How Minds Actually Change

The Argument You Won

Lesson 5.2 left you with a question about every signal you send: could you defend it as true? Hold that, because this lesson assumes you are honest and asks a harder question. What do you do when you are right, your evidence is good, your signals are all true, and the person still does not move?

Every leader has the memory. You had the numbers. You had the sequence of events. You made the case cleanly, and you watched the other person's face close somewhere in the second minute. Maybe he conceded a point to end the conversation. Maybe he went quiet and you told yourself that was progress. Then a week later he was exactly where he started, or further away, and cooler toward you than before. You won the argument and lost the person, and you have probably repeated the experience enough times to have quietly concluded that people are irrational.

That conclusion is comfortable and it is wrong, and it is worth seeing why it is attractive: if people are simply irrational, you are excused from getting better at this. Here is the more accurate account. Beliefs are not stored like files. Some of them are load-bearing — they hold up a person's sense of who he is, which group he belongs to, whether his life so far has been defensible, and whether the people he loves were right. When you attack one of those, you are not correcting a file. You are pulling on a wall, and every instinct he has will make him hold the wall.

Daniel Kahneman's framework in Thinking, Fast and Slow gives part of the mechanism. Fast, automatic processing runs constantly and produces immediate reactions; slow, effortful reasoning is expensive and gets recruited mostly to defend what the fast system already delivered. In practice, when a threatening claim arrives, the reaction lands first and the reasoning shows up afterward as counsel for the defense. You have watched a smart person become brilliant at exactly the moment he was most wrong. That is not a failure of intelligence. That is intelligence in the service of the wall.

So the strategic error most leaders make is not choosing bad arguments. It is choosing argument at all when the obstacle is not informational. If a person lacks a fact, give him the fact — that works and this lesson is not telling you otherwise. But if his objection is that accepting your fact would cost him standing, or a self-image, or a community, then more facts are simply more attacks on the wall, and every one of them makes him better at defending it.

This lesson gives you what actually works, and it starts by cleaning up the evidence, because this particular research area has produced two famous embarrassments and both of them are teaching moments you cannot afford to miss.

Two Findings You Have Probably Been Quoting Wrong

Start with the one you have almost certainly heard. In 2010, Brendan Nyhan and Jason Reifler published a study in Political Behavior reporting a backfire effect: for ideologically committed subjects, corrections could actually strengthen the misperception. The finding traveled further and faster than almost any political science result of the decade, and I want you to notice why. It flattered a cynicism people already held. It gave everyone permission to conclude that the other side is unreachable, which is an extremely comfortable conclusion because it excuses you from the work.

Then it was tested at scale. Thomas Wood and Ethan Porter ran five experiments with more than ten thousand participants and published 'The Elusive Backfire Effect' in Political Behavior in 2019. What they found was that citizens generally moved toward accuracy when corrected, including on issues tied to their partisan commitments. The current state of play is that backfire is real but rare and hard to reproduce, and that correcting a factual error usually helps at least somewhat. If you have been telling yourself that facts never change anybody's mind, the evidence does not support you, and the belief has been doing something convenient for you.

The second embarrassment is worse and more instructive. In 2014, Science published a striking field experiment reporting that brief conversations with gay canvassers durably changed voters' views on same-sex marriage. In 2015 the paper was retracted after David Broockman, Joshua Kalla, and Peter Aronow, attempting to build on it, found irregularities indicating the data had been fabricated. Sit with what that means. The finding was beautiful, it was published in one of the two most prestigious journals on earth, it was covered everywhere, and it was not real.

Now the part that redeems the story. Those same researchers went and did the work properly. In 2016, Broockman and Kalla published a genuine randomized field experiment in Science: fifty-six canvassers held roughly ten-minute, non-judgmental, perspective-taking conversations at the doors of 501 voters in South Florida. The intervention reduced measured prejudice toward transgender people, and the effect was still present three months later, surviving exposure to counterarguments. Later work by Kalla and Broockman replicated the general approach across additional field experiments. This academy's convictions differ from the canvassers' aims on the underlying question, and I am including the finding anyway, at full strength, because that is what intellectual honesty requires and because the mechanism is what matters here: it was not a debate, not a rebuttal, and not a fact sheet. It was a person asking questions, listening to the answer, and sharing an experience.

Third, the framework that explains the pattern. Dan Kahan's research program on identity-protective cognition argues that on certain culturally charged questions, people process information in whatever way protects their standing in the group whose respect they depend on. His work with colleagues found that greater science literacy and numeracy were associated with more polarization on climate risk, not less, and a later study found that subjects who solved a data-interpretation problem competently when it concerned skin cream performed worse, and split along political lines, when the identical numbers concerned gun control. The unsettling implication is that reasoning ability can function as ammunition rather than as a corrective.

And now the honest caveat, because Kahan's interpretation is genuinely contested. Druckman and McGrath argued in 2019 that the same patterns are also consistent with people rationally weighting source credibility differently based on prior beliefs — a Bayesian story rather than a motivated one. Tappin, Pennycook, and Rand have made related arguments that the standard experimental designs cannot cleanly separate motivated reasoning from ordinary updating on different priors, and related work finds that analytic thinking tends to track better truth discernment regardless of partisanship. The behavior is well documented. The mechanism is a live dispute. A leader can act on the first without pretending the second is settled, and if you hear someone confidently explain that the other side is neurologically incapable of reason, you are listening to a person who has read a headline.

The Nathan Sequence

Now the model, and it is three thousand years old. Second Samuel 12. David has committed adultery with Bathsheba and arranged the death of her husband Uriah, and he has been living with it. This is a king with absolute power, a prophet's word already in his life, and every incentive to defend himself. He is precisely the man no argument will reach.

Nathan is sent, and here is what he does not do. He does not open with the accusation. He does not cite the law. He does not bring evidence, name witnesses, or warn of consequences — all of which were available to him and all of which would have failed, because the moment David hears the charge, the king in him takes the field and the man in him goes behind the wall.

Instead: 'There were two men in one city, the one rich and the other poor.' The rich man had many flocks. The poor man had one little ewe lamb that he had bought and raised, which ate from his own bread and drank from his own cup and lay in his bosom and was like a daughter to him. A traveler came to the rich man, and the rich man spared his own flock and took the poor man's lamb and prepared it for his guest.

David's reaction is immediate and it is not defensive at all: 'Then David's anger burned greatly against the man, and he said to Nathan, As the LORD lives, surely the man who has done this deserves to die.' The Hebrew behind 'deserves to die' is ben-mavet — literally a son of death, the strongest available idiom for a man whose life is forfeit. That matters. David does not issue a measured judgment. He convicts in the harshest possible terms, and he does it out loud, in front of a witness, as king. Then Nathan says four words: 'You are the man.'

Take apart why it worked, because the mechanism is the lesson. Move one: approach on unguarded ground. The story is about someone else, in another city, so no defense ever activates — David is not a defendant, he is a judge, which is the one role his identity is fully comfortable occupying. Move two: let him render the verdict. The conclusion is generated by David, not supplied by Nathan, and a self-generated conclusion is held far more tightly than a delivered one and cannot be attacked later without self-contradiction. He would have to reverse his own sentence to escape it. Move three: close the gap. Four words, once, no elaboration, no piling on. Nathan does not enumerate the offenses first; he lets the identity link land in a single sentence and then keeps speaking the Lord's words rather than his own. Move four, and it is the one most leaders skip: leave a way to stand up. When David says, 'I have sinned against the LORD,' Nathan answers immediately, 'The LORD also has taken away your sin; you shall not die.' The consequences are severe and Nathan states them without softening. But there is a path forward that does not require David to cease being a person. This is not sentimentality. It is structural. A man with no way to remain a person will defend the sin in order to keep the self, and you will have made repentance more expensive than denial.

Call it the Nathan Sequence, and notice how exactly it maps to what the canvassing research found: no rebuttal, no fact sheet, a story, a question, and space for the person to arrive somewhere himself. Notice also what it is not. It is not a trick, and it survives Lesson 5.1's Daylight Test completely. Nathan discloses in the middle of the method — 'you are the man' is the disclosure — and the persuasion does not collapse when he does. It completes. That is the signature of influence rather than manipulation: full disclosure does not kill it, it finishes it.

Before you go on, apply it. Think of the person you most want to change and have most conclusively failed to change by argument. What is the unguarded ground for him — the case he could judge without being the defendant? Do not answer in general. Name the story.

Paul on Their Ground, King to His Friends

Acts 17 gives the same principle in a hostile intellectual setting. Paul is in Athens, provoked by a city full of idols, and he is brought to the Areopagus to explain himself. His opening line, NASB 1995: 'Men of Athens, I observe that you are very religious in all respects.'

The argument here rests on a Greek word, so take it directly: deisidaimonesterous. Its semantic range genuinely runs from reverent or religious to superstitious, which is why the King James rendered it 'too superstitious' while the NASB 1995 has 'very religious.' Lexicographers acknowledge the ambiguity rather than resolving it, and the majority reading among commentators is that Paul chose a deliberately double-edged word — a courteous opening that his hearers could take as recognition, without Paul having conceded that their worship was true. That is not evasion. That is finding the most honest available common ground and standing on it.

Then watch the sequence. He walks their city and cites their own altar, 'TO AN UNKNOWN GOD,' making his subject something they had already admitted they lacked. He quotes their poets — 'for we also are His children' — using their authorities rather than his own, in a room where his Scriptures carried no weight at all. He builds from creation and providence toward repentance and resurrection. And at no point does he pretend to agree; the whole speech is a frontal challenge to Athenian religion delivered in Athenian terms.

Then the result, which is the part leaders need most. 'Now when they heard of the resurrection of the dead, some began to sneer, but others said, We shall hear you again concerning this.' Some sneered. Some deferred. And a few joined him and believed, among them Dionysius the Areopagite and a woman named Damaris. Paul does not chase the sneerers, does not stay to litigate, and does not measure the day by them. He takes the maybes and the fews and moves on. If you cannot accept an outcome shaped like that, you will keep aiming at the wrong people for the rest of your leadership.

John Stuart Mill supplies the discipline underneath all of this in the second chapter of On Liberty: he who knows only his own side of the case knows little of that. The application here is not epistemic humility as a pose. It is that you cannot find a person's unguarded ground unless you have taken the trouble to understand what he is actually protecting — and most people who lose arguments repeatedly have never once accurately restated their opponent. Which is where the difference between winning and persuading finally becomes concrete: winning is producing a state in which the other person cannot answer you; persuading is producing a state in which he no longer wants to. They feel similar in the room and they diverge completely by the following week.

History's best-documented instance of choosing the right audience is Martin Luther King's Letter from Birmingham Jail, written in April 1963. Read who it is addressed to. It responds to a public statement by eight local white clergymen who had called the demonstrations unwise and untimely — men who were not segregationists and thought of themselves as allies of a slower approach. King wrote to sympathizers, not to enemies. He conceded their good intentions, argued in the vocabulary of their own tradition, and reserved his sharpest passage for the moderate who prefers a negative peace to a positive justice — naming that figure, not the committed opponent, as the greater obstacle. The letter is a persuasion document aimed with precision at the movable people nearest the center of gravity, and it is the reason it still works sixty years later.

The Persuadable Middle

Now put it in the campaign and in the office, where the audience question decides everything.

Here is the discipline in one sentence: you are almost never persuading your opponent, and treating him as your target is how you lose the people who were actually available to you. On a debate stage the man across from you will not change his mind, cannot afford to change his mind in public, and is not there to. The audience is the persuadable middle — the people who have not decided, who dislike both of you a little, and who are watching how you handle him more closely than they are following the substance. Every minute you spend trying to defeat him is a minute you are not spending on them, and your visible desperation to win the exchange is itself information they are recording.

The research here should also calibrate your expectations, because it is bracing. Kalla and Broockman's 2018 meta-analysis of forty-nine field experiments found that the best-estimated average effect of campaign contact on candidate choice in general elections is approximately zero. Not small. Zero, on average, in that setting. Persuasion is more achievable in primaries and on ballot measures, and persuasive effects from early contact tend to decay well before election day. Hold that alongside the deep-canvassing result and the picture becomes clear rather than contradictory: brief, argumentative, partisan contact about a candidate moves almost nobody, while long, non-judgmental, personal conversation about an issue can move people durably. The two findings are not in tension. They are telling you which activity is which.

Which brings you to Greg Koukl's goal, and it is exactly right for this office. He calls it putting a stone in the other person's shoe: you are not trying to win outright, you are trying to leave one thought the person cannot stop turning over. For a candidate that is a strategic correction, not a modesty pose. You will not convert a hostile moderator on live television, and the attempt makes you look like a man who needs something from him. One clear, true, memorable thing that survives the news cycle beats winning six exchanges nobody remembers by Thursday. And the stone must be a specific idea rather than a slogan, because a slogan is discarded on contact and an idea keeps working after you leave the studio.

Two practical instruments before we close. First, ask for mechanism instead of justification. Rozenblit and Keil documented the illusion of explanatory depth — people believe they understand how ordinary things work far better than they do — and Fernbach and colleagues found in 2013 that asking people to explain, step by step, how a policy would actually produce its intended results reduced the extremity of their position, while asking them to justify why they held it did not. That is a usable question in a hearing, a meeting, or a doorway: not 'why do you support that,' which invites the wall, but 'walk me through how it would work,' which invites the person to look. It is Proverbs 6:6 in a modern room — go and look at the ant. Second, remember what Module 4 established: accurately restating a person's position before you answer is the highest-yield opening move available to you, because almost nobody has ever had it done to them by an opponent.

Then the constraint that keeps all of this honest. Some minds will not change, and some should not change, and you are not owed a result. Nathan's method worked on a king who had a conscience left; the same story told to a man determined to keep his sin would have produced a shrug and an execution order. Paul kept his integrity on Mars Hill and most of the Areopagus sneered. So do the work, choose the right audience, tell the truth in the order that lets it be heard — common ground, then context and evidence, then the point — and leave the outcome alone. You are responsible for the method and the message. You are not responsible for the verdict. Lesson 5.4 takes up what makes any of this land in the first place, and it is not technique: it is whether the person listening believes you.