Scientists identify distinct brain patterns linked to mental health symptoms

A new study published in Nature Mental Health has found that patterns of brain activity can help predict different types of mental health symptoms, and that these brain-based predictors are more similar within symptom categories than between them. In other words, the brain features linked to behaviors like anxiety or depression are more alike with each other than they are to behaviors like aggression or rule-breaking—and vice versa. This pattern held true in children, adolescents, and adults, suggesting that brain connectivity plays a consistent but distinct role in different types of mental health issues across development.

Researchers conducted this study to address a long-standing question in mental health: whether internalizing behaviors and externalizing behaviors are supported by shared or unique patterns in the brain. These categories are often used in psychiatry to help understand a wide range of psychological problems, but little is known about how the brain’s network architecture relates to each one.

“Mental health symptoms can be classified into two broad categories: internalizing and externalizing problems,” explained study authors Yueyue Lydia Qu and Avram J. Holmes, a PhD candidate at Yale University and an associate professor at Rutgers University, respectively.

“Internalizing problems, such as anxiety, withdrawal, and somatic complaints, are directed “internally” towards the individual. Externalizing problems, such as disruptive and aggressive behaviors, are directed “externally” towards the environment or other people. In this study, we explored whether brain-based predictors differ between these two categories of mental health symptoms across development.”

The researchers analyzed resting-state functional magnetic resonance imaging (fMRI) data , which captures brain activity while participants are not performing any specific task. The main goal was to examine whether certain brain connectivity patterns—how different regions of the brain interact at rest—could reliably predict internalizing or externalizing symptoms.

The study included three large and independent samples spanning children, adolescents, and adults. The primary sample came from the Adolescent Brain Cognitive Development (ABCD) study and included 5,260 children around age 10. Two smaller samples were used to test whether the findings from the children would generalize: one of 229 adolescents aged 12 to 18 from the Healthy Brain Network, and another of 423 young adults with an average age of 29 from the Human Connectome Project. All participants completed mental health questionnaires designed to assess levels of internalizing and externalizing problems, and all underwent high-quality resting-state brain scans.

The researchers analyzed functional connectivity across 419 regions of interest in each person’s brain. They then applied a machine learning model known as kernel ridge regression to try to predict each participant’s symptom levels based on their brain connectivity. While prediction accuracy in children was modest, it was statistically better than chance. However, the predictions did not generalize as well to the adolescent and adult samples, likely due to the smaller size of those groups.

Even so, across all three age groups, one consistent finding stood out: brain-based predictors of internalizing behaviors were more similar to each other than to those predicting externalizing behaviors, and the same held true in reverse. For instance, the patterns in the brain that helped predict a child’s anxiety also looked more like those that predicted withdrawal than they did the patterns predicting aggression. This supports the idea that internalizing and externalizing symptoms are biologically distinct in the brain, even if they sometimes co-occur or share risk factors.

The researchers also examined which specific brain networks were involved in these predictions and whether they changed across age. In children and adolescents, externalizing behaviors were more strongly associated with connectivity between the brain’s visual network and other areas, while internalizing symptoms were more associated with connections to the subcortical regions—deep brain structures involved in emotion and motivation. In adults, however, internalizing and externalizing symptoms were better predicted by connections within large-scale networks such as the limbic and temporal-parietal systems.

One particularly interesting finding was that the same brain networks can play different roles depending on age. For example, connectivity between the subcortical regions and the temporal-parietal network was linked to externalizing behavior in children and adolescents but linked to internalizing behavior in adults. This suggests that the way different parts of the brain communicate may shift in meaning across development, reflecting how the brain reorganizes itself from childhood to adulthood.

The researchers also found some shared brain patterns across symptom categories. This reinforces earlier work showing that many mental health symptoms share overlapping features, but the new findings go a step further by showing that more fine-grained distinctions—such as the difference between anxiety and aggression—can still be detected in brain connectivity patterns.

“We discovered that brain-based predictors of mental health symptoms are more similar within the internalizing and externalizing categories than between them, across independent samples of children, adolescents, and adults,” Qu and Holmes told PsyPost. “Overall, there are both shared and unique brain-based predictors for internalizing and externalizing categories of mental health symptoms across these samples.”

Despite the strengths of this study—including large sample sizes, rigorous data processing, and replication across age groups—there are some limitations to keep in mind. “The three independent samples used in our study were cross-sectional and did not follow participants over time,” the researchers noted. “Therefore, it remains unclear whether our findings hold true within the same individuals as they grow and develop from children to adults. Additionally, the strength of the brain-behavior relationship observed is modest, suggesting that non-brain factors also play a significant role in predicting mental health symptoms.”

Looking ahead, the researchers hope to build on this work by studying how brain-based predictors evolve within the same individuals over time. Longitudinal research could help clarify whether early differences in brain connectivity predict later mental health outcomes, and whether interventions can alter these patterns.

“The long-term goal of this line of research is to identify distinct brain signatures that predict distinct classes of mental health symptoms,” Qu and Holmes said. “This would enable early identification of at-risk individuals and allow for the personalization of prevention strategies.”

The study, “Distinct brain network features predict internalizing and externalizing traits in children, adolescents and adults,” was authored by Yueyue Lydia Qu, Jianzhong Chen, Angela Tam, Leon Qi Rong Ooi, Elvisha Dhamala, Carrisa V. Cocuzza, Shaoshi Zhang, Tianchu Zeng, Connor Lawhead, B. T. Thomas Yeo, and Avram J. Holmes.