Migraine is a complex and common episodic neurological condition characterised by a combination of exteroceptive (symptoms resulting from sensory perception from the world external to the body such as photo- or phono-sensitivity) and interoceptive phenomena (symptoms resulting from sensory perception from the body itself such as premonitory auras). The impact of migraine extends beyond the excruciating pain and discomfort experienced during migraine attacks, influencing various aspects of an individual’s life, including emotional well-being, quality of life, and cognitive functioning (1). While research into migraine biology and management has made considerable advances, understanding of those individuals with chronic migraines and migraine co-morbidities remains limited (2, 3).

Interoception refers collectively to the body-to-brain axis of signals originating from the internal body and visceral organs, and encompasses the predictive representation and control of one’s physiological state—the state of “minimal selfhood” (4). Interoception, therefore, provides insights into how the body’s internal state and sensations are perceived and regulated. Changes in the body’s internal state trigger automatic physiological reactions and mental sensations like hunger, thirst, pain, and emotions. There is clinical and neuroimaging evidence that supports the involvement of interoceptive systems in migraine (5, 6). During the prodromal phase individuals with migraine may experience specific interoceptive feelings (e.g., fatigue, hunger) (7), while the headache phase is characterised by other interoceptive phenomena such as pain (8). Functional neuroimaging studies highlight the important role played by brain structures such as the brain stem, insula, and default mode network in migraine, which are also all important interoceptive hubs (1). Finally, discrepancies between objective and subjective aspects of interoception underpin the integrity of self-representation, a cornerstone of dissociation (9). We have previously demonstrated the relevance of this relationship in patients with functional seizures (10). Many individuals with migraine also suffer from functional neurological disorders and dissociative symptoms (11), including both psychoform (disconnection from one’s thoughts, emotions, or surroundings) and somatoform (disconnection from one’s body) dissociation (12, 13).

Given the frequent co-morbidity of functional seizures, dissociation, and migraine and the interoceptive abnormalities identified in other chronic pain disorders such as fibromyalgia and chronic low back pain (8, 11), we hypothesised that patients with migraine would have interoceptive abnormalities, and that these may correspond to levels of dissociation. While migraine is typically an episodic neurological disorder, dissociation has been shown to play an important role in the development of chronic migraine (14, 15). We therefore also hypothesised that there may be dissociative and interoceptive differences between individuals with chronic and episodic migraine. In this study, we sought to explore interoception and dissociation using cardiac interoceptive paradigms and self-report questionnaires.

Interoceptive function was assessed using a dimensional framework which included several self-report questionnaires, and cardiac interoceptive tests. The experiment was conducted in a quiet room with no way of telling the time. Participants were sat upright facing away from the experimenter and wore a finger pulse oximeter (Nonin Xpod 3012LP) connected to a laptop, providing an objective measurement of heart rate. The finger pulse oximeter method has been validated against electrocardiogram (ECG) for heartbeat timing (17, 18).

In this study we have identified for the first time abnormalities of interoception in individuals with migraine with a moderate to large effect size. These individuals had poorer interoceptive accuracy as indexed by the HTT, greater interoceptive sensibility, and greater ITPE in comparison to individuals who did not suffer from migraines. Greater levels of dissociation including psychoform and somatoform dissociation, as well as anxiety and depression, were also found in individuals with migraine. However, differences were also identified between individuals with episodic and chronic migraine. Greater somatoform dissociation was found in individuals with chronic versus episodic migraine. These findings in relation to dissociation are broadly in keeping with the published literature (13, 34). A few studies have investigated differences in dissociation between individuals with chronic and episodic migraine and have typically focused on psychoform dissociation, which is noted to be elevated, and an important mediator of the development of chronic migraines (15, 35). Importantly, greater ISPE and greater ITPE for the HTT and HDT were identified in individuals with chronic migraine, although only in the former case was this significant. Our GLMs highlight significant differences in ITPE between migraine and control groups and in ISPE between clinical migraine subtypes, while allowing for confounding variables. These findings overlap with those reported in patients with FMD and FS (10, 36), and may therefore be relevant to our mechanistic understanding of why migraines and functional movement disorders/functional seizures are frequently co-morbid (37, 38). Our findings of a correlation between dissociation and ITPE across all individuals with migraine are also in keeping with this finding. The overall task performances and questionnaire scores of healthy controls in the current study are similar to those in our previous study (10), supporting the validity of our current interoception and dissociation assessment methods, and emphasising the value of these findings.

The interoceptive findings in this study can be understood through a framework of predictive processing, given that subjective experience, including pain perception, has been recently re-conceptualised from the viewpoint of these frameworks (39). Historically, models of subjective experience were based on the idea that they emerge from bottom-up processing of sensory signals by the brain. In contrast, in predictive processing models, subjective experience emerges as perceptual inference based on the discrepancy between bottom-up sensory signals and top-down predictions (priors), and the process to minimise this discrepancy. The model has been used to explain exteroceptive modalities, such as vision, but has recently been extended to interoception (23, 24). In an extended interoceptive model, the sense of body ownership is determined by the minimization of the discrepancy between top-down predictions about the interoceptive state of the body and bottom-up incoming interoceptive signals evoked (directly) by autonomic control signals and (indirectly) by bodily responses to afferent sensory signals. Conversely, this model maintains that disorders of body ownership, such as dissociation, result from pathologically imprecise interoceptive predictive signals and increased discrepancy between top-down and bottom-up signals (9).

ITPE is a metric of the divergence between the subjective longer-term belief/self-model of general interoceptive ability and objective sensory sensitivity to interoceptive information. It can also be conceived as the discrepancy between the precision (weighting) afforded to subjective belief/self-model of general interoceptive ability, and the ability to adjust or update these predictions in the light of new bottom-up sensory information. Differences in ITPE between individuals with and without migraine may therefore reflect disordered bodily predictions by the brain. These may arise because the occurrence of migraine, regardless of whether it is episodic or chronic, contributes to increased levels of both the attention that an individual pays to their body, and their subjective perception of their accuracy in perceiving those bodily signals. However, there may be a lack of a corresponding increase in the objective accuracy of detection of bottom-up bodily signals due to structural changes in interoceptive networks in patients with migraine (1). Indeed, patients with migraine scored lower on objective tests of interoceptive accuracy and higher on interoceptive sensibility measures which index attention or weighting afforded to subjective bodily signal appraisal, compared to individuals without migraine. Furthermore, as we have demonstrated in previous work in patients with FS (10), there was a trend toward a correlation between ITPE scores and both psychoform and somatoform dissociation in individuals with migraine.

In contrast to ITPE, ISPE differed between migraine subgroups, and was greater in individuals with chronic migraine. ISPE represents the discrepancy between interoceptive accuracy and a participant’s trial-by-trial subjective opinion of their interoceptive accuracy. Prediction errors in this setting refer to state moment-to-moment discrepancies between expected and actual interoceptive signals, rather than trait-based differences between objective and subjective performance (as indexed by ITPE). Within predictive coding models, subjective bodily experiences arise from the discrepancy (i.e., interoceptive prediction error) between bottom-up interoceptive signals and top-down predictions of internal bodily states. This model proposes that adaptive, normal functioning relies on the relative precision (weighting) of error signals and predictions to adjust expectations and perceptions for any given context. In this model the persistence of imprecise error signals can lead to situations where highly precise but inaccurate priors or predictions (e.g., the presence of a headache in an individual who suffers with frequent headaches) may dominate the generative model for somatic stimuli, leading to perceived symptoms in individuals which are not a perfect correlate of bodily, physiological activity. This finding may therefore contribute to the development of central sensitization and chronic pain in individuals with chronic migraine, where the pain experienced by individuals is not in proportion to relevant bodily physiological activity, such as the activation of trigeminovascular neurons and subsequent inflammatory cascades, typically seen in episodic migraine (40). This mechanism may also be partly responsible for medication overuse being a risk factor in the transformation from episodic to chronic migraine and medication overuse headaches (MOH) (14). Here, frequent use of analgesics, including over-the-counter medications, will dampen bottom-up interoceptive signaling pathways (41), reducing opportunities for the brain to recalibrate priors in light of actual bodily states. This can result in an over-reliance on maladaptive top-down predictions of pain and a heightened sensitivity to interoceptive prediction errors when medication effects subside. Such a mechanism may perpetuate headache frequency and drive further analgesic use, culminating in MOH as a maladaptive allostatic response. In contrast, individuals who do not frequently use analgesics may avoid the maladaptive cycle underlying MOH. In such individuals, the nervous system continues to receive intact bottom-up interoceptive signals from nociceptive pathways, enabling accurate updating of top-down expectations and maintaining an appropriate precision balance between predictions and sensory input. The system remains better able to adapt to normal fluctuations in pain and avoids developing a self-reinforcing cycle of increased prediction error and increased top-down expectation of pain.

The findings in this study parallel our work in individuals with functional seizures where, ITPE predicted dissociation, and ITSE predicted seizure frequency significantly (10). These parallel findings may therefore reflect shared mechanisms underpinning the frequent co-occurrence of chronic migraines and functional seizures (11). Furthermore, this importance of interoceptive prediction errors in migraine is compatible with the recently proposed theory that migraine can be understood as an allostatic reset triggered by unresolved interoceptive prediction errors (5).

The inclusion of many individuals with very frequent or chronic daily headaches, all of whom met ICHD-3 criteria for chronic migraine, is an important feature of this study. While classical episodic migraine has well-characterised neurobiological mechanisms, patients with chronic and particularly daily headache often present with a more complex clinical profile that may not be fully explained by these models alone. Our data suggest that this subgroup is characterised by elevated dissociation and altered interoceptive prediction error, pointing to additional mechanisms involving self-representation and bodily awareness. These features overlap with those seen in functional neurological disorders and may reflect a broader set of transdiagnostic processes relevant to symptom chronification and central sensitisation. Rather than representing a source of diagnostic ambiguity, patients with daily headache may instead offer key insights into the multidimensional nature of migraine, particularly the transition from episodic to chronic forms. This supports the view that chronic migraine is not merely an extension of episodic migraine in frequency, but may involve qualitatively different underlying processes. As such, studying individuals with daily headache is essential for identifying mechanisms that contribute to migraine refractoriness, symptom amplification, and co-occurring functional symptoms. These findings may help to bridge biological and psychophysiological models of migraine and open new avenues for therapeutic intervention.

These findings have several potential clinical implications. First, they suggest that in patients with frequent or chronic migraine, especially those reporting dissociative experiences or high symptom burden not fully explained by classical mechanisms, clinicians might consider assessing aspects of interoceptive awareness or bodily self-perception. This could be as simple as including questions about how accurately patients perceive bodily signals or how distressed they are by them. Second, recognising altered interoceptive prediction errors as a possible contributor to symptom maintenance highlights a target for intervention. Emerging approaches such as interoceptive training, body-focused mindfulness, or even neuromodulatory techniques like vagus nerve stimulation might help recalibrate interoceptive processing, potentially reducing symptom amplification and preventing chronification. For example, non-invasive vagus nerve stimulation (VNS) represents a non-pharmacological treatment option for migraine. Interestingly, recent studies have revealed that VNS facilitates interoceptive processing and reduces an individual’s susceptibility to dissociation and disembodiment (42–44). The findings in this study suggest that one potential mechanism of action of VNS in migraine may be the modulation of interoceptive systems. VNS may therefore have a role in the prevention and treatment of chronic migraine and its comorbidities, such as dissociation. Finally, understanding that dissociation and altered bodily awareness may be intertwined with migraine could guide clinicians to screen for co-occurring functional neurological symptoms, inform psychoeducation, and foster a more integrated biopsychosocial treatment plan.

This is the first study to investigate interoception in individuals with episodic and chronic migraine, and in doing so, it highlights potential links to functional movement disorders/functional seizures. However, several limitations remain. First, the small sample size is a significant limitation. Although a statistically significant group difference in ISPE was observed, the post-hoc power for this comparison was relatively low indicating a limited ability to consistently detect this effect. Therefore, the findings in individuals with episodic and chronic migraine need to be replicated and extended with a broader spectrum of migraine patients. Second, although individuals taking medications with direct cardiac effects were excluded, some individuals were taking other medications, including as-required pain killers, and the effect of medication on our findings cannot be controlled. Since even simple over-the-counter analgesic medications may block bottom-up interoceptive signaling pathways and thereby alter prior beliefs and precision weighting, they should be considered when exploring the interoceptive mechanisms of migraine chronification. Further studies are therefore needed to clarify the influence of analgesic medication use on interoceptive processing. Third, while none of the participants included in the study experienced a migraine either during the assessments or earlier on the same day, no information was collected regarding the interval between their last (or subsequent) migraine attacks and the assessments. This makes it challenging to determine the direct influence of migraine attacks on our findings. Since headache attacks themselves may transiently alter interoceptive processing, further studies investigating the temporal dynamics of interoception following attacks are warranted. Fourth, knowledge and beliefs regarding one’s own heart rate may affect the HTT results, making it less representative as a measure of true interoception (45). In order to address this, we included the time tracking task as a control task. However, future studies should consider using other interoceptive tasks, such as heartbeat-evoked potentials or tasks assessing cardiac modulation of cognition, which may provide more implicit and objective indices (46, 47). Fifth, we acknowledge that certain potential confounding factors, such as educational level and socioeconomic status, were not collected in this study. These variables may plausibly influence interoceptive functioning, and this limitation should be considered when interpreting the findings. Sixth, the cross-sectional nature of the study limits causality-related conclusions. Seventh, while the finger pulse oximeter method used in the tasks has been validated against ECG for heartbeat timing, it does not provide the same level of temporal precision as ECG (17, 18). Finally, we investigated patients referred to a neurology clinic for their headaches, who by definition may represent a selective group with more disordered forms of migraine, compared to more common community-based individuals with migraine who do not require hospital referrals. Indeed, analysis of a broader spectrum of individuals would enable the assessment of whether those individuals with less disordered forms of migraines have better interoceptive accuracy. Therefore, the generalizability of our findings is limited, and further research is needed to address this issue.

Although the cross-sectional nature of the study precludes causal inference, interoceptive abnormalities may represent transdiagnostic mechanisms relevant to the chronification of migraine, and also to frequently occurring co-morbidities such as dissociation. In the future, interoception could represent a novel transdiagnostic biomarker and therapeutic target for patients with migraine.