Reframing fibromyalgia
It is well documented that pain can lead to significant changes in the nervous system whereby sensations become persistent. This leads to many diagnoses with chronic pain at their core – including fibromyalgia, which I have had myself since age 19.
A key mechanism underlying these changes is known as “central sensitisation”. The nervous system essentially undergoes “wind up” processes, leading to persistent pain amplification. As a result, stimuli such as light touch can evoke severe pain, and stimuli that would normally be mildly painful can be excruciating. Sensitised nervous systems further perpetuate pain cycles, for example by disturbing sleep (side note: I suspect that the Princess in “The Princess & The Pea” had fibromyalgia).
If central sensitisation is such a key culprit in fibromyalgia, what’s behind it?
I have read many papers on fibromyalgia (I am a Neuroscientist and Psychologist, and have had an interest in chronic pain since my undergraduate degree – ironically I did my dissertation on this topic before my diagnosis). Although there is a relatively scant amount of research considering the number of people with fibromyalgia (according to the NHS, 1 in 20 people may have some degree), a myriad of physiological and psychological changes have been implicated in central sensitisation – particularly involving the immune system and neural pain pathways. Central sensitisation appears to be a complex process with many threads to tease and pull at…with little funding available to do so.
What’s my theory? The context:
“Predictive processing” has emerged as a transformational theory in Neuroscience. This theory has radically reshaped our understanding of the mind, and has recently been adapted to catalyse advancements in AI. At its core, predictive processing is about how brains processes information. It posits that your brain is an extraordinary prediction machine, constantly creating and updating a mental model of your world. It uses this world model to predict what will happen next.
Your brain actively works to ensure its world model is as accurate as possible, by measuring the difference between what it expected to happen, and what it actually experienced. This difference is known as the “prediction error”. Your brain uses this to update its world model and increase the accuracy of predictions.
Expanding this, the Active Inference Framework (AIF) posits that we don't just passively predict the world, but actively engage with it to minimise prediction errors (also known as “surprise”). Our drive to reduce surprise is a key principle in the AIF. Ultimately, our brains choose beliefs and actions that align our sensory experiences and predictions – reducing the “surprise” encountered.
The details:
I believe that "central sensitisation" – the process widely accepted to cause pain amplification – can be effectively and usefully conceptualised under these two frameworks. The primary culprit is prediction error, caused by a backdrop of chronic missing information/misinformation about the body. While the resulting “wind up” processes are the brain’s attempt to minimise these errors, in order to enhance survival and reduce the amount of surprise experienced.
Let’s walk it through.
Prediction errors are a normal part of everyday life - it’s very hard for your brain to infer information about the world outside of its dark skull. However, some factors may make it particularly hard for the brain to reconcile the difference between what it expected to happen and what it experienced (i.e. reduce its prediction error). Theoretically, intense periods of physiological or psychological stress, and underlying information deficits may contribute to prediction errors.
It is noteworthy that infections and stressful life events are commonly reported as a trigger of fibromyalgia, and many of the conditions commonly found alongside fibromyalgia include profound information deficits. I believe that a chronic lack of bodily information/misinformation is a key precipitator of the central sensitisation mechanisms underlying fibromyalgia.
Conceptualising chronic pain under predictive processing frameworks is not new (see further reading below), however I have not come across any articles presenting chronic missing information and misinformation as a core mechanistic feature.
The evidence for the “missing information” in fibromyalgia lies in its comorbidities. For example, an especially common comorbidity is joint hypermobility (one study found around 80% of people with fibromyalgia have this). A core feature of hypermobility is inaccurate and unreliable body sensations, caused by abnormalities within the proprioceptive system (which sends information to the brain about where your body is in space). Research also suggests that approximately half of people with fibromyalgia have small-fiber polyneuropathy (damage to the small sensory nerves in the skin, peripheral nerves, and organs). Additionally, a recent study identified alexithymia (difficulty identifying feelings and distinguishing feelings from bodily sensations) in around half of people with fibromyalgia. Diminished interoception (problems sensing internal signals from the body), and a high prevalence of ADHD (indicating attentional deficits) are also common in people with fibromyalgia.
I suggest that such profound gaps in information lead to the brain being unable to create a sufficiently detailed and valid model of the body. As a result, it is overwhelmed by “surprising” bodily signals, and ultimately, unable to determine when and how it is in danger.
So, how might the brain deal with such a chronic and profound threat to its existence? By increasing the weight of information vital to its survival – pain signals. Essentially, the brain responds by amping pain up to 11. I believe this would be more likely to happen in systems where there are true indications of damage present – whether due to inflammation caused by arthritis, ligament damage from hypermobility, or immune signals from infections or toxins. All of which are common comorbidities in fibromyalgia.
Importantly, when acute damage or inflammation occurs in people with a chronic lack of bodily information/misinformation, then enhancing and persisting pain signals may reduce prediction errors (i.e. surprise). In this sense, chronic pain is learned pain (similarly to how anxiety has recently been conceptualised as learned uncertainty within the active inference framework).
What might this mean, practically?
This theory is suggestive of an approach to fibromyalgia treatment which combines addressing any underlying pain/damage signals (such as from joint issues or infections), and recalibrating predictions about the body and threat levels.
In particular, identifying any potential underlying informational deficits relating to the body and threat, and working to improve or counteract these. Practically, this means screening for common comorbidities, and choosing treatment strategies which address their information deficits/imbalances.
For example:
Comorbid hypermobility – physical exercises and therapies to improve balance and body awareness.
Comorbid small fibre neuropathy – addressing any underlying causes and taking medications to improve small fibre sensory nerve function.
Comorbid interoceptive deficits – exercises to improve interoceptive awareness (such as mindfulness and breathing techniques).
Comorbid alexithymia – exercises to improve emotional awareness (such as journaling).
Comorbid anxiety and depression – therapies and exercises focused on reducing threat biases.
Generally, it has been suggested that model recalibration takes place during sleep. Therefore practices to improve sleep quality are also strongly implicated. It has also been suggested that meditation and self-talk (similar in the practice to affirmations – e.g. “The pain is fading”) may help recalibrate models.
Please note, this article is entirely theoretical and does not constitute medical advice. I will work on a more detailed and better written peer-reviewed version! In the meantime, I welcome feedback and comments from researchers, practitioners working with patients, and people with experience of chronic pain.
Further reading
Hierarchical models of pain: Inference, information-seeking, and adaptive control
Agency and Expectations in Pain Treatment: An Investigation of the Active Inference Model