Current neurobiological models of functional neurological disorder (FND) conceptualize symptoms as the product of disordered brain network functioning rather than structural damage. These models emphasize aberrant integration of sensory, motor, cognitive, and affective processes within large-scale brain networks. FND is therefore framed as a disorder of brain–mind interface and predictive processing, in which the brain’s expectations, prior beliefs, and attentional biases shape perception and motor output in a way that becomes maladaptive and clinically disabling.
A central theme is the role of abnormal predictive coding. In predictive coding frameworks, the brain continuously generates predictions about incoming sensory information and updates these predictions based on prediction errors. In FND, overly precise prior beliefs about bodily states, movement, or threat may dominate perception, while actual sensory evidence is down-weighted. When this imbalance occurs in motor systems, internally generated predictions can override normal motor commands, resulting in symptoms such as weakness, tremor, or gait disturbance despite preserved motor pathways. When it occurs in interoceptive and perceptual systems, patients may experience sensations or symptoms that are not explained by peripheral pathology yet are subjectively real and distressing.
Neurobiological models also highlight altered self-agency. Sense of agency—the feeling that one’s actions are self-generated—is thought to rely on accurate comparison between predicted and actual sensory consequences of movement. In FND, mismatches in this comparator system may be interpreted not as self-produced errors but as actions occurring without intentional control. This disruption in agency aligns with clinical descriptions of involuntary-feeling movements and the common patient report that symptoms are “happening to them” despite preserved motor capacity. Abnormal connectivity between motor regions and higher-order control networks such as the prefrontal cortex and parietal cortex is proposed as a neural substrate of this impaired agency.
Another important element of current models is the integration of emotion, attention, and motor control. The salience network, anchored in the anterior insula and dorsal anterior cingulate cortex, is thought to be overactive or dysregulated, leading to heightened detection of bodily signals and threat-related cues. This may bias attention toward symptoms and amplify their perceived intensity. Simultaneously, the default mode network and executive control networks appear to interact atypically with sensorimotor areas, allowing affective and cognitive states to directly shape motor output. In this framework, emotional arousal, trauma-related memories, or chronic stress can repeatedly trigger or maintain functional symptoms via their impact on these interconnected networks.
Neuroimaging-based models further emphasize that FND involves abnormal top-down modulation of sensorimotor circuits rather than a simple bottom-up failure of peripheral inputs. For example, increased prefrontal influence on motor regions could inhibit or distort motor commands, while abnormal limbic modulation might alter the gain on sensory processing. Structural mri studies have suggested subtle changes in gray matter volume or cortical thickness in regions involved in emotion regulation and interoception, such as the insula and anterior cingulate, whereas functional imaging points to dynamic, state-dependent alterations in network communication rather than fixed lesions.
Stress and early life adversity are incorporated into these models as factors that shape brain development and network organization. Repeated exposure to stress may sensitize limbic and salience networks, modify hypothalamic–pituitary–adrenal axis functioning, and alter connectivity with prefrontal regulatory regions. Over time, this can create a neural milieu in which bodily sensations and emotions are experienced as overwhelming, poorly regulated, and more likely to be expressed through motor and sensory systems. This “kindling” of networks supporting arousal and threat detection is proposed to lower the threshold for the emergence of functional symptoms when individuals encounter new stressors, illness, or injury.
Current frameworks also consider learning and reinforcement mechanisms. Once functional symptoms appear, they may be strengthened through associative learning, attentional focusing, and behavioral avoidance. For instance, repeated pairing of specific contexts or internal states with symptom onset can create powerful conditioned associations that perpetuate symptoms even in the absence of the original trigger. Negative reinforcement, such as temporary relief from obligations or distress when symptoms occur, can further consolidate these patterns. Neurobiologically, these processes are linked to plastic changes in cortico-striato-thalamo-cortical loops that support habit formation and action selection.
Contemporary models seek to reconcile the presence of voluntary motor system activation with patients’ experience of lack of control by positing dissociations between lower-level motor execution and higher-level conscious intention. Motor commands may be generated or modulated by networks involved in habitual or emotionally driven actions, while systems responsible for conscious intention and monitoring are bypassed or receive altered feedback. Disturbed connectivity between supplementary motor area, parietal cortex, and prefrontal regions has been proposed as one mechanism by which intention fails to align with action, contributing to the characteristic phenomenology of FND.
Another component of these models is the role of interoception and body representation. The insula and somatosensory cortices construct an integrated map of the body’s internal and external states. In FND, distortions in this map may arise from persistent prediction errors and maladaptive attention to bodily sensations. This can manifest as altered body ownership, distorted perception of limb position, or inconsistent sensory findings on examination. Neurobiological accounts suggest that these abnormalities reflect dynamic changes in network-level processing rather than static peripheral dysfunction, consistent with fluctuating clinical presentations.
Neurobiological theories also integrate the concept of metacognition, referring to how individuals monitor and interpret their own thoughts, sensations, and actions. In FND, heightened self-focused attention, catastrophic interpretations of symptoms, and rigid beliefs about illness may interact with abnormal neural signals arising from prediction and agency networks. This interaction can stabilize maladaptive explanatory models (“my leg just stops working”) that become deeply embedded and resistant to disconfirmation, reinforcing symptom persistence. At the neural level, aberrant coupling between default mode, salience, and executive networks is posited to underlie these metacognitive distortions.
Collectively, these neurobiological models converge on the idea that FND is a disorder of brain network function and information processing, in which mechanisms of prediction, emotion regulation, attention, and agency become dysregulated. Rather than viewing functional and organic disorders as binary opposites, current perspectives place FND along a spectrum of brain-based conditions, characterized by distinctive patterns of network-level dysfunction that are increasingly being delineated through neuroimaging, connectivity analyses, and translational research on the underlying mechanisms.
Structural and functional neuroimaging findings
Structural and functional neuroimaging studies in functional neurological disorder have moved the field beyond the assumption of “normal brains” and instead reveal distributed, condition-relevant alterations. Conventional structural MRI is typically unremarkable in routine clinical practice, which historically contributed to diagnostic uncertainty and at times to misattribution of symptoms as purely psychological. However, research employing voxel-based morphometry, cortical thickness analyses, and diffusion tensor imaging has identified subtle but reproducible group-level differences in gray and white matter. These changes tend to cluster in regions implicated in emotion regulation, interoception, sensorimotor integration, and self-agency rather than primary motor or sensory pathways alone.
Gray matter alterations have been observed in the anterior insula, dorsal anterior cingulate cortex, orbitofrontal cortex, and limbic structures including the amygdala and hippocampus. Some studies report increased gray matter volume or cortical thickness in insular and cingulate regions, whereas others find reductions, with variability likely reflecting differences in symptom profiles, illness duration, and comorbidities. In patients with functional motor symptoms, volume changes have also been described in supplementary motor area, premotor cortex, and parietal regions involved in movement planning and body representation. These findings align with neurobiological models that emphasize abnormal weighting of salience, interoceptive, and motor prediction signals, rather than frank tissue loss.
Diffusion-based techniques provide complementary evidence of microstructural abnormalities in white matter tracts connecting limbic, frontal, and sensorimotor areas. Reduced fractional anisotropy and altered diffusivity have been reported in the uncinate fasciculus, cingulum bundle, and corpus callosum, suggesting compromised integrity of pathways that support emotional processing, cognitive control, and interhemispheric coordination. Some studies indicate altered connectivity within corticospinal and cerebellar tracts in functional motor weakness or tremor, although these changes are often modest and heterogeneous. The emerging picture is one of subtle, distributed disruptions in communication between networks rather than focal lesions, consistent with a systems-level disorder.
Functional MRI has provided the most compelling evidence that FND is characterized by abnormal brain network dynamics during rest and during symptom provocation. Resting-state connectivity analyses show altered coupling between salience, default mode, executive control, and sensorimotor networks. For instance, patients frequently demonstrate increased connectivity between amygdala or insula and motor regions, alongside decreased connectivity between prefrontal control areas and motor or limbic structures. This pattern suggests heightened bottom-up emotional or salience-driven influence over motor systems, coupled with weakened top-down regulatory control, which may facilitate the emergence of involuntary-feeling movements or sensory experiences.
Task-based functional neuroimaging has been particularly informative in delineating mechanisms of impaired agency and abnormal motor control. During voluntary movement tasks, individuals with functional weakness or paralysis often show reduced activation in primary motor cortex contralateral to the symptomatic limb, with relatively preserved activation of premotor and supplementary motor areas. This underactivation is not seen when the same limb is moved passively by an examiner, indicating that peripheral pathways and basic sensorimotor responsiveness are intact. In some paradigms, increased activation is observed in prefrontal regions, temporoparietal junction, or limbic structures during attempted movement, suggesting recruitment of networks associated with monitoring, attention, and emotion at the expense of efficient motor execution.
Studies examining involuntary symptom expression, such as functional tremor or non-epileptic attacks, reveal distinct activation patterns compared with both voluntary movements and organic counterparts. Functional tremor is often associated with increased activity in prefrontal and anterior cingulate regions, as well as heightened coupling between limbic and motor networks. These findings support the idea that symptoms are driven by aberrant top-down influences and altered weighting of internally generated predictions. In functional seizures, neuroimaging during or shortly after episodes has shown increased limbic and insular activation, abnormal thalamocortical engagement, and disruptions in connectivity within networks supporting arousal, self-awareness, and motor control, helping differentiate these events from epileptic seizures that originate from focal cortical hyperexcitability.
Functional imaging of sensory processing in FND provides convergent evidence for altered gating and modulation of incoming information. Pain-related paradigms in patients with functional sensory symptoms or overlapping chronic pain disorders demonstrate enhanced activation in insula, anterior cingulate, and prefrontal areas, even when objective sensory stimuli are comparable to controls. At the same time, there may be reduced activity in primary somatosensory cortex or altered connectivity between somatosensory areas and higher-order networks, consistent with a mismatch between sensory input and its conscious representation. In functional visual or auditory symptoms, similar dissociations have been observed in modality-specific cortices and higher-order association areas, again highlighting that symptoms are linked to abnormal integration and interpretation rather than absence of basic sensory responses.
Neuroimaging paradigms examining sense of agency employ tasks in which visual or sensory feedback is manipulated to create discrepancies between intended and perceived actions. In such tasks, individuals with FND show altered activation in regions involved in self-monitoring and agency, including the temporoparietal junction, inferior parietal lobule, and prefrontal cortex. For example, when sensory feedback is experimentally perturbed, patients may fail to show the typical neural signatures of detecting mismatch, or they may display exaggerated responses in salience and limbic regions. These findings support the hypothesis that comparator mechanisms linking predicted and actual sensory consequences of movement are disrupted, contributing to the characteristic experience that movements or sensations are not fully self-generated.
Another important line of work has examined the interaction between emotional processing and motor or sensory systems. Using emotional faces, trauma-related cues, or stress-inducing tasks, researchers have shown heightened amygdala activation and increased insula and anterior cingulate responses in FND, often in conjunction with altered connectivity to motor, premotor, and brainstem regions. In functional motor symptoms, emotional stimuli can modulate motor cortex excitability and network dynamics more strongly than in healthy controls, indicating that affective states more readily influence motor output. This emotion–motor coupling provides a plausible neurobiological route by which stress, interpersonal conflict, or trauma reminders can precipitate or exacerbate attacks and motor or sensory disturbances.
Resting-state studies have further highlighted abnormal organization of intrinsic networks. Increased connectivity within the salience network and between salience and motor areas has been found in several cohorts, alongside decreased connectivity between prefrontal executive regions and motor or limbic networks. Some investigations report abnormally strong coupling between default mode network nodes and symptom-relevant sensorimotor regions, suggesting that self-referential and internally focused processing may be more tightly linked to bodily representations in FND. These network-level alterations are consistent with the prominence of symptom-focused attention, ruminative thinking, and heightened bodily awareness reported by many patients.
Neuroimaging research has also investigated state versus trait characteristics by comparing patients during symptomatic and asymptomatic periods. In functional motor disorders, transitions from symptom-free to symptomatic states are accompanied by dynamic shifts in network connectivity, including increased limbic–motor coupling and decreased prefrontal–motor connectivity at symptom onset. In functional seizures, preictal changes in limbic and salience network connectivity have been observed, followed by widespread disruptions during the event and gradual normalization afterward. These temporal dynamics underscore that FND involves flexible but maladaptive reconfiguration of networks, which can help explain symptom fluctuation and variability over time.
Electrophysiological techniques, such as EEG and magnetoencephalography, complement MRI findings by offering higher temporal resolution. In functional seizures, EEG often lacks the characteristic ictal epileptiform discharges, yet quantitative analyses reveal abnormal oscillatory patterns, altered functional connectivity, and atypical responses to sensory or cognitive tasks. In functional movement disorders, EEG and transcranial magnetic stimulation studies demonstrate preserved corticospinal conduction but altered cortical excitability, impaired inhibition, and changes in readiness potentials preceding movement. These findings reinforce the notion that motor pathways are structurally intact but subject to abnormal higher-order modulation.
Multimodal approaches that integrate structural MRI, functional connectivity, and electrophysiology are beginning to delineate more comprehensive signatures of FND. Combining diffusion imaging with resting-state fMRI, for instance, reveals that microstructural alterations in fronto-limbic tracts are associated with the degree of functional connectivity abnormalities between these regions. Similarly, linking TMS measures of cortical inhibition with fMRI markers of prefrontal and motor network engagement can help clarify how specific circuit-level mechanisms translate into observable symptoms. These converging data sets support the view that FND arises from aberrant interactions among distributed networks rather than isolated dysfunction in a single locus.
An important consideration across studies is the heterogeneity of patient populations. Functional motor symptoms, dissociative seizures, sensory disturbances, and functional cognitive complaints may share overlapping but non-identical neuroimaging signatures. Comorbid conditions such as anxiety, depression, chronic pain, and post-traumatic stress disorder can further influence brain structure and connectivity, complicating interpretation. Nonetheless, across different symptom types and cohorts, several recurring themes emerge: abnormal salience and limbic network engagement, disrupted prefrontal control of motor and sensory areas, altered interoceptive processing in the insula, and impaired integration within networks supporting self-agency and body representation.
These converging structural and functional findings have important implications for the conceptualization of FND. They demonstrate that the absence of gross lesions on standard clinical MRI does not imply an absence of brain-based pathology; rather, the relevant abnormalities lie at the level of network organization and dynamic modulation. As neuroimaging methods become more sophisticated, with improved spatial and temporal resolution and advanced analytic techniques such as graph theory and machine learning, it is becoming increasingly feasible to identify patterns of connectivity and activation that distinguish FND from both healthy controls and other neurological or psychiatric conditions. This neurobiological characterization provides a foundation for developing objective biomarkers of neural network dysfunction and for designing mechanistically informed interventions that target specific circuits and processes.
Biomarkers of neural network dysfunction
The search for biomarkers of neural network dysfunction in functional neurological disorder has prioritized modalities capable of capturing distributed, dynamic alterations in brain function rather than focal lesions. Neuroimaging-derived metrics, electrophysiological signatures, and psychophysiological measures are being evaluated both as state markers, which fluctuate with symptom expression, and trait markers, which reflect enduring vulnerability. Across these approaches, candidate biomarkers aim to index abnormalities in predictive processing, salience attribution, emotion–motor coupling, and sense of agency, while maintaining sufficient reliability and specificity to be clinically useful.
Functional MRI has generated some of the most promising candidate biomarkers through measures of resting-state and task-based connectivity. Resting-state analyses frequently demonstrate increased coupling between salience network hubs such as the anterior insula and dorsal anterior cingulate and motor or premotor cortices, together with reduced connectivity between dorsolateral prefrontal regions and limbic structures. Quantitative indices of this imbalance, such as salience–motor connectivity strength or prefrontal–amygdala coupling, have been associated with symptom severity, frequency of attacks, and functional impairment. In some studies, connectivity patterns differentiate individuals with functional seizures from those with epileptic seizures, suggesting potential utility in diagnostic classification. Importantly, these network-based markers are typically identified at the group level, and ongoing research focuses on improving their robustness and predictive power at the individual-patient level.
Task-based neuroimaging paradigms provide complementary biomarkers by probing specific mechanisms thought to be central to symptom generation. During voluntary movement attempts in patients with functional weakness, underactivation of primary motor cortex combined with overactivation of prefrontal and limbic regions yields composite activation profiles that distinguish them from both healthy controls and patients with organic hemiparesis. Quantitative metrics such as the ratio of prefrontal to primary motor activation, or the degree of limbic–motor coupling during attempted movement, have been proposed as biomarkers of impaired agency and top-down interference with motor execution. Similarly, paradigms that manipulate sensory feedback or introduce prediction errors reveal altered engagement of temporoparietal junction and inferior parietal lobule, producing measurable deviations in the neural correlates of mismatch detection that correlate with dissociation and symptom chronicity.
Biomarkers have also been sought in structural imaging, though here the emphasis is on network-relevant anatomy rather than gross pathology. Voxel-based morphometry and cortical thickness studies have identified reproducible gray matter differences in insula, anterior cingulate, orbitofrontal cortex, and temporoparietal regions. While these alterations lack diagnostic specificity when considered in isolation, multivariate pattern analysis combining regional volumes can yield structural “fingerprints” that partially separate individuals with functional motor symptoms or dissociative seizures from healthy and clinical controls. Diffusion tensor imaging adds microstructural markers of fronto-limbic and interhemispheric tract integrity; for example, reduced fractional anisotropy in the uncinate fasciculus and cingulum has been linked to early life adversity, emotional dysregulation, and symptom burden. Such measures may serve as trait biomarkers reflecting developmental influences on network configuration that predispose to FND.
Electrophysiological techniques have generated candidate biomarkers that leverage high temporal resolution to capture rapid network dynamics. In functional seizures, routine scalp EEG often appears normal during events, yet advanced analyses reveal alterations in spectral power, phase synchrony, and directed connectivity between frontal, limbic, and parietal regions. Quantitative EEG features, including atypical theta–gamma coupling or reduced long-range coherence, can distinguish functional from epileptic events in some cohorts with encouraging sensitivity and specificity. For functional movement disorders, movement-related potentials and Bereitschaftspotentials (readiness potentials) preceding tremor or jerks are frequently present but show abnormal timing, topography, or amplitude. Instead of the tightly localized, stereotyped pattern seen in voluntary movement, functional tremor may be preceded by more diffuse, variable premotor activity coupled with heightened frontal midline theta, suggesting increased reliance on attention and monitoring networks. These signatures have been proposed as biomarkers of altered voluntary control and agency.
Transcranial magnetic stimulation has yielded additional circuit-level biomarkers related to cortical excitability and inhibition. Patients with functional motor symptoms often exhibit preserved corticospinal conduction but abnormal intracortical inhibition or facilitation, as measured by paired-pulse paradigms. Short-interval intracortical inhibition may be reduced, or long-interval intracortical inhibition altered, indicating dysregulation of GABAergic interneuronal circuits. Moreover, measures of cortical silent period duration and input–output curves can demonstrate atypical modulation of motor cortex in response to cognitive or emotional tasks, indexing enhanced susceptibility of motor circuits to non-motor influences. When combined with neuroimaging, TMS-based metrics help map physiological abnormalities onto specific networks, reinforcing their potential as mechanistically grounded biomarkers.
Psychophysiological measures provide a further tier of biomarkers reflecting altered salience processing, arousal regulation, and interoception. Autonomic indices such as heart rate variability, skin conductance responses, and pupillary reactivity are frequently abnormal in FND, especially during symptom provocation or exposure to emotional stimuli. Reduced heart rate variability, indicative of impaired parasympathetic regulation, has been associated with greater symptom severity and poorer functional outcomes. In functional seizures, exaggerated or blunted autonomic responses preceding episodes may serve as prodromal state markers of imminent attacks. When synchronized with neuroimaging or EEG, these autonomic signals help delineate how changes in salience network activity translate into bodily arousal patterns that can precipitate or reinforce symptoms.
Beyond central and autonomic markers, peripheral biomarkers are being explored as indicators of stress-system dysregulation and inflammatory processes that might interact with neural networks. Altered diurnal cortisol rhythms, elevated inflammatory cytokines, and changes in neurotrophins such as brain-derived neurotrophic factor have been reported in subsets of patients, particularly those with comorbid chronic pain, depression, or post-traumatic stress. While none of these measures is specific to FND, their integration with neuroimaging-based connectivity markers may help identify biologically distinct subgroups characterized by heightened stress responsivity or impaired neuroplasticity, which could have implications for treatment selection and prognosis.
An emerging area of work focuses on multivariate and machine-learning approaches that combine diverse biomarkers into composite signatures. By integrating functional connectivity metrics, structural MRI features, EEG indices, and clinical variables, classification algorithms have achieved moderate to high accuracy in distinguishing FND from healthy controls and from some neurological comparators in research settings. These models often highlight the combined importance of salience–motor connectivity, prefrontal–limbic coupling, and measures of self-referential processing within the default mode network. Importantly, such multimodal approaches also facilitate the identification of latent dimensions underlying symptom clusters, providing a more nuanced representation of pathophysiology than categorical diagnoses alone.
Longitudinal studies have begun to test the stability and plasticity of proposed biomarkers, addressing whether they track symptom change and response to treatment. In patients undergoing physiotherapy, cognitive-behavioral therapy, or multidisciplinary rehabilitation, improvements in motor function or seizure frequency have been accompanied by normalization of salience–motor connectivity, increased prefrontal regulation of limbic regions, and changes in movement-related potentials. The degree of biomarker change often correlates with clinical improvement, suggesting that these markers index modifiable network mechanisms rather than fixed abnormalities. Such findings support the use of biomarkers as surrogate endpoints in clinical trials and as tools to monitor engagement of targeted neural circuits during intervention.
Despite these advances, several challenges limit the immediate clinical deployment of biomarkers of neural network dysfunction. Many findings are based on small, heterogeneous samples; protocols and analytic pipelines differ widely across studies; and there is substantial overlap between FND and other conditions in many candidate measures. Motion artifacts, medication effects, and comorbid psychiatric or pain disorders can confound neuroimaging and electrophysiological results. Standardization of acquisition and analysis, preregistration of hypotheses, and replication in independent cohorts are essential steps toward establishing robust, generalizable biomarkers. Moreover, the field must grapple with ethical issues surrounding the communication and use of biomarkers, ensuring that they enhance rather than undermine patient trust and that they are not misinterpreted as definitive lie-detection or willfulness tests.
Conceptually, the most promising biomarkers are those that are tightly linked to specific mechanisms—such as aberrant predictive coding, disrupted agency, or heightened emotion–motor coupling—and that can be modulated by targeted interventions. Network-based neuroimaging measures, physiologically grounded TMS and EEG indices, and integrated psychophysiological profiles are well positioned to fulfill these criteria. As these markers are refined and validated, they offer the prospect of stratifying patients according to underlying network dysfunction, predicting who is most likely to benefit from particular therapies, and guiding the development of novel treatments that directly engage the circuits implicated in symptom generation and maintenance.
Clinical applications of neuroimaging and biomarkers
Clinical use of neuroimaging and other biomarkers in functional neurological disorder currently centers on clarifying diagnosis, improving patient education and engagement, guiding treatment planning, and monitoring change over time. While most applications remain adjunctive rather than decisive, the accumulating data on network-level mechanisms have begun to influence how clinicians structure assessments and communicate with patients. Rather than using mri or EEG solely to “rule out” structural or epileptic pathology, clinicians increasingly conceptualize these tests as tools that can also help “rule in” a brain-based disorder of network function, provided they are interpreted in the context of positive clinical signs and established diagnostic criteria.
In everyday practice, structural mri and standard EEG retain a primary role in excluding alternative neurological conditions such as stroke, tumor, inflammatory disease, or epilepsy. However, awareness of typical FND patterns—such as normal structural scans despite marked disability, non-epileptiform EEG during attacks, and preserved corticospinal conduction on neurophysiology—supports diagnostic confidence when positive clinical signs are present. Some centers incorporate more advanced imaging or quantitative EEG on a case-by-case basis to address persistent diagnostic uncertainty, particularly in patients with overlapping organic and functional pathology or atypical presentations. In these situations, findings such as preserved or only subtly altered connectivity in primary motor pathways, in contrast to marked alterations in salience or limbic networks, may help clinicians argue against progressive degenerative or demyelinating disease.
One of the most powerful clinical applications of neuroimaging and related biomarkers lies in communication of the diagnosis. Many patients struggle with explanations that seem to divide conditions into “physical” and “psychological” in a way that feels invalidating. Illustrative use of neuroimaging—whether by describing or, when available, showing anonymized activation maps from research studies—allows clinicians to frame FND as a disorder of brain networks and predictive processing rather than one of structural damage or malingering. Explaining that functional scans in FND demonstrate overactivity in regions processing threat and attention, with under-recruitment of efficient motor control circuits, can make the diagnosis more tangible and reduce stigma. Patients may better understand statements such as “your brain is working too hard in the wrong places at the wrong times,” which can open the door to acceptance and engagement with rehabilitative strategies.
Diagnostic conversations can also draw on electrophysiological and autonomic markers to emphasize the involuntary nature of symptoms. For example, in functional tremor, demonstration of entrainment or distractibility on accelerometry or EMG, alongside preserved motor evoked potentials on TMS, helps show that the motor system is intact but misregulated. In dissociative seizures, video-EEG recordings that capture events without epileptiform activity, accompanied by postictal responsiveness patterns distinct from epileptic seizures, provide objective evidence that episodes are real, brain-based events driven by abnormal network states rather than intentional behavior. When carefully explained, such findings validate symptom experience while reinforcing the message that the condition is reversible through targeted therapy.
Neuroimaging and biomarkers are increasingly influencing treatment planning by helping clinicians conceptualize which circuits and processes to prioritize in intervention. Evidence of heightened salience network engagement and strong limbic–motor coupling has encouraged the development and refinement of therapies that explicitly target attention, interoception, and emotion–motor interactions. Physiotherapy protocols for functional motor disorders, for example, now commonly emphasize external focus, distraction, graded exposure to feared movements, and retraining of automatic rather than effortful control. These strategies are directly informed by neuroimaging research showing that excessive self-monitoring and prefrontal overengagement interfere with efficient motor execution, whereas more automatic, goal-directed movement relies on distributed sensorimotor networks that can be re-engaged.
Similarly, neurobiological models and connectivity findings have supported the integration of psychological treatments that focus on predictive processing, threat perception, and agency. Cognitive-behavioral and psychodynamic approaches can be framed as methods to recalibrate the brain’s predictions about bodily states, reduce catastrophic interpretations of sensations, and modify rigid illness beliefs that reinforce maladaptive neural patterns. Clinicians can explain that by addressing trauma, chronic stress, or maladaptive attentional habits, therapy is literally helping to “rewire” circuits linking limbic, salience, and motor systems. This framing is particularly helpful for patients who are initially skeptical of psychological interventions but are receptive to the idea of changing brain networks through learning.
Neuromodulation techniques such as repetitive TMS and transcranial direct current stimulation represent another area where biomarkers guide clinical application. Although still largely in the realm of controlled trials and early implementation in specialized centers, these interventions are designed to modulate network excitability in regions implicated by neuroimaging, such as supplementary motor area, dorsolateral prefrontal cortex, or insula. Target selection, stimulation parameters, and outcome evaluation increasingly rely on connectivity-based maps and physiological readouts. For example, abnormal prefrontal–motor connectivity on resting-state fMRI or reduced intracortical inhibition on TMS can inform which cortical targets might be most suitable, while changes in these same measures during a treatment course can provide mechanistic evidence that the intervention is engaging the intended circuit.
Biomarkers are also beginning to play a role in prognostication. Several studies suggest that certain baseline features—such as more normalized connectivity between prefrontal and limbic regions, higher heart rate variability, or more flexible movement-related potentials—are associated with better response to physiotherapy or psychotherapy. Conversely, pronounced structural or microstructural alterations in fronto-limbic tracts, very low autonomic flexibility, or rigid, highly stereotyped EEG patterns might indicate greater chronicity or the need for more intensive, multidisciplinary intervention. Although these associations are not yet strong enough to dictate individual-level decisions, they offer clinicians additional factors to consider when setting expectations and designing follow-up plans.
In certain clinical contexts, particularly epilepsy monitoring units and movement disorder clinics, quantitative biomarkers are directly used to distinguish functional from organic events and to determine eligibility for procedures. Video-EEG monitoring, for example, combines behavioral observation, ictal semiology, and EEG signatures to differentiate dissociative seizures from epileptic seizures before patients undergo invasive evaluations or surgery. In movement disorders, analysis of tremor frequency spectra, coherence across limbs, and responses to distraction help prevent unnecessary deep brain stimulation or ablative procedures. Here, the clinical application is not simply negative—avoiding interventions that target presumed structural pathology—but also positive, by directing patients earlier to FND-informed rehabilitation.
Beyond individual diagnostics, neuroimaging and other biomarkers increasingly shape the design and evaluation of clinical services for FND. Specialized multidisciplinary clinics draw on research evidence about shared and distinct mechanisms across symptom types to co-locate neurologists, psychiatrists, psychologists, physiotherapists, and occupational therapists. For instance, recognition that heightened salience processing and impaired emotion regulation cut across functional motor symptoms, dissociative seizures, and functional sensory complaints supports the development of transdiagnostic group programs teaching attention reorientation, grounding, and interoceptive skills. At the same time, awareness of symptom-specific network patterns encourages tailoring of motor retraining tasks, seizure-management strategies, or sensory integration exercises to the predominant circuit-level disturbances in each patient.
Biomarkers are beginning to be incorporated into routine outcome monitoring in some research-oriented clinics, where repeated measures of resting-state connectivity, quantitative EEG, or autonomic function are collected alongside clinical scales. Although not yet standard of care, these approaches illustrate how objective indices of network function might eventually complement symptom reports and functional scores in tracking progress. For patients, seeing changes in physiological measures—such as improved heart rate variability, reduced hyperconnectivity between limbic and motor regions, or more normalized movement-related potentials—can reinforce a sense of agency and hope, demonstrating that intensive rehabilitation is producing measurable changes in the brain and body even before full symptomatic recovery.
Educational initiatives for clinicians also draw heavily on neuroimaging and biomarker research to improve recognition and management of FND. Training materials often include case-based vignettes paired with imaging or EEG examples that highlight typical patterns, emphasizing that “normal” structural scans do not equate to absence of pathology but rather point toward network-level dysfunction. This helps reduce underdiagnosis and misdiagnosis, encourages earlier discussion of FND as a primary working diagnosis rather than a late “diagnosis of exclusion,” and fosters collaboration across neurology, psychiatry, and rehabilitation services. As understanding of mechanisms deepens, teaching can move beyond broad reassurance that the condition is real and brain-based to more specific explanations of how prediction, attention, and emotion interact to shape motor and sensory output.
At the interface of clinical care and research, biomarkers are increasingly used to stratify participants in clinical trials and to test mechanism-based interventions. Stratification according to baseline connectivity profiles, autonomic markers, or TMS-derived measures of cortical inhibition allows investigators to identify which subgroups respond best to particular treatments, accelerating progress toward personalized care. For example, individuals with pronounced limbic–motor coupling might be prioritized for interventions combining trauma-focused therapy with motor retraining, whereas those with prominent attentional dyscontrol might receive modules emphasizing metacognitive strategies and external focus. By feeding trial results back into clinical algorithms, this bidirectional flow between research and practice gradually refines how biomarkers are deployed in real-world settings.
Despite these promising applications, clinicians must remain cautious in how they present and interpret neuroimaging and biomarker data with patients. Overemphasis on subtle or group-level findings can inadvertently imply a degree of diagnostic certainty or individualized precision that current science does not yet support. Conversely, absence of visible abnormalities on routine scans should not be framed as evidence against a brain-based disorder, but rather as characteristic of conditions rooted in network dynamics rather than structural lesions. Thoughtful communication requires acknowledging both the strengths and limitations of current biomarkers, using them to support a coherent biopsychosocial explanation that integrates life history, psychological factors, and neural mechanisms without reducing the person to their scan or their connectivity pattern.
Future directions and research priorities
Future work in this field needs to systematically address the heterogeneity of functional neurological disorder and move toward biologically informed subtyping. Large, multicenter cohorts with harmonized protocols will be essential to disentangle symptom-specific from transdiagnostic features and to separate effects of comorbid depression, anxiety, chronic pain, or post-traumatic stress from FND-specific mechanisms. Deep phenotyping that integrates detailed clinical characterization, neuropsychological profiling, life history, and standardized assessment of dissociation, trauma exposure, and illness beliefs with neuroimaging and electrophysiology will allow clustering of patients based on shared network alterations rather than on descriptive symptom labels alone. Such an approach is critical for developing targeted interventions and for understanding why ostensibly similar presentations respond differently to the same treatment.
Standardization of neuroimaging and electrophysiological methods represents another high priority. At present, variation in MRI acquisition parameters, preprocessing pipelines, connectivity metrics, and task paradigms hinders replication and cross-study comparisons. Consensus guidelines for core imaging protocols in FND—covering resting-state fMRI, structural sequences, diffusion imaging, and basic task batteries probing motor control, agency, and emotional processing—would greatly facilitate pooled analyses and meta-analytic work. Parallel efforts are needed in EEG and magnetoencephalography to define minimal reporting standards, reference tasks, and analytic frameworks for spectral, connectivity, and event-related measures. Such standardization would accelerate the validation of candidate biomarkers and clarify which findings are robust enough to warrant translation to clinical settings.
Advances in analytic approaches will also shape future directions. Graph-theoretical measures of network topology, dynamic functional connectivity analyses, and computational modeling based on predictive coding frameworks offer promising avenues to capture the temporal evolution of abnormal network states. Machine-learning techniques can be applied not only to classification but to discovery of latent dimensions that cut across traditional categories such as functional motor disorder and dissociative seizures. Incorporating priors about circuit architecture and known physiological constraints into these models may improve interpretability and link statistical patterns more directly to plausible neurobiological mechanisms. A key goal is to develop models that can generate testable predictions about how specific interventions should modify network dynamics and, in turn, symptoms.
Longitudinal research is crucial for distinguishing vulnerability markers from changes secondary to chronic illness and for understanding trajectories over time. Prospective studies following individuals from the earliest presentation, or even from high-risk states such as acute stress reactions or medically unexplained transient neurological symptoms, could identify which combinations of psychological, social, and neural factors predict persistence versus resolution. Repeated neuroimaging, EEG, and psychophysiological assessments over months to years would allow characterization of how networks reorganize with treatment, relapse, or spontaneous improvement. Embedding such longitudinal designs within routine care pathways, for example in epilepsy monitoring units or functional movement clinics, may be a pragmatic way to accrue sufficiently large and representative samples.
Another major frontier lies in mechanistically informed intervention development. Existing therapies—physiotherapy, occupational therapy, cognitive-behavioral therapy, psychodynamic psychotherapy, and multidisciplinary rehabilitation—are effective for many patients, but their active ingredients at the neural level remain poorly specified. Future studies should use neuroimaging and physiological measures as proximal readouts of target engagement, explicitly testing whether treatments normalize predicted abnormalities such as excessive salience–motor connectivity, impaired prefrontal regulation of limbic regions, or disrupted agency-related networks in parietal and frontal cortices. Adaptive trial designs that iteratively refine interventions based on intermediate biomarker changes can help converge more rapidly on optimally targeted protocols.
Neuromodulation provides a particularly promising arena for such mechanism-driven work. Repetitive transcranial magnetic stimulation, transcranial direct current stimulation, and emerging approaches such as transcranial focused ultrasound can be directed at specific cortical or thalamic nodes within networks implicated by prior research. Future trials should systematically compare different targets and stimulation parameters, guided by connectivity maps and individualized structural data, to determine which configurations most effectively shift maladaptive network states. Combining neuromodulation with concurrent psychological or physical therapy may prove especially powerful, with stimulation used to transiently enhance plasticity or reduce pathological network dominance while behavioral interventions consolidate new patterns of perception and action.
Integration of peripheral and central biomarkers is another area requiring sustained effort. Autonomic indices, endocrine markers, and inflammatory profiles offer windows into how systemic physiology interacts with brain networks to shape symptom expression. Future studies should routinely include measures such as heart rate variability, skin conductance, cortisol dynamics, and circulating cytokines alongside neuroimaging and EEG, with analyses aimed at modeling bidirectional influences between central salience and interoceptive circuits and peripheral arousal systems. Multimodal datasets of this type can clarify why certain patients exhibit pronounced bodily manifestations of stress or dissociation and may highlight new targets for pharmacological or behavioral interventions focused on autonomic regulation.
From a translational standpoint, developing clinically usable tools based on research findings is an urgent priority. Many candidate biomarkers are currently derived from complex connectivity analyses or bespoke experimental paradigms that are impractical for routine use. Future work should focus on simplifying and operationalizing these measures into protocols that are feasible in standard clinical environments, such as brief resting-state scans with automated analysis pipelines, short EEG tasks with real-time classification, or portable psychophysiological assessments. Validation in real-world clinical cohorts, with attention to sensitivity, specificity, cost-effectiveness, and interpretability for non-specialists, will determine which tools can realistically augment diagnostic and treatment workflows.
Implementation science frameworks will be essential to bridge the gap between research and everyday practice. Studies should evaluate how neurobiologically informed explanations and use of illustrative imaging or physiological data affect patient understanding, engagement, and outcomes in diverse healthcare settings. Trials comparing different communication strategies—for example, purely verbal explanations versus explanations supplemented with anonymized neuroimaging examples—could quantify the added value of “showing the brain” in building trust and reducing stigma. Similarly, pragmatic evaluations of multidisciplinary service models informed by network-based conceptualizations can clarify which configurations of care yield the best functional outcomes and are sustainable within varying health systems.
Ethical and sociocultural considerations must remain central as the field progresses. As neuroimaging and biomarkers become more sophisticated, there is a risk that they will be perceived as definitive arbiters of diagnosis or intent, despite their probabilistic and group-derived nature. Future guidelines should address how to communicate biomarker results transparently, including their limitations, and how to avoid reinforcing dualistic notions that only visually demonstrable abnormalities are “real.” Research should also examine how different populations—across cultures, socioeconomic backgrounds, and healthcare systems—interpret brain-based explanations, ensuring that advances do not exacerbate disparities in access to care or understanding of the disorder.
Collaborations across disciplines and conditions will play a pivotal role in shaping future priorities. FND shares overlapping mechanisms with other disorders characterized by altered interoception, prediction, and agency, including chronic pain syndromes, depersonalization and derealization, somatic symptom disorders, and certain anxiety-spectrum conditions. Comparative studies that explicitly test shared and distinct neural signatures can refine transdiagnostic models and encourage cross-fertilization of treatment approaches. Likewise, partnerships with computational neuroscientists, engineers, and data scientists will be needed to handle large multimodal datasets, develop interpretable models, and design novel digital tools for assessment and intervention.
Active involvement of people with lived experience of FND in setting research agendas, designing studies, and interpreting findings is critical. Co-production models can help ensure that priorities such as reducing diagnostic delay, improving communication, and addressing functional impairment and quality of life guide the choice of research questions and outcomes. Incorporating patient perspectives into the development of neuroimaging-based explanations and decision aids will increase their acceptability and relevance. As the field moves toward more precise characterization of mechanisms and personalized applications of biomarkers, grounding these advances in the realities and preferences of patients and families will be essential to achieving meaningful impact on care.
