ENHANCED PREFRONTAL LOSS SENSITIVITY IN EXCESSIVELY SUBMISSIVE FRAGILE X SYNDROME MODEL RATS

Social deficits are core features of neurodevelopmental disorders including Fragile X Syndrome (FXS), yet the neural mechanisms underlying these impairments remain poorly understood. Here we combine behavioral tracking, computational modeling, and calcium imaging to reveal how altered learning from social outcomes drives maladaptive submissiveness in FXS model rats. Using a dominance tube test where two rats compete to push each other out of a narrow tube, we tracked behaviour across 30 sessions over 4 months in knockout (KO) and wild-type (WT) rat pairs. We tracked rats' poses using DeepLabCut and developed a model to estimate bout evolution on a moment-by-moment basis. KO rats displayed excessive submissiveness that increased progressively over time, while WT rats became more dominant. To identify underlying mechanisms, we developed a reinforcement learning model where a rat's effort depends on perceived dominance status learned from previous wins and losses. To examine how social dynamics are represented in the brain, we performed calcium imaging of prelimbic medial prefrontal cortex (mPFC) during competition. Using linear regression and SVM classifiers, we found that mPFC population activity accurately encodes trial evolution, including rats' position inside the tube and whether the imaged rat is being pushed. Surprisingly, KO rats encoded
these signals with higher accuracy than WT rats, suggesting enhanced neural sensitivity to negative outcomes. Our findings suggest a computational principle whereby asymmetric learning rates produce maladaptive social behaviors, demonstrating how neural systems that more accurately encode negative outcomes may lead to worse behavioral adaptation in social contexts.