After PISA 2025, What Can Schools Still Do to Improve Learning?

The results of PISA 2025 reveal an unprecedented decline in student performance in France and across OECD countries, with average scores reaching their lowest levels in 20 years. More concer-ning still, the proportion of students failing to achieve minimum proficiency levels is rising in ma-ny countries, while inequalities between countries and students persist.

The publication of these results has prompted a wave of opinion pieces, reactions, and policy pro-posals. What can academic research contribute to this debate? Probably not a ready-made set of solutions, but certainly a longer-term perspective, critical distance, and tangible evidence from educational experiments.

One such experiment is Motiv'action, a project jointly led by Yann Algan (HEC Paris), Elise Huillery, and Laila Souali (Paris Dauphine-PSL). It explores ways to fundamentally transform stu-dents' relationship with learning. Through teacher training, the project seeks to strengthen students' beliefs in their own abilities.

At a time when education systems are being challenged to help every student succeed, this ap-proach offers promising possibilities — provided it forms part of a broader transformation of the education system.

As early as 1966, American sociologist James Coleman attributed most differences in educational achievement to students' socioeconomic backgrounds, as well as those of their classmates. His conclusions, based on a cohort of more than 600,000 American students, profoundly shaped politi-cal and academic debates.

More recently, PISA 2025 confirmed that socioeconomic inequalities continue to have a major influence on educational achievement. The performance gap between advantaged and disadvanta-ged students remains substantial, particularly in science, where it averages 85 points across OECD countries and reaches 100 points in France.

Does this mean education systems are powerless in the face of socioeconomic disadvantage? Not entirely. The extent to which inequality affects academic performance is not fixed: it varies across countries and over time.

Moreover, a number of educational experiments have demonstrated that inequalities can be ad-dressed. In the United Kingdom, for example, researchers examined 88 educational interventions with comparable characteristics and found that some had a greater impact on disadvantaged stu-dents — those eligible for Free School Meals — than on their peers.

If change is possible, where exactly should educators intervene? And how?

Public debate often revolves around competing visions of education, pitting advocates of "traditio-nal" teaching methods against supporters of more progressive approaches.

Recent developments appear to support the traditionalists' position. Finland, once a pioneer of open and progressive educational methods, has seen its PISA performance decline since 2012 — a trend confirmed in 2025.

Countries such as Sweden and New Zealand are implementing educational reforms that place renewed emphasis on foundational knowledge. In the United States, "classical education" is gai-ning ground, favoring structured curricula, major works of the Western intellectual tradition, me-morization, and recitation.

Yet other developments lend support to more progressive approaches. Countries known for their highly disciplined education systems are also introducing changes.

China, for instance, launched a major curriculum reform in the 2000s. Without fundamentally challenging the existing system, the Chinese government has long acknowledged the negative ef-fects of excessive academic pressure and has sought to reduce student workloads.

The introduction of artificial intelligence education in Chinese schools is also prompting renewed attention to skills such as problem-solving, independent thinking, communication, and coopera-tion.

Even Singapore does not rely on a purely teacher-centered model: its educational practices com-bine direct instruction with elements of active learning.
It is therefore difficult to identify a single formula or ideal model. Foundational learning, like the effectiveness of any teaching approach, depends on multiple factors.

Students' ability to engage in learning varies according to their living conditions, technological environment, and interpersonal experiences, among other influences.

There is no "miracle" teaching method, but rather a range of levers that can be activated at diffe-rent levels and adapted to specific contexts.

Some of these factors fall under what researchers call social and behavioral skills: abilities that influence how we perceive ourselves and others, adapt our behavior to different situations, and interact with those around us.

These skills therefore affect our ability to learn. Numerous school-based interventions designed to develop social and behavioral skills have been scientifically evaluated, producing consistently positive findings.

There is good reason for optimism: we now have at least some understanding of how these skills can be strengthened.

One particularly important example is metacognition.

Among so-called "transferable skills," metacognition is one of those with the strongest evidence of impact. Indeed, metacognitive abilities appear to predict subsequent academic performance.

Metacognition refers to the ability to reflect on one's own thinking and learning processes. It in-volves planning work, monitoring comprehension, identifying mistakes, and adjusting learning strategies accordingly.

Metacognition could also help reduce educational inequalities from an early age, as suggested by an article highlighted by Nature in 2024.

Beliefs play an important role in metacognition — and crucially, those beliefs can be changed.

This idea is supported by substantial scientific evidence, particularly the work of Carol S. Dweck and Ellen L. Leggett in the late 1980s.Their research distinguishes between two conceptions of intelligence: intelligence as a fixed trait, and intelligence as something that can develop and improve over time.
Students who hold the latter belief — known as a growth mindset — are more likely to seek out challenges, persevere, and view difficulties as opportunities to learn.

By contrast, students who see intelligence as fixed may be more inclined to avoid mistakes, reduce their efforts, or even give up.

Changing students' beliefs about intelligence therefore represents a promising avenue for develop-ing metacognition, strengthening learning abilities, and ultimately improving academic achieve-ment.

The Motiv'action project provides a compelling illustration of this approach.

Jointly led by Yann Algan (HEC Paris), Elise Huillery, and Laila Souali (Paris Dauphine-PSL), in partnership with the French Ministry of National Education and HEC Paris's HOPES program, Mo-tiv'action examines the effects of training teachers in social and behavioral skills across five French regional education authorities.

The training module focusing on how the brain works has produced particularly interesting results.

By teaching educators that the brain is malleable and capable of change, the program encourages them to focus their feedback on the learning strategies students use rather than solely on the results they achieve.

Within this framework, most students show improvements in self-confidence. Their beliefs about their own ability to learn also become more positive — a shift researchers des-cribe as the development of a "growth mindset." The effects on academic performance are still being evaluated.

Tracking these results over time is essential.

Although the positive effects of social-emotional skills are now well documented, the extent to which improvements in these skills translate into better academic performance varies depending on the intervention and the conditions under which it is implemented.

The most sustainable and consistent results are likely to emerge from a more holistic understan-ding of education systems — one that considers how different components interact and how inter-ventions can combine several complementary levers.

For now, research has made considerably more progress in assessing the effectiveness of individual interventions than in identifying the optimal combinations of educational approaches.

An emerging field of research is therefore exploring a more systemic approach to education, brin-ging together explicit instruction, student autonomy, metacognition, social skills, school climate, and family engagement.

This is one of the ambitions of the RISE program. It also underpins approaches advocated by the World Health Organization and the European Union, often referred to as the "whole-school ap-proach."

The challenge now is to determine which combinations of interventions work best, for which students, and under what conditions.