Psychology’s Greatest Misses (Part 2/3)

An extended tour of your favorite wrong ideas.

In Part I of this series, we examined six popular psychological proposals that are either wrong or widely misunderstood: the 3-Cueing System, Attachment Styles, Stages of Grief, Power Posing, Stereotype Threat, and Transference. In Part II, we add six more misses and turn briefly to a deeper puzzle: why do so many people still believe these ideas? Why, once research has shown them to be false, do most people continue to believe them?

Learning Styles—JZ

You’ve probably heard that some people are “visual learners” while others are “auditory” or “kinesthetic.” Entire industries of educational consultants are built on this claim, despite the fact that study after study has shown it to be false: instruction matching a student’s preferred “learning style” does not improve learning outcomes. Worse, catering to these preferences can end up legitimizing avoidance, helping students sidestep the kind of challenges that are prerequisites to growth.

Consider a student who struggles with reading and therefore leans into an “auditory” learning style. As one paper puts it:

By 5th grade and beyond, an individual’s preference for auditory learning may reflect difficulty in learning to read…Thus, contrary to the learning style hypothesis, it may be particularly important to focus on strengthening reading skills.

In other words, a “preference” is often just a weakness in disguise. Not only have students been underserved by this confusion, but teachers have spent decades designing lesson plans that were, at best, a complete waste of time.

To the extent that differing presentations of material do improve learning, the content, not the learner, is probably the relevant factor. That is, diagrams help with spatial problems, sounds with tonal ones, words with verbal reasoning, and movement with athletic skills.¹ Basically, use the right tool for the job, and don’t worry about the right vibe for the learner. We could have lifted this advice from a general contractor; instead, we entrusted psychologists.

Implicit Bias—RK

When I was in graduate school, there were whispers of what would become a chorus of an obsession with issues to do with identity politics, especially race and racism. In the group of psychologists studying the vast complexity of human social behavior—from abortion attitudes to zoophilia—for some reason most of them came to feel as if the single most important topic they could set their scholarly sites on was how bigoted people are.

Because in the 90s most people were quick to assert that they weren’t at all racist, especially where I was, southern California, a technique designed to measure secret, hidden racism became wildly popular. It was known as the Implicit Association Task. The general idea was that a research subject might not say they were racist, but clever psychologists could tap deep into their psyche to see if they actually were.

Here’s how the task works. Suppose I ask you if you like Star Wars more than Star Trek and you say no, I like them both equally. I then have you do the same sort of task two times. I show you some words and your job is to press the correct button when you see the word. I tell you there are two categories, A and B. Category A consists of words that are either related to Star Wars or positive terms, like “good” or “happy.” Category B consists of words that are either related to Star Trek or negative terms, like “bad” or “sad.” The intuition is that if you associate Star Wars with goodness, then this task is relatively easy and you can press the correct button quickly. But then I reverse matters, pairing Star Wars with negative words and Star Trek with positive words. If you really—subconsciously, implicitly—associate Star Wars with goodness, this task will be slower because (good) Star Wars is mixed with negative (bad) words. We’re quick with lists that are somehow consistent with one another, but it takes us a beat to work on lists with mixed things.

But they didn’t use science fiction franchises, they used races.

The notion of implicit bias got everywhere, including, as Paul Bloom pointed out, a presidential debate. Unfortunately, although the task does show that most people do show a difference in speed on the task when you pair good or bad words with words to do with race, age, sex, and other categories, the conclusions researchers have drawn don’t hold up to further scrutiny. Gregory Mitchell and Phil Tetlock recently addressed the disconnect between what people believe and what the test shows, writing: “The attentive public widely believes a false proposition, namely, that the race Implicit Association Test (“IAT”) measures unconscious bias within individuals that causes discriminatory behavior.”

Read that piece for an excellent treatment, but broadly, here are some reasons to ignore the results of implicit association tasks. First, when people take the test multiple times, their scores vary… a lot. That variation implies it’s not a good measurement tool. Would you continue to use a ruler if it kept giving you different lengths for the same object? Second, even if some people have negative associations with a race, that might be because they believe that a person’s race can lead to negative treatment, not that members of a particular race have negative qualities. Third, the IAT is a reaction time measure, which almost certainly is not a good way to measure racism or discrimination because the associations people have with different kinds of people does not necessarily cause them to treat people in those categories any worse². Fourth, people are actually pretty good at predicting how their results will turn out, making the claim that the association is implicit suspicious.

My experience suggests that social psychologists, like many in the soft sciences and humanities, are happiest when they are piously accusing people of this or that flavor of bigotry. They seem to like it especially when they have a tool they can point to and say, oh, sure, you deny being racist, but I have reaction time data that say different!

They were less concerned, however, about whether that is, in fact, what the data say.

Trauma—JZ

The average person likely believes two things about trauma. First, that “the body keeps the score,” meaning traumatic experiences lodge themselves somewhere in our bones and tissues, our organs and nervous systems, and continue to wreak havoc in ways we often cannot perceive or control.³ Second, that the effects of trauma can be mitigated—perhaps even cured—by “processing” it, whether through conversation, bodily techniques, or both.

The first belief is false. The second is a bit more complicated.

To dispense with the first, I like the way Abigail Shrier puts it in Bad Therapy: “Does the body keep the score? Literally, no. Figuratively, also no.”⁴ Trauma does not seep into tissues or organs in any meaningful explanatory sense, nor does it lie dormant in the body awaiting somatic release. For readers who want more detail, here is a concise critique.

As for processing trauma, here are a few key points. First, although figures like Bessel van der Kolk, author of The Body Keeps the Score, helped popularize the idea that body-based treatments are uniquely effective, the evidence doesn’t support it. At their best, body-based interventions function as adjuncts—not replacements—for approaches with stronger empirical backing, such as cognitive behavioral therapy for trauma (CBT-T). Personally, I suspect that many body-based treatments, from Somatic Experiencing to Tension and Trauma Release, are 100% organic snake-oil.⁵

However, not all talk therapy is productive, either, and some approaches may even be harmful. A particularly instructive episode is Critical Incident Stress Debriefing (CISD), which emerged in the late 1970s in emergency services and first-responder settings. Proponents claimed that prompting people to recount and emotionally process traumatic events shortly after they occurred would reduce the likelihood of developing post-traumatic stress disorder (PTSD). The evidence told a different story: not only did CISD fail to prevent PTSD, but in some studies it made symptoms worse. As a result, many official guidelines now explicitly forbid its use.⁶

Truth be told, I find even the gold-standard trauma treatments—CBT-T and Eye Movement Desensitization and Reprocessing (EMDR)—somewhat disappointing. For one, I have yet to come across a problem that CBT cannot supposedly solve.⁷ It seems only a matter of time before it is recommended in lieu of an oil change or as an alternative to plastic. More importantly, CBT (like all talk therapy) is expensive, time-intensive, and inaccessible for many people, which makes it a poor candidate for universal prescription.

As for EMDR, it’s weird and nobody knows why it works. Seriously, watch this video and tell me with a straight face that it inspires confidence:

Amid the surge of interest in trauma, I wonder if there is a simpler way to think about it. Is there good reason to believe that trauma is any more complicated than learning, or that recovery from trauma is any more complicated than extinction?

Suppose Mary goes to a park late on a Wednesday night and is mugged by a teenager. In the months that follow, she develops symptoms of PTSD: intrusive memories, avoidance of parks and teenagers, negative beliefs about her neighborhood, diminished interest in seeing friends and family, and a state of physiological hypervigilance that disrupts her sleep.

Isn’t this suite of responses, uh, kind of adaptive? Hasn’t Mary learned something genuinely useful? Undergone an experience that she is desperately eager not to repeat? Indeed, in her effort not to repeat the experience, shouldn’t she overreact to details that may seem irrelevant—such as the fact that it was Wednesday night or that she was wearing sneakers—given that in matters of life and death, it is better to be safe than sorry?⁸

Now let’s ask what would cure Mary of her troubles. “Time” is not quite the right answer, but it’s close. The more precise answer is benign exposure, which time often provides. Mary can gradually unlearn her negative associations if she has repeated encounters with the same or similar cues—teenagers, darkness, her neighborhood—that do not end in disaster. With enough safe repetitions, the original learning weakens (i.e., becomes extinct).⁹ Of course, exposure only makes sense when a person’s perception of danger is wildly out of proportion to the actual danger. We should never advise Mary to take more midnight strolls in an area where there is a decent likelihood the mugging will happen again.

All in, if evolution had assigned YOU the task of designing a system that learns from danger, could you do better than one that overgeneralizes by default and relaxes only after ample evidence of safety? If so, let me know.

Facilitated Communication—RK

When I was in graduate school, I was a teaching assistant for Introductory Psychology several times when my advisor at the time, Leda Cosmides, taught the class. In addition to the two days of lecture, every week students met in smaller study sections.

The very first study section was always given over to having the students watch a Frontline episode that investigated a technique called “facilitated communication.” The idea was that people with symptoms of autism and other disorders had capabilities beyond what their limited capacity for spoken language could communicate. Facilitated communication was a technique in which a facilitator—a nurse, caregiver, or teacher—would gently touch the patient’s hand and the patient could, with this minor connection, point to letters to spell out words on a specially designed keyboard. People previously believed to have sharply circumscribed mental abilities were, using the technique, answering sophisticated questions and composing complex sentences, ripe with imagery and metaphor. (See this transcript.)

The episode, entitled Prisoners of Silence, takes a dark turn. Some patients begin to use their newfound communication skills to level allegations of sexual abuse and other harms. Arrests are made. Parents are separated from their children.

Eventually, a third party was brought in to run tests to see if facilitated communication was genuine. One Dr. Howard Shane devised a series of clever tests. He presented the patient with a picture and, at the same time, presented the facilitator with a different picture, but in a way that made it difficult for the facilitator to know that they were not seeing the same one that the patient saw. After the two images were shown, the patient was instructed to spell out what they saw. In every case, the patient spelled out what the facilitator had seen. This result shows that the facilitator was producing the words, the sentences… the accusations. Not the patient. (See also “recovered memories.”)

You would think—you really would—that would be that. The evidence showed that the technique doesn’t work.

In 2024, Netflix released the documentary, Tell Them You Love Me, the story of Anna Stubblefield who “learned,” through the use of facilitated communication with a young, developmentally disabled student, that he harbored amorous feelings for her. She acted on these protestations of love, which eventually landed her in jail for sexual assault because she was not, in fact, talking to the young student, but using him as a human Ouija board.

There is something boosting about the idea that someone with a disability has an active but untapped vivid mental life. But facilitated communication is a myth, and the cost of continuing to believe it has real, tangible consequences for everyone involved.

Personality Tests—JZ

The personality literature is an example of psychology getting something genuinely right, and then squandering it by overextending the finding.

The pride and joy of personality research is the Big Five or Five Factor model, which describes how people differ along five broad dimensions: Openness to Experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.¹⁰ Unlike many personality theories, these traits were not dreamed up in an armchair or inferred from clinical intuition—they emerged from decades of empirical work that asked a simple question: when people describe themselves and others, which descriptions tend to cluster together? Using large datasets and statistical techniques, researchers found that thousands of everyday personality descriptors reliably collapse into a small number of broad dimensions.

The Big Five is empirically grounded, cross-culturally robust, and modestly predictive. But its power is still limited in the following ways:

1. Descriptions, not explanations. Personality research largely describes behavioral patterns; it does not explain them. An “extravert,” for example, is someone who tends to go out on a typical Friday night. Why does this person go out? Because they’re extraverted. Rob and Jason Weeden coined a term for this kind of tautology in The Hidden Agenda of the Political Mind: DERPing—Direct Explanation Renaming Psychology.

2. Gradations, not categories. Personality traits exist on adaptive spectra, meaning people can fall anywhere along the range and remain evolutionarily viable. There is no natural cutoff at which someone becomes an extravert.¹¹ Yet personality research—and especially personality tests—train us to think in bins rather than continua, producing endless, content-free conversations of the form: “I’m sort of extraverted but, like, not completely. Sometimes I like time to myself.” Cool.

3. Modest predictability. Personality predicts behavior modestly because situations matter more. Whether someone goes out on a Friday night depends more on context than their level of extraversion. Are they single, exhausted, sick, stressed, broke, hungover, or new to town? Big Five research captures real patterns, just not at a level that reliably explains behavior in specific situations. At best, it adjusts probabilities. All else equal—which it never is—Patty goes out more than Sue.

So, Big Five research is limited, but genuinely useful. It’s a hit. The miss is how the concept of personality is popularly, or even professionally, used: either by overextending Big Five findings or, more egregiously, by relying on personality tests that distort or ignore them altogether.

At their best, personality tests attempt to operationalize Big Five traits. This includes the NEO or Big Five Inventory. At their worst, they’re just trash. The Myers–Briggs Type Indicator, for example, is based on Carl Jung’s book, Psychological Types. That is, a single theorist’s speculative framework from over a century ago.¹² Meanwhile, the Enneagram seems cobbled together from whatever theory, religion, or perspective came into the creator’s mind in the moment.

All this said, I think the primary harm of the personality concept lies in the mistaken beliefs—shared by laypeople and professionals alike—that 1) personality types exist and 2) that personality can reliably explain or predict behavior in any strong sense. The best research we have does something far more modest: it adjusts probabilities in a system dominated by context, incentives, and circumstance. Personality nudges the odds in one direction or another, but the equation is far too complex—and too situation-dependent—for types or traits to do the explanatory work people want them to.

Love Languages—JZ&RK

Just no.

Why Do Bad Ideas Travel So Far?

In this article, we’ve drawn on psychological research to undermine a set of popular hypotheses which, in most cases, psychological research spawned. So why isn’t the public up to date? Why is it that, as Jonathan Swift put it, “Falsehood flies, and the Truth comes limping after it”?

The staying power of false ideas raises two big and related questions. First, why do so many bad ideas emerge in the first place and, second, why do they spread? In our view, the answer to the first is incentives and the answer to the second lies in cultural epidemiology, the study of the forces that determine why some ideas spread while others don’t.

As discussed in It’s All Academic, academia incentivizes novelty. No one gets tenure for saying the way you pose doesn’t increase testosterone. Everyone already assumed that was the case, so that claim is dog bites man. The incentives in academia are to publish a new claim, with bonus points if the finding is counterintuitive. Because there are no real penalties for being even luminously wrong—Amy Cuddy is doing just fine, thank you—academics spew six zany ideas before breakfast. When they get rewarded for them—promotions, tenure, TED talks—they naturally do it again.

The second part, why some ideas spread more than others, is really complicated. In a sense, it’s the fundamental question about culture. And there are no shortage of proposals about how culture works. We obviously can’t work through all of these proposals, so we’ll just point to a few we think are especially relevant and add some of our favorite sources for more in this footnote.¹³

A main takeaway from cultural epidemiology is that people have preferences for certain kinds of explanations. For example, explanations that are counterintuitive—but not too counterintuitive, as Pascal Boyer has pointed out. You wouldn’t believe Amy Cuddy if she said that power posing gives you X-ray vision. Too far. But if she says it gives you a hormone boost… maybe?

Thomas Sowell thinks people want explanations that give us good guys to cheer for and bad guys to root against:

The reason so many people misunderstand so many issues is not that these issues are so complex, but that people do not want a factual or analytical explanation that leaves them emotionally unsatisfied. They want villains to hate and heroes to cheer - and they don’t want explanations that fail to give them that.

Social psychologists have been eager to provide villains. Do you know what harm people who ask you to put your race at the top of an exam are doing? <clutches pearls> This fact explains why so much work in that field is to do with the bad, bad way that people treat others based on their group membership. Stereotypes, prejudice, and discrimination are the red meat of this crew. They can’t get—or rather, give—enough of it.

There are plenty of other kinds of explanations humans prefer, including those that make a good story or are simple. It would be impossible to name them all.

Still, why doesn’t the truth override errors when new scholarship undermines the prior claims? That question is also complicated, but one possibility is that the truth doesn’t care about what kinds of explanations humans prefer. So, much of the time, the truth is boring. The fact that power posing doesn’t do much isn’t the kind of tidbit you want to share at the water cooler or make a short video about. Many facts that are true don’t make anyone or any group a hero or a villain. Some true things are counterintuitive, but many are not because humans have, in some areas, pretty good intuitions.

The truth, we might say, doesn’t care about what kinds of ideas people find interesting. But the human searching for truths very much does.

In this respect, at least, AI offers some hope. In the past, people had to “keep tabs” on what the latest research said, which hardly anyone did. Now, when there is an update to the research, AI will, from that point forward—cross our fingers—start spitting out the right answer. This ought to give researchers more hope that the playing field between Falsehood and Truth will be a bit more even.

1. None of this is to deny that presenting information in multiple formats can improve learning. Combining words with diagrams, for example, often leads to better retention than either alone. But this is not evidence for “learning styles.” It reflects well-understood principles of memory and encoding: redundancy, elaboration, and dual coding help most learners, not different styles for different people.

2. You might think that a particular group is good at, say, baking, but that doesn’t mean you’ll discriminate against them. You might be more likely to hire them if you need a cake.

3. Freud’s theory of repression—the idea that unspoken anguish will eventually find some distorted or displaced way to express itself—is a close ancestor of this view and still exerts considerable influence on popular and clinical thinking.

4. p.113.

5. I tried Tension and Trauma Release once. A practitioner in my building offered a few free sessions so that I could recommend it to clients if I found it helpful, which I did not. She was a very nice woman, but my strong impression was that the technique is bogus.

6. In the linked guideline for the UK’s National Institute for Health and Care Excellence, the language is explicit: “Do not offer psychologically-focused debriefing for the prevention or treatment of PTSD.”

7. For example, I learn from Jean Twenge’s book, The Impatient Woman’s Guide to Getting Pregnant, that in some of Dr. Alice Domar’s research, women who underwent CBT were more likely to get pregnant (p. 140). Give me a break.

8. It is also hard to know which details are relevant. This is ultimately a problem of induction, as W. V. O. Quine would put it. The world is complicated, and the only way to determine which variables are causal or predictive is to gather more data or to have a sound theory. Does Mary have either?

9. An emphasis on exposure—and systematic discouragement of avoidance—is the central mechanism of CBT-T and likely explains much of its effectiveness. The same may be true of EMDR, whose benefits may derive primarily from repeated, structured exposure rather than from any distinctive process of “reorganizing” memories by moving someone’s eyes around.

10. Some researchers favor a six-factor model (HEXACO) that adds Honesty–Humility to the Big Five.

11. By contrast, some human traits do have natural categories—for example, biological sex—where intermediate values are rare and usually reflect atypical developmental processes rather than normal variation.

12. And I really like Jung, too!

13. See especially Boyer’s book, The Naturalness of Religious Ideas. See this classic chapter by Dan Sperber and Richerson & Boyd’s book, Not by Genes Alone.