Standard mental health tests may be inaccurate for highly intelligent people
Psychologists and the public alike have frequently debated whether exceptional cognitive ability comes with a cost to emotional well-being. A new analysis suggests that we may not be able to answer this question because the standard tools used to diagnose depression and distress may become inaccurate when applied to highly intelligent people.
The researchers found that as intelligence scores rise, the questions on common mental health surveys lose their ability to consistently measure the underlying psychological condition. These results were published in the journal Intelligence.
The concept of the “tortured genius” is a cultural staple. It suggests that high intelligence is accompanied by social isolation, existential anxiety, or other psychological difficulties. Previous research on this topic has produced conflicting results. Many large-scale studies indicate that intelligence generally correlates with better health and happiness.
However, other researchers argue that this relationship might not be a simple straight line. They propose a “nonlinear” relationship. This means intelligence could be protective up to a certain point, but extremely high levels might eventually lead to negative outcomes. This phenomenon is sometimes called the “too-much-of-a-good-thing” effect.
Stanisław K. Czerwiński, Roman Konarski, and Paweł A. Atroszko from the University of Gdańsk in Poland conducted this investigation. They sought to clarify these contradictions by using advanced statistical modeling. They wanted to see if the link between intelligence and mental health changes shape at the high end of the IQ spectrum. They also aimed to investigate whether this dynamic differs between men and women.
Social environments often react differently to high intelligence depending on a person’s gender. The authors hypothesized that highly intelligent women might face unique stressors. These could include feeling alienated due to being in the minority in certain professional fields or facing social disapproval for prioritizing a career. If true, the “curvilinear” drop in mental health might be steeper for women than for men.
To test these ideas, the team analyzed data from two massive American datasets. These were the National Longitudinal Survey of Youth 1979 (NLSY79) and the National Longitudinal Survey of Youth 1997 (NLSY97). These government-funded studies have tracked thousands of individuals over several decades.
The researchers used a cognitive test called the Armed Forces Qualification Test to estimate intelligence. To measure mental health, they looked at responses to the Center for Epidemiological Studies Depression Scale and the Mental Health Inventory-5. These are standard questionnaires asking people how often they feel sad, nervous, or downhearted.
The team first used a method called polynomial regression. This statistical approach allows for curved lines rather than just straight ones. Initially, the data seemed to support the “tortured genius” hypothesis. The models showed a U-shaped pattern.
In this initial pattern, mental health appeared to improve as intelligence increased, but only up to a certain threshold. After that point, higher intelligence scores were associated with worse mental health. This inflection point appeared at the upper end of the intelligence distribution. The researchers also used a technique called piecewise regression to confirm this breakpoint.
However, the researchers did not stop with the initial finding. They chose to investigate the validity of the mental health tests themselves. They applied a sophisticated technique known as local structural equation modeling. This method allowed them to check if the survey questions functioned the same way for people of all intelligence levels.
This step is designed to test for “measurement invariance.” Measurement invariance is a assumption in psychological testing. It implies that a test measures the same construct in the same way for everyone, regardless of their background or ability. If a test lacks invariance, it is like using a rubber ruler that stretches or shrinks depending on what is being measured.
The investigation revealed a fundamental problem with the data. The researchers found a lack of measurement invariance across intelligence levels. This was true for both the depression scale used in the 1979 cohort and the mental health inventory used in the 1997 cohort.
Specifically, the “factor loadings” for the test items decreased as intelligence increased. A factor loading is a number that indicates how well a specific question correlates with the overall concept it is supposed to measure. For example, answering “yes” to “I felt sad” should strongly indicate the presence of depression.
For participants with average intelligence, the questions were strong indicators of their mental health state. But for participants with high intelligence, the link between the specific questions and the general psychological condition became weaker. The items lost their diagnostic power at the high end of the spectrum.
This phenomenon occurred for both men and women. The researchers observed that the model fit—a statistic showing how well the data matches the theory—deteriorated significantly at high intelligence levels. This suggests that the standard questions might not mean the same thing to a highly intelligent person as they do to others.
There are several reasons why this might happen. Highly intelligent individuals might interpret the wording of questions differently. They might engage in overthinking or semantic analysis of simple phrases like “trouble keeping my mind on what I was doing.”
It is also possible that behaviors labeled as symptoms of pathology in the general population are merely characteristic traits of giftedness. For instance, intense focus or “hyperfixation” could be mistaken for obsessive behavior or attention deficits. Alternatively, gifted individuals might be better at masking symptoms, or their symptoms might manifest in ways these specific tests do not catch.
Because of this lack of invariance, the researchers could not rely on the initial U-shaped curves they found. The apparent drop in mental health among the highly intelligent might be a statistical artifact. It could be a mirage caused by the failure of the measurement tool rather than a real psychological trend.
The study implies that comparing mental health scores across different levels of intelligence is scientifically unsound with these current instruments. If the “ruler” changes length for smart people, researchers cannot say with certainty whether they are happier or sadder than average. This casts doubt on previous studies that claimed to find linear or nonlinear relationships without checking for invariance.
The authors note several limitations to their work. The study relied solely on samples from the United States. Cultural factors heavily influence gender roles and the social experience of intelligence. The results might differ in other countries or cultures.
Additionally, the study examined only two specific mental health scales. It is possible that other diagnostic tools might remain valid across intelligence levels. The researchers also pointed out that the intelligence test was taken under “low-stakes” conditions, which might affect effort and scoring for some participants.
Future research needs to address the measurement problem before drawing conclusions about the “mad genius.” Scientists may need to develop new mental health assessments specifically designed or validated for the gifted population. These tools would need to account for the unique cognitive processing styles of highly intelligent people.
Furthermore, future studies should investigate whether this measurement issue applies to other psychological traits. If intelligence changes how people answer questions about depression, it might also change how they answer questions about personality, anxiety, or social attitudes. Understanding this could lead to a substantial reassessment of how psychological research is conducted across the ability spectrum.
The study, “Lack of measurement invariance in mental health assessment across intelligence levels: Investigation into nonlinearity reveals a broader issue,” was authored by Stanisław K. Czerwiński, Roman Konarski, and Paweł A. Atroszko.
