The Adolescent Brain and Schizophrenia: What Research Reveals About Pruning, Dopamine, and Myelin

Many psychiatric disorders, particularly schizophrenia, peak in onset during late adolescence, yet direct tissue-level data on brain changes during this period remained limited, said Bruno Averbeck, PhD. Much of the available human evidence to this effect is indirect, such as structural magnetic resonance imaging measures of myelination. Nonhuman primates, whose adolescent developmental period spans approximately ages 2 to 5, provides a model for Averbeck’s research to measure these changes directly in tissue.

An adolescent increase in dopamine has long been hypothesized, but supporting human data relied largely on indirect measures such as brain iron, Averbeck noted. In primates, dopamine availability in the striatum rose from the youngest ages measured and peaked around age 7—roughly ages 25 to 30 in humans—before declining later in life, with no adolescence-specific surge. Comparisons of 2-year-old and 5-year-old monkeys showed decreased excitation amplitude; inhibition showed a nonsignificant but correlated decrease, so the excitation-to-inhibition ratio did not meaningfully change.

Looking at synaptic pruning, first documented in postmortem human tissue in 1979 and characterized mainly in dorsolateral prefrontal cortex, was replicated in primate neocortex. Limbic cortical areas and the striatum, however, showed no synaptic pruning across adolescence. Myelin increased steadily from the earliest ages measured through the human equivalent of the 30s or 40s. Of the 4 changes, "maybe only the pruning is kind of locked to adolescence. The rest of them are ongoing developmental changes," Averbeck said.

Computational modeling aimed to link these tissue-level changes to shifts in behavior and, potentially, to the onset of psychiatric disorders. The models indicated that pruning, and likely myelination and dopamine, altered attractor dynamics. During cognitive processes, prefrontal neural activity evolved along a trajectory measurable at the single-trial level; when perturbed, activity returned to that trajectory. "It's attracted back to these mean trajectories and the strength of that attraction is increasing during adolescent development," Averbeck said. This effect emerged first in recurrent neural networks trained on cognitive tasks and then pruned, and it was subsequently observed in electroencephalography measures in human participants.^1,2

Dr Averbeck is a principal investigator at the National Institute of Mental Health.

References

1. Averbeck BB. Pruning recurrent neural networks replicates adolescent changes in working memory and reinforcement learning. Proc Natl Acad Sci U S A. 2022;119(22):e2121331119.

2. Liuzzi L, Pine DS, Fox NA, et al. Changes in behavior and neural dynamics across adolescent development. J Neurosci. 2023;43(50):8723-8732.