Functional Neurological Disorder as a Disorder of Network Integration and 
Predictive Inference

Author: Olivia Taylor    Date: 10/13/2025

Functional Neurological Disorder (FND) presents with genuine, often disabling 
neurological symptoms—e.g., limb weakness, tremor, seizures—without a structural lesion that 
accounts for them. Modern practice emphasizes positive clinical signs (e.g., Hoover’s sign, 
tremor entrainment) rather than a “rule-out-everything” diagnosis (Hallett et al., 2022; Voon et 
al., 2016). Historically cast as “psychological,” contemporary models frame FND as disrupted 
integration across large-scale brain networks involved in bodily awareness, movement planning, 
agency, attention/salience, and executive control (Stone et al., 2021; Perez et al., 2021). This 
paper argues that FND symptoms emerge from aberrant network interactions, not a single 
damaged node (Stone et al., 2021).  

A useful unifying perspective is predictive processing: symptoms reflect maladaptive 
priors/expectations and attentional weighting applied to interoceptive and sensorimotor signals. 
In FND, abnormal precision (confidence) may be assigned to misleading internal cues, 
destabilizing perception of the body and control of action. Here, “precision” (or “confidence”) 
refers to how strongly the brain weights a signal or prediction when deciding what is happening; 
too much precision on noisy internal cues can overpower corrective sensory evidence and bias 
experience toward non-movement or non-agency (Stone et al., 2021; Perez et al., 2021).  

The Salience (Cingulo-Insular) Network (SN/CIN)

 
The salience network acts like the brain’s “relevance filter,” deciding what deserves 
attention right now and shifting resources accordingly. Its hubs, the anterior insula and anterior 
cingulate cortex (ACC), link body signals (like heartbeat, muscle tension, visceral sensations) 
with emotion and goal-directed control (Craig, 2009; Stone et al., 2021). In FND, this filter 
appears overreactive for internal threat cues: interoceptive sensations are marked as highly 
important, attention is drawn inward, and competing goals (like fluid movement) get crowded 
out. Empirically, this shows up as atypical coupling between salience hubs and 
sensorimotor/control networks and aligns with common clinical features such as anxiety and 
difficulties identifying/labeling internal states (alexithymia) (Hallett et al., 2022; Perez et al., 
2021; Aybek et al., 2015). The issue is not a linear cause-and-effect sequence but a broader 
miscalibration in how the system assigns importance to internal versus external cues, 
especially under uncertainty (Stone et al., 2021).  

Supplementary Motor Area (SMA) and Movement Initiation

 
The supplementary motor complex—encompassing pre-SMA and SMA  
proper—coordinates the preparation, sequencing, and initiation of voluntary movement, 
particularly when actions are internally generated rather than purely stimulus-driven. In a 
predictive-processing frame, these regions help specify forward models of the intended action 
and issue efference copies that predict the sensory/proprioceptive consequences of the movement 
(Stone et al., 2021). Successful execution depends on a tight communication loop among 
pre-SMA/SMA, primary motor cortex, basal ganglia, cerebellum, and parietal comparators, with 
top-down precision signals determining which motor intentions come to fruition (Voon et al., 
2016; Perez et al., 2021).  

In FND, available evidence is most consistent with an instability at the interface between 
intention and enactment of movement. Preparatory activity can be present but expressed weakly 
in behavior, suggesting a decoupling between motor readiness and the subjective experience of 
willing the movement (Edwards et al., 2011; Maurer et al., 2016). Altered coupling between 
SMA/pre-SMA and parietal–temporal nodes (including agency-related regions) fits a picture in 
which prediction and feedback are not being integrated with the usual confidence. When the 
salience network biases attention toward internal threat signals, this mis-calibration is amplified: 
the system prioritizes monitoring and “checking” over fluid execution, and the intended 
movement is either inhibited or experienced as not truly self-generated (Hallett et al., 2022; 
Perez et al., 2021).  
 

This account supports why physiotherapy that minimizes self-focused monitoring and restores 
automaticity can be effective. External cueing (rhythm, targets), dual-tasking, graded 
complexity, and an external focus of attention reduce the maladaptive precision assigned to 
interoceptive “error” signals and allow the SMA–parietal–cerebellar communication loop to  
re-stabilize. When movement is shaped under conditions that discourage excessive internal error 
monitoring, individuals more readily re-experience actions as intended and voluntary (Stone et 
al., 2021; Hallett et al., 2022).  

Temporoparietal Junction (TPJ) and Sense of Agency (SoA)  
 

The temporoparietal junction (TPJ) contributes to the sense of agency (SoA) by integrating 
multisensory predictions (derived from efference copies) with the reafferent feedback, which is 
the sensory information that returns to the brain following actions. Positioned at the confluence 
of visual, vestibular, and somatosensory streams, the TPJ helps maintain a coherent body schema 
and evaluates whether outcomes match intentions (Perez et al., 2021). Agency, in this view, 
emerges when predicted and observed consequences are sufficiently aligned and weighted with 
appropriate precision (Baek et al., 2022; Stone et al., 2021).

 
Findings in FND indicate abnormal TPJ participation in these comparisons, with connectivity to 
sensorimotor and frontal control regions altered across cohorts and phenotypes. Rather than a 
single direction of effect, the pattern suggests a mismatch in how predictive and feedback 
signals are integrated and trusted. When salience circuitry up-weights bodily noise and control 
networks fail to recalibrate, TPJ computations are performed on noisy inputs with skewed 
precision, increasing the likelihood that self-generated actions feel alien. The clinical 
phenomenology of “I can’t make it move,” or “it moves but not by my will,” naturally follows 
from this imbalance: the comparator does not recognize the action as self-produced, even when 
the motor system can execute it (Weber et al., 2025; Baek et al., 2022; Hallett et al., 2022).  
 

Therapeutic strategies that restore reliable prediction–feedback contingencies therefore 
make mechanistic sense. Techniques that provide accurate, synchronous visuo-proprioceptive 
feedback (e.g., action observation, video feedback, mirror-based tasks when appropriate) and that 
grade exposure to self-initiated movement under an external focus help re-calibrate TPJ 
computations. As the system repeatedly experiences consistencies between intended and 
observed outcomes without threat-biased monitoring, agency strengthens and symptoms subside 
(Stone et al., 2021; Hallett et al., 2022). 
 

Dorsolateral Prefrontal Cortex (DLPFC) / Frontoparietal Control  

 
The dorsolateral prefrontal cortex (DLPFC) is an important node within the 
frontoparietal control network that provides top-down control, helping the system decide how 
much weight to give interoceptive and sensorimotor signals at any moment. In a 
predictive-processing frame, this involves calibrating the precision of competing hypotheses 
about bodily states and actions. When this control is compromised, misleading internal cues 
(amplified by salience circuitry) retain excessive influence, and corrective evidence is 
underweighted. Clinically, that looks like persistent convictions of “something is wrong with my 
body” and a persistent sense of non-agency, even when motor pathways are capable of producing 
movement (Stone et al., 2021; Perez et al., 2021).

 
Converging work places the DLPFC as the link between attention allocation, conflict 
monitoring, and metacognitive updating, processes that should suppress maladaptive prior 
expectations and support re-estimation when predictions fail. In FND, alterations within this 
network can weaken the system’s ability to down-regulate threat-biased interoception and to 
reconcile prediction errors arising from disrupted interactions among the salience network, 
supplementary motor regions, and temporoparietal nodes. The result is not simply a failure to 
move but a failure to re-experience movement as intended and self-generated. This model 
complements evidence of abnormal SMA–parietal/TPJ coupling by specifying how 
DLPFC-mediated control would ordinarily stabilize intention–action coherence and re-establish 
confidence in voluntary control (Baek et al., 2022; Perez et al., 2021).

 
Therapeutically, this positioning of DLPFC within the control network explains why 
explanatory models, attention-shifting physiotherapy, and CBT-informed strategies can be 
effective: they all work, in part, by retuning precision and re-weighting misleading bodily 
predictions under active, goal-directed control. It also motivates cautious exploration of 
adjunctive neuromodulation aimed at control or agency nodes as potential stabilizers of the 
network dynamics that support accurate bodily inference, as a complement to behavior-based 
retraining (Perez et al., 2015; Stone et al., 2021). 

Discussion: A Systems-Level Account  
 

FND appears to be an emergent property of disrupted integration among salience, 
sensorimotor/agency, and control networks. Explored through predictive processing, symptoms 
arise when maladaptive priors and attentional weighting are applied to interoceptive and 
sensorimotor evidence, producing compelling but misleading inferences about the body—for 
example, that a limb cannot move or that a tremor is entirely involuntary. Clinically, 
abnormalities within the salience network bias attention toward internal signals and threat, 
differences in coupling between SMA/pre-SMA and parietal/TPJ nodes destabilize the coherence 
between intention and action and erode the SoA, and alterations in frontoparietal control limit the 
system’s capacity to re-weight misleading priors or to update beliefs in the face of contradictory 
evidence (Stone et al., 2021; Hallett et al., 2022). This account does not require the nervous 
system to be “normal” in every respect, but it emphasizes that the characteristic symptoms reflect 
network-level dysfunction rather than a focal lesion (Perez et al., 2021; Voon et al., 2016).  

Future Directions and Conclusion  

Progress depends on moving beyond static group comparisons toward methods that 
reveal how signals flow through the system. Dynamic connectivity analyses, coupled with 
computational models of predictive coding and precision weighting, can show when and where 
salience, sensorimotor, and control networks miscommunicate during symptom expression and 
recovery. Studies should also stratify participants by phenotype, such as functional weakness, 
tremor, or seizures, because connectivity patterns and treatment responses likely differ across 
these groups (Hallett et al., 2022). Practically, that means planning studies in advance with clear 
hypotheses, ensuring adequate sample sizes, and measuring both how the brain–behavior 
mechanisms change and whether patients actually improve (Perez et al., 2021).

 
Within this framework, neuromodulation can be tested as an adjunct to physiotherapy and 
CBT-informed rehabilitation rather than as an independent intervention, with protocols targeted 
to specific nodes (for example, rTPJ, motor, or control centers) and evaluated for their ability to 
stabilize network interactions that support accurate bodily inference and voluntary control (Perez 
et al., 2015; Stone et al., 2021). 

 
References  
 

Aybek, S., Nicholson, T. R., O’Daly, O., Zelaya, F., Kanaan, R. A., et al. (2015). 
Emotion–motion interactions in conversion disorder: An fMRI study. PLOS ONE, 10(4), 
e0123273. doi: 10.1371/journal.pone.0123273  

Baek, K., Donamayor, N., Morris, L. S., Strelchuk, D., Mitchell, S., Mikheenko, Y., ... & Voon, 
V. (2022). The self-agency network: Neural correlates of sense of agency in healthy 
controls and patients with functional neurological disorder. Brain and Behavior, 12(9), 
e2733. doi: 10.1002/brb3.2733  

Craig, A. D. (2009). How do you feel—now? The anterior insula and human awareness. Nature 
Reviews Neuroscience, 10(1), 59–70. doi: 10.1038/nrn2555  

Edwards, M. J., Moretto, G., Schwingenschuh, P., Katschnig, P., Bhatia, K. P., & Haggard, P. 
(2011). Abnormal sense of intention preceding voluntary movement in patients with 
psychogenic tremor. Neuropsychologia, 49(9), 2791–2793. doi:  
10.1016/j.neuropsychologia.2011.05.021  
 

Hallett, M., Aybek, S., Dworetzky, B. A., McWhirter, L., Staab, J. P., & Stone, J. (2022). 
Functional neurological disorder: New subtypes and shared mechanisms. The Lancet 
Neurology, 21(6), 537–550. doi: 10.1016/S1474-4422(21)00422-1  
 

Maurer, C. W., LaFaver, K., Ameli, R., Epstein, S. A., Hallett, M., & Horovitz, S. G. (2016). 
Impaired supplementary motor area inhibition in functional movement disorders: A 
preliminary study. PLOS ONE, 11(7), e0158556. doi: 10.1371/journal.pone.0158556  
 

Perez, D. L., Dworetzky, B. A., Dickerson, B. C., Leung, L., Cohn, R., Baslet, G., & Silbersweig, 
D. A. (2015). An integrative neurocircuit perspective on psychogenic nonepileptic seizures and 
functional movement disorders: Neural functional unawareness. Clinical EEG and Neuroscience, 
46(1), 4–15. doi: 10.1177/1550059414555905 
 
Perez, D. L., Nicholson, T. R., Asadi-Pooya, A. A., Begue, I., Butler, M., Carson, A. J., David, 
A. S., Deeley, Q., Diez, I., Edwards, M. J., Espay, A. J., Gelauff, J. M., Hallett, M., Horovitz, S. 
G., Jungilligens, J., Kanaan, R. A., Tijssen, M. A., & Aybek, S. (2021). Neuroimaging in 
functional neurological disorder: State of the field and research agenda. The Lancet Psychiatry, 
8(10), 826–841. doi: 10.1016/S2215-0366(21)00303-3  
 

Stone, J., Carson, A., & Hallett, M. (2021). A framework for understanding the pathophysiology 
of functional neurological disorder. The Lancet Psychiatry, 8(9), 819–829. doi: 
10.1016/S2215-0366(21)00343-4  
 

Voon, V., Cavanna, A. E., Coburn, K., Sampson, S., Reeve, A., & LaFrance, W. C., Jr. (2016). 
Functional neuroanatomy and neurophysiology of functional neurological disorders 
(conversion disorder). Journal of Neuropsychiatry and Clinical Neurosciences, 28(3), 
168–190. doi: 10.1176/appi.neuropsych.14090217  
 

Weber, S., Cartoni, E., Gaggioni, L., Turri, F., Wiest, R., & Aybek, S. (2025). Altered brain 
network dynamics in functional neurological disorder: The role of the temporoparietal 
junction. Brain, 148(1), 1–15. doi: 10.1093/brain/awae999