Social cognition as an under-used behavioural endpoint in neurodevelopmental psychopharmacology: A conditional...

Social cognition, encompassing theory of mind (ToM), facial affect recognition, and mentalising under social load, is impaired across several neurodevelopmental disorders (NDDs), including autism spectrum disorder (ASD), attention-deficit/hyperactivity disorder (ADHD), developmental language disorder (DLD), and intellectual disability (ID). These impairments are plausibly downstream of neurochemical processes, particularly serotonergic, dopaminergic, neuropeptidergic, and excitatory-inhibitory modulation of prefrontal-temporal-limbic circuits, although this pathway is indirect and moderated by age, language, cognitive ability, and comorbidity. Despite this, performance-based social cognition is rarely positioned as a primary or co-primary endpoint in NDD pharmacotherapy trials. The predominant standard, the Aberrant Behavior Checklist Irritability subscale and its caregiver-rated relatives, indexes reactive behavioural output that may be construct-distant from the mechanisms through which several contemporary agents are hypothesised to act. This narrative review advances a conditional thesis: endpoint-mechanism misalignment is one plausible contributor to the uninterpretability of certain null trials, alongside insufficient power, sample heterogeneity, unverified target engagement, dosing limitations, high placebo response, and genuine inefficacy. We synthesise condition-specific social-cognitive profiles with graded evidence, the neurobiological substrates across mechanism classes, and a three-tier framework positioning social cognition as a co-primary endpoint for mechanism-matched agents and as a prespecified secondary or moderator variable otherwise. We set out the measurement and regulatory preconditions fit-for-purpose validation, adequate test-retest reliability, practice-effect control, and defined meaningful change that must precede such repositioning. We do not claim ToM should replace existing endpoints or that it suits all NDD trials.