseudo profound bullshit—vague jargon masquerading as insight—as a distinct category that exploits receptivity to obscurity rather than genuine meaning.

The conversation then turns to intervention. Pennycook describes experiments where large language model chatbots reduce conspiracy belief by roughly 20–25% through factual, evidence-based countering of specific claims—without flattery or persuasion tactics—and this effect persists one to two months later, which is exceptionally durable by psychological standards. A striking finding emerges: talking someone out of one conspiracy produces a modest transfer effect, reducing belief in unrelated conspiracies as well. Pennycook addresses a counterintuitive observation: many conspiracy believers possess detailed knowledge of their topics and actively seek engagement, suggesting the problem is not closed-mindedness but rather tumbling down the wrong informational rabbit hole. The practical upshot challenges prevailing pessimism: if unthinkingness, not identity defense, drives false belief, then supplying better, more engaging accurate information can work—the real problem is an information environment that fails to surface it.

Sharpest takeaway

Pennycook argues that susceptibility to bullshit, misinformation, and conspiracy theories stems primarily from 'unthinkingness'—a failure to engage effortful reflective reasoning—rather than from motivated reasoning, and that supplying good evidence (e.g. via patient AI chatbots) can therefore durably change minds.

  • People who rely on intuition over reflection are worse at distinguishing true from false claims regardless of ideology
  • AI chatbots reduce conspiracy belief mainly through facts and evidence, not flattery, with effects persisting months later
  • The misinformation problem is driven by the information environment, not by people's unwillingness to accept truth

The argument · threads11 threads · 42 claims
0.72

More reflective, analytically-disposed people distinguish truth from falsehood regardless of ideology.

5 pointscentrality 5/5
  • The mind processes information in two fundamentally different ways—an automatic intuitive mode (System 1) that is fast and often useful, and an effortful deliberative mode (System 2) that is optional and engaged afterward—and they are highly interconnected rather than being two separate brain systems.

    “there's two fundamental different ways in which our brain kinda works when we're processing information. We can respond intuitively and automatically... But also there are times in which our intuitions are wrong... we have to stop and engage in effortful deliberation sometimes.”

  • Once a person has deeply deliberated within a domain (e.g. earning a PhD), they immediately recognize complexity and uncertainty and can be reflective within it, but outside their domain they revert to less reflective intuitive judgment—showing reflective skill is partly domain-bound.

    “Once you have really dug down into a topic... you see the difficulties and uncertainties in the thing... But then if you read something that's outside of the domain, then you can kind of see where you're not being as reflective with that.”

  • People who rely more on intuitions and gut feelings rate pseudo-profound statements as more profound and are more likely to endorse pseudoscience and alternative medicine, while more analytic people who question their intuitions resist these.

    “people who tend to rely more on their intuitions, and like their gut feelings tend to think that these things are more profound. People who are more likely to kind of like, go with alternative medicines and believe in pseudoscience”

  • The disposition to think analytically does not correlate strongly with demographics or personality traits like openness or extroversion, but is highly related to whether one values questioning evidence and values accuracy; reflective people also tend to have higher cognitive capacity, and the combination produces the most pro-science individuals.

    “One thing that is very highly related to whether you have the disposition to think in an analytic way is your stance on the kind of importance of questioning evidence... it's more about whether you value evidence, whether you kind of value accuracy.”

  • People who are more analytic and question their intuitions are less likely to interpret sleep paralysis as a demon on the chest, and crucially they experience less distress in the days following such episodes because they use reasoning to contextualize the frightening event.

    “the people who are more analytic, who question their intuitions have less distress following sleep paralysis... because they're using their thinking to be basically kind of like contextualize the event to deal with the emotions”

0.68

Conspiracy belief is driven by unmotivated reasoning and failure to engage analytic thought.

4 pointscentrality 5/5
  • Contrary to the popular and influential view that people fall for misinformation and conspiracies because of motivated reasoning (preserving identity, feeling good, group belonging), the primary driver is 'unmotivated reasoning' or unthinkingness—lazy failure to expend cognitive effort—evidenced by the fact that more reflective people better distinguish true from false headlines regardless of whether the content fits their ideology.

    “the reason why people are susceptible to misinformation, is because they're not really thinking that much about what they are engaging with”

  • The practical value of the unthinkingness account is that it refutes the implication of motivation theories that truth doesn't matter; if identity and bias drove belief, one would have to undermine people's political identities to change minds (which no one can do), whereas the evidence shows supplying good information works, so the real battle is flooding the environment with accurate, engaging information.

    “what it combats is the idea that the truth doesn't matter... in order to really get people to be more rational, you have to kind of undermine the motivation somehow... And no one's really offering ways to do that.”

  • If society is losing the misinformation war, it is not because people are irrational or want falsehoods, but because of the information environment itself failing to get good information into the market—implying the solution is better, more engaging supply of accurate information rather than changing human nature.

    “if we're losing the misinformation war it's not because of people, it's because of the information environment itself.”

  • We do not have divine access to truth—people can only know information they are exposed to—so propaganda works: with enough control over exposure, those in power can strictly shape beliefs, meaning truth does not automatically prevail and power can overwhelm truth, though this can be countered by controlling exposure to good information too.

    “Propaganda does work... We don't have divine access to the truth. If people can only know information that they're exposed to, so you have enough control over that, then you can influence in a very strict way what people believe.”

0.58

AI chatbots can durably reduce conspiracy belief by supplying specific evidence to counter claims.

5 pointscentrality 4/5
  • Despite AI's known tendency to hallucinate, in the conspiracy-debunking study external fact-checkers found about 99 of 100 sampled AI claims true (with one arguably misleading), because the task—debunking conspiracies on topics heavily documented online—is exactly what the internet-trained model is good at, though it might be less reliable on dense scientific topics.

    “we had external fact checkers. There was like 100 claims and 99% of them were true or something like that... it was right in line with what it was good at.”

  • In experiments that selectively disabled features of the debunking AI, a blunt non-persuasive fact-delivery had essentially the same effect as the original, but an AI that tried to persuade without using any facts had no effect—demonstrating that facts and evidence, not flattery or politeness, are what change conspiracy beliefs.

    “we made the AI be less polite. Basically, all it was doing was just providing facts and evidence... That has more or less the same effect... If you get the AI to try to persuade, but you say you can't use any facts, it doesn't work.”

  • AI chatbots can reduce conspiracy belief: participants write their conspiracy in their own words, the AI counters the specific reasons with detailed evidence over an ~8-minute conversation, producing roughly a 20% drop in confidence and about 25% of believers no longer believing—an effect that does not decay one to two months later, which is exceptionally durable by psychology standards.

    “the conspiracy theorists had a 20% decrease in their confidence in the belief... After the conversation, which lasts about eight minutes, 25% of them don't believe it anymore.”

  • Chatbots succeed at debunking where humans fail because they have near-infinite patience and access to the specific obscure information needed to counter a conspiracy's 'gish gallop'—rapid jumps between obscure facts—whereas a human debunker would have had to go down the same rabbit hole to acquire the relevant counter-evidence.

    “They have access to all the information that you need in most cases... it's called the gish gallop, where they're giving new different... they're just like jumping around to all these obscure facts. And it's very difficult to win that debate.”

  • Talking someone out of one specific conspiracy via AI produces a small but real transfer effect, reducing belief in unrelated conspiracies as well—an unusual finding in psychology where transfer effects are rarely seen—suggesting the intervention breeds a bit of additional general skepticism even though it does not make people broadly more reflective.

    “you have some carryover effects. People believe the other conspiracy is a little bit less, now, it's a much smaller effect, but it's still there. And again, in psychology, you never see what we would refer to as like a transfer effect.”

0.55

Conspiracy believers overestimate how widely others share their beliefs due to echo chambers and silence.

4 pointscentrality 4/5
  • Many conspiracy believers are 'consumers' who passively absorb information drifting through YouTube videos, and by the same mechanism that took them down the rabbit hole—globally accepting information—they can be brought back out by giving them comprehensive, engaging alternative information, which is why debunking works for many even if not all.

    “if you just give them the alternative information, especially in a way that's very comprehensive or even engaging, then by the same mechanism that they went down the rabbit hole, they can come back out of it.”

  • Conspiracy believers overestimate consensus partly because social and online interactions rarely produce direct disagreement—people walk away, change the subject, or politely agree, and online posts get a few likes and no pushback—so silence and the echo of like-minded online communities are misread as widespread agreement.

    “Very infrequently will people say you're out to lunch... even if you post a conspiracy on Facebook, most of the time you're gonna get a few likes, and then no one's gonna say anything about it. And so that feels like agreement”

  • Conspiracy believers are not closed-off; they are often interested and want to talk their beliefs through, which means their misalignment is better understood as a failure of science communication—they fell down the wrong rabbit hole rather than a more interesting true one—rather than a wilful preference for falsehood.

    “It's a failure of science communication that they fell down that hole instead of a different one. They could be learning about the Big Bang or quarks or whatever.”

  • Contrary to the idea that conspiracy believers cherish being a unique minority who alone know the truth, they massively overestimate agreement: in one study only 8% endorsed the Sandy Hook false-flag conspiracy yet believers estimated 61% of people agreed with them—an exceptionally large false-consensus effect amplified by general overconfidence.

    “8% of people thought it was more likely to be true than false... In reality, what they said was 61% of people they thought agreed with them. So, they thought they were in the majority.”

0.51

Overconfidence is a foundational bias that prevents people from questioning whether they are wrong.

3 pointscentrality 4/5
  • It is a psychological truism that we detect bias in others far more easily than in ourselves—almost definitionally, since seeing one's own bias would dissolve it—and our own undetected bullshit is the most consequential to detect; overconfidence is 'the mother of all biases' because it stops us from questioning whether we are wrong.

    “it's way easier to see bias in others than bias in ourselves. It's almost definitionally. 'Cause like if you saw the bias then you wouldn't have the bias.”

  • A need for cognitive closure or certainty is a genuine individual difference that is generally not beneficial, because the world is not simple; a scientific stance requires both a drive to find answers and the ability to be comfortable not knowing, which ultimately comes down to intellectual humility versus overconfidence.

    “There is some research on what's need for cognitive certainty, cognitive closure... having the need for certainty is not usually generally good.”

  • Using a novel perceptual task with indiscernible fuzzy images where everyone is effectively guessing, some people consistently believe they are performing well when they are not; this 'general overconfidence' is brought to the task rather than caused by it, and overconfident people are more likely to believe conspiracy theories.

    “We give people like a fuzzy image that's hard to discern... is there a chimpanzee or a baseball player? And people don't know, their guesses are random... some people do think they're doing better, and they have a feeling that like, they at some level know that they're guessing, but they also really feel like they could do it”

0.38

People prone to conspiracy thinking hold contradictory beliefs simultaneously without epistemic scruple.

3 pointscentrality 3/5
  • People prone to believing one conspiracy tend to believe others, even directly contradictory ones, and to simultaneously hold religious and counter-religious occult beliefs, because they are not scrupulous—they accept what comes to mind without putting effort into distinguishing true from false.

    “you often see people who are likely to believe one conspiracy are also likely to believe a conspiracy that might even directly contradict it.”

  • Conspiracy believers often possess far more detailed knowledge of their topic than skeptics—9/11 truthers know more about the tensile strength of steel beams than most people—so the problem is not ignorance of details but assembling the big picture in a distorted way; they have fallen down the wrong rabbit hole.

    “they have fallen down the wrong rabbit hole... they are putting together pieces that are like... Shouldn't actually go together.”

  • Misinformation and conspiracy theories overlap but are not identical—not all misinformation involves a conspiracy and some conspiracies are true (Tuskegee, MKUltra)—yet the conspiracy theories typically discussed are the unverified speculative kind, and the underlying psychological process that matters is detachment from reality and reliance on bad evidence.

    “some conspiracies are true like the Tuskegee syphilis trials and MKUltra... It's just that most of the conspiracies that we talk about as like conspiracy theories are the unverified speculative sort.”

0.38

Bullshit and misinformation work because they exploit intuitive processing without regard for truth.

3 pointscentrality 3/5
  • Following Harry Frankfurt, bullshit is distinct from lying: a liar cares about truth enough to subvert it, whereas a bullshitter has no regard for truth at all and is merely trying to elicit a reaction (look smart, sell a product), which means a bullshit statement can even happen to be true.

    “if you're lying, that implies that you care about the truth to some extent. 'Cause you care enough about it to try to subvert the truth. Bullshitting is kind of almost the opposite, where if you're bullshitting, that means you don't really care about the truth.”

  • People who rate randomly generated profound-sounding sentences as profound are psychologically the same people who rate actual Deepak Chopra tweets as profound, showing the response is driven by receptivity to vague profundity rather than by actual content.

    “the people who believe that the random sentences are profound are the same people who think that the tweets are profound.”

  • Pseudo-profound bullshit is a category of bullshit where communication is geared not toward producing meaning but toward obscuring it—taking a trite observation and dressing it in jargon to make it sound important—which is the opposite of good science communication that distills complexity into understanding.

    “instead of communicating in a way that actually produces meaning for the other person, it obscures it... You take usually some sort of like basic trite observation, and then you make it sound like it's really important and then you can sell more books or whatever.”

0.38

Deliberative habits can be instilled early but are difficult to change in adults after decades.

3 pointscentrality 3/5
  • The jury is out on whether reflective thinking can be trained in adults; early-developmental interventions to instill habits of questioning intuitions may be feasible, but changing how someone has thought for decades is non-trivial, and the necessary long-term heavy intervention studies have not been done.

    “taking someone who's thought about the world in a particular sort of way for decades and changing the way they think, this is not a trivial thing... we haven't really done the sorts of long-term heavy intervention experiments that you would need to do to really test that.”

  • Some people not only fail to deliberate but explicitly devalue deliberation, identifying as non-rational and preferring feeling and emotion; such people are unlikely to become more reflective because they do not see value in it, and valuing deliberation correlates with valuing evidence and accuracy.

    “some people actually literally value deliberation more than others. And those people also tend to value evidence more and getting it right.”

  • Simply going through the exercise of writing out one's reasons for believing something reduces certainty in the belief, because it requires reflecting on those beliefs; this is a small effect, while providing counter-evidence afterward produces a much larger effect, so reflection and evidence work hand in hand.

    “simply going through the exercise of writing out the reasons for why you believe something is enough to kind of decrease how certain you are about it. Now, that's only a pretty small effect.”

0.30

Psychological research faces intractable constraints studying durable behavioral change in humans.

9 pointscentrality 1/5
  • A general methodological problem in psychology is that researchers typically only have participants for 15-30 minutes, so causal-mechanism interventions must be done in minutes; this short time horizon, combined with the inability to house humans like lab animals, makes studying durable change on human nature an almost intractable problem.

    “for psychologists, we have people for 15 minutes, 30 minutes... the sorts of interventions you can do if you wanna really test causal mechanisms are things that you can do in a matter of minutes. And there's not... That takes away a lot of bullets outta the chamber”

  • The Dunning-Kruger problem is that measuring overconfidence with a skill test confounds confidence with ability (someone bad at math appears overconfident on a math test but not a humor test); a task designed so there is no relationship between perceived and actual performance isolates pure confidence from performance, circumventing this confound.

    “if I wanna measure how overconfident somebody is, if I give them a math test and they happen to be bad at math... they'll appear overconfident. But if I gave them a test of like how good they're identifying humor, they might be good at that and they won't appear overconfident”

  • The Ig Nobel Prize honors research that first makes you laugh and then makes you think, awarded both to work being gently mocked and to legitimate amusing-yet-important research, without disclosing which is which.

    “it's for research that makes you laugh and then makes you think.”

  • Intuition is not necessarily innate instinct but can be developed through learning, as exemplified by chess grandmasters who, through rote study of tens of thousands of board positions, come to recognize configurations as instantly as faces—an intuitive capacity acquired through prior analytic effort.

    “chess grand masters. They can immediately identify just through rote memorization and playing the game... That's purely intuitive, but of course, it's not like they were born with that capacity. They have to learn it by thinking analytically.”

  • When people share content online, accuracy is often not what comes to mind; simply prompting people to think about accuracy beforehand—through questions or small ads—makes them better at distinguishing true from false content and reduces reflexive sharing of falsehoods, a small but real effect that makes that specific choice more rational without making people more reflective in general.

    “we've done these experiments where we just remind people about accuracy... then they're better at distinguishing between the true and false stuff when they're deciding what to share.”

  • It is evolutionarily sensible for the brain to rely mostly on intuition rather than constant deliberation, because deliberation consumes cognitive resources and energy; the problem is that people vary in when they deploy effortful thinking, sometimes failing to use it when needed and sometimes overthinking when they shouldn't.

    “if you have a process that requires resources, cognitive resources, energy, then it wouldn't be that adapted to be doing that all the time. The problem is that people vary in how much they do it.”

  • AI chatbots tend to reflect majority/accurate views because they are trained with mechanisms identifying good versus bad sources; even when an owner tries to bias the model (e.g. Musk with Grok), it requires laborious hard-coding because reality is a constraint—training on data connected to reality pushes outputs toward accuracy.

    “even with Grok... Musk has put his thumb on the dial to change, but they have to kind of hard code that in. It's really a lot of work to try to get it to espouse the views that Musk wants... if you train things with some connection to reality, then you're gonna face that constraint.”

  • The shift from a few shared news sources to fragmented, choice-driven media means people now live in different realities with no going back, so scientists and communicators can no longer assume unique access to people's attention and must put more work into reaching audiences—an adjustment Pennycook thinks scientists have been too slow to make.

    “now you have choice and you have all these different things. Now people are living in different realities. And there's no going back to that... you cannot assume that you have unique access to people's attention.”

  • Susceptibility to bullshit is both dispositional and situational: everyone relies on intuition and has blind spots, but people vary in how much they question their intuitions, and these tendencies can be context-dependent.

    “everybody relies on their intuitions and probably everybody could spend more time questioning their intuitions and thinking and reflecting... Everybody needs to question their intuitions. And there are cases in which we have blind spots.”

0.20

Modern partisan asymmetry in misinformation reflects decades of deliberate anti-science political effort.

2 pointscentrality 2/5
  • The modern partisan asymmetry in misinformation is not natural but the product of a 'war on science' that began in earnest around Reagan, driven significantly by climate change denial, with historical party realignment (e.g. anti-evolution Democrats becoming Republicans) part of the story.

    “my historical take on this is that it's Reagan's fault. And the kind of war on science really started earnest around that time. A lot of it has to do with climate change.”

  • In the US there is an asymmetry: more misinformation circulates on the political right than the left, a finding robust across dozens of studies using fact-checker reports, journalist reports, and politically balanced samples, so right-leaning people believing more false headlines partly reflects exposure (the supply of misinformation) rather than only susceptibility.

    “There's way more misinformation on the right than on the left. And this is not just... That sounds like a political statement. There's been like dozens of studies... even based on that, there's more falsity on the right.”

0.17

Some people explicitly reject deliberation and value feeling over evidence, resisting reflective change.

1 pointcentrality 2/5
  • Being a sports fan need not be inconsistent with valuing analytic engagement: it can be a rational choice to allow oneself to be irrational in a circumscribed domain because it is more fun, illustrating that even reflective people deliberately maintain intuitive blind spots.

    “for me, it's a rational choice to allow myself to be irrational in this domain. It's more fun if you just watch it and you hope for the best.”