A person can keep working, caring for others and learning while facing substantial adversity. Calling that “resilience” may recognize real capacity. It can also conceal the costs of coping or imply that people should tolerate preventable harm. The 2025 Nature Medicine review Resilience and brain health in global populations asks researchers to study the biological, psychological, social and environmental conditions together, especially where the existing evidence base has been thin.
This article is a critical guide to Udeh-Momoh and colleagues’ review. It is distinct from a general entry on resilience or a guide to allostatic-load measurement. Its focus is how this particular paper expands the frame of brain-health research, what concepts it brings together and which claims remain proposals rather than demonstrated interventions. The paper’s global ambition makes careful interpretation especially important: a measure designed in one setting does not automatically describe another.
The problem the review addresses
Much published work on brain resilience comes from high-income settings. The authors argue that this imbalance leaves out populations whose exposures, resources, care systems and cultural understandings differ. They use “majority world” to direct attention to people living outside the settings that dominate the literature, including many low- and middle-income countries. Their point is not that those places form one uniform category. It is that a global theory built from a narrow sample risks missing important determinants and misreading others.
The review describes resilience as a dynamic capacity to maintain or regain function in the face of adversity. It spans biological processes, cognitive and emotional resources, relationships and institutions. A broad definition can make research more inclusive, yet it also creates a measurement challenge. If a study records only one questionnaire score, an imaging marker or an educational credential, it has not captured every level of the construct. The authors propose a multilevel agenda rather than a single ready-to-use scale.
The paper is a narrative review. It integrates studies and conceptual work, proposes priorities and offers examples. It is not a global randomized trial of a resilience program. Some of the mechanisms it discusses have stronger evidence than others, and evidence from one population may not transfer to another. A source guide should keep those differences visible even while explaining the value of the synthesis.
There is also a question of language. “Resilient” can describe an observed outcome under an exposure, an underlying capacity, a social resource or a desirable policy goal. These are not the same. A person may do well on a cognitive test despite pathology, remain socially connected despite disrupted care, or report hope despite severe hardship. Each observation requires its own definition and comparison group. The review’s broad framework invites researchers to name the level they mean.
Reserve, maintenance and adaptation
One part of the review concerns reserve. Brain reserve refers to features of the brain’s structure or capacity that may be related to tolerance of pathology. Cognitive reserve concerns how people can sustain performance despite brain changes, often discussed through efficiency, flexibility or alternative strategies. Brain maintenance concerns delaying or limiting the development of age-related changes. Stern and colleagues’ whitepaper distinguishes these constructs because they make different predictions.
Educational attainment is often used as a proxy for cognitive reserve, but a proxy is not the mechanism itself. Years of school can reflect learning opportunities, socioeconomic position, discrimination and access to care. Two people with the same years of education may have had very different experiences. If a study interprets education as a clean measure of inner capacity, it may confuse social opportunity with biology. The 2025 review urges a wider view that includes cultural and community resources.
Neuroplasticity and compensation are also discussed. A brain may change with experience, and alternative networks may contribute when a usual route is impaired. But an imaging pattern does not by itself prove successful compensation. The person must be shown to preserve or regain a meaningful function under a relevant challenge. A new activation can also reflect greater effort without better outcome. The comparison between brain signal and lived function is essential.
Allostasis adds a different layer: how physiological systems adjust to demand. Long-running exposure may burden regulation, but a single cortisol value does not measure a whole history of adversity. The companion allostasis paper in this Library addresses biomarker and technology questions in detail. Here the important point is that bodily adaptation takes place within social and environmental conditions. A theory of resilience that ignores those conditions risks attributing unequal outcomes entirely to individuals.
Exposome and the conditions around a person
The review uses the exposome to bring cumulative environmental and social exposures into view. Nutrition, pollution, infection, education, work, safety, housing and care access can shape brain health across a life course. These exposures do not act independently in every setting. A person facing poor air quality may also face limited health services and less control over work hours. A study that isolates one variable without measuring the others can misstate what it explains.
Exposure is not destiny. The paper describes how biological, psychological and social factors may moderate outcomes. The challenge is to test moderation rather than announce it. If a community program is said to build resilience, researchers should measure who can access it, what changes, how long benefits last and whether harmful exposure itself was reduced. An intervention that helps some people cope while leaving a dangerous environment intact addresses only part of the problem.
The authors discuss cultural reserve and community resilience as additions to an individual-centered picture. Shared caregiving, mutual aid, local knowledge and social belonging may offer practical resources. These ideas should not be romanticized. Not every communal arrangement is protective, and a community cannot replace equitable clinical care, income or accessible infrastructure. A study must ask what people in that setting themselves value and whether the proposed resource improves outcomes without shifting unpaid labor onto them.
Climate disruption, infectious disease burden and migration can also reshape risk and support. Their effects vary by place and time. A global model should use locally meaningful exposure measures and involve local researchers and participants in interpretation. Pooling datasets is useful only when the underlying concepts and instruments remain comparable. Otherwise a large sample can reproduce a narrow worldview at greater scale.
Measuring brain health across settings
The World Health Organization’s brain-health position paper describes determinants across physical health, healthy environments, safety, learning, social connection and access to quality services. That public-health frame is compatible with the review’s multilevel argument. It also reminds readers that brain health is broader than the absence of a neurological diagnosis. Function, participation and well-being matter.
Measurement should match the question. A memory test may be affected by language, schooling, sensory impairment or familiarity with timed tasks. A survey item about support may mean something different in a multigenerational household and in a setting where formal services dominate. A biomarker may vary with nutrition or infection exposure. Researchers need translation, local validation, transparent norms and consultation with the people whose lives are represented in the data.
Cross-sectional correlations are limited. If a group with stronger social ties has better cognitive scores, the ties may help; healthier people may also find connection easier; or both may be influenced by wealth and safety. Longitudinal studies can improve temporal interpretation, but retention and changing measures still matter. Interventions can test whether changing one condition improves a meaningful outcome, provided they are designed and evaluated in the population where they will be used.
The 2025 review calls for better operationalization. That is a substantive finding about the field, not merely a methodological footnote. A claim that one population is more resilient than another may be meaningless when measures, exposures and outcomes differ. Research should define adversity, record relevant resources and specify what continued or recovered function means. It should also report whose data were missing and whose interpretation shaped the study.
Sex, gender, age and social position
The review examines how life stage, sex-linked biology, gendered roles and structural inequalities interact with brain health. These dimensions cannot be reduced to one variable each. Pregnancy-related conditions, caregiving demands, discrimination, education and access to treatment may all matter, but their pattern differs across people and places. A biologically framed result must be read beside the social processes that determine exposure and care.
The authors explicitly note research gaps for transgender and gender-diverse people. The appropriate response is not to fill the gap with assumptions from studies that excluded them. It is to design inclusive research with clear questions, privacy protection and valid measures. Similar caution applies to race: it is not a simple biological category. Patterns associated with racialized groups may reflect histories of exposure, exclusion and unequal resources that research must measure rather than hide behind the label.
Age changes the balance of opportunities and vulnerabilities. Early nutrition, safety and learning can influence development; later-life services, social connection and disease management shape functioning. Yet a life-course approach does not mean every adverse early event has an inevitable neural outcome. Plasticity and changing conditions matter. Researchers should avoid deterministic narratives that stigmatize children or older adults.
These distinctions make the review more demanding than a generic message to “be resilient.” They ask who bears a risk, who controls the conditions and who has access to repair. At the individual level, a person may benefit from support and treatment. At the institutional level, services can reduce avoidable barriers. At the population level, policy can alter exposures. The relevant outcome must be measured at the level of the action.
From research map to policy
The authors propose culturally grounded interventions, more inclusive cohorts and reciprocal research partnerships. WHO’s brain-health position paper likewise organizes action across sectors rather than placing all responsibility on clinical care. Those are reasonable directions, but the effectiveness of a particular program still requires evaluation. A community initiative can be promising in one location and impractical or inequitable in another.
The 2024 Lancet Commission on dementia prevention, intervention and care assembles evidence about potentially modifiable risks across the life course. Its estimates concern populations and assumptions about risk reduction, not a guarantee that one person can prevent dementia by following a checklist. The global-resilience review goes wider than dementia and asks how social and cultural context changes the research agenda. Reading the two together can sharpen questions about prevention while preserving uncertainty.
For the Institute, a useful application is to design studies and services around actual opportunities. If a person is advised to exercise but the neighborhood is unsafe, a personal-capacity message is incomplete. If cognitive testing is offered only in a language the person uses poorly, the score may misrepresent function. If a community network is praised while caregivers lack support, “resilience” can become a way to overlook labor. These are design and policy hypotheses drawn from the review’s framework, not specific intervention findings established by it.
Equitable research also requires governance. Local partners should help define the problem, choose outcomes, interpret data and share benefits. A large international dataset does not by itself create such a partnership. The review’s call for reciprocal exchange matters because people in underrepresented regions are sources of theory and method, not simply new subjects for established models.
What the evidence does not show
The review does not provide a universal resilience score, a precise prediction for one person’s future brain health or a single intervention that works across cultures and resource settings. It synthesizes heterogeneous evidence and argues for a more inclusive framework. Some proposed integrations, including cultural reserve and multi-domain precision approaches, need further operationalization and testing.
It does not show that adversity is beneficial, or that strong community ties make pollution, malnutrition or poor care harmless. Coping can coexist with injury and exhaustion. Calling a person or group resilient should not become a reason to tolerate preventable exposure. Likewise, a poorer average outcome must not be read as evidence of a group’s inherent lack of capacity.
Neither the WHO framework nor the Lancet Commission turns the narrative review into direct causal proof for every pathway. Disease risks, social resources and biomarkers are studied with different designs and uncertainties. Researchers and decision-makers should state which claim is association, which is mechanism, which is intervention evidence and which is a proposal for future work.
Return to the person
The person who kept working and caring under strain may indeed show remarkable adaptation. A good account also asks what it cost, which relationships helped, which services failed and which exposures could have been reduced. Udeh-Momoh and colleagues’ review widens brain-health science to those questions and challenges the field to test them across populations that have been underrepresented.
Its most useful lesson is a discipline of scale. Do not infer a person’s worth from a biomarker, a community’s strength from one survey or a policy’s success from a framework. Define the outcome, listen locally, measure the conditions and test the change. Brain health is shaped in bodies, relationships and the places that make care possible.
Research references
- Udeh-Momoh, C. T., Migeot, J., Blackmon, K., et al. (2025). Resilience and brain health in global populations. Nature Medicine, 31, 2518–2531. View source
- Stern, Y., Arenaza-Urquijo, E. M., Bartrés-Faz, D., et al. (2020). Whitepaper: Defining and investigating cognitive reserve, brain reserve, and brain maintenance. Alzheimer’s & Dementia, 16(9), 1305–1311. View source
- World Health Organization. (2022). Optimizing brain health across the life course: WHO position paper. View source
- Livingston, G., Huntley, J., Liu, K. Y., et al. (2024). Dementia prevention, intervention, and care: 2024 report of the Lancet standing Commission. The Lancet, 404(10452), 572–628. View source
- World Health Organization. (2021). Global status report on the public health response to dementia. View source