Autism research contains two influential but only partially connected explanatory traditions. Predictive-processing accounts describe what goes wrong at the computational level, namely aberrant precision-weighting and, in some formulations, attenuated priors. Excitation–inhibition accounts describe where at least part of the dysfunction is expressed, namely in glutamatergic and GABAergic circuitry, with N-methyl-D-aspartate receptor signaling occupying a central position. To our knowledge, no single endogenous molecular regulator has yet been established that could plausibly sit upstream of both levels of description. This hypothesis paper proposes that agmatine depletion is a candidate upstream biochemical driver of core distress-related features of autism and that this depletion may link predictive-processing abnormalities, N-methyl-D-aspartate receptor dysregulation, neuroinflammation, autonomic instability, altered social reward processing, and gut dysfunction into one mechanistic framework. The direct human evidence is currently limited. One small study reported lower plasma agmatine in autistic children, and several animal studies reported that exogenous agmatine attenuated autism-like phenotypes in valproate models and stress-related behavioral and neurophysiologic abnormalities in prenatal-stress offspring models. On that basis, and on agmatine's known pharmacology, we hypothesize the following mechanistic chain: depleted agmatine leads to insufficient N-methyl-D-aspartate receptor modulation, impaired precision-weighting, and downstream emergence or amplification of sensory flooding, anxiety, cognitive rigidity, reduced social reward salience, autonomic dysregulation, and gastrointestinal disturbance. We further introduce three novel theoretical constructs. The first construct is the proposal that agmatine is the specific candidate molecule whose depletion helps explain the precision-weighting deficit. The second construct is the Stress Flywheel, which names a nonlinear threshold at which stress no longer merely burdens the system but begins to generate additional stress by eroding the molecular buffer required for sensory and autonomic regulation. The third construct is a proposed resolution of the "BDNF paradox," in which elevated peripheral brain-derived neurotrophic factor in autism is reframed as a compensatory signal that accumulates because utilization is impaired when N-methyl-D-aspartate-dependent plasticity machinery is dysregulated. This paper is explicitly a hypothesis paper rather than a treatment recommendation. The central claims remain unverified in autistic people and require prospective biomarker, pharmacokinetic, and interventional testing before any clinical inference can be justified.
Danny Raede (2026) studied this question.