Bulverism
Explaining why an opponent holds a view is offered in place of showing that the view is wrong, and people do assume opponents have causes where they have reasons.
Explaining why an opponent holds a view is offered in place of showing that the view is wrong, and people do assume opponents have causes where they have reasons.
Lewis's charge is about the order of operations. A claim is answered by an account of how its holder came to make it, before anything has established that the claim is false. He set the rule against it in one line: 'You must show that a man is wrong before you start explaining why he is wrong.' His arithmetic example carries the argument.
If a man's bank balance is in dispute, the only way to learn whether his figure is wishful thinking is to work the sum. A psychological account of the error is admissible after the sum comes out against him, not instead of it. Lewis then draws the distinction the name rests on. All beliefs have causes, and a reason is a special kind of cause, and 'Bulverism tries to show that the other man has causes and not reasons and that we have reasons and not causes.' That asymmetry is the part psychology has measured.
Replicated No experiment on Bulverism by that name was located. The premise underneath it, that opponents have causes where we have reasons, holds in nine American samples and again in three Swedish ones by another group.
Stanley and colleagues report the reasons gap on five issues and two independent participant pools, with effects from d = 0.47 to d = 0.93 and none in the wrong direction. Eight of their nine samples were preregistered and the largest analysed 798 people. That is repetition within one paper by one team. The independent check is Bäck and Lindholm, whose Swedish participants rated opponents as less rational and more externally driven, partial eta squared .14 and .11 in a sample of 311, and the rationality gap turned up again in samples of 135 and 109, F(1, 124) = 35.97 and, among the two positions taken separately in the third study, F(1, 103) = 58.70 and F(1, 103) = 3.91.
The version of the asymmetry that concerns motives for conflict rather than reasons for beliefs has an explicit replication of its own. Waytz, Young and Ginges reported across 661 American partisans, 995 Israelis and 1,266 Palestinians that people trace their own group's aggression to ingroup love and the other group's to outgroup hate. Hughes and Stewart, who describe themselves as the first to replicate it, recruited 666 American partisans and found the interaction at F(1, 633) = 219.997, partial eta squared .258, then found it again three months apart in a second sample, F(1, 639) = 227.896 at the first wave and F(1, 498) = 239.033 at the second. What no located study tests is Lewis's own claim, which is normative: that the explanation is worthless until the error is established.
The figure nobody checksillustration
A treasurer's account of the shortfall is met with the observation that he has always wanted the festival cancelled. Whether the shortfall is real is settled by working through the accounts, and the motive becomes admissible only once the sum has come out against him.
The objection answered by a diagnosisillustration
A colleague objects to a reorganisation and is told that people of his age always resist change. The reply explains the objection without touching it, so the objection stands exactly where it was.
The letter about the bypassillustration
A campaigner's traffic counts are answered in the local paper by an account of who owns land near the proposed route. If the counts are wrong, showing that comes first. The landholding explains an error nobody has established.
Cases marked as illustrations describe nobody in particular. They are there to make the pattern recognisable, not to report that a particular thing happened.
The essay was read in two web transcriptions rather than in any printing Lewis authorised. Both carry the same wording for the passage everyone quotes. Ezekiel Bulver's destiny was settled at five, when he heard his mother tell his father, who was maintaining that two sides of a triangle are together greater than the third, 'Oh, you say that because you are a man.' From that came 'the great truth that refutation is no necessary part of argument.
Assume that your opponent is wrong, and explain his error, and the world will be at your feet.' Lewis's own target in 1941 was Freudian and Marxist debunking, thoughts 'psychologically tainted at the source' or 'ideologically tainted' at it. His objection is that the move costs the user his own footing: 'Until Bulverism is crushed, reason can play no effective part in human affairs.'
Nothing measured carries the name. What has been measured is the premise: that the other side holds its position for causes and this side holds its own for reasons.
Stanley, Whitehead, Sinnott-Armstrong and Seli ran nine samples on Amazon Mechanical Turk across five issues, eight of them preregistered. In Study 1, 239 participants rated whether people on each side of standardised testing, police body cameras and animal testing had good reasons. Own side beat other side every time: Mdiff = 0.54, t(238) = 8.06, d = 0.52 for standardised testing, Mdiff = 1.22, t(238) = 12.65, d = 0.82 for body cameras, and Mdiff = 0.63, t(238) = 7.22, d = 0.47 for animal testing, all p < .001.
Study 4, on 244 participants and two polarised issues, gave Mdiff = 1.49, t(243) = 14.55, d = 0.93 for universal healthcare and Mdiff = 0.93, t(243) = 10.33, d = 0.66 for concealed carry. The belief tracked contempt. Participants who thought their opponents lacked good reasons were likelier to call them unintelligent, irrational and ignorant, r = -.29, -.40 and -.46 by issue, and to call them unethical and immoral, r = -.21, -.32 and -.40.
The same team then handed people the reasons. Participants shown compelling arguments for the other side rated their opponents as less lacking in intellect than participants shown arguments for their own side or none at all: d = 0.56 and 0.54 on body cameras in Study 2b, N = 594, and d = 0.43 and 0.62 in Study 3a, N = 767.
Being shown one's own side's reasons did nothing, d = 0.01 against the no-reasons condition in Studies 2a and 2b. On the two polarised issues the effect held for intellect and thinned for character. Study 5a, N = 791, gave F(3, 787) = 3.89, p = .009 for intellect and F(3, 787) = 1.23, p = .30 for moral character.
Bäck and Lindholm measured the same asymmetry in Sweden with the causal language Lewis used. Their participants rated the rationality of holders of each position and how far external factors drove them. In Study 1, on 311 high school students, own side was rated more rational, F(1, 294) = 46.56, and less externally driven, F(1, 298) = 37.38, both p < .001, with partial eta squared of .14 and .11.
People arguing against the status quo showed the bias more strongly than people defending it, F(1, 294) = 20.25 and F(1, 298) = 22.57. Study 2, N = 135, repeated both effects, F(1, 124) = 35.97 for rationality and F(1, 121) = 7.18, p = .008 for externality, and found nothing from an experimental manipulation of whether the participant's side was in the majority, Fs < 0.74.
Reeder, Pryor, Wohl and Griswell put the question in terms of motives. Their four studies asked American and Canadian respondents about the motives behind the Iraq war and then about abortion and gay marriage. Respondents who disagreed with a position attributed more selfish motivation to those holding it, and the effect was concentrated among those most involved in the issue. This entry quotes that paper from its Europe PMC abstract, which carries no numbers.
4 primary research2 replication and reanalysis4 commentary and critique2 background and reference
The name arrives in a longer essay than the name-giving passage suggests, and its printing history is untidy. Arend Smilde's annotated survey of Lewis's shorter writings records the piece as first published under the standing rubric 'Notes on the Way' in Time and Tide on 29 March 1941, then revised as 'Bulverism', read to the Oxford Socratic Club on 7 February 1944, and printed with notes added by the Club's secretary in The Socratic Digest, vol. II, June 1944.
Joel Heck's chronology of Lewis's 1941 publications gives the same date and the same rubric, listing it as "'Bulverism'" (as 'Notes on the Way') in Time and Tide, Vol. XXII, on March 29, 1941. Smilde's own bibliography carries the dating conflict inside one document: his chronological list of first publications dates the essay 4103, March 1941, and his alphabetical list dates it 4406, June 1944, the month of the Socratic Digest.
The collected texts differ too. Smilde notes that 'Bulverism' as printed in First and Second Things (1985) and Compelling Reason (1996) lacks the closing passage of Socratic Club minutes that the 1971 Undeceptions text added. Neither the 1941 magazine nor the 1944 Digest was read for this entry.
The claim, and the replication this entry reads its figures from.
The name has no literature. A Crossref search for 'bulverism' returns no records, a Europe PMC search for the same string returns none, and a PubMed search returns none. Searches also ran over OpenAlex, which refused the request, and over Semantic Scholar, which rate-limited it, so neither was checked. The essay itself was found through Smilde's bibliography and read in web transcriptions of the Eerdmans God in the Dock text, which Smilde reports carries at least one uncorrected typo in this essay.
The measured work sits beside the idea rather than on it. Stanley's outcome is how people rate opponents on six adjectives, not whether anyone substituted an explanation for a refutation in an argument. His own limitations section records that the composites for lack of intellect and lack of moral character were never significantly above the scale midpoint in any study, so most participants did not endorse the derogation at all. All nine samples came from one crowdsourcing platform.
Bäck and Lindholm's samples are high school and university students, 311, 135 and 109. Their asymmetry is uneven: in Study 1 defenders of the status quo did not significantly differ in the externality they ascribed to the two sides, F(1, 298) = 1.79, p = .18, and in Study 2 the same comparison gave F(1, 121) = 0.13, p = .72. The bias in those samples belongs mainly to the side arguing for change.
Hughes and Stewart's discussion reports their two effect sizes as .258 and .261, while their results sections give .258, .263 and .324. The figures above are taken from the results sections.
Two papers relevant to the attribution half were not obtained. Reeder and colleagues on negative motives, and Kennedy and Pronin's bias-perception conflict spiral, are both behind paywalls, and both are described here from their Europe PMC abstracts, which print no effect sizes. Chiu, Hong and Dweck's lay dispositionism, which the brief for this entry named as a lead, is about who makes strong trait inferences rather than about explaining away disagreement, and it is quoted here from its PubMed abstract. A Crossref bibliographic search and a Europe PMC search for the exact phrase 'lay dispositionism', which returns nine records, located no direct replication of it.
Each of these is named in this entry's own prose as something it gets taken for. Where the two came out differently, that is worth knowing before you quote either.
The commonest error is treating Bulverism as a synonym for the genetic fallacy, and this site's own genetic fallacy entry lists it as an alias. Lewis's target is narrower. He does not say that the origin of a belief is irrelevant. He says the psychological account is admissible after the error has been shown and worthless before it, which is a rule about sequence rather than about relevance.
The essay is misdated in retellings. It is regularly cited as coming from 'the 1941 collection of essays God in the Dock', as The European Conservative's reprint of the passage has it. God in the Dock was published in 1970, nine years after Lewis's death, and the 1941 publication was a weekly magazine column that did not carry the word Bulverism as its title.
The research is easy to over-read in the Bulverist's favour. What Stanley's participants did was assume that opponents lack good reasons. Nothing in that shows the assumption is wrong in any particular case, and Lewis's complaint is not that opponents always have reasons but that finding out requires doing the sum.
The cure is cheaper than the diagnosis suggests. Showing people three or four vetted arguments for the other side moved their ratings of opponents' intelligence by about half a standard deviation, while showing them arguments for their own side moved nothing, d = 0.01 in Studies 2a and 2b. Lewis proposed no remedy beyond refusing the move.
Every claim on this page was held against these sources on 2026-08-17. Nothing here is written from memory. Written and checked by Krishna Chagti; how entries are written and corrected. Found a mistake? Report it, with the sentence and a source.