Storm and stress: trajectories of youth mental health – Nature Mental Health

Nearly 100 years ago, anthropologist Margaret Mead captured global attention with her book Coming of Age in Samoa¹. In it, she detailed her fieldwork conducted in three villages of the Manu’a Archipelago in Polynesia, interviewing adolescent girls and documenting their experiences in their communities during the transitional years between childhood and adulthood. Mead’s aim was to compare adolescence as it was experienced in Samoa with the narrative of American adolescent ‘storm and stress’. She posited that rather than biological changes during puberty, it was cultural forces, such as more-restrictive sexual mores and societal expectations, that made adolescence in the USA inherently more stressful.

Credit: Serenat Keskin / iStock / Getty Image Plus and A-Digit / DigitalVision Vectors / Getty and Marina Corral Spence. Design: Marina Corral Spence

For the many faults of Mead’s accounts, including outmoded terminology and a tendency to romanticize and downplay some of the social ills of the groups she studied, she was an early proponent of acknowledging the impact of experiencing stress and trauma in early life. Mead pointed to the range of stressors that young people experienced as bound to inequities in their immediate environments, suggesting that “the conditions which vex our adolescents are the flesh and bone of our society”¹.

This recognition that childhood experiences shape later outcomes laid the groundwork for decades across many fields of research seeking to understand how adversity during this sensitive window of heightened neuroplasticity becomes biologically and psychologically embedded across development. The twentieth century saw several waves of influential theoretical frameworks emerge. Theories of attachment highlighted how stress and trauma can disrupt infant–caregiver bonds and affect emotion regulation and interpersonal relationships; the diathesis–stress model emphasized interactions between pre-existing vulnerabilities and environmental stressors; and the concept of allostatic load described the physiological consequences of repeated stress exposure.

Building on those frameworks, researchers increasingly sought ways to measure adversity systematically. It was the joint US Centers for Disease Control and Prevention–Kaiser Permanente health group study in 1998, led by Felitti and Anda, that was instrumental in quantifying early-life stress in a structured and reproducible way². The study used electronic health records and questionnaire data to identify detrimental childhood exposures, or ‘ACEs’ (adverse childhood experiences). Later expanded from seven categories to ten, ACEs are measured across two domains: child maltreatment (psychological abuse, physical abuse, sexual abuse, physical neglect and emotional neglect); and household dysfunction (the presence of substance abuse, mental illness, intimate partner violence, criminal behavior and parental separation).

Since then, the ACEs literature has grown enormously, spanning cultures and populations and examining associations with biological assessments, physical and mental health conditions and life-course outcomes. Out of this vast literature, several overarching themes have emerged: ACEs are exceedingly common but differentially affect certain groups, including women and ethnic and sexual minorities; ACEs can have long-lasting negative effects on opportunities in life, as well as on physical and mental health; and there is a dose–response relationship, whereby increased exposure is associated with increased risk of adverse outcomes.

At the same time, the field has moved beyond measuring exposure alone. The ACEs framework captures one dimension of stress, but many other instruments and modalities have since been developed to assess related processes, including perceived stress, psychophysiological reactivity, neurofeedback, hormonal responses and epigenomic changes. Together, these approaches continue to advance the understanding of how stress shapes developmental trajectories, the neurobiological, psychosocial and environmental mechanisms involved, and the factors, such as resilience, that may provide protection against risk.

This shift toward a more nuanced and predictive science of mental health is reflected in the August 2026 issue of Nature Mental Health. The studies and commentaries collected here illustrate efforts to move beyond identifying mental health conditions to anticipating them, with a particular focus on stress during childhood and the uniquely formative period of adolescence. Rather than examining discrete disorders, much of this work adopts a transdiagnostic perspective, considering symptom dimensions and stratification by brain-based or mechanistic features.

Prediction of mental health outcomes frequently relies on the presence of subclinical symptoms long before diagnostic thresholds are reached. An Articleby Glynn et al. demonstrates that unpredictability, including inconsistency in routines, caregiving and household environments, provides information beyond traditional ACE measures, substantially improving the identification of children vulnerable to depression, anxiety, sleep disorders and externalization of symptoms. Their findings reinforce the notion that adversity is defined not only by traumatic events but also by the absence of reliable and predictable environments in which children can develop.

The importance of caregiving environments is echoed in a Perspective by Gee and colleagues. Drawing on developmental neuroscience, the authors suggest that two parental processes — filtering of a child’s environment, and buffering of a child’s fear response — may provide mechanistic links between neural development and anxiety risk while also informing more personalized interventions to improve outcomes.

Other contributions underscore the enduring impact of childhood adversity across the lifespan. In a systematic review and meta-analysis, Coleman et al. find that retrospective self-reports of childhood maltreatment are more stable than often assumed, particularly in clinical samples.

The interaction between adversity and other vulnerabilities is further highlighted by Wilson et al., who demonstrate that childhood attention deficit hyperactivity disorder and socioeconomic deprivation jointly shape multimorbidity in young women. Their analyses reveal that disadvantage and neurodevelopmental conditions amplify each other rather than acting independently, producing patterns of psychiatric and physical illness that demand more holistic and integrated care. These findings serve as an important reminder that mental health trajectories are shaped not only by symptoms and neurobiology but also by social determinants. If structural disadvantage, poverty and inequitable access to resources are not addressed, even the most sophisticated biomarkers can provide only a partial view of risk and resilience.

Technology, meanwhile, is expanding the sources of information available for assessing risk. An Article by Antonacci et al. demonstrates that the words children use to describe stressful experiences contain rich predictive information about future psychopathology. Automated analyses of naturalistic speech outperformed traditional expert assessments of stress severity, identifying linguistic patterns associated with both vulnerability and resilience years before clinical illness emerged. Notably, these predictive signals were derived not simply from emotional vocabulary but from the broader structure and content of children’s narratives, which suggests that language itself provides a window into developmental processes that conventional clinical assessments may overlook.

Looking ahead, Giampetruzzi and coauthors argue for technology-driven approaches such as digital phenotyping through wearable devices and machine learning, which could enable continuous monitoring of depression risk during adolescence. Such tools could shift mental healthcare from episodic assessment toward ongoing monitoring and prevention. Yet as predictive algorithms become increasingly powerful, questions of privacy, equity, interpretability and implementation become equally important. The challenge is not simply whether risk can be identified earlier, but how predictions can be translated into interventions that are effective, ethical and accessible.

As Margaret Mead wrote, “The solution to adult problems tomorrow depends in large measure upon how our children grow up today”¹. Prediction is valuable only if it informs action. The same research that illuminates biological and behavioral pathways involved in mental health conditions repeatedly highlights the importance of social determinants and the environments in which children develop. Effective interventions rely not only on elucidating the variations in stress response and resilience in children and adolescents but also on addressing and correcting systemic inequities to positively influence wellbeing and mental health trajectories across the life span.