Note (added 9/29/24) - I do not think this explains most Long COVID symptomology. Even when I wrote this, I was thinking this theory was maybe only relevant in ~50% of cases. The main drivers are probably persistent virus, Epstein-Barr reactivation, immune dysregulation, and mitochondrial damage. Another point I would like to stress here at the start is this is not a psychological theory. This theory has to do with unconscious structures deep in the brain which are responsible for self modeling and fatigue generation. While I hypothesize that conscious thoughts may be able to interact with this system, that is pure speculation on my part and I have little idea if it is true.
Epistemic status: speculative
“… a very deep, primal part of your nervous system protects you from damage and shuts you down… After a while, they (your symptoms) become self-fulfilling. The symptoms shut you down, and then you feel the symptoms, and that tells your brain that you’re ill and it thinks your in danger, and that feeds your perception of the symptoms.” - Dr. Paul Garner, speaking with Amy Engkjer.
In this post I’m going to discuss and elaborate on a specific theory for Long Covid fatigue. I got this theory from Dr. Paul Garner, who in turn got it from Dr. Vegard Wyller.
In a nutshell, the theory says that Long Covid fatigue is caused by faulty wiring in the brain’s fatigue-generating circuitry. This may be conceptualized as the brain having a faulty internal model of the body’s energy state which generates “false alarm fatigue” that the body is running out of energy. Recovery involves reconditioning these circuits by gradually increasing activity levels so the brain learns that the body can handle more exertion.
Note: I’m not saying this theory explains all fatigue in all Long Covid patients, but that it might explain some of what’s going on. Long Covid is a syndrome that appears to have many potential causes at play. See my brief review of potential biological contributors.
I am publishing this in conjunction with a second post, “Psychosomatic causes of Long Covid suffering”. In appendices to that post, I summarize my personal Long Covid experience and discuss biological potential causes for Long Covid such as Epstein-Barr virus reactivation and persistant SARS-CoV-2 virus.
What is fatigue?
First, let’s be clear that fatigue is different from sleepiness (sometimes also called “tiredness”). Fatigue is what you feel after completing an especially heavy work out or doing mentally taxing work for several hours. It’s also what you feel when you get sick. Fatigue should also be distinguished from lack of drive, lack of motivation, or akrasia.
The neurobiology of fatigue is poorly understood. A common intuition about fatigue is that it is generated by “the body running out of energy” — basically a battery running down. Of course, there must be some truth to this. However, fatigue is much more complex. For instance, fatigue is one of the most common symptoms of depression, but depression is not generally thought to be related to the body running out of energy. Also, many psychiatric drugs can also cause fatigue via actions on neurotransmitter systems in the brain.
Before we dive into the “false alarm” theory, let’s look at some problems with the naive view that fatigue is directly caused by “the body running out of energy”, also called the “biological limits” theory:
Problems with the “biological limits” theory of fatigue
What follows I’ve cribbed from the book Endure by Alex Hutchinson. Hutchinson asks us to consider runners in an Olympic race like the 10,000 meter. Since it’s the Olympics, an event the athletes trained years for, it's safe to assume that the runners will be trying as hard as possible to run as fast as they can. So the nice thing about considering this situation is we can disregard the possibility that runners are holding back for psychological reasons like anxiety or lack of willpower.
Under the “biological limits” theory of how fatigue works, fatigue is a direct consequence of biological limits being hit, such as lactate build up, failing muscle fibers, muscle glycogen reserves running low, low blood glucose, or low blood oxygen. We should expect these sources of fatigue, if present, to increase monotonically over time during the race. So, as fatigue mounts we should expect the speed of the athletes to decrease over time - maybe not linearly, but their speed should always be decreasing as time goes on. For most of the race, this is indeed what happens. However, toward the end of the race, athletes invariably speed up. More specifically, studies show that athletes consistently increase their pace towards the last 5 to 10% of a race.
If you’ve ever run a race, you know that subjectively fatigue increases over time, but then suddenly and mysteriously decreases as the finish line comes into view. This strange phenomenon is remarkably consistent, and I have experienced it myself at nearly every 5k race I’ve run. The question is, where does that “boost of energy” for that final sprint come from?
Tim Noakes, a professor of exercise and sports science at the University of Cape Town, developed the central governor theory to explain this and other phenomena the “biological limits” view of fatigue has trouble with. The central governor theory posits that fatigue is generated by circuitry in the brain that creates a mental model of the body. Using that model, the central governor can estimate how much energy the body has in reserve and how close organs are to being damaged due to biological limits being reached (for instance by lack of oxygen). One the basis of this model, the central governor carefully limits the body’s use of energy so as to leave an emergency reserve of energy and to prevent irreversible muscle or organ damage. All of this happens unconsciously, outside of our direct control.
Crucially, the central governor extrapolates into the future when making its calculations. When the finish line comes into view, the central governor is smart enough to realize that the race is about to end, and you’ll soon be resting, eating, and drinking. So, the central governor “eases off the breaks” towards the end of the race.
The central governor theory can explain a lot of interesting phenomena that the “biological limits” view can’t explain:
Unexplained feats of strength. Under special circumstances people exhibit feats of strength that defy explanation. There are numerous stories, for instance, of people being able to lift a car to save a person trapped underneath. The central governor theory provides an explanation - in rare circumstances the governor may “remove the breaks” to enable a person to save the life of another or to save their own life.
Gunshots improve exercise performance. Studies show unexpected gunshots can improve how much weightlifters can lift.1 It’s possible the gunshot causes the governor to think that the body is in imminent harm, so it releases the brakes in response.
Complete collapse is rare in sports. Under the “biological limits” view of fatigue, we should expect that ultra-long distance runners and other endurance athletes should push themselves to complete exhaustion, causing them to collapse because they are literally out of energy. In practice, this is very rare. When runners do collapse, it’s more often due to dehydration or heat stroke, not that they’ve run out of energy. In fact, it’s common for athletes to still walk around after a long race, or even take a victory lap, showing they hadn’t hit or surpassed any biological limit.
Certain placebo effects. A study showed that swishing a sugary drink in the mouth and spitting it out increased running speed in athletes trying to run a 12.8 km run as fast as possible. In another study, endurance cyclists started pedaling harder after ingesting a placebo drink they were told had carbohydrates in it. Both these studies can be explained as cases where the central governor is tricked into thinking it will get more energy soon, so it releases the brakes.
Chronic fatigue syndrome. Despite decades of research and hundreds of studies, researchers have not found a consistent bodily cause for fatigue in CFS/ME. Speaking very speculatively, Prof. Noakes hypothesizes that CFS/ME may be caused by a central governor with a faulty/ broken model.
Crucially, the central governor’s model is learned, based on prior experience, and responds to incoming sensory stimuli. The model also adjusts over time in response to physical training which changes the body’s abilities. As Hutchinson discusses, training yourself to run faster involves not only conditioning your body but conditioning your brain as well.
The “false fatigue alarms” theory for Long Covid fatigue
It’s normal to feel fatigue in response to an infection. This is because the catabolism (synthesis) of antibodies and mobilization of white blood cells requires a lot of energy. The brain regions that generate fatigue after infection may be different from the regions where the central governor is located.2 It isn't really understood how the immune system communicates to the brain that an infection has started, or when it has ended. It is generally thought that some of the communication happens via inflammatory markers called cytokines, but the precise mechanism and brain areas involved seems almost completely unknown.3
In his article in The Guardian about his Long Covid recovery, Dr. Paul Garner references the work of Oslo-based physician Vegard Wyller and his concept of “false fatigue alarms”, which he developed in the context of researching chronic fatigue syndrome. Here’s how he describes it:
“Where pain perception is an alarm about tissue damage, fatigue may be an alarm about excessive energy expenditure. Activating the fatigue alarm ensures that the individual rests, but at the same time – in common with pain – activates a stress response characterized by cognitive, neuroendocrine and immune changes. Evolution dictates that both alarms must be plastic – they must be modified by learning to ensure that maladaptive behaviours are not repeated.” - Vegard Wyller, M.D.
That last sentence is key — what causes fatigue is plastic, subject to learning/conditioning.
One can distinguish two slightly different scenarios:
During the acute phase of viral infection, the bodies energy resources may be drastically lower due to energy being sapped by the immune system. In response, the brain builds up a model that body has less energy. This model is maintained later after the infection has cleared, perhaps via some some sort of feedback loop involving stress and other reactions.
During infection, certain activities like exercising may cause severe “push-back” or “post-exertional” fatigue. The brain’s circuits then become conditioned to generate fatigue when these activities are performed, even if the fatigue response is not in line with biological reality. This is the same sort of conditioning as Pavlov’s dog — replace the dog’s food with “running out of energy”, salivating with “generating fatigue”, and the “bell” with the activity that caused the fatigue. Just as the Pavlov’s dog continues to salivate in response to the bell even when there is no food, the brain may generate fatigue in response to exercise even when the infection and low energy state are gone.
It’s quite possible that false fatigue alarms are self-reinforcing — that they maintain themselves via some sort of feedback loop. A false fatigue alarm could generate stress, anxiety, and depression, all of which may create more fatigue. Paul Garner references this:
“A vicious cycle is set up, of dysfunctional autonomic responses being stimulated by our subconscious. These neural tracks become established like tyre tracks in mud.” - Dr. Paul Garner
As referenced by Dr. Wyller, the same mechanisms that create false alarm pain, a fairly well established phenomena, may also create false alarm fatigue. In the case of pain, the false alarm is about bodily damage, in the case of fatigue it is about energy stores.
A new therapy, Pain Reprocessing Therapy (PRT), trains patients to use their conscious mind to re-conceptualize and cure chronic pain that exists in the absence of tissue damage. I find this almost too strange to believe, but there are several randomized controlled trials (RCTs) showing this sort of treatment works. At least two RCTs have shown the effectiveness of PRT in the context of back pain. In one study 66% of patients who underwent PRT emerged pain-free compared to 20% randomized to placebo and 10% randomized to usual care.4 Here is how the authors describe the theory underlying PRT:
“In constructionist and active inference models, pain is a prediction about bodily harm, shaped by sensory input and context-based predictions.18,19,22-26 Fearful appraisals of tissue damage can cause innocuous somatosensory input to be interpreted and experienced as painful.22,24,27,28 Such constructed perceptions can become self-reinforcing: threat appraisals enhance pain, which is in turn threatening, creating positive feedback loops that maintain pain after initial injuries have healed”. - Yoni Ashar et al., 2022.
Again, replace “pain” with “fatigue” and “bodily harm” with “bodily energy stores” to recover the “false alarm” theory for Long Covid fatigue.
A similar therapy is psychophysiologic symptom relief therapy, which has also been shown to be effective for treating chronic back pain in an RCT. There is a preprint reporting a small non-randomized study that looked at using this therapy to help Long Covid patients, with promising (although obviously still preliminary) results. (Update 8/8/23 - has been published in Mayo Clinic Proceedings).
A lot of this makes sense within the context of recent neuroscience theories such as predictive processing and Karl Friston’s active inference theory. Under these theories, rather than being a passive recipient of incoming sensory signals, the brain actively tries to model the causes of the incoming signals, in a sort of “controlled hallucination”. These theories are revolutionizing how we understand brain-based illnesses.5
Further useful intuition pumps to think about here are the rubber hand illusion (see this demonstration on YouTube) and phantom limb pain.
Relation to functional neurological disorder
“The brain abandons its old predictive schema and creates a new one, patterned after the crisis at hand: a man lies in a hospital bed, his arm healed of injury, but still unable to move. The brain learns the injury, absorbs it, and then it re-enacts it, forever.” - @FNDPortal, in “Cadenza for Fractured Consciousness: A Personal History of the World’s Most Misunderstood Illness”
Functional neurological disorder (FND) is broad category of conditions where patients experience significant disability in the absence of tissue damage. It is said to be one of the most common reasons that patients visit a neurologist. The prevalence of FND in the general population is unknown but one study found one third of neurology patients have medically unexplained symptoms which may be caused by some type of FND.6 Up to 20% of patients in pediatric neurology clinics have FND. Another study found that “functional/psychological disorder” is the second most common diagnosis in neurology clinics after headache. FND is a “software problem”, not a hardware problem. While FND may manifest in many different ways, a simple example of FND is when a patient injures a limb but then months later still cannot move it, even though it has healed. Generally a clinician can coax the limb to move involuntarily via certain tricks, but the limb won’t respond to the patients conscious will to action. It appears the brain’s model (“body schema”) was updated during the illness but then never updates again to reflect that the limb has healed.
As a side note, it is sometimes said that FND used to be called hysteria. I think a more accurate statement is that the diagnostic category of “hysteria” was abandoned and broken into several different more specific diagnostic entities - mainly somatic symptom disorder (aka somatoform disorder or conversion disorder), FND, chronic fatigue syndrome, and fibromyalgia.
Several recent articles have suggested that Long Covid should be treated, at least in some cases, as a type of functional neurological disorder. Given the theory outlined here, this makes some sense. The main difference is that FND is typically a movement disorder, whereas Long Covid is primarily a fatigue disorder, although both FND and Long Covid often involve unexplained pain as well, among other symptoms.
Treatment implications
Cognitive behavioural therapy (CBT)
“Cognitive behavioural therapy can be thought of as a method for unlearning a false fatigue alarm.” - Vegard Wyller, M.D.
“I stopped my constant monitoring of symptoms. I avoided reading stories about illness and discussing symptoms, research or treatments by dropping off the Facebook groups with other patients. I spent time seeking joy, happiness, humour, laughter, and overcame my fear of exercise. I started slowly with some graded physical activity on a bicycle.. I began to build back my strength.” - Dr. Paul Garner
We’ve discussed about how an unconscious model (possibly the “central governor”) can create fatigue based on a faulty model and faulty interpretation of incoming data (“false fatigue alarms”). To treat this, feedback loops which maintain the problem must be broken and the model must be unlearned. How can this be done? I believe that cognitive behavioural therapy (CBT) and graded exercise therapy (GET) are useful treatments, for both theoretical and empirical reasons.
CBT deals for maladaptive beliefs and behaviors which perpetuate illness. There is fairly sophisticated CBT theory for chronic fatigue which describes predisposing, precipitating, and perpetuating factors. Describing the entire theory would require a separate blog post, but in a nutshell, the core CBT theory for CFS/ME posits that the illness is maintained by unhelpful and stress-inducing beliefs about disability. Examples of such beliefs include “exercise is harmful”, “there is no cure, I need to adjust to being like this the rest of my life”, and “if I am fatigued, I must rest, I can’t do anything when fatigued”. CBT also helps with perfectionism, which has found to be a risk factor for CFS/ME. (Examples of unhelpful beliefs in this vein include “my self-worth is determined by my productivity” or “I must make be hyper active when I’m feeling well to make up for my lack of productivity when fatigued”). In patients with chronic fatigue, unhelpful beliefs can lead to a “boom-and-bust” cycles where patients alternates between extreme inactivity when fatigued and bouts of too much activity when feeling well. Unhelpful beliefs also lead to a chronic always-on stress response, depression, anxiety, and insomnia, all of which contribute to fatigue. There are also CBT protocols and techniques for dealing for any co-morbid anxiety, depression, and sleep disturbance.
CBT teaches people to challenge the thoughts that are troubling them to see if they really align with empirical data. CBT practitioners my ask their patients to complete “homework”, which may take the form of activity logs, sleep logs, and thought logs. Data generated by the patient is then analyzed by the therapist in conjunction with the patient. CBT is focused on finding specific actions that can be taken right away, rather than spending many sessions probing for “deep insights” into the person’s psychology. Thus, relative to other psychotherapies the benefits of CBT tend to take effect faster.
While CBT operates solely within the realms of conscious beliefs and consciously-directed behaviors, it seems reasonable to assume that conscious processing and unconscious processing are bidirectionally linked, like everything in the brain. Thus it is not crazy to hypothesize that the conscious mind can interdict and modify unconscious models, reducing the frequency of false fatigue alarms and tampering down unconscious physiological stress responses.
Graded exercise therapy (GET)
WARNING: do not attempt graded exercise in the early stages of Covid! There is a lot of anecdotal evidence that Long Covid is sometimes triggered by people rushing back to their normal activity levels or trying to exercise to early. The most important thing to do when recovering from Covid is to rest, at least initially. I do not recommend attempting GET until three months after infection.
To retrain the false fatigue alarm circuits, it makes sense to gradually increase exercise levels, so the faulty model can be re-calibrated. In addition, after months of inactivity the body quickly becomes deconditioned. Obviously for general health it is important to at some point reverse this deconditioning. Many Long Covid patients worry a lot about exercise, fearing it may cause “post exertional malaise” (PEM). I personally developed a very intense fear of exercise which took several months to overcome. In one online survey of those who self-identified as Long Covid sufferers, nearly all of them reported that exercise worsens their symptoms. PEM is normal and expected during the beginning of any viral illness, and indeed exercise should be avoided then. However, the belief that exercise is harmful can become a factor that prolongs Long Covid recovery. As Dr. Garner and two other doctors wrote in a letter to The Guardian, there is no direct scientific evidence that exercise is harmful for Long Covid patients. Still, graded exercise should be done with care. It is easy to over-exert oneself when recovering from Long Covid.
Personally, I trust the results of RCTs that show a benefit for GET in CFS/ME (discussed below) rather than what CFS/ME advocacy organizations say. I trust the two doctors I worked with during Long Covid, both of whom recommended easing back into exercise. I prefer to read the stories of people who have successfully recovered from Long Covid, like Dr. Garner, Amy Engkjer, and Rachel Whitfield. Nearly all the stories I have read from people who have recovered from Long Covid say that getting back into exercise and slowly increasing their activity levels was correlated with the turning point in the course of their illness.
The science on CBT and GET for CFS/ME
CBT and GET are the only treatments for CFS/ME that have been validated in multiple decently sized RCTs. There have been at least 12 RCTs on CBT and/or GET for CFS/ME, most of which showed a statistically significant benefit.78 The largest of these was the monumental PACE trial which had four arms - CBT, GET, Adaptive Pacing Therapy (APT), and Supportive Medical Care (SMC).9 It took 10 years and two grant applications to obtain the necessary funding, and the trial itself took eight years to run. The results of the trial were published in The Lancet. While the interpretation of PACE trial results have been controversial, all of the findings of paper stand and it has not required any correction. The study found that both CBT and GET were beneficial relative to SMC in reducing subjective symptom ratings, but APT was not. A follow-up study found that the beneficial effects of CBT and GET were sustained after follow-up several years later.10 The PACE trial has been relentlessly attacked by many in the CFS/ME community and many of the researchers involved have been harassed and some some have even received death threats. I believe this has led to a chilling effect where doctors are afraid to discuss the PACE trial, CBT, and GET publicly. It is true that the effect sizes in the PACE trial and other RCTs are small, and they have only been shown to help those with mild or moderate CFS/ME, not severe. However, the small effect sizes mask the fact that for a small number of patients CBT and GET completely cure their CFS/ME.
In 2021, the National Institute for Health and Care Excellence (NICE) in the UK updated their guidelines for CFS/ME, removing the recommendations for CBT and GET for mild/moderate CFS/ME. Ultimately, as Flottorp et al. point out in a detailed letter to The Lancet, this move was not in line with the science and appears to have been done under pressure from advocacy groups. NICE consulted heavily with CFS/ME patients in re-writing the guideline but did not consult with people who had recovered from CFS/ME. NICE has also cautioned against GET in patients with Covid-19, a move which was widely reported and has led to a general narrative online that exercise is not recommended for Long Covid patients.
A small RCT is already underway to test if GET helps patients with Long Covid, so hopefully we will have more direct data on whether and how much GET helps patients with Long Covid soon.
Further reading
“It was like being buried alive’: battle to recover from chronic fatigue syndrome”. The Guardian, 2016
Papers (updated 10-6-23)
Donnino, Michael, et al. “Psychophysiologic Symptom Relief Therapy for Post-Acute Sequelae of Coronavirus Disease 2019.” Mayo Clinic Proceedings: Innovations, Quality & Outcomes, vol. 7, no. 4, Aug. 2023, pp. 337–48.
The Oslo Chronic Fatigue Consortium, et al. “Chronic Fatigue Syndromes: Real Illnesses That People Can Recover From.” Scandinavian Journal of Primary Health Care, Sept. 2023, pp. 1–5.
Ikai, Michio, and Arthur H. Steinhaus. “Some Factors Modifying the Expression of Human Strength.” Journal of Applied Physiology, vol. 16, no. 1, Jan. 1961, pp. 157–63.
All mammals exhibit the so-called“sickness response” of which fatigue is just one part. Other parts of the response are isolating from others, reducing grooming, and not eating as much. Which brain region generates this response is poorly understood but it is thought by some scientists to involve the hypothalamus. The “Eyam hypothesis” postulates that social isolation and not eating as much are genetically-programmed responses that serve to lower the chance that the infected organism spreads their disease to others. The evidence for this theory mainly comes from observing other social mammals, rather than humans, where it’s hard to disentangle cultural factors. Mice, for instance, quarantine themselves when sick.
Yamato, Masanori, and Yosky Kataoka. “Fatigue Sensation Following Peripheral Viral Infection Is Triggered by Neuroinflammation: Who Will Answer These Questions?” Neural Regeneration Research, vol. 10, no. 2, 2015, p. 203.
Ashar, Yoni K., et al. “Effect of Pain Reprocessing Therapy vs Placebo and Usual Care for Patients With Chronic Back Pain: A Randomized Clinical Trial.” JAMA Psychiatry, vol. 79, no. 1, Jan. 2022, p. 13.
Henningsen, Peter, et al. “Persistent Physical Symptoms as Perceptual Dysregulation: A Neuropsychobehavioral Model and Its Clinical Implications.” Psychosomatic Medicine, vol. 80, no. 5, June 2018, pp. 422–31.
Carson, A. J. “Do Medically Unexplained Symptoms Matter? A Prospective Cohort Study of 300 New Referrals to Neurology Outpatient Clinics.” Journal of Neurology, Neurosurgery & Psychiatry, vol. 68, no. 2, Feb. 2000, pp. 207–10.
Ingman, Tom, et al. “A Systematic Literature Review of Randomized Controlled Trials Evaluating Prognosis Following Treatment for Adults with Chronic Fatigue Syndrome.” Psychological Medicine, vol. 52, no. 14, Oct. 2022, pp. 2917–29.
Kim, Do-Young, et al. “Systematic Review of Randomized Controlled Trials for Chronic Fatigue Syndrome/Myalgic Encephalomyelitis (CFS/ME).” Journal of Translational Medicine, vol. 18, no. 1, Dec. 2020, p. 7.
White, Pd, et al. “Comparison of Adaptive Pacing Therapy, Cognitive Behaviour Therapy, Graded Exercise Therapy, and Specialist Medical Care for Chronic Fatigue Syndrome (PACE): A Randomised Trial.” The Lancet, vol. 377, no. 9768, Mar. 2011, pp. 823–36.
Sharpe, Michael, et al. “Rehabilitative Treatments for Chronic Fatigue Syndrome: Long-Term Follow-up from the PACE Trial.” The Lancet Psychiatry, vol. 2, no. 12, Dec. 2015, pp. 1067–74.
Nice overview of fatigue. Enjoyed reading.