[EN] Research Spotlight: The Math Gender Gap – Martinot et al. (2025)
Differences in Math - socially made?
Research on gender differences is both important and necessary. There are still many domains in which women face disadvantages that we should not accept and must actively address. However, in recent years, a nearly dogmatic stance seems to have taken hold when it comes to research on cognitive differences. Differences between the sexes that relate to the brain or cognitive abilities are almost always assumed to be sociocultural in origin. Biological explanations are frequently dismissed a priori (Pinker, 2002).
Likely out of concern that biological differences could be misunderstood as immutable or used to justify discrimination, their potential existence is often entirely denied. Yet a biological difference is neither automatically unchangeable nor does it justify differential treatment.
This pattern appears to be reflected in a recent large-scale study published in Nature (Martinot et al., 2025). The data are excellent, but the interpretation seems constrained by the very concerns just described.
The Study
The methodological strength of the study by Martinot et al. (2025) is impressive. The researchers tracked the mathematical and verbal performance of over 2.6 million French first- and second-graders over four years.
The core findings:
Equal starting point: At school entry, average math performance is nearly identical for boys and girls.
Rapid divergence: A significant performance gap favoring boys emerges within the first four months.
Widening gap: This difference quadruples by the start of second grade.
Nature News shows this divergence cleary:
The Authors’ Interpretation
Schooling as the cause: Since performance was equal at the outset, the authors argue that the cause must lie in schooling or sociocultural influences.
Internalization of stereotypes: They propose that the gap reflects girls internalizing the stereotype "girls are bad at math," triggered when classroom activities become explicitly labeled as mathematics.
Critique of the Interpretation
While the study’s empirical observations are robust, the interpretation is not the only possible one. Other explanations, also compatible with the data, are overlooked in favor of a familiar narrative.
The Fallacy of the "Equal Start"
The study assumes that equal initial performance rules out biological factors. But that’s a logical fallacy. Consider this analogy:
Suppose researchers find that boys and girls are the same height at age 11. Two years later, boys are significantly taller. If we applied the logic of the math study, we would conclude that the cause of the height difference lies in school—or perhaps in a stereotype like "men are taller than women."
Clearly, that’s incorrect. A biologically programmed difference, in this case, puberty, emerges only after a time delay. Applied to the math findings: just because no difference is visible on simple entry tests doesn’t mean cognitive differences aren’t present. They may only become apparent with increasing task complexity.
An Overlooked Explanation: An Alternative Causal Chain
The study itself contains data that challenge its main thesis, but these are not given much attention.
The data also support another interpretation; one that includes biological differences without ignoring social influences. It's possible that boys and girls differ cognitively in ways relevant to mathematics, but that these differences aren’t yet detectable at school entry due to the simplicity of early tasks. As the complexity of math increases (around time point T2), performance differences begin to emerge.
A second dynamic adds to this: Even at the first test point, boys are overrepresented at both the top and bottom ends of the distribution (Martinot et al., 2025). This matches the Greater Male Variability Hypothesis, which holds that many cognitive and physical traits show greater variance among males with more outliers at both extremes. This phenomenon has been documented in multiple international education studies (Baye & Monseur, 2016).
Taken together, a possible mean difference and greater variability, these mechanisms could explain why more boys are visible as "math prodigies." This visibility may then produce the stereotype "boys are better at math." In this scenario, the stereotype is not the cause of the gap but a result of observed patterns. Social and educational influences may amplify such differences - but they are not their origin.
One might object that stereotypes act primarily through psychological mechanisms like stereotype threat, where fear of confirming a negative group stereotype leads to reduced performance. But this once-celebrated explanation is now facing serious challenges. Many of the foundational stereotype threat studies have failed to replicate in large-scale replications or show much smaller effects.
In short, the authors’ preferred explanation relies on a mechanism whose reliability is increasingly in doubt, therefore making the alternative account (stereotype as consequence, not cause) more plausible.
Another plausible dynamic deserves attention: Children tend to engage more with subjects they’re good at or where they receive positive feedback. If boys are slightly ahead in math, this could lead them to invest more time and effort in the domain, while girls may gravitate toward language-related subjects. This kind of self-selection, described by
(2018), can reinforce early differences without the need for external stereotyping.Conclusion: It’s Complicated
The study by Martinot et al. (2025) is valuable. It provides a strong empirical foundation. But when it comes to explaining why the observed gap emerges, it dismisses alternative explanations—unconvincingly, in my view. A serious and open debate must acknowledge that biological and social factors interact in complex ways (Tucker-Drob & Bates, 2016).
Effective solutions require a proper diagnosis—and that means embracing multicausal models. Biological dispositions, social dynamics, classroom practices, and cultural norms don’t act in isolation. Ignoring any one of them risks treating symptoms (such as stereotypes) instead of root causes. Only when we allow for biological differences as a legitimate part of the picture can we design interventions that actually work—for girls and boys.
Sources:
Baye, A., & Monseur, C. (2016). Gender differences in variability and extreme scores in an international context.1 Large-scale Assessments in Education, 4(2). https://doi.org/10.1186/s40536-016-0016-x
Flore, P. C., Mulder, J., & Wicherts, J. M. (2018). The influence of gender stereotype threat on mathematics test scores of Dutch high school students: A registered report. Comprehensive Results in Social Psychology, 3(2), 140–174. https://doi.org/10.1080/23743603.2018.1559647
Martinot, P., Colnet, B., Breda, T., Sultan, J., Touitou, L., Huguet, P., Spelke, E., Dehaene-Lambertz, G., Bressoux, P., & Dehaene, S. (2025). Rapid emergence of a maths gender gap in first grade. Nature. [https://doi.org/10.1038/s41586-025-09126-4](https://doi.org/10.1038/s41586-025-09126-4)
Organisation for Economic Co-operation and Development. (2023). PISA 2022 results2 (Volume I): The state of learning and equity in education. OECD Publishing.
Pinker, S. (2002). The Blank Slate: The Modern Denial of Human Nature. Penguin Books.
Stewart-Williams, S. (2018). The Ape That Understood the Universe. Cambridge University Press.
Tucker-Drob, E. M., & Bates, T. C. (2016). Large cross-national differences in gene × socioeconomic status interaction on intelligence. Psychological Science, 27(10), 1347–1359. https://doi.org/10.1177/0956797616652754