The boy was out after curfew. He had turned his sweater inside out to hide the yellow star, and he kept to the walls. A man in the black uniform of the SS waved him over. The boy went. There was nothing else to do.
The soldier picked him up and held him. He spoke in German, with feeling, and the boy understood none of it. Then the man set him down, opened his wallet, and showed him a photograph of a boy. He pressed some money into his hand and sent him on his way. Daniel Kahneman (March 5, 1934 to March 27, 2024) walked home along the dark street and thought his mother had been right. She had told him that people were complicated and worth the trouble of figuring out. A man who had been trained to kill him had instead missed his own son and given a Jewish child money on the street. Kahneman carried that lesson the rest of his life. People do not run on a single rule. The work was to find the rules anyway.
He was born in Tel Aviv during the British Mandate, on a visit, and grew up in Paris, where his father ran research for a cosmetics firm. The occupation arrived. The family hid, moved, and waited. The father was caught in a roundup and held at Drancy, then released when his employer intervened. He was a diabetic, and the war wore him down faster than the disease should have. He died in 1944, weeks before the area was freed. In 1948 the family went to the new state of Israel. Kahneman studied psychology and mathematics at the Hebrew University of Jerusalem and finished in 1954.
The army gave him his first laboratory. A second lieutenant, twenty-one years old, he drew the job of evaluating officer candidates. The standing method put recruits through a field test. They had to move a heavy log over a wall without letting it touch the ground or letting any man touch the wall. The assessors watched who took charge, who gave up, who pushed through. Then they ranked the men with confidence.
Kahneman checked the rankings against how the officers later performed in training. The link was near zero. The assessors had felt sure each time. They felt sure the next time too. He named the feeling later: the illusion of validity. Confidence is a feeling, and a feeling is not evidence.
He redesigned the interview. He broke the judgment into separate traits and made the interviewers score each one on a scale before they were allowed to form a general impression. They hated it. One of them told him the new forms stripped the work of everything they were good at. They could see the man in the room. A column of numbers could not.
“The numbers beat you,” Kahneman told them. “We tested it.”
The structured method outpredicted the seasoned eye. The lesson held for the rest of his career. A dull formula, applied the same way every time, often beats a confident expert. Years later he would say he never trusted his own first impression of a person without checking it.
He took his doctorate at Berkeley in 1961 and returned to teach in Jerusalem. His early research had nothing to do with economics. He studied perception, attention, and the cost of mental effort. He measured pupils. The pupil widens as a task grows harder, and he used it to track the load on the mind in real time. He worked from a single idea about the mind. Attention is scarce. The mind spends it carefully, and to spend it carefully the mind takes shortcuts.
Then Amos Tversky (1937-1996) came to give a talk.
Tversky was a star, quick and certain, a logician’s logician. He presented work that treated people as decent intuitive statisticians who updated their beliefs in roughly the right direction. Kahneman, asked to comment, said the premise was wrong. He had watched trained psychologists botch the simplest questions about sample size. People were not conservative Bayesians. People were systematic, and systematically off.
The two of them began to argue, and the argument did not stop for fourteen years. They shared an office and a single mind. They wrote each sentence together, out loud, fighting over every word and laughing until they cried. On their first joint paper they flipped a coin to decide whose name came first. Tversky won the toss. After that they alternated.
The first paper carried a title that reads now like a prophecy. They called it “Belief in the Law of Small Numbers.” Researchers, they showed, treat a small sample as if it carried the same authority as a large one. A scientist runs twenty subjects, finds an effect, and trusts it as though the number twenty meant something. The number twenty means very little. A small sample is mostly noise wearing the costume of a result. The statistician Andrew Gelman (b. 1965) has made the point that this paper anticipated the replication crisis by forty years, and that the field would have spared itself a great deal of grief had it read the paper closely. The same man who wrote it would later forget it.
The collaboration produced the program that made their name. People judge probability by resemblance, so a quiet man who likes poetry seems more likely to be a librarian than a farmer, never mind that farmers outnumber librarians many times over. People judge frequency by what comes to mind, so they fear the rare death that makes the news and shrug at the common one that does not. People anchor on whatever number they heard last, even a number they know to be random, and adjust too little from it. Kahneman and Tversky gave these errors names and clean demonstrations, and the demonstrations were hard to argue with because the reader fell for them in real time.
The most famous was Linda. The subjects read a sketch of a woman, single, outspoken, concerned with discrimination and social justice. Then they ranked the odds of several statements. Most ranked “Linda is a bank teller and active in the feminist movement” as more likely than “Linda is a bank teller.” The conjunction of two things cannot be more probable than one of them alone. People said it was anyway, because the fuller story fit the picture.
Not everyone read the result the way Kahneman did. Gerd Gigerenzer (b. 1947) argued that the errors thinned out when you changed the wording. Ask the question in plain frequencies, how many of a hundred women like Linda are bank tellers, and many people who failed the first version pass the second. Gigerenzer held that the shortcuts Kahneman called biases were often fast and frugal rules that served people well in the world they evolved for, and that stripping a question of its context and then scoring the answer against a textbook rule stacked the deck. The exchange ran for years. The press called it the rationality wars. Gigerenzer disliked the word war. He had tried, he said, to keep the disagreement sharp and the personal respect intact, which is harder than it sounds.
There is also the question of who got there first. Herbert Simon (1916-2001) had described much of this in the 1940s and called it satisficing. People do not optimize; they search until they find something good enough, because they lack the time and the horsepower to do more. Simon took the economics prize in 1978 for the idea. Pierre-Simon Laplace (1749-1827) had cataloged cognitive illusions early in the nineteenth century, though he blamed them on hope and fear rather than on flaws in the machinery, and thought education could clear them away. None of this erases what Kahneman and Tversky did. It places them in a line.
Their theoretical peak came in 1979. Prospect theory took apart the standard economic account of choice. The textbook said people weigh outcomes by expected utility. Kahneman and Tversky showed that people judge from a reference point, that they feel a loss about twice as hard as they feel the matching gain, and that they overweight small chances and underweight near-certainties. The theory explained why a man buys a lottery ticket and an insurance policy in the same afternoon, and why he holds a losing stock long past the point of sense rather than admit the loss by selling. It became one of the most cited papers in economics. In 2002 Kahneman took the Nobel Memorial Prize in Economic Sciences, shared with the experimental economist Vernon Smith (b. 1927). Tversky had died of melanoma in 1996. The prize does not go to the dead. Kahneman said often that the honor was half stolen.
His point was never that people are fools. The crude version of Kahneman holds that human judgment is a junk heap. The careful version holds that intuition is powerful and conditional. It works in a stable world with quick, clear feedback. A firefighter reading a burning building and a nurse reading a sick infant have earned their hunches, because the world corrects them fast and often. A stock picker and a political forecaster have not, because the feedback is slow, noisy, and easy to explain away. He worked this out in an adversarial collaboration with Gary Klein (b. 1944), a researcher who had spent his career defending expert intuition. The two men disagreed on purpose, in print, and found the line that separated the cases where each was right.
He kept adding to the catalog. The planning fallacy, where every project runs late and over budget because we plan from the best case and forget the record. Hindsight bias, where the moment a thing happens we feel we knew it all along. The peak-end rule, where the memory of an ordeal hangs on its worst moment and its final one, not on how long it lasted. And the split he prized most, between the experiencing self that lives through each minute and the remembering self that files the report and runs the next decision. The two selves want different things, and the remembering self wins, because the remembering self is the one who chooses.
In 1993 he moved to Princeton University. In 2011 he published Thinking, Fast and Slow and reached a readership most scientists never touch. He organized the work around two characters. System 1 is fast, automatic, and sure of itself. System 2 is slow, effortful, and lazy, and it signs off on whatever System 1 hands it. The frame was a simplification and he said so. It gave millions of readers a way to talk about their own minds. He dedicated the book to Tversky.
The book also held a flaw that would mark his last years.
One chapter leaned hard on social priming, the claim that small cues steer behavior below the level of awareness. The signature study had subjects read words about old age and then walk more slowly down the hall. Kahneman told his readers that disbelief was not an option. The studies were sound, he wrote, and the reader had no choice but to accept them.
The reader did have a choice. The priming results came from small samples run through flexible analyses, the precise trap Kahneman and Tversky had described in 1971. The effects were tiny when they appeared at all, and often they did not appear. At a 2014 meeting the psychologist Hal Pashler laid out a long string of failed replications. Disbelief, it turned out, remained an option. Andrew Gelman put Kahneman beside Alan Turing (1912-1954), who in 1950 had called the statistical evidence for telepathy overwhelming. Two brilliant men, each insisting that the rest of us had no choice but to believe, each fooled by numbers that could not bear the weight. When a careless man overstates his evidence, Gelman noted, no one learns anything. When a careful man does it, the error is worth studying.
To his credit, Kahneman did not dig in. In 2012 he sent an open letter to the social priming researchers and warned them that a train wreck was coming unless they cleaned up their methods and replicated their headline findings. He later admitted he had placed too much faith in underpowered studies and should have known better, given what he himself had taught. One researcher who worked with him recalled asking why the message of the field had become a list of human defects. Kahneman answered that the defects were never the point. People simply liked hearing about them.
Gelman’s skepticism ran wider than priming. He has argued for years that the curved utility function economists reach for to explain caution does not in fact explain it, that a man who prefers a sure thirty dollars to a coin flip between twenty and forty is not revealing a bend in his utility for money but a dislike of the gamble, and that the profession keeps confusing the two. Whatever one makes of that fight, it points at the same soft spot. The shortcut that lets you model a person can quietly replace the person.
His last book pressed on a different error. Noise, written in 2021 with Olivier Sibony and Cass Sunstein (b. 1954), drew the line between bias and noise. Bias is the systematic miss, the scale that always reads three pounds heavy. Noise is the scatter, the scale that reads a different weight each time you step on it. Two judges hand down different sentences for the same crime. Two doctors read the same scan and disagree. Two underwriters price the same risk miles apart. Institutions obsess over bias, he argued, and ignore the scatter, though the scatter does as much damage. His fixes were the old fixes in new clothes. Break the judgment into parts. Score the parts before you form the whole. Use the dull, reliable formula. He knew the danger in his own prescription. A rigid system handles the standard case and fails the case the data never saw, and the cases that never saw the data are the ones that ruin people.
He married twice. His first wife was the psychologist Irah Kahneman, and they had two children. In 1978 he married Anne Treisman (1935-2018), a major figure in the study of attention, and they stayed together until her death. In his last years his companion was Barbara Tversky, Amos’s widow, a psychologist in her own right. The circle closed where it had opened.
He died on March 27, 2024, at ninety. The plain facts came out the next year. He had gone to Switzerland and ended his life by his own arrangement. He was not dying. He felt the edges of his mind beginning to soften and he chose to leave while the man making the choice was still the man he recognized. He had spent his life on the two selves, the one who lives the minutes and the one who remembers them and decides. At the end he handed the decision to the remembering self and let it write the last line.
There is a temptation to file Kahneman as the scientist who proved that people fool themselves, and to leave it there. The fuller story is better. He spent sixty years mapping the ways a confident mind goes wrong, and then he went wrong in one of the exact ways he had mapped, and then he said so in public. The map was real. The mapmaker walked into his own terrain and got lost like everyone else. What he did next, the warning letter, the admission, the refusal to defend a result because his name was on it, is the part worth keeping. He taught that confidence is a feeling and not a proof. He proved it last on himself.
If John J. Mearsheimer’s anthropology is right, the groundbreaking work of psychologist and Nobel laureate Daniel Kahneman (1934–2024) provides the precise cognitive map for why liberalism fails. Kahneman’s lifework, summarized in Thinking, Fast and Slow, dismantled the classical economic myth of Homo economicus — the rational, atomistic actor who processes information flawlessly to maximize personal utility.
Kahneman proved that human decision-making is dominated by System 1: an automatic, fast, and unconscious mode of thinking driven by heuristics, biases, and emotional shortcuts. System 2, our slow, deliberate, and logical reasoning capacity, is lazy and often acts merely to rationalize the snap judgments already made by System 1.
In a liberal framework, Kahneman’s insights are used as a tool for technocratic optimization. Liberal policy designers use behavioral economics to “nudge” individual actors toward better choices, operating on the assumption that if we can just correct for these cognitive blind spots, the marketplace of ideas and free commerce will function smoothly.
Mearsheimer’s anthropology reveals that Kahneman did not discover a set of random individual software glitches. He documented the evolutionary engine of tribal survival.
Mearsheimer argues that reason is the least important way we determine our preferences, ranking far behind intense childhood socialization and innate sentiments. Kahneman’s System 1 is the cognitive mechanism through which that socialization operates. The heuristics Kahneman identified—like availability bias, loss aversion, and in-group favoritism—are not design flaws. They are defensive mechanisms engineered to keep the individual tightly bound to his tribe.Consider Kahneman’s concept of loss aversion: the psychological reality that the pain of losing something is twice as powerful as the pleasure of gaining it. In a liberal economic model, this is an irrational glitch in investment strategy.
In Mearsheimer’s realist framework, loss aversion is the foundational logic of group survival. Tribes prioritize security, territory defense, and sovereignty above all else because the loss of those assets means destruction.
Furthermore, Kahneman’s tracking of how easily the human mind is manipulated by framing effects confirms Mearsheimer’s view of public discourse. Liberalism relies on the belief that a free society engages in rational debate over universal rights. Kahneman showed that human preferences are highly malleable based on how information is presented, and that System 2 simply rubber-stamps intuitive reactions. Mearsheimer argues that political leaders use this exact psychological reality to mobilize their populations, utilizing fear and tribal narratives to override abstract reason and justify aggressive state actions.
If Mearsheimer is right, Kahneman’s behavioral science is a direct assault on the philosophical foundations of liberalism. Kahneman provided the empirical evidence that humans are not wired for the atomistic, rational individualism required by liberal theory. By proving that human reason is subordinate to fast, intuitive, and contextual survival responses, Kahneman showed why the universalist dreams of liberalism inevitably collapse when they collide with the tribal nature of man.
If David Pinsof is right, the monumental career of Daniel Kahneman (1934–2024) presents a fascinating irony: he became the intellectual patriarch of the “misunderstanding myth” while personally demonstrating the exact self-serving, strategic logic that Pinsof describes.
Kahneman’s pathbreaking work with Amos Tversky (1937–1996), popularized in his book Thinking, Fast and Slow, established the field of behavioral economics by cataloging dozens of cognitive biases and heuristics. To the global intellectual class, Kahneman provided the definitive proof that human beings are fundamentally broken, irrational animals. This research became the foundation for the rationality movement, behavioral “nudges,” and endless policy interventions designed to fix public stupidity and curb cognitive errors.
A Pinsofian analysis strips away this high-status academic mission statement and points directly to Kahneman’s own candid admissions. Pinsof notes that Kahneman himself acknowledged in interviews that learning about cognitive biases did not improve his own behavior or make him more “rational” in his daily life. From a Pinsofian view, this lack of motivation makes perfect sense. Kahneman’s brain subconsciously understood that these so-called biases—such as loss aversion, overconfidence, and confirmation bias—are actually savvy, evolved heuristics designed to help human animals win arguments, secure resources, and protect alliances in a hostile social marketplace. Becoming perfectly unbiased would be a Darwinian disaster.
The ultimate irony, through a Pinsofian lens, is how the intellectual class weaponized Kahneman’s research. By transforming Kahneman’s catalog of heuristics into a list of 265 “cognitive errors,” social scientists and elite professionals created a powerful tool to claim moral and intellectual superiority over the masses. Framing the public’s political choices, financial decisions, and social behaviors as mere “brain-farts” and “systematic biases” allowed the academic elite to position themselves as the necessary technicians who must nudge, educate, and govern a confused populace.
Kahneman did not uncover a tragic flaw in human rationality; he mapped the highly optimized, self-serving architecture of the human mind. While his work was used by others to build the ultimate secular church of “saving the world from misunderstanding,” the raw success of his paradigm demonstrates the exact competitive logic Pinsof outlines. His brilliant insights functioned as the highest-status currency in modern academia, securing him a Nobel Prize, immense wealth, and unparalleled institutional dominance within the global intellectual hierarchy.
Kahneman and the Tacit
The interviewers could see the man in the room. That was the whole of their claim. They watched a recruit take command of the log or fail to, and they felt they knew him. The feeling ran strong and the feeling ran wrong. When Kahneman checked their confident rankings against how the officers later performed, he found almost nothing. These men had a real skill at reading other men. They could not say what they read, and what they read did not predict. The skill was silent and the silence was the point.
Michael Polanyi (1891-1976) gave the thing its name. We know more than we can tell. The surgeon’s hands know the resistance of tissue. The chess master takes in the board at a glance and the good move presents itself. The native speaker hears the wrong preposition without consulting a rule. In each case the knowing outruns any account the knower can give of it. Kahneman built a life’s work in this gap. System 1 is the tacit fitted with a name and a diagram. The heuristics are the tacit caught in the act. His subject was the part of the mind that arrives at an answer before the owner of the mind can say how.
Stephen P. Turner (b. 1951) has spent decades pressing on this subject, and his pressure falls on the place where Kahneman stands. In The Social Theory of Practices and again in Understanding the Tacit, Turner grants that the tacit is real and then asks what it can be made to explain. His answer is: less than people want. The trouble starts when the tacit stops being one man’s silent skill and becomes a shared possession, a hidden content that many heads are said to carry in common. Turner calls that move unearned. You see two men behave alike. You infer a shared rule beneath the behavior, sitting in both heads, the same in each. But the rule is tacit, so no one has seen it, in either head, and no one can state it for comparison. The likeness of the behavior is the only evidence, and the likeness of the behavior is also the whole of what the shared rule was invented to explain. The explanation and the thing explained are the same observation wearing two hats.
Turner adds a second problem, the problem of transmission. If the content is tacit, never stated, how does it pass from one man to the next? It cannot pass as content, because content that can be handed over is not tacit. Each man acquires his habits through his own history of trial and correction. The habits of separate men come to resemble one another because they meet the same world and get filed down by it in similar ways. There is no underground delivery of a shared rulebook. There are only individual dispositions that happen to mesh.
Now turn this on Kahneman. The representativeness heuristic, the availability heuristic, anchoring. He writes them as if each names a fixed thing carried in common by all the subjects, a rule in every head that produces the error on cue. Turner’s question lands here with full weight. What licenses the leap from “most subjects gave the same wrong answer” to “all these subjects ran the same hidden rule”? The Linda problem shows many people ranking the fuller story above the plainer one. It does not show that one rule, identical across all those minds, drove the ranking. Many separate paths can reach a convergent answer. The shared heuristic is Kahneman’s inference, not his finding, and it is the inference Turner has argued against his whole career.
Kahneman’s project is to take the tacit and make it speak. He hands the reader the rule the silent mind was supposedly following. Here Turner’s sharpest point applies. When you articulate the tacit, the words you produce are a reconstruction, a model built after the fact, and the model is a different object from the silent process it claims to capture. The expert who confabulates a reason for his hunch produces such a reconstruction. Turner’s move is to ask why the theorist’s reconstruction should stand any higher than the expert’s. Kahneman believed he could state the true rule behind the confabulation, the actual heuristic under the false reason. Turner would say the heuristic is his articulation too, a story the theorist tells about a process he cannot see any more directly than the subject can. The map is not the silent territory, and Kahneman drew the map.
Kahneman half conceded this. He called System 1 and System 2 a simplification. He said the heuristics worked as if, that the mind behaves as though it followed them. As if is the language of a man who knows he is modeling something he cannot open up and read. Turner only takes the as if seriously and asks what is left once you stop mistaking the model for the contents of the head. What is left is behavior, regular and predictable, and a theorist’s vocabulary laid over it.
The strongest evidence for Turner’s reading comes from Kahneman’s own opponent. Gerd Gigerenzer (b. 1947) showed that the errors thin out when the question changes format, that people who fail the probability version pass the frequency version. Read through the tacit, that result says the answer was never a stable content sitting in the mind, portable across forms. The answer was a response to the form, bound to the task and the wording, the way a craftsman’s skill is bound to his tools and his shop. Change the setting and the silent disposition does something else. A fixed rule in the head would travel. This one did not travel. It belonged to the situation as much as to the man.
Kahneman came closest to Turner’s terrain in the work he did with Gary Klein (b. 1944). The two men set out to learn when a hunch is real knowledge and when it is the warm feeling the interviewers had. They settled on a condition. Trust the tacit where the world is regular and the feedback is fast and honest. The firefighter and the nurse earn their silent skill because the building and the infant correct them within the hour. The stock picker and the pundit do not, because the feedback is slow, noisy, and easy to talk away. This is a fine practical finding and it sits oddly beside the heuristics program. The heuristics treat the tacit as a uniform engine of error, the same in everyone. The Klein work treats the tacit as something local, trained, sometimes sound and sometimes empty, that has to be judged case by case against the world that shaped it. The second picture is Turner’s picture. A man’s silent competence is the residue of his particular history, and you cannot grade it in the abstract, only against the patch of world it grew in.
Kahneman trusted the social priming studies. He told his readers that disbelief was not an option. That judgment was itself a tacit one, an expert’s feel for which results ring true, the same kind of silent verdict the interviewers reached about their recruits. It came from his long immersion in a field, and it felt like knowledge, and it was wrong. The studies were thin and they did not replicate. The point a reader takes from Turner is not that Kahneman was careless. The point is that the tacit is not a rulebook a man can open and audit, not even when the man is Kahneman and the rulebook is his own. His skill ran ahead of his account of it, in the failure as in the triumph. He could catalog the silent mind in others and reconstruct its rules in a book. He could not stand outside his own silent mind and check it. No one can. That is what makes the tacit tacit, and it is why the great cartographer of intuition walked into a hole he had himself drawn on the map and did not see it until he had fallen in.
Andrew Gelman wrote Dec. 3, 2025:
Daniel Kahneman and Amos Tversky’s joint papers from the 1970s and 1980s . . . turned statistical thinking–previously a niche interest–into a major research focus. . . . In their joint work, known as the heuristics-and-biases program, Kahneman and Tversky argued that human judgment systematically deviates from the norms of probability and logic, resulting in predictable cognitive biases. These biases were attributed to heuristics–mental shortcuts–which led to a broader narrative in behavioral economics and psychology that emphasized human fallibility in decision-making. . . .
In his article, Gigerenzer offers an interesting perspective. One thing I like is that he talks about the science and the sociology of science, both of which are important, and he treats Tversky and Kahneman as human beings rather than as abstract heroes (as here, for example).
I had four thoughts in reaction to Gigerenzer’s article.
1. Staying out of war
I appreciate that Gigerenzer pushes against the metaphor of scientific debate as war:
The intellectual exchange between Kahneman & Tversky and myself has been dubbed the “rationality wars.” I am not partial to the term war, given that I tried hard to separate scientific disagreements and personal respect. It is easy to contest someone’s ideas if you dislike the person, but it is emotionally demanding to disagree with mutual respect. It was not always easy for either side, but Kahneman and I both did our best.
I’d put it in even stronger terms. Sometimes I’ve had scientific disputes with people who I think have behaved very badly and whom I strongly dislike–but I still don’t think such disputes should have the flavor of “war.” Even with a scientist who is doing misguided work and who is also a bad person, it’s rare for there to be a negative-sum “war” scenario, and I think we should do our best to avoid such settings. I wrote something about this once, entitled There is a war between the ones who say there is a war, and the ones who say there isn’t.
2. Don’t forget Laplace
Just a reminder that many of the ideas of cognitive illusions, heuristics and biases were in a book by Laplace from the early 1800s; see here. In his article, Gigerenzer mentions Laplace in the context of the gambler’s fallacy and the idea of probabilistic reasoning being a form of logic. But there’s a lot more to the Laplace book than that! I say this not to disparage the important work of Tversky and Kahneman, but just to trace the history of these ideas. Among other things, Laplace formulated the concept of cognitive illusions.
3. Economist are confused about rationality and psychology
As Gigerenzer points out, one of Kahenman’s important contributions was to bring some modern ideas of cognitive psychology and decision analysis to the attention of the field of economics. But economists remain very confused about concepts of rationality and psychology. One place I’ve seen this is in the very misused term, “risk aversion”: see here or, for more links, here.
Given this persistent misunderstanding on the part of nearly all of the economics profession, one thing I have always appreciated about Kahneman and Tversky is that they pushed against naive interpretations in which probabilities and utilities can be directly deduced from decisions. Even if, Gigerenzer argues, they went too far in many cases, I think they moved the discussion in the right direction.4. The progress of Tversky and Kahneman’s early research
Gigerenzer talks about Tversky and Kahneman’s experimentation using simple questions (Linda, etc.) rather than randomization devices. I’m not sure on this, but it’s my impression that there was an intermediate step, that before asking questions about Linda etc., Tversky and Kahneman were asking questions of psychology researchers about sample size and replication. For example see the studies describes in pages 107-109 of their Belief in Small Numbers paper. They anticipated a lot of the concerns of the replication crisis, which is ironic given Kahneman’s later dive into poorly supported pop-psychology of the Ariely variety.
My impression, without studying the literature carefully, is that Tversky and Kahneman started with the then-surprising realization that professional psychology researchers had systematic confusions about probability and uncertainty, and then they moved to probability reasoning questions that were not tied to psychology…
Andrew Gelman wrote May 23, 2021:
By now, we’re all familiar with the three modes of thought. From wikipedia:
System 1 is fast, instinctive and emotional.
System 2 is slower, more deliberative, and more logical.
System 3 is when you say things that sound good but make no sense.
System 3 can get activated when you trust what someone tells you rather than figuring it out yourself.
I thought about this after someone pointed out this post by Rachael Meager, who pointed out this erroneous claim in the new book, Noise, by Daniel Kahneman, Olivier Sibony, and Cass Sunstein.
We must, however, remember that while correlation does not imply causation, causation does imply correlation. Where there is a causal link, we should find a correlation. If you find no correlation between age and shoe size among adults, then you can safely conclude that after the end of adolescence, age does not make feet grow larger and that you have to look elsewhere for the causes of differences in shoe size. In short, wherever there is causality, there is correlation.
As Rachael points out, “this is not a case of experts simplifying a claim for a lay audience. This claim is just outright incorrect.”
It’s an interesting formulation when someone says, “We must remember X,” where X is a false statement. What is it exactly that we’re supposed to remember??
Rachael gives an example where there is causation but no correlation: “Imagine driving a car, reaching a hill and pumping the gas as you begin to go up so that your speed is constant. The correlation between pressing on the gas and the speed of the car is zero but they’re obviously causally related, it’s that the agent is optimizing speed!”
Strictly speaking, if your speed is constant, the correlation is not zero, it’s undefined. But, once you allow the speed to vary, you can get the correlation between speed and the position of the accelerator pedal to be positive, negative, or zero, even though in all cases pushing the accelerator makes the car go faster.
You can also get causation without correlation from a non-monotonic relationship or from plain old selection bias. So let me just emphasize that Rachael’s example is fine and there are a zillion others too. Causation and correlation are different things; it’s just not true to say that one implies the other.
“Why did we think they could get that one right?”
The question is, how could the authors of this book have made such a clear mistake?
To answer this question, we can turn to Cass Sunstein, one of the authors, who in an interview about the book says:
When a forecaster is wrong, we think, “Why did they make that mistake?” The better question is: “Why did we think they could get that one right?”
Well put.
The authors of this new book are a psychologist, a law professor, and some dude who describes himself as “a professor, writer and keynote speaker specializing in the quality of strategic thinking and the design of decision processes.” Between them, there’s no reason to think they’d have any particular expertise in correlation, causation, or statistics. You might as well ask me to have an opinion on the non-accelerating inflation rate of unemployment or the theory of operant conditioning. If I were to write a book and include categorical statements about such things, I’d check with the experts first. The relevant skill for Kahneman here was not to be an expert on statistics or econometrics but rather to realize that his coauthors are not experts either. A Washington Post reviewer called the authors an “all-star team,” but you wouldn’t want a baseball all-star team to play basketball (unless it included, I dunno, Jackie Robinson, Michael Jordan, and Danny Ainge), and I don’t know that you’d want a psychology/biz-school/law-school all-star team to be playing statistics. Again, though, maybe this is part of the problem. These guys get too much deference, more than is good for you. In sports, you ultimately have to face the music. In celebrity academia, once you’re high enough in the stratosphere, you can stay afloat forever…
So I think the answer to Sunstein’s question, “Why did we think they could get that one right?”, is that, like that famously well-dressed emperor, they were surrounded by yes-men. And remember that Sunstein’s earlier reaction to being questioned was to liken the skeptics to the former East German secret police. Take someone who gets too much positive feedback, and who actively resists negative feedback, and that’s a recipe for overconfidence, which is, ironically, one of the biases that Kahneman discussed in his earlier book.
The chain of trust
We discussed this general issue a few years ago in the context of the unstable mix of skepticism and trust that was characteristic of the Freakonomics franchise. The skepticism came because one of the main themes of Freakonomics was how everything you thought was right, was wrong. Drunk walking is worse than drunk driving, global cooling rather than global warming, etc. The trust came because, after their first book, which was mostly based on author Levitt’s research, the Freaknomics franchise pretty much ran out of original research and was reduce to promoting the work of Levitt’s friends and various randos on the internet.
Something similar seems to have happened with Kahenman. His first book was all about his own research, which in turn was full of skepticism for simple models of human cognition and decision making. But he left it all on the table in that book, so now he’s writing about other people’s work, which requires trusting in his coauthors. I think some of that trust was misplaced.
The question then arises, how is it that luminaries such as Philip Tetlock, Max Bazerman, Robert Cialdini, Rita McGrath, Annie Duke, Angela Duckworth, Adam Grant, Jonathan Haidt, Steven Levitt, and Esther Duflo thought this book was so brilliant, essential, masterful, eye-opening, important, etc.
Start with the gap. A subject reads about Linda and ranks the feminist bank teller above the bank teller. The conjunction of two claims cannot beat one of them alone. The subject has broken a rule of probability, and Kahneman records the break as a bias, a defect in the human machine. The whole heuristics-and-biases program runs on gaps of this shape. There is a norm, the answer the subject gave, and the distance between them is the finding. Pull on the norm and the program shifts under your feet.
Stephen P. Turner has spent his career pulling on norms. His anti-normativism is not the claim that people lack standards or that anything goes. It is a claim about what a norm can be made to do in an explanation. When a theorist says a subject’s answer is wrong, the theorist has imported a standard and granted it the authority to sit in judgment. Turner asks where that authority comes from and what work it does. His answer runs against the grain. The norm explains nothing. It is a label the analyst lays over the behavior after the behavior is in, and the label carries the analyst’s commitments, not the subject’s. Strip the label and you still have everything you actually observed. You have lost only the verdict.
Hold that against the conjunction error. Kahneman treats the probability axioms as the standard the answer should have met. But who appointed the axioms judge over a sentence about Linda? The subject was handed a paragraph rich with meaning, a person sketched in enough detail to invite a reading, and asked to rank statements about her. The subject did what people do with a person. He read her. The axioms of probability are one tool for one kind of question, and the subject was answering a different question, the one the paragraph posed. Turner’s point is that calling the answer wrong requires you to insist that the probability question was the real question all along, and that insistence is the analyst’s, not the world’s. The norm did not come up out of the data. Kahneman brought it with him and set it down on top.
This is the move Turner distrusts most, the move from is to ought smuggled in as description. Kahneman presents the biases as facts about the mind. They are facts about the mind measured against a rule, and the rule is doing quiet normative work the whole time. Take it away and the facts change their character. The subject did not fail. The subject responded, in a regular and predictable way, to the material in front of him. Regular and predictable is the most a science of behavior can ask for. The failure is an addition, and the addition is a value judgment dressed as a measurement.
Turner presses further on where the standard lives. For a norm to explain why the subject erred, the norm has to be in force for that subject, binding on him, present in his situation. The probability axioms are binding in a seminar room, among people who have agreed to be bound by them and trained to feel their pull. The subject in the experiment never entered that agreement. He brought the standards of ordinary reading and ordinary talk, where the fuller, more vivid description of a person is the more informative one and the cooperative listener treats it as such. By the standards actually in force for him, his answer was sound. Kahneman judged him by a standard in force somewhere else, in the logician’s room, and reported the mismatch as a flaw in the man rather than a clash of two settings with two different rules. Turner’s anti-normativism is the refusal to let one room’s standard travel into another room and keep its authority on arrival.
The same blade cuts the anchoring work and the availability work. A man told a high number gives a high estimate. Called a bias, against the norm of an estimate uncontaminated by the irrelevant figure. But the norm of the contamination-free estimate is itself a posit. In a world where the numbers people say to you usually carry information, leaning on the number you just heard is not a defect. It is a reading of the ordinary case. Kahneman strips the number of its usual informativeness, in the lab, and then faults the subject for treating it the way the world has taught him to treat such numbers. The defect appears only once the analyst has decided which features of the situation count and which do not, and that decision is a normative one wearing the coat of a control condition.
Kahneman did not merely describe the norms. He endorsed them. The arc of the work bends toward correction. Learn the biases, install System 2, debias the judgment, improve the decision. That program assumes the norm is the right standard and the human answer the thing to be fixed. Turner’s question is blunt. By what authority does the theorist crown the textbook rule the goal of human reasoning? The rule earns its keep in narrow settings, in a casino, in an actuarial table, where the world has been made to match the axioms. Outside those settings the rule is one option among several, with no standing to demand obedience. Kahneman wrote as if the rational ideal were fixed and human nature the deviation. Turner reverses the load. The ideal is the artifact. Human judgment is the baseline, and the question worth asking is not how far people fall from the norm but how the norm got built, who built it, and why anyone should answer to it.
Gerd Gigerenzer (b. 1947) supplied the evidence, though Turner would use it for a colder purpose than Gigerenzer did. Gigerenzer showed that the errors shrink when the question is posed in frequencies instead of single-case probabilities. Read through the norm, that result is not a tweak. It is a confession. It shows the standard was never neutral. A different but equally defensible framing of the same situation produces a different verdict on the same subject, which means the verdict was tracking the framing, the analyst’s choice of standard, and not a stable flaw in the head. When the norm moves, the bias moves with it. A property of the man would hold still while you changed the words. This one did not hold still. So it was never a property of the man. It was a property of the comparison, and the comparison belonged to Kahneman.
Kahneman judged the social priming studies sound and told his readers that disbelief was not an option. He was applying a norm there too, a working scientist’s sense of which results meet the bar, which evidence counts, what a real effect looks like. The studies did not replicate. The norm he applied, his felt standard for sound evidence, returned the wrong verdict in his own hands. This is the heart of the anti-normative reading and it spares no one, least of all the analyst. There is no view from above the norms, no neutral perch from which the theorist grades the subject and stays ungraded himself. Kahneman stood inside a set of standards while he measured everyone else against them, and his standards, applied to a real case, failed the way standards do. He had spent a career scoring human answers against a fixed rule and calling the distance a defect. The priming episode is the rule turning to face him. Measured against the outcome, his expert judgment was the deviation. There was no higher norm waiting to certify that he, unlike his subjects, had gotten it right. There never is. That absence is the whole of Turner’s case, and Kahneman lived it out without ever conceding the point in those terms.