The developmental trajectory of the prefrontal cortex is not linear. Gray matter volume peaks in late childhood and then declines through adolescence and early adulthood as pruning refines the architecture. White matter continues to expand, with myelination proceeding last in the prefrontal regions. This sequencing means that adolescents often show peak reactivity in reward regions while the regulatory prefrontal circuitry is still catching up — a temporary imbalance that contributes to adolescent risk-taking under normal conditions and becomes acutely relevant under AI's reward-dense stimulation.
The asymmetry between reward system reactivity and prefrontal regulation during adolescence is not a bug but a feature of human development — it likely supports the exploration that shapes adult identity. Under natural conditions, the imbalance resolves as regulatory circuitry matures. Under supernormal-stimulation conditions, the imbalance may calibrate the reward system to parameters the eventual regulatory circuitry cannot modulate.
Clinical implications are concrete. Children's responses to AI engagement cannot be treated as informed choices by agents equipped for self-regulation, because the self-regulatory architecture is incomplete. The adult framework of 'let users choose their own exposure' does not apply to a population whose capacity to choose is literally under construction.
The principle of scaffolded incompleteness in AI design — building tools that withhold capability calibrated to the user's developmental stage — is grounded in this neuroanatomy. A tool optimized for an adult prefrontal cortex provides support for a twelve-year-old prefrontal cortex at a level that system cannot yet handle.
The extended-development framework was consolidated by Jay Giedd's longitudinal neuroimaging work at the National Institute of Mental Health in the 1990s and 2000s, which produced the growth curves showing prefrontal myelination continuing into the mid-twenties. B.J. Casey's work on the adolescent imbalance between reward and regulatory systems provides the mechanistic account.
Extended developmental timeline. Prefrontal myelination continues into the mid-twenties — by far the longest developmental trajectory of any brain region.
Reward-regulatory imbalance. Adolescent reward systems are fully reactive before regulatory systems are fully mature; the imbalance is a feature of development, not a disorder.
Supernormal-stimulation risk. AI's reward density interacts with the reward-regulatory imbalance to produce behaviors adults recognize as compulsive but adolescents cannot yet regulate.
Clinical implication. Treating children as fully autonomous agents regarding their AI exposure misreads the neuroanatomy; scaffolding is required during construction.
Design implication. Tools that modulate capability based on developmental stage align with the neuroanatomy in a way that uniform-adult-design tools do not.