Neural dynamics and personality traits under the predictive coding framework in Functional Neurological Disorders

Functional Neurological Disorders (FND) present a long-standing challenge at the intersection of neurology, psychiatry, and cognitive neuroscience: patients exhibit disabling motor, sensory, and interoceptive symptoms in the absence of structural pathology, yet the mechanisms that generate and maintain these experiences remain only partially understood. This thesis addresses this gap by integrating large-scale brain network analyses with predictive-processing accounts of perception, action, and bodily self-representation. The work is structured in two complementary parts. The first examines altered brain dynamics in FND through a comprehensive synthesis of recent neuroimaging, electrophysiological, structural, genetic, and connectivity findings. Building on this foundation, an original EEG microstate study provides the first evidence that individuals with FND display a selective instability of microstate G (associated with somatosensory and sensorimotor networks) and atypical transition probabilities among microstates related to arousal, imagery, and self-referential processing. These results suggest that FND are characterized not only by aberrant activity in specific networks but also by a disruption in the temporal “syntax” through which the brain spontaneously organizes functional states. The second part situates these findings within a predictive-coding framework, exploring how altered inferential mechanisms, weighting of sensory evidence, and precision of prior expectations may contribute to symptom formation. Three empirical studies examine behavioural and neural correlates of predictive styles in FND: (i) schizotypal and autistic-like traits as dimensional proxies of inferential bias; (ii) frontal theta–alpha asymmetry as a possible neural signature of predictive imbalance; and (iii) individual differences in interoceptive prediction and their relevance for symptom experience. Across these studies, results converge on the view that FND reflect a bias toward rigid or overly influential high-level priors, paired with unstable or down-weighted sensory evidence. Taken together, the thesis advances a unified account in which FND emerge from disturbances in the dynamic coordination of large-scale networks and from mismatches in predictive inference. By combining high-resolution neural dynamics with computationally informed models of perception and action, this work contributes to a growing reconceptualization of FND as disorders of dynamic inference rather than disorders of intent, volition, or structural integrity.