.png)
Reasoning or Rationalising: What the Research Says About Dissonance and the Decisions You Carry.
Updated: Sep 28
It is twenty past six and you are drafting the email. This morning, in the room, you said you thought the restructure was being rushed, that the timing would land badly with the team who had barely finished the last one, and that nobody had asked finance what it would do to the quarter. You said it carefully, and you were thanked for it, and the decision went the other way. Now the announcement has your name in the sender line, and somewhere around the third paragraph you notice that you have started to believe it. The timing is bold, not rushed. The team are resilient. Finance will cope. By the time you press send you have found four good reasons for a decision you argued against at nine o'clock.
Nothing about that is dishonest. It is one of the most studied things the human mind does, and it happens to people in support roles more often than to almost anyone, because the job keeps asking you to carry, defend and deliver decisions that were not yours. Last March I wrote about four places this shows up in leadership support and asked whether you were reasoning or rationalising. This is the research underneath that question: what the classic dissonance studies actually found, which of them held up when they were run again, how the mind builds a case after the verdict, and the handful of tools that reliably help you tell the two apart in your own head.
What you will get out of this
What the famous dissonance experiments showed, and what happened when 39 laboratories in 19 countries tried to repeat one of them
Why preaching something you do not practise can be the most useful kind of discomfort you ever feel
What arguing for a position does to your own belief in it, measured at international debating competitions
The study in which most people defended a moral opinion that was the opposite of the one they had written down minutes earlier
How the mind quietly improves a decision once it has been made, and how long that lasts
The difference between asking "can I believe this?" and "must I believe this?", and why it decides most arguments before they start
The experiment in which more than half of people rejected their own argument when it was shown to them as someone else's
Five tools that hold up in the evidence, and one popular version that backfires
The reasoning check: a tool that looks at the decision you are carrying and tells you which way your head is leaning
The Reasoning Pages: three pages for catching a rationalisation before it becomes your opinion
The knot, and what the famous studies really found
The idea is older than most of the people reading this. In 1957 a psychologist proposed that holding two thoughts that do not fit together, or doing something that does not fit with what you believe, produces an uncomfortable state, and that people are motivated to make the discomfort go away. He called it cognitive dissonance, and the three ways out he and his successors described are the ones you would recognise from any office: change the belief, add a new belief that bridges the gap, or decide the gap does not matter very much.
Three experiments made the theory famous, and all three are worth knowing properly, because you will hear them quoted.
In the first, from 1959, people did an hour of mind-numbing tasks and were then paid to tell the next participant that it had been enjoyable. Some were paid one dollar and some twenty. Afterwards, the people paid a single dollar rated the dull task as more enjoyable than the people paid twenty. The explanation was that twenty dollars was a good enough reason to have said it; one dollar was not, so the mind supplied a better reason, which was that perhaps it had not been so bad after all.
In the second, also from 1959, people who went through an embarrassing initiation to join a discussion group liked the group more, even though the discussion was deliberately dull, than people whose initiation was mild. A 1966 study replicated the effect with a different kind of unpleasantness. The more it costs you to get in, the better the room looks once you are there.
In the third, from 1956, people rated household appliances, chose one to take home, and rated them again. The chosen appliance went up in their estimation and the rejected one went down. Researchers call this the spreading of alternatives, and it became the basis of hundreds of experiments.
That is the story as it is usually told. The modern chapter is more interesting, and it matters for how much weight you put on any of this.
In 2010 two researchers showed that the appliance design had a statistical flaw: if people's ratings are an imperfect measure of what they prefer, and their choices are guided by what they prefer, the method will show spreading even when nobody's preferences have changed at all. For a while it looked as if half a century of choice research might be measuring an artefact. Then the corrected designs arrived. In one, people chose between holiday destinations without being able to see which was which, and still came to prefer the one they had picked, while a computer choosing for them changed nothing. The same team found the shift still there two and a half to three years later. A 2021 meta-analysis of 43 studies and 2,191 people, using only designs that avoid the flaw, found a moderate effect with no sign of publication bias. Choosing does change what you prefer. It simply changes it by less than the old studies suggested.
The one-dollar paradigm had a harder time. In 2024, 39 laboratories in 19 countries and 14 languages ran a carefully planned replication with 4,898 people. Writing an essay against your own view under high choice did not shift attitudes any more than writing it under low choice, which was the core prediction. Writing it at all did shift attitudes compared with writing something neutral. A reanalysis of the same data found that when people's own sense of how free they had been was measured, rather than the experimenters' manipulation of it, those who felt free did move their attitudes towards the essay. The replication team's own conclusion was that the classic paradigm is less reliable than the textbooks assumed and that the field needs more rigorous methods.
So here is my honest reading. The feeling is real; experiments have measured it as discomfort that eases once people change their attitude, and have watched it in the brain. The tendency to improve a choice after making it is real and lasting. The famous twenty-dollar effect is shakier than its fame suggests, and the part that seems to matter most is whether you felt you chose. And a 2011 synthesis across five research paradigms found that the thing most strongly predicting how much people shifted was guilt, which fits what most of us would have guessed about our own evenings.
The most useful modern idea, though, is about what the feeling is for. The action-based model argues that dissonance is aroused specifically when two thoughts pull you towards different actions, and that the discomfort is a signal that something needs resolving before you can act well. Another researcher argues that inconsistency works as a cue that there is an error somewhere in what you believe. Both readings turn the knot into information. The problem is not the discomfort; it is how quickly and how cheaply we get rid of it.
The knot in your stomach is not the problem. The problem is how fast you can talk yourself out of it.
Try this. Next time you feel the knot, name the two thoughts that are pulling against each other before you do anything else, in one line each. "I think this is rushed. I am the one sending it." Writing the pair down does not resolve anything, and that is the point: it holds the gap open long enough for you to decide which side should move, rather than letting the more comfortable one win by default.
The research behind this section
Festinger and Carlsmith (1959) paid participants either one dollar or twenty dollars to tell a waiting participant that a boring task had been enjoyable, and found that those paid one dollar later rated the task as more enjoyable, which they explained as the person who is induced to say something convincing himself. Aronson and Mills (1959) found that people who underwent a severe initiation to join a discussion group liked the group more than those with a mild initiation, and Gerard and Mathewson (1966) replicated the effect with a different form of initiation. Brehm (1956) had participants rate and choose between products and found that chosen alternatives were re-rated more favourably and rejected ones less favourably. Bem (1967) proposed self-perception as an alternative explanation that needed no aversive drive. Elliot and Devine (1994) found in two induced-compliance experiments that dissonance was experienced as psychological discomfort that was alleviated once people changed their attitude. Harmon-Jones, Brehm, Greenberg, Simon and Nelson (1996) found that dissonance occurred even when a counterattitudinal act produced no aversive consequences, with increased physiological arousal, and Harmon-Jones (2000) found more negative affect after freely choosing to write a counterattitudinal statement, reduced after attitude change. Scher and Cooper (1989) argued that felt responsibility for aversive consequences was the key motivator. Kenworthy, Miller, Collins, Read and Earleywine (2011) synthesised studies across five dissonance paradigms and found that none of the tested models supported discomfort as the mediator, while guilt strongly predicted dissonance effect sizes. Chen and Risen (2010) showed mathematically and experimentally that the free-choice paradigm will measure spreading even if preferences are stable. Izuma and colleagues (2010) used a control condition and brain imaging and found that making a choice changed both self-reported preference and its neural representation. Sharot, Velasquez and Dolan (2010) found that preferences changed after a blind choice but not after a computer made the choice, and Sharot and colleagues (2012) found that choices altered preferences immediately and 2.5 to 3 years later, only when people made the choices themselves. Salti and colleagues (2014) found genuine choice-induced preference change related to episodic memory of the choice. Enisman, Shpitzer and Kleiman (2021) meta-analysed 43 artefact-free free-choice studies (N = 2,191) and found an overall effect of d = 0.40, with no evidence of publication bias. Vaidis and colleagues (2024) ran a multilab replication of the induced-compliance paradigm in 39 labs across 19 countries and 14 languages (N = 4,898) and found no significant attitude difference between high-choice and low-choice counterattitudinal essays, but a significant difference between counterattitudinal and neutral essays. Pauer, Linne and Erb (2024) reanalysed the replication data using self-reported choice and found that people who wrote a counterattitudinal essay aligned their attitudes only if they reported high freedom of choice. Sleegers and colleagues (2024) responded to the commentaries and concluded that the replication and a review of the wider literature raised substantial concerns about the reliability of the paradigm. Van Veen, Krug, Schooler and Carter (2009) found that dissonance engaged the dorsal anterior cingulate cortex and anterior insula and that this activity predicted subsequent attitude change. Egan, Santos and Bloom (2007) found decision rationalisation in preschool children and capuchin monkeys. Harmon-Jones, Harmon-Jones and Levy (2015) set out the action-based model, in which dissonance arises when cognitions with conflicting action implications make it difficult to act. Gawronski (2012) argued that inconsistency serves as an epistemic cue for errors in one's system of beliefs. Cooper (2019), McGrath (2017) and Hinojosa, Gardner, Walker, Cogliser and Gullifor (2017) reviewed dissonance theory, the strategies people use to reduce dissonance, and its use in management research.
When you preach it and do not practise it
There is one kind of dissonance that almost everybody in a support role knows intimately. You tell your team to log off at six and take the weekend, and then you answer the executive's message at eleven at night. You champion honest feedback and then say nothing when the feedback would be awkward. The gap between what you say and what you do is visible to you, and you suspect it is visible to them.
Researchers have been deliberately creating that exact feeling for thirty years, because it turns out to be one of the few forms of dissonance that pushes behaviour in a good direction.
The procedure is called induced hypocrisy. People first advocate for something they already believe in, publicly. Then they are privately reminded of the times they have not done it themselves. In a field experiment at a university swimming pool, people who urged others to take shorter showers and were then reminded of their own long showers took significantly shorter showers themselves. People who were only reminded, or who only made the public statement, did not change. An earlier study on safer sex found the same pattern with a purchase at the end of it. A 2019 meta-analysis of 38 studies found that hypocrisy reliably increased both the intention to change and the change itself, and, interestingly, that being reminded of your own lapses sometimes did the work on its own.
The point for your working life is that the discomfort of not practising what you preach is less a character flaw to be managed away than a lever. People resolve it by changing what they do, when the easier routes are closed off.
Here is where the easier routes come in, because the mind is very good at finding them. A 2025 study of consultants in a firm whose public commitments did not match its practice found three ways staff made peace with the gap: they avoided looking at the inconsistency, they decided it was somebody else's responsibility, and they added new, comforting thoughts that bridged it. Those three map almost exactly onto the three tactics you will recognise from your own head: I am not really working late, I am getting ahead; this week is exceptional; everybody does it.
And there is a finding about how other people read the gap that is worth holding onto. People judge hypocrites more harshly than they judge people who lie outright about their own conduct, because the condemnation sends a false signal that the speaker behaves well. But hypocrites who admitted to the lapse were not judged negatively at all. The order matters too: saying the standard and then breaking it looked more hypocritical than breaking it first and then setting the standard, because the second version reads as someone who has changed.
Try this. When you catch yourself preaching something you are not doing, choose one of two honest routes and say it out loud. Either close the gap ("I've told you all to log off at six, so I'm going to do it too, starting Thursday") or name it ("I'm terrible at this myself and I'm working on it, but I still want it for you"). Both keep your credibility. The route the research says costs you is the silent one.
The research behind this section
Stone, Aronson, Crain, Winslow and Fried (1994) had young adults publicly advocate safer sex and then made them mindful of their own past failures, and found that more participants in the hypocrisy condition bought condoms and bought more of them. Dickerson, Thibodeau, Aronson and Miller (1992) ran a field experiment at a campus recreation facility and found that people who made a public commitment urging shorter showers after being reminded of their own past water use took significantly shorter showers than controls, while those who were only reminded or only made the commitment did not. Stone and Fernandez (2008) reviewed the hypocrisy procedure and found its greatest effect on behaviour when people publicly advocate a course of action and are then privately reminded of their own recent failures. Priolo and colleagues (2019) meta-analysed 29 published and nine unpublished induced-hypocrisy studies and found that hypocrisy increased behavioural intention and behaviour, that evidence on psychological discomfort was inconclusive, and that the transgressions-only condition could itself increase intention and behaviour. Murcia, del Rio and Annosi (2025) studied employees in a multinational consulting firm with compartmentalised social responsibility commitments and identified three dissonance reduction strategies: blinding (selective exposure to the inconsistency), outsourcing (denial of responsibility) and imagining (adding consonant cognitions). Jordan, Sommers, Bloom and Rand (2017) found across five studies that hypocrites were judged more negatively than people who directly made false statements about their own morality, because condemnation falsely signals moral goodness, and that honest hypocrites who admitted the transgression were not judged negatively. Barden, Rucker and Petty (2005) found in three studies that a statement followed by a violating behaviour was judged more hypocritical than the reverse order, because the reverse order was attributed to a change in the person.
Defending a decision you argued against
This is the scenario from the opening, and it is the one that makes support roles a special case. You are not only living with decisions you disagreed with; you are frequently the person who explains them, announces them, answers the questions about them and makes them work. And each of those acts, the research suggests, pulls your own view a little towards the decision.
The oldest version of this is called saying is believing. In a 1978 experiment, people described someone to an audience who either liked or disliked that person, and naturally tailored the description to suit. Later, their own memory of the person and their own liking for them had drifted towards what they had said. A 2005 set of studies found that the drift happened when people felt they had created a shared view with the audience, and particularly when they trusted the audience's judgement. So the effect is strongest with exactly the people you work most closely with.
The most striking modern evidence comes from outside the office entirely. At international debating competitions, experienced debaters are randomly assigned a side to argue. Researchers measured their factual beliefs and their confidence before and after, and found that arguing a side persuaded the debaters of it. The effect was smaller than in the laboratory, but it survived an hour of hearing the other side's arguments and a tenfold increase in the reward for being accurate. These were people who know perfectly well that the side was assigned by a coin. It moved them anyway.
Then there is a finding I find almost unsettling. In 2012 researchers built a survey with a magic trick in it: after people had rated their agreement with moral statements, some of the statements were secretly reversed, so the person was now looking at their own mark next to the opposite position. Most of the reversals went unnoticed, 69% of people failed to spot at least one of two changes, and many then gave fluent, confident arguments for the view that was the opposite of the one they had marked moments earlier. An earlier version of the same trick with photographs of faces found the same thing: people defended choices they had not made, with reasons.
Three smaller findings complete the picture, and they explain why the pull is strongest in exactly the work you care about most. Effort raises value: people who build something rate it more highly than an identical thing they did not build, but only if they finish it, and effort increased the value of a reward when people felt in control of getting it. Doing someone a favour can make you like them more. And people change their attitudes when they watch someone they identify with argue against their own view, which researchers call vicarious dissonance. You do not even have to be the one who said it.
None of this means you should refuse to defend decisions. Carrying a decision well is a large part of what the job is, and the drift is the mind's way of making committed action possible. It means that your later opinion of a decision you had to carry is not independent evidence about whether it was right. It is partly the residue of having carried it.
Your opinion of a decision you had to defend is not a second opinion. It is partly the echo of the defending.
Try this. Before you draft the announcement, write down your own view of the decision in two sentences and date it. It takes a minute. Then carry the decision properly, with all the conviction the role needs. When someone asks you in three months whether it worked, read your two sentences first, and you will know which part of your answer is judgement and which part is loyalty.
The research behind this section
Higgins and Rholes (1978) found that communicators who tailored a description of a person to their audience's attitude later showed memory and liking for the person consistent with their tailored message, the saying-is-believing effect. Echterhoff, Higgins and Groll (2005) found in three studies that this audience-tuning bias in memory occurred when feedback signalled a successful shared view and for in-group but not out-group audiences, mediated by trust in the audience's judgement. Schwardmann, Tripodi and van der Weele (2022) ran field experiments at international debating competitions that randomly assigned experienced debaters to one side of a motion and found self-persuasion in factual beliefs and confidence, smaller than in the laboratory but robust to a one-hour exchange of arguments and a tenfold increase in incentives for accuracy. Schwardmann and van der Weele (2019) studied deception and self-deception, finding that the prospect of having to persuade others made people more confident about themselves. Hall, Johansson and Strandberg (2012) used a self-transforming survey to reverse participants' stated moral attitudes and found that the majority of reversals went undetected, that 69% of participants failed to detect at least one of two changes, and that participants often constructed coherent arguments for the opposite of their original position. Johansson, Hall, Sikström and Olsson (2005) found that people failed to notice mismatches between the face they chose and the face they were given, while offering reasons for their choice, which they called choice blindness. Norton, Mochon and Ariely (2012) found across four studies that people valued products they had assembled more highly, but only when the labour resulted in successful completion. Harmon-Jones, Matis, Angus and Harmon-Jones (2024) found that high effort increased neural reward valuation when people had high perceived control over the outcome, and decreased it when they had low control. Jecker and Landy (1969) found that people liked a person more after doing him a favour. Norton, Monin, Cooper and Hogg (2003) found in three studies that people changed their attitudes after witnessing an in-group member make a counterattitudinal speech, especially when the speaker was known to disagree, and that this vicarious discomfort eased once participants expressed their revised attitudes. Cooper (2019) discusses vicarious dissonance as part of the move from the individual to the social group.
After the decision, the story gets better
Here is the scene every one of us has lived. The decision is made, the email has gone, and the conversation in the corridor afterwards is noticeably more confident than the conversation in the room before. Nobody has learned anything new in the intervening twenty minutes. The decision has simply become ours.
The cleanest demonstration is still one of the oldest. At a racetrack in 1968, researchers asked people how confident they were that their horse would win, either moments before they placed a two-dollar bet or moments after. On a seven-point scale, the median rating was 3.48 before betting, which meant a fair chance, and 4.81 after, which meant a good chance. Same horse, same race, thirty seconds apart.
The modern research adds two things that matter for anyone who keeps records. The first is memory. In a 2014 study, people who made a difficult choice between two phones were more likely to misremember their original rating of the chosen phone as more favourable than it had been. The second is hindsight. Once people know how something turned out, they overestimate how predictable it was, and a review of the field describes three parts to that: memory distortion, beliefs about how likely the outcome was all along, and inflated faith in their own foresight. A study of 85 investment bankers found that the more hindsight-biased among them had lower performance. The mind does not only defend the decision; it edits the record of how the decision felt at the time.
The heavy version of this is escalation of commitment, the tendency to keep putting resources into a failing course of action because you were the one who chose it, and it has a post of its own in this series, so I will keep to the part that belongs here. A 2012 set of four experiments tested the popular idea that people responsible for a failing decision go looking for evidence that supports it. They did not; their search was as even-handed as anyone's. What differed was how they evaluated the information once they had it, and that biased evaluation partly explained why they reinvested. The bias lives in the weighing, not the looking, which means that seeking out the other side is necessary and nowhere near sufficient.
There is a hopeful finding here too. When the need to defend the decision is really a need to protect your sense of yourself as competent, affirming yourself in some unrelated way can take the pressure off. A 1983 set of experiments found that giving people the chance to affirm an important value removed the usual attitude change after dissonance, and a 2008 study examined the same route as a way to reduce escalation of commitment, along with its limits. Culture shapes the urge as well: in studies comparing North American and East Asian participants, the need to justify a choice was strongest for choices made for oneself in the first group, and for choices made for a friend or with other people in mind in the second.
Try this. Make a habit of writing the prediction before the outcome. For any decision worth a second thought, put three lines somewhere you will find them again: what you expect to happen, how confident you are, and what would tell you that you were wrong. It is the only reliable way to argue with your own hindsight, because it is the one piece of evidence the mind cannot quietly revise.
The research behind this section
Knox and Inkster (1968) asked 69 bettors at a racetrack to rate their horse's chance of winning before placing a two-dollar bet and 72 bettors immediately afterwards, and found a median rating of 3.48 before and 4.81 after on a seven-point scale, replicated at a harness race, indicating that dissonance reduction can occur very rapidly after commitment. Rodriguez and Strange (2014) found that participants who made a difficult choice between two smartphones showed the spread of alternatives and were more likely to misremember their initial evaluation of the decision more favourably than they had rated it. Fischhoff (1975) found that knowing an outcome increased its perceived likelihood in hindsight and that people were largely unaware of the effect. Roese and Vohs (2012) reviewed hindsight bias as memory distortion, beliefs about events' objective likelihoods and beliefs about one's own prediction abilities, and noted that encouraging people to consider alternative causal explanations can reduce it. Biais and Weber (2009) found in an experiment with 85 investment bankers in London and Frankfurt that more hindsight-biased agents had lower performance. Schultze, Pfeiffer and Schulz-Hardt (2012) found across four experiments that decision makers responsible for a failing initial decision did not search for information more selectively than non-responsible decision makers, but that their evaluation of the information was biased in favour of reinvestment, partially mediating the effect of responsibility on escalation. Steele and Liu (1983) found in three experiments that affirming an important self-relevant value eliminated dissonance-reducing attitude change. Sivanathan, Molden, Galinsky and Ku (2008) examined the promise and peril of self-affirmation in de-escalation of commitment. Kitayama, Snibbe, Markus and Suzuki (2004) found that Japanese participants justified their choices only when self-relevant others were primed, whereas European Americans did so regardless, and Hoshino-Browne and colleagues (2005) found that European Canadians justified choices made for themselves more, while Asian Canadians and Japanese participants justified choices made for a friend more; culture shapes when the need to justify a decision arises.
The lawyer in your head
Most of what we call rationalising is something quieter than lying to ourselves, and far more respectable-looking, and the research on it is the most useful part of this whole literature for anyone whose job involves weighing things up for someone else.
The founding paper, from 1990, made a careful claim. People are more likely to arrive at the conclusions they want to arrive at, but their ability to do so is limited by their ability to build a justification that looks reasonable. We do not believe whatever we like. We believe what we like whenever we can make a respectable case for it, and we are extremely good at making respectable cases.
The mechanism was pinned down two years later, and it is the single idea I would most like you to take from this post. In three experiments, people needed far less information to accept a conclusion they wanted than one they did not. Given an unwelcome medical test result, people took longer to decide the result was complete, were more likely to retest it, listed more reasons it might be wrong and rated the test as less accurate. Nobody refused the evidence. They simply applied a much higher standard of proof to it.
Psychologists sometimes summarise this as the difference between two questions. For a conclusion you like, you ask: can I believe this? One reasonable argument is enough. For a conclusion you do not like, you ask: must I believe this? And one flaw is enough to escape.
The same pattern shows up everywhere it has been looked for. People on both sides of a contested question, shown the same mixed evidence, rated the studies that supported their view as more convincing and came away more polarised than before. In a meta-analysis of selective exposure, people showed a moderate preference for information that agreed with them, and the preference weakened when their confidence was high, and reversed when the unwelcome information was useful for a goal they were working on. A 2019 meta-analysis of 51 experiments and more than 18,000 people found that partisan bias was robust and almost exactly equal on both sides of the political divide. Nobody gets to feel superior here.
Two findings about intelligence are worth knowing, because the natural hope is that thinking harder protects you. Research across many forms of what is called myside bias finds that its size has very little relation to intelligence. And a well-known 2017 experiment reported that people with the strongest number skills were the most polarised when the same data came labelled with a politically charged topic, as if their skill was being used to protect their conclusion. A preregistered replication in 2021 did not find good evidence for that particular pattern, though it did still find people interpreting the numbers in line with their politics. So the honest summary is that being clever does not rescue you from motivated reasoning; whether it makes it worse is disputed.
There is a real disagreement underneath all of this, and it is worth being fair to it. Several researchers argue that much of what gets labelled motivated reasoning could be people reasoning sensibly from different starting beliefs or different levels of trust in a source, and that the classic designs cannot tell the two apart. That is a serious point, and it is why the most convincing evidence comes from designs like the unwelcome medical test, where the only thing that differs is what the person would like to be true. My reading is that motivated reasoning is real, that it is smaller and more specific than the popular version, and that its signature is the asymmetry of standards rather than the conclusion itself.
One more idea reframes the whole thing kindly. A widely cited theory argues that human reasoning evolved less for finding truth alone than for producing and evaluating arguments with other people. On that view, we are naturally lazy about checking our own arguments and sharp at checking other people's, which is exactly why reasoning works best in company.
Try this. When a conclusion arrives that you are pleased about, deliberately ask the question you would ask of one you disliked: must I believe this? What would I need to see before I accepted it if I were hoping for the opposite? If the evidence you have clears that bar, you are reasoning. If it only clears the lower one, you have found the lawyer.
The research behind this section
Kunda (1990) proposed that motivation affects reasoning through biased strategies for accessing, constructing and evaluating beliefs, and that people's ability to reach desired conclusions is constrained by their ability to construct seemingly reasonable justifications. Ditto and Lopez (1992) found in three experiments that information consistent with a preferred conclusion was examined less critically than inconsistent information, so less information was needed to reach the preferred conclusion, and that participants given an unfavourable medical test result took longer to decide the result was complete, were more likely to retest it, cited more irregularities and rated the test as less accurate. Lord, Ross and Lepper (1979) found that supporters and opponents of capital punishment rated the study confirming their view as more convincing and became more polarised after seeing the same mixed evidence. Taber and Lodge (2006) found a prior attitude effect, a disconfirmation bias and a confirmation bias in evaluating political arguments, with polarisation strongest among those with the strongest priors and highest sophistication. Hart and colleagues (2009) meta-analysed selective exposure and found a moderate preference for congenial over uncongenial information (d = 0.36), weaker when confidence was high, and an uncongeniality bias when uncongenial information was relevant to a current goal. Ditto and colleagues (2019) meta-analysed 51 experimental studies with more than 18,000 participants and found robust partisan bias (r = .245) with no difference between liberals (r = .235) and conservatives (r = .255). Stanovich, West and Toplak (2013) reported that the magnitude of myside bias shows very little relation to intelligence. Kahan, Peters, Dawson and Slovic (2017) found that polarisation over data about gun control increased rather than abated among the most numerate participants. Persson, Andersson, Koppel, Västfjäll and Tinghög (2021) ran a preregistered replication and did not find good evidence for motivated numeracy, finding ideologically congruent responding that was not moderated by numeracy and, where moderation appeared, only at the highest levels. Tappin, Pennycook and Rand (2020) argued that paradigmatic designs cannot separate politically motivated reasoning from differences in prior beliefs, and Druckman and McGrath (2019) argued that the evidence for motivated reasoning in climate change preference formation is less clear than assumed for the same reason. Kruglanski and Webster (1996) described the need for closure, in which urgency leads people to seize on an early answer and then freeze on it. Mercier and Sperber (2011) proposed that the function of reasoning is argumentative, explaining confirmation bias as a feature of producing arguments.
Reasoning or rationalising: how to tell in your own head
This is the question the original post ended on, and the research has more to offer on it than I expected. The difficulty is that from the inside, the two feel identical. Both arrive as good reasons. Both feel like thinking. And, uncomfortably, we are unusually bad at seeing the difference in ourselves.
People see bias far more readily in other people than in themselves. In the original studies, participants rated themselves as less biased than their classmates and fellow travellers, insisted their self-assessments were objective even after reading how they might have been affected, and judged a peer's self-serving explanation of a test result as biased while seeing their own identical explanation as fair. Researchers call this the bias blind spot. It is not cured by being clever: in a 2012 study, cognitive ability did nothing to reduce it, and if anything the more able had a larger blind spot. A 2015 measure of it predicted, among other things, how much people ignored advice from others.
Introspection is not much help either. A classic 1977 review concluded that when people explain why they did something, they often report a plausible theory about what would have caused it rather than any direct observation of their own mind. And a well-known account of moral judgement argues that the reasoning usually arrives after the verdict, built to support a conclusion that intuition has already reached.
So the test cannot be "does this feel like reasoning?" It has to be something you can check from the outside. The research offers five checks that work, and two of them are the tests the original post suggested.
Would I accept this argument from a colleague? This is the first question from the original post, and it turns out to have been tested almost exactly. In a 2016 pair of experiments, people solved reasoning problems and wrote their arguments, then evaluated other people's arguments. Without their knowledge, one of the arguments shown to them was their own. Among those who did not notice, 56% in one experiment and 58% in the other rejected it. They were also better at telling valid from invalid arguments when the arguments belonged to someone else. We are lenient judges of our own reasoning and good judges of other people's, which means borrowing the second judge is the fix.
Would I advise a friend to do this? The second question from the original post has its own evidence. In three experiments with 693 people, participants reasoned more wisely about another person's problem than their own: more aware of the limits of what they knew, more open to compromise, more attentive to other perspectives. When they were asked to think about their own problem from a distance, as an observer would, the gap disappeared. The effect was the same for adults in their twenties and in their seventies.
Did the criteria move? In a study of hiring decisions, people favoured a candidate from the group they preferred and then redefined what the job required to match whatever that candidate happened to have. When they committed to the hiring criteria before learning who the candidates were, the bias disappeared. A related series of six studies found the same move in admissions and employment: people quietly inflated the value of their preferred candidate's qualifications. If the reasons for a decision were chosen after the decision, they are probably rationalisations.
Would I judge it the same if they did it? People judged their own unfair act as fairer than the identical act by someone else, and extended the same leniency to members of their own group. In a 1997 study, people given a coin to decide fairly who got the pleasant task still gave it to themselves, and still rated what they had done as moral. This is the check for the fourth scenario in the original post, the rule that applies to everyone except the person you work for: swap the name and see whether your verdict survives.
Is the conclusion suspiciously convenient? When the answer you reach is also the one that is easiest for you, most flattering to you or least awkward to deliver, that proves nothing on its own, but it is a good reason to run the other four checks before you commit to it.
And the counterweight, because it matters. One researcher has argued that rationalisation is not simply an error. The mind runs on habit, instinct, social norms and emotion as well as reasoning, and building a reason after the fact can be a useful way of extracting what those other systems knew and making it available to deliberate thought. On that view, producing reasons after the fact is normal; the trouble starts when you treat those reasons as the verdict rather than as a draft to be checked.
Try this. Pick the five checks that suit you and put them where you make decisions: the colleague test, the friend test, the moving criteria, the swapped name, the convenient answer. Two of them will usually settle it. The colleague test is the fastest: rewrite your reason as if someone else had sent it to you, and read it as you would read theirs.
The research behind this section
Pronin, Lin and Ross (2002) found across three studies that people saw biases more in others than in themselves, insisted their self-assessments were accurate after reading how bias could have affected them, and judged a peer's self-serving attributions as biased but their own similar attributions as unbiased. West, Meserve and Stanovich (2012) found replicable bias blind spots for many classic cognitive biases, not attenuated by cognitive ability or thinking dispositions, with a larger blind spot, if anything, among those with higher cognitive ability. Scopelliti and colleagues (2015) developed a measure of the bias blind spot that predicted, among other things, the extent to which people ignored the advice of others. Nisbett and Wilson (1977) reviewed evidence that people have little direct introspective access to higher-order cognitive processes and report instead on implicit causal theories. Haidt (2001) proposed that moral reasoning is usually a post hoc construction generated after a judgement has been reached. Trouche, Johansson, Hall and Mercier (2016) used a choice blindness manipulation to present people with their own argument as someone else's, and found that among those who accepted the manipulation, 56% and 58% in two experiments rejected their own arguments, and that people were better at distinguishing valid from invalid arguments when they were someone else's. Grossmann and Kross (2014) found in three experiments (N = 693) that people reasoned more wisely about another person's problems than their own, that self-distancing eliminated the asymmetry, and that the effects were comparable for younger and older adults. Uhlmann and Cohen (2005) found in three studies that people redefined the criteria for success in a job to match the credentials of the applicant from the desired group, and that committing to hiring criteria before learning the applicant's gender eliminated the discrimination. Norton, Vandello and Darley (2004) found across six studies that people justified decisions biased by social category by inflating the relative value of their preferred candidates' qualifications. Valdesolo and DeSteno (2007) found that people judged a transgression as fairer when they committed it themselves than when another person did, extending the leniency to in-group members. Batson and colleagues (1997) found across three small studies that participants deciding who would get a positive task tended to assign it to themselves, even when a coin was available to decide fairly, while still seeing their behaviour as moral. Shalvi, Gino, Barkan and Ayal (2015) described self-serving justifications before and after ethical violations that allow people to do wrong while feeling moral. Lowell (2012) argued that dissonance-induced self-justification can create an amplifying feedback loop in managers' ethical behaviour. Cushman (2020) argued that rationalisation extracts useful information from non-rational influences on behaviour and makes it available to later reasoning.
Tools that hold up, and one that backfires
The encouraging news is that the evidence on correcting this is better than the evidence on most workplace advice. The less encouraging news is that some of the most popular versions are weaker than their reputation, and one of them can make things worse. Here are the five I would use, in the order I would reach for them.
Decide the criteria first. This is the strongest tool in the list because it removes the opportunity rather than relying on willpower. Committing to criteria before seeing the options eliminated the moving-goalposts bias in the hiring study. A classic 1979 paper found that even crude scoring rules, with the factors given equal weight, outperformed expert intuition at prediction. For anyone who shortlists, triages or recommends, the practical form is to write down what good looks like before the options arrive.
Consider the opposite, briefly. The founding 1984 study found that asking people to consider the opposite reduced their biased reading of evidence more than an instruction to be fair and unbiased did. Earlier work found that listing a reason why your answer might be wrong improved people's confidence judgements, while listing reasons for it did nothing. Later research found that considering the opposite reduced anchoring among real experts, and that any plausible alternative works, not only the direct opposite.
Now the backfire. In a 2002 pair of studies, people asked to list two ways an event could have turned out differently showed no hindsight bias. People asked to list ten found it hard, took the difficulty as a sign that there were not many alternatives, and showed more hindsight bias than before. So consider the opposite, and keep it to two or three reasons. Ten is not more rigorous; it is a trap.
Keep a decision journal. This is the prediction habit from earlier, made regular. Among 194 neuropsychologists, those who had to list a reason each diagnosis might be correct were less prone to hindsight bias. In a two-year forecasting tournament, the teams that trained people, let them discuss their reasoning and tracked their accuracy produced markedly better forecasts. A journal is how you track: prediction, confidence and what would change your mind, written before, read after.
Invite a red team, and make it real. Groups primed to think about how things could have turned out differently searched for more information that challenged their preferred option. You can even run a small version alone: a technique called dialectical bootstrapping asks you to make an estimate, assume it is wrong, make a second from a different angle and average the two, which reduced individual error. The conformity post in this series covers the research showing that real disagreement beats assigned devil's advocacy, so the best red team is someone who actually doubts the plan.
Know who you will have to answer to, before you decide. This one is subtle and important. A long review of accountability research found that it depends when it arrives. Being accountable before a decision, to someone whose view you do not know, led people to think more self-critically and consider more sides. Being accountable after you have committed tended to produce defensive bolstering of the choice you had already made. So the moment to picture the tough question is before the decision, not after it.
The best time to argue with yourself is before you have anything to defend.
Two supporting tools are worth a line each. Self-affirmation, reflecting on a value that matters to you, made people more open to evidence that challenged their views, though across 144 tests in health the effects were small. And debiasing training can work: one-shot training games reduced several biases by around a third immediately, with effects still visible two months later, and in a field test 290 graduate students who had received training were 19% less likely to choose the inferior, hypothesis-confirming answer to a business case modelled on a famous launch decision.
There is one thing that makes almost all of this harder, and your job is full of it. Time pressure raises the need for closure, which pushes people to seize on an early answer and then freeze on it. The tools above are fast by design for that reason. None of them needs more than five minutes.
Try this. Build a five-minute routine for any decision you will have to defend: two lines of criteria before the options, two reasons it could be wrong, one line of prediction with a confidence figure, and the name of the person who would disagree. The page does not need to be seen by anyone else. It needs to exist before the decision does.
The research behind this section
Uhlmann and Cohen (2005) found that committing to hiring criteria before learning the applicant's gender eliminated discrimination. Dawes (1979) summarised evidence that proper linear models outperform clinical intuition and that even improper models, including equal weights, are superior to intuition for predicting numerical criteria. Lord, Lepper and Preston (1984) found that instructions to consider the opposite reduced biased assimilation of evidence more than instructions to be fair and unbiased. Koriat, Lichtenstein and Fischhoff (1980) found that only listing reasons contradicting a chosen answer improved the appropriateness of confidence judgements. Mussweiler, Strack and Pfeiffer (2000) found that listing arguments against a provided anchor reduced the anchoring effect, including among experts in a real-world setting. Hirt and Markman (1995) found that considering any plausible alternative outcome, not only the opposite, debiased likelihood judgements. Sanna, Schwarz and Stocker (2002) found that listing ten counterfactual thoughts was experienced as difficult and increased hindsight bias, while listing two produced no significant hindsight effects. Arkes, Faust, Guilmette and Hart (1988) found among 194 neuropsychologists that listing one reason why each possible diagnosis might be correct reduced the hindsight bias. Mellers and colleagues (2014) found in a two-year geopolitical forecasting tournament that probability training, team collaboration and tracking improved both calibration and resolution. Kray and Galinsky (2003) found that counterfactual mind-sets increased groups' search for disconfirming information in group decisions. Herzog and Hertwig (2009) showed that averaging a first estimate with a second one drawing on different knowledge, dialectical bootstrapping, improved accuracy beyond reliability gains. Tetlock, Skitka and Boettger (1989) identified conformity, self-critical complexity and defensive bolstering as distinct strategies for coping with accountability, and Lerner and Tetlock (1999) reviewed the conditions under which accountability attenuates, has no effect on, or amplifies cognitive biases. Cohen, Aronson and Steele (2000) found that self-affirmed participants were more persuaded by evidence impugning their views on capital punishment and more critical of an advocate who confirmed their views. Cohen and Sherman (2014) reviewed self-affirmation theory and interventions. Epton, Harris, Kane, van Koningsbruggen and Sheeran (2015) meta-analysed 144 experimental tests and found small but reliable positive effects of self-affirmation on message acceptance (d = .17), intentions (d = .14) and behaviour (d = .32). Morewedge and colleagues (2015) found in two longitudinal experiments that single training interventions reduced biases immediately (games at least 31.94%, videos at least 18.60%) and that effects persisted at least two months later. Sellier, Scopelliti and Morewedge (2019) found that 290 graduate students who had received a one-shot debiasing training were 19% less likely than untrained participants to choose the inferior hypothesis-confirming solution to a business case modelled on the Space Shuttle Challenger launch decision. Milkman, Chugh and Bazerman (2009) reviewed strategies for improving decisions. Kruglanski and Webster (1996) described how urgency raises the need for closure.
Saying it out loud without making it a fight
Most of the time, catching the rationalisation is the private half. The public half is saying something, and the original post had good instincts about how: frame the concern as a practical risk rather than an ethical accusation, and remember that disagreement is not disloyalty.
The research explains why the framing matters so much, and it is the same mechanism as everything above, running in someone else's head. When a leader experiences a colleague's challenge as a threat to their competence, two vignette studies with 269 and 208 people found they were more likely to reject the idea, by finding fault with the message and with the messenger. A good mood beforehand weakened that response. Your executive has a lawyer in their head too, and it wakes up when the challenge sounds like a verdict on them.
So the practical framing is doing real psychological work. "Could we face a reputational risk if this comes out?" invites the "must I?" question about the risk rather than the "must I?" question about the person. And the most persuasive move of all borrows the colleague test: "If another team brought us this plan, what would we ask them?" It lets everyone use the sharper judgement we reserve for other people's arguments, without anyone having to admit they were being lenient with their own.
Try this. Before you raise a concern, write it as a question about the plan rather than a statement about the person, and choose your moment for when the executive is not already under pressure. One sentence, one risk, one question. The aim is to let them change their mind without having to lose the argument.
The research behind this section
Popelnukha and colleagues (2022) ran two vignette studies (N = 269 and N = 208) and found that a leader's perceived threat to competence from a follower's voice was related to voice rejection, partially through message derogation and source derogation, and that positive affect beforehand weakened these effects while negative affect strengthened them. Trouche, Johansson, Hall and Mercier (2016) found that people evaluate others' arguments more critically and more accurately than their own. Cohen, Aronson and Steele (2000) found that affirming the self reduced defensive resistance to challenging evidence.
The reasoning check
Every idea in this post comes down to a moment you will recognise: there is a decision in front of you, you have a view about it, and you have a role in carrying it. The question is how much of your current view is judgement and how much is the job talking. So here is a tool for that moment.
Tell it what the decision is, what part you played in it, how your view has moved since, and what you have and have not done to test it. It will tell you which way your head is likely to be leaning, which of the checks from this post to run first, and what to look for, and it gives you a short note to keep for later.
Where to start
If you take one thing from all of this, make it the two questions. When you like a conclusion, you ask whether you can believe it; when you do not, you ask whether you must. Most rationalising lives in the gap between those two standards, and the single most reliable way to close it is to ask "must I?" of the conclusions you are pleased about.
After that, there are three moves worth making.
Write your own view down before you carry a decision, two sentences and a date, so your later opinion has something to be checked against. Before you decide anything you will have to defend, set the criteria first and name two reasons it could be wrong, and no more than a handful. And when a reason arrives that you like very much, rewrite it as though a colleague had sent it and read it the way you would read theirs.
I have put the three pages I would use into a download: a two-sentence record for any decision you are asked to carry, a five-minute pre-decision page with criteria, two reasons against and a prediction, and a reasoning or rationalising check you can run on any conclusion before it becomes your opinion.
Having spent years as the right hand to some remarkable leaders, I think this is one of the quiet skills that separates the good people in this profession from the great ones. It has nothing to do with being difficult or keeping your distance from the decisions you support, and everything to do with carrying them wholeheartedly while keeping one small, honest record of what you really thought, so that when it matters, your judgement is still your own and still worth asking for.
The part of this that shows up in how you are seen at work sits inside what the Connected Leader Assessment calls Judgement and Decision-Making: how well you weigh evidence you would rather not hear, how you set criteria before the options arrive, and whether you can tell the difference between a view you reached and a view you were handed. It is the dimension that decides whether an executive asks you what you think, or only asks you to make it happen.
The assessment is free, takes about twelve minutes and gives you a read across all five dimensions. If you want to go further on this one, the Skill Sprints are short, focused programmes on a single dimension, and the Irreplaceable Audit is the place to look at which of your judgements the organisation relies on most, which are exactly the ones most worth keeping honest.
Stay savvy.
Meg ✌️
Free tools
The reasoning check, the interactive tool in this post, free
The Reasoning Pages, the download in this post, free
The Connected Leader Assessment, twelve minutes, free, with a read on Judgement and Decision-Making, the dimension this post sits in
P.S. Want to Go Deeper on Reasoning and Rationalising?
If this resonated and you want to go further, here is the evidence the piece is built on. These are the papers that shaped my thinking on how the mind defends a decision once it is made, and how to tell reasoning from rationalising, and they are well worth your time.
Every DOI below has been checked. Where I could not confirm a figure from the paper itself, I have left the figure out rather than guess.
📃 Festinger and Carlsmith (1959) Paid one dollar rather than twenty to call a dull task enjoyable, people later rated it more enjoyable. https://doi.org/10.1037/h0041593
📃 Aronson and Mills (1959) A severe initiation made people like a dull discussion group more. https://doi.org/10.1037/h0047195
📃 Brehm (1956) After choosing, the chosen product rose in people's estimation and the rejected one fell. https://doi.org/10.1037/h0041006
📃 Elliot and Devine (1994) Dissonance measured as psychological discomfort that eased once attitudes changed. https://doi.org/10.1037/0022-3514.67.3.382
📃 Bem (1967) Self-perception: the alternative explanation that needs no uncomfortable drive. https://doi.org/10.1037/h0024835
📃 Egan, Santos and Bloom (2007) Decision rationalisation in preschool children and capuchin monkeys. https://doi.org/10.1111/j.1467-9280.2007.02012.x
📃 Harmon-Jones, Harmon-Jones and Levy (2015) The action-based model: dissonance signals conflicting pulls that make it hard to act. https://doi.org/10.1177/0963721414566449
📃 Hinojosa and colleagues (2017) A review of cognitive dissonance theory in management research. https://doi.org/10.1177/0149206316668236
📃 McGrath (2017) A review of the strategies people use to reduce dissonance. https://doi.org/10.1111/spc3.12362
📃 Cooper (2019) Sixty years of dissonance research, including dissonance felt on behalf of the group. https://doi.org/10.5334/irsp.277
📃 Chen and Risen (2010) The statistical flaw that can make the classic choice studies show change where there is none. https://doi.org/10.1037/a0020217
📃 Vaidis and colleagues (2024) 39 labs, 19 countries, 4,898 people. The core one-dollar-style prediction did not replicate. https://doi.org/10.1177/25152459231213375
📃 Kenworthy and colleagues (2011) Across five paradigms, guilt predicted how much people shifted; discomfort did not mediate it. https://doi.org/10.1080/10463283.2011.580155
📃 Stone and Fernandez (2008) Reviewing hypocrisy: preach publicly, recall your lapses privately, and behaviour changes. https://doi.org/10.1111/j.1751-9004.2008.00088.x
📃 Stone and colleagues (1994) Advocating safer sex and then recalling lapses led more people to buy condoms. https://doi.org/10.1177/0146167294201012
📃 Dickerson and colleagues (1992) Urging shorter showers after recalling long ones led to shorter showers. https://doi.org/10.1111/j.1559-1816.1992.tb00928.x
📃 Priolo and colleagues (2019) 38 studies. Induced hypocrisy raised both intentions and behaviour. https://doi.org/10.1177/0146167219841621
📃 Jordan and colleagues (2017) Hypocrites are judged worse than liars; honest hypocrites who admit it are not. https://doi.org/10.1177/0956797616685771
📃 Higgins and Rholes (1978) Saying is believing: tailoring a message to an audience changed the speaker's own memory. https://doi.org/10.1016/0022-1031(78)90032-x
📃 Echterhoff, Higgins and Groll (2005) The saying-is-believing drift depended on trust in the audience and a shared view. https://doi.org/10.1037/0022-3514.89.3.257
📃 Norton and colleagues (2003) Watching an in-group member argue against their view changed observers' attitudes. https://doi.org/10.1037/0022-3514.85.1.47
📃 Norton, Mochon and Ariely (2012) The IKEA effect: labour leads to love, but only when the thing is finished. https://doi.org/10.1016/j.jcps.2011.08.002
📃 Johansson and colleagues (2005) Choice blindness: people defended faces they had not chosen, with reasons. https://doi.org/10.1126/science.1111709
📃 Hall, Johansson and Strandberg (2012) 69% missed at least one reversal of their own moral views, and argued for the opposite. https://doi.org/10.1371/journal.pone.0045457
📃 Fischhoff (1975) Knowing the outcome made it seem more predictable than it had been. https://doi.org/10.1037/0096-1523.1.3.288
📃 Roese and Vohs (2012) Hindsight bias reviewed: memory, likelihood and faith in one's own foresight. https://doi.org/10.1177/1745691612454303
📃 Kunda (1990) Motivated reasoning: we reach the conclusions we want when we can justify them. https://doi.org/10.1037/0033-2909.108.3.480
📃 Ditto and Lopez (1992) Unwelcome conclusions faced a higher standard of proof than welcome ones. https://doi.org/10.1037/0022-3514.63.4.568
📃 Lord, Ross and Lepper (1979) Both sides rated the study that agreed with them as better, and polarised. https://doi.org/10.1037/0022-3514.37.11.2098
📃 Taber and Lodge (2006) Congenial arguments were judged stronger; contrary ones were argued against. https://doi.org/10.1111/j.1540-5907.2006.00214.x
📃 Kahan and colleagues (2017) The most numerate were the most polarised when the data touched politics. https://doi.org/10.1017/bpp.2016.2
📃 Hart and colleagues (2009) Meta-analysis of selective exposure: a moderate preference for agreeable information, d = 0.36. https://doi.org/10.1037/a0015701
📃 Ditto and colleagues (2019) 51 experiments, 18,000 people. Partisan bias was robust and equal on both sides. https://doi.org/10.1177/1745691617746796
📃 Tappin, Pennycook and Rand (2020) Classic designs cannot separate motivated reasoning from different starting beliefs. https://doi.org/10.1016/j.cobeha.2020.01.003
📃 Druckman and McGrath (2019) The evidence for motivated reasoning on climate is less clear than assumed. https://doi.org/10.1038/s41558-018-0360-1
📃 Mercier and Sperber (2011) Reasoning evolved to argue, which explains why we are lazy with our own arguments. https://doi.org/10.1017/s0140525x10000968
📃 Pronin, Lin and Ross (2002) People saw bias in others far more readily than in themselves. https://doi.org/10.1177/0146167202286008
📃 Scopelliti and colleagues (2015) A measure of the bias blind spot, which predicted ignoring advice. https://doi.org/10.1287/mnsc.2014.2096
📃 West, Meserve and Stanovich (2012) Cognitive ability did not shrink the bias blind spot. https://doi.org/10.1037/a0028857
📃 Grossmann and Kross (2014) 693 people. Wiser about others' problems than their own, until they stepped back. https://doi.org/10.1177/0956797614535400
📃 Uhlmann and Cohen (2005) People redefined the job to fit the candidate they preferred; fixing criteria first stopped it. https://doi.org/10.1111/j.0956-7976.2005.01559.x
📃 Norton, Vandello and Darley (2004) Six studies. People inflated the value of their preferred candidate's qualifications. https://doi.org/10.1037/0022-3514.87.6.817
📃 Batson and colleagues (1997) Moral hypocrisy: appearing fair while still taking the better task. https://doi.org/10.1037/0022-3514.72.6.1335
📃 Lerner and Tetlock (1999) Accountability before a decision prompts self-criticism; after it, bolstering. https://doi.org/10.1037/0033-2909.125.2.255
📃 Tetlock, Skitka and Boettger (1989) Conformity, complexity and defensive bolstering as ways of coping with accountability. https://doi.org/10.1037/0022-3514.57.4.632
📃 Lord, Lepper and Preston (1984) Considering the opposite beat an instruction to be fair and unbiased. https://doi.org/10.1037/0022-3514.47.6.1231
📃 Mussweiler, Strack and Pfeiffer (2000) Arguing against an anchor reduced anchoring, including among experts. https://doi.org/10.1177/01461672002611010
📃 Hirt and Markman (1995) Any plausible alternative debiased judgement, not only the opposite. https://doi.org/10.1037/0022-3514.69.6.1069
📃 Sanna, Schwarz and Stocker (2002) Listing ten alternatives backfired and raised hindsight bias; listing two did not. https://doi.org/10.1037/0278-7393.28.3.497
📃 Morewedge and colleagues (2015) One-shot training cut several biases, with effects lasting two months. https://doi.org/10.1177/2372732215600886
📃 Sellier, Scopelliti and Morewedge (2019) 290 graduate students. Trained people were 19% less likely to pick the confirming answer. https://doi.org/10.1177/0956797619861429
📃 Herzog and Hertwig (2009) Averaging two of your own estimates made from different angles cut error. https://doi.org/10.1111/j.1467-9280.2009.02271.x
📃 Cohen, Aronson and Steele (2000) Affirming the self made people more open to evidence against their views. https://doi.org/10.1177/01461672002611011
📃 Cohen and Sherman (2014) Self-affirmation as a social psychological intervention, reviewed. https://doi.org/10.1146/annurev-psych-010213-115137
📃 Epton and colleagues (2015) 144 tests. Self-affirmation had small but reliable effects on accepting health messages. https://doi.org/10.1037/hea0000116
📃 Steele and Liu (1983) Affirming an important value removed the usual attitude change after dissonance. https://doi.org/10.1037/0022-3514.45.1.5
📃 Koriat, Lichtenstein and Fischhoff (1980) Only listing reasons against your answer improved confidence judgements. https://doi.org/10.1037/0278-7393.6.2.107
📃 Mellers and colleagues (2014) Training, teaming and tracking improved forecasts in a two-year tournament. https://doi.org/10.1177/0956797614524255
📃 Gawronski (2012) Inconsistency as a cue that something in what we believe is wrong. https://doi.org/10.1521/soco.2012.30.6.652
📃 van Veen and colleagues (2009) Brain activity during dissonance predicted later attitude change. https://doi.org/10.1038/nn.2413
📃 Harmon-Jones and colleagues (1996) Dissonance without aversive consequences, with rising physiological arousal. https://doi.org/10.1037/0022-3514.70.1.5
📃 Scher and Cooper (1989) Felt responsibility for a bad outcome aroused dissonance even without inconsistency. https://doi.org/10.1037/0022-3514.56.6.899
📃 Enisman, Shpitzer and Kleiman (2021) 43 flaw-free studies, 2,191 people. Choosing changed preferences, d = 0.40. https://doi.org/10.1037/pspa0000263
📃 Hoshino-Browne and colleagues (2005) Culture shaped whether choices for oneself or for a friend were justified more. https://doi.org/10.1037/0022-3514.89.3.294
📃 Shalvi and colleagues (2015) Self-serving justifications before and after doing wrong. https://doi.org/10.1177/0963721414553264
📃 Stanovich, West and Toplak (2013) Myside bias shows very little relation to intelligence. https://doi.org/10.1177/0963721413480174
📃 Kruglanski and Webster (1996) The need for closure: seizing on an answer, then freezing on it. https://doi.org/10.1037/0033-295x.103.2.263
📃 Dawes (1979) Even crude equal-weight scoring rules beat expert intuition at prediction. https://doi.org/10.1037/0003-066x.34.7.571
📃 Cushman (2020) Rationalisation as a useful fiction that feeds later reasoning. https://doi.org/10.1017/s0140525x19001730
📃 Haidt (2001) Moral reasoning as a post hoc construction after the judgement. https://doi.org/10.1037/0033-295x.108.4.814
📃 Nisbett and Wilson (1977) We report plausible theories, not direct observations, of our own minds. https://doi.org/10.1037/0033-295x.84.3.231
📃 Gerard and Mathewson (1966) The severity-of-initiation effect, replicated with a different kind of unpleasantness. https://doi.org/10.1016/0022-1031(66)90084-9
📃 Knox and Inkster (1968) Racetrack bettors rated their horse's chances higher straight after betting than moments before. https://doi.org/10.1037/h0025528
📃 Sharot, Velasquez and Dolan (2010) A blind choice still shifted preferences; a computer choosing did not. https://doi.org/10.1177/0956797610379235
📃 Valdesolo and DeSteno (2007) The same unfair act looked fairer when people did it themselves. https://doi.org/10.1111/j.1467-9280.2007.01961.x
📃 Izuma and colleagues (2010) With a proper control, making a choice changed stated preference and its trace in the brain. https://doi.org/10.1073/pnas.1011879108
📃 Barden, Rucker and Petty (2005) Saying then doing looks more hypocritical than doing then saying. https://doi.org/10.1177/0146167205276430
📃 Biais and Weber (2009) 85 investment bankers. The more hindsight-biased performed worse. https://doi.org/10.1287/mnsc.1090.1000
📃 Kitayama and colleagues (2004) Japanese participants justified choices only when other people were brought to mind. https://doi.org/10.1111/j.0956-7976.2004.00714.x
📃 Milkman, Chugh and Bazerman (2009) How can decision making be improved? https://doi.org/10.1111/j.1745-6924.2009.01142.x
📃 Persson and colleagues (2021) A preregistered replication found no good evidence for motivated numeracy. https://doi.org/10.1016/j.cognition.2021.104768
📃 Sivanathan and colleagues (2008) Self-affirmation as a route out of escalating commitment, and its limits. https://doi.org/10.1016/j.obhdp.2007.12.004
📃 Kray and Galinsky (2003) Thinking about what might have been made groups search for disconfirming information. https://doi.org/10.1016/s0749-5978(02)00534-4
📃 Schultze, Pfeiffer and Schulz-Hardt (2012) Responsible decision makers searched fairly but weighed the evidence in their own favour. https://doi.org/10.1037/a0024739
📃 Rodriguez and Strange (2014) After a difficult choice, people misremembered their first view as more favourable. https://doi.org/10.1080/20445911.2014.925459
📃 Salti and colleagues (2014) Genuine choice-induced preference change, tied to memory of the choice. https://doi.org/10.1371/journal.pone.0108579
📃 Pauer, Linne and Erb (2024) Reanalysis: people who felt free to choose did shift towards their essay. https://doi.org/10.1177/25152459241265002
📃 Sleegers and colleagues (2024) The replication team's reply: real concerns about the induced-compliance paradigm. https://doi.org/10.1177/25152459241268197
📃 Murcia, del Rio and Annosi (2025) Employees facing corporate hypocrisy looked away, passed on responsibility, or added comforting thoughts. https://doi.org/10.1002/csr.70023
📃 Lowell (2012) Self-justification as an amplifying loop in managers' moral dissonance. https://doi.org/10.1007/s10551-011-0931-9
📃 Harmon-Jones and colleagues (2024) Effort raised the brain's valuation of a reward when people felt in control. https://doi.org/10.1111/psyp.14536
📃 Jecker and Landy (1969) Doing someone a favour made people like them more. https://doi.org/10.1177/001872676902200407
📃 Popelnukha and colleagues (2022) Leaders who felt their competence threatened rejected the idea and the messenger. https://doi.org/10.1108/pr-09-2019-0508
📃 Sharot and colleagues (2012) Choice-induced preference change was still there two and a half to three years later. https://doi.org/10.1177/0956797612438733
📃 Arkes and colleagues (1988) 194 neuropsychologists. Listing a reason for each diagnosis reduced hindsight bias. https://doi.org/10.1037/0021-9010.73.2.305
📃 Harmon-Jones (2000) Freely chosen counterattitudinal statements raised negative feeling, eased by attitude change. https://doi.org/10.1177/01461672002612004
📃 Schwardmann, Tripodi and van der Weele (2022) Debaters randomly assigned a side came to believe it more. https://doi.org/10.1257/aer.20200372
📃 Schwardmann and van der Weele (2019) The prospect of persuading others made people more confident about themselves. https://doi.org/10.1038/s41562-019-0666-7
📃 Trouche and colleagues (2016) Shown their own argument as someone else's, more than half rejected it. https://doi.org/10.1111/cogs.12303
P.P.S. Why Do I Even Have the Nerve to Write This?
Fair question.
I am someone who has spent years working alongside remarkable executives, navigating chaos, translating vision into structure and figuring things out, sometimes beautifully and sometimes the very hard way. I have had the privilege of being the steady right hand to leaders who move fast, take risks and expect a lot. That environment has shaped me more than any classroom ever could.
I also happen to have an MBA, and I am studying psychology because people fascinate me. How we work. Why we disconnect. What actually holds us together when the pressure rises.
But really, none of that is the point.
I am here because I have lived the erosion and rebuilt from it. I have seen the cost of carrying too much and the relief that arrives when you finally stop disappearing inside your own competence.
So I share what I have learned in case it helps someone else. Take what is useful. Leave the rest. And remember this one truth that professionals like you often forget:
You are probably doing far better than you give yourself credit for.
Comments