Genetic Study Challenges the “Mad Genius” Link Between Creativity and Mental Illness

For centuries, the idea of the “mad genius” has captivated writers, philosophers, and scientists alike: the notion that creativity and mental illness are two sides of the same coin, that the same restless mind capable of artistic brilliance is also vulnerable to psychiatric disorder. From Aristotle’s observation that “no great mind has ever existed without a touch of madness” to modern biographical studies of tortured artists, the hypothesis has proven remarkably durable. Now, a new genetic study published in BMC Psychiatry puts this seductive idea to one of its most rigorous tests yet—and the results suggest the story is far more complicated than the popular myth would have us believe.

Researchers led by Guilai Zhan and Zhipeng Cao of the Shanghai Xuhui Mental Health Center and Wenzhou Medical University turned to Mendelian randomization, a powerful statistical technique that uses naturally occurring genetic variants as proxies to probe whether one trait causally influences another. Because genetic variants are randomly allocated at conception—much like a randomized trial—this approach can, in principle, distinguish cause from mere correlation, something that decades of observational studies linking creativity and psychopathology have struggled to do. Previous genetic evidence for the creativity–mental illness connection has largely rested on genetic correlations and polygenic associations, which reveal shared genetic architecture but say little about the direction of any causal arrow.

To measure creativity in a way that genetics could interrogate, the team used a machine-learning-derived genome-wide association study (GWAS) of occupational creativity, built from data on 241,736 participants of European ancestry in the UK Biobank. Rather than asking people whether they consider themselves creative, this approach quantified creativity at the level of occupations, drawing on established measures of how creative different jobs are. The researchers then tested the bidirectional relationships between genetically predicted occupational creativity and the genetic liability to six psychiatric conditions: schizophrenia, bipolar disorder, major depressive disorder (MDD), attention-deficit/hyperactivity disorder (ADHD), autism spectrum disorder (ASD), and anorexia nervosa.

The primary analysis used random-effects inverse-variance weighted Mendelian randomization, a method that combines the effects of many genetic variants scattered across the genome into a single causal estimate while accounting for variability between individual variant estimates. Crucially, the team did not stop there. They deployed a battery of sensitivity analyses designed to detect the classic pitfalls of Mendelian randomization: heterogeneity among variant-specific estimates, horizontal pleiotropy (where variants influence the outcome through pathways other than the exposure of interest), directional bias, and the confounding influence of education- and cognition-related genetic architecture, which looms large over any study connecting intelligence, creativity, and psychiatric traits.

The headline finding was striking precisely because it ran counter to the mad genius narrative. Genetically predicted occupational creativity was inversely associated with liability to ADHD (β = −0.518; 95% confidence interval, −0.821 to −0.214; false discovery rate q = 0.0031) and with liability to MDD (β = −0.273; 95% confidence interval, −0.445 to −0.102; q = 0.0049). In other words, people whose genetic profile predicted greater occupational creativity showed, on average, lower genetic liability to these two conditions—not higher. The estimate for autism spectrum disorder was positive but proved sensitive to the choice of analytical methods, failing the robustness checks that the ADHD and MDD signals passed. For schizophrenia, bipolar disorder, and anorexia nervosa, the forward analyses found no robust evidence of any association at all.

The reverse direction—asking whether psychiatric liability influences later occupational creativity—told a subtler story. The analyses suggested inverse associations of ADHD and MDD liability with genetically predicted occupational creativity, mirroring the forward results. However, the diagnostics for heterogeneity and pleiotropy raised caution flags here: the reverse estimates showed signs that some genetic variants might be influencing the outcome through pathways unrelated to psychiatric liability itself, meaning these findings should be interpreted with considerable care. In Mendelian randomization, pleiotropy is the great saboteur, and the authors were explicit that their reverse results did not withstand scrutiny as cleanly as they might have hoped.

Perhaps the most consequential finding came when the researchers adjusted for educational attainment. Using multivariable Mendelian randomization—a framework that simultaneously estimates the effects of multiple exposures to disentangle their independent contributions—they found that the apparent protective associations between occupational creativity and both ADHD and MDD liability were attenuated once education-related genetic architecture was taken into account. This means the univariable estimates could not be clearly separated from the well-documented genetic overlap between educational attainment, cognitive performance, and psychiatric traits. Educational attainment is itself genetically correlated with many psychiatric conditions, and occupations requiring high creativity typically require high education, creating a tangled web of selection effects that a simple two-variable model cannot easily untangle.

Taken together, the results challenge the simple genetic version of the mad genius hypothesis. Rather than a direct causal pathway in which genes predisposing to creativity also predispose to mental illness—or vice versa—the study suggests that the observed associations are heterogeneous across disorders and partly reflect education-related and occupational-selection pathways rather than effects of creativity itself. The romantic image of the tortured artist whose suffering fuels their genius finds little support in this genetic data; if anything, the genetic architecture of occupational creativity appears modestly protective against the two most common psychiatric conditions, and even that signal may be an echo of educational genetics rather than creativity per se.

The study’s strengths lie in its scale and methodological rigor. The occupational-creativity GWAS, derived from machine-learning models applied to a quarter of a million UK Biobank participants, represents a novel way of operationalizing a notoriously slippery construct. The bidirectional design, the random-effects primary estimator, and the layered sensitivity analyses—including MR-PRESSO, a method that detects and corrects pleiotropic outliers—reflect the current best practice in the field. The authors also transparently reported which findings survived scrutiny and which did not, a refreshing degree of candor in a literature often accused of overclaiming.

Still, important limitations remain, and the authors acknowledge them implicitly through their cautious conclusions. The GWAS data come exclusively from individuals of European ancestry, limiting generalizability to other populations. Occupational creativity, however cleverly modeled, is not the same as the psychological trait of creativity, and the machine-learning derivation introduces its own assumptions. Mendelian randomization estimates can be biased by pleiotropy, population stratification, and winner’s curse, and the attenuation of key findings under education adjustment underscores how fragile causal claims in behavioral genetics can be. What the study decisively demonstrates is that the next chapter of the creativity–psychiatry story will not be written in simple correlations, but in carefully disentangled causal pathways—ones that account for the educational and occupational scaffolding on which both creativity and mental health are built. The mad genius myth, it seems, will have to survive on its literary charm rather than its genetic credentials.

Subject of Research: Bidirectional Mendelian randomization analysis of the relationship between occupational creativity and genetic liability to six psychiatric disorders

Article Title: Occupational creativity and liability to psychiatric disorders: a bidirectional Mendelian randomization study

Article References: Zhan, G., Bao, L., Qin, J., & Cao, Z. (2026). Occupational creativity and liability to psychiatric disorders: a bidirectional Mendelian randomization study. BMC Psychiatry. https://doi.org/10.1186/s12888-026-08556-2

Image Credits: AI Generated

DOI: 10.1186/s12888-026-08556-2

Keywords: occupational creativity, Mendelian randomization, psychiatric genetics, ADHD, major depressive disorder, schizophrenia, bipolar disorder, autism spectrum disorder, educational attainment, UK Biobank, GWAS, mad genius hypothesis