Towards the unity of pathological and exertional fatigue: A predictive coding model, 2020, Greenhouse-Tucknott et al.

Preprint:
https://psyarxiv.com/nqxz2

Neil Harrison is involved in ME/CFS; possibly others also.

 

I see what you mean, @Trish...I can't make a lot of sense of it either!

The only thing I've learnt about fatigue is that it's part of a continuum with pain. I only usually take a pain med at night, but if it's really getting to me during the day, I'll occasionally take a dose of paracetamol or tramadol. It relieves fatigue more completely than it does pain.

 

eta: see also
https://en.wikipedia.org/wiki/Allostasis

(I'm not trying to pretend I understand any of it but someone might!)

eta2:
I see all the researchers are in the Sussex area (eg Brighton), Sussex ME Soc territory.

 

I suspect you haven't missed the point, and summed it up well.

 

This looks to be based on the idea of "circular causal and feedback mechanisms in biological and social systems". It seems to be a popular idea on the south coast http://users.sussex.ac.uk/~inmanh/C...in,to ‘thing’ and ‘man’ to ‘organism’ for the

There is a slight problem with this. It seems to be an idea promoted in 1946 by the Macy Foundation. Everything it touched should be viewed with caution. It was used by the CIA for channelling money to various researchers, including Ewen Cameron. This does not necessaily invalidate everything, but it shifts the burden of proof.

 

This is either a convoluted way of saying that wrong signals are being sent, or that it's entirely a matter of perception expecting something else than normal, which is the same tired old model as usual. I think it's the first but then I have no idea what point there is to argue that perception matters because when the signals are wrong, expectations of a correct signal are entirely irrelevant.

If someone has a fever it's completely superfluous to be puzzled that the temperature isn't as expected. It isn't, deal with it. "But the temperature is supposed to be lower" is simply not relevant, the fact that it is higher is. Lethargy is a central feature of infections because metabolic energy is used by the immune system. Absolutely no need to bring perception to the equation other than as a part of the process that reacts to something wrong, and even then all that happens far below consciousness.

 

The constant thread is that everything is normal but something in someone's psychology is wrong so in some magical way this changes how the body reacts.

But as you say about fever, our bodies work in ways that have evolved to protect us from parasites for millions and millions of years. They are so deeply evolved that they happen without thought being involved. We may take to bed with a hot water bottle and a toddy when we think we are getting a cold but that feeling is actually our immune system already revved up and working. Conscious (or unconscious!) thought has little to do with the process except in a very limited and macro way. Improving our conscious response does very little unless it is the result of scientific study of the innate response.

We feel pain because feeling pain helps survival. We know how important it is and that it is not something we have evolved beyond because of the dreadful outcomes in people who feel no pain. Since being in pain is not conducive to reproduction it must be a very necessary response.

Feeling fatigue is very common with illness and injury so it must have evolutionary roots. Interfering with the protection it gives must be done with a constant awareness that it could be dangerous and we know that using cocaine so that they do not feel fatigued has killed competitive cyclists.

The trite psychology of the BPSers is being forced onto deep evolutionary protections with little thought for the consequences. The evidence that these mechanisms are not appropriate in certain circumstances should be proven by careful robust unbiased studies before they are used in medicine, the exact opposite of the current situation.

 

On reflection I have found this paper immensely valuable, not because of what it says-I pretty much gave up on it with the "coding" metaphor- but for where it leads in terms of understanding.

We have been led down the track of psychiatrisation of ME back through personal vulnerability, neurasthenia, hysteria etcetera and it seemed that somatisation and central sensitivity were merely aspects of that, in some difficult to understand sort of way.

I have not heard of reference to the alternate strand originating in the post-war Tavistock Institute and Macy Lectures but this could be where the big unexplained and unreferenced ideas about dysfunctional cognitions and maladaptive behaviour originated. It would certainly explain those loopy diagrams beloved by Wessely and Sharpe demonstrating the circularity of their explanations. There seems to be or have been a strand of "cybernetics" which may now have taken hold in Brighton, Sussex and Southampton, and these circular causal and feedback mechanisms seem to explain some of the thinking which we encounter.

An early book review contained the joke "Circular causal mechanisms, or arguing in circles".
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2599565/?page=2

Perhaps this is all clear to those who have studied cognitive psychology, but there seems to have been little explicit reference to it for the benefit of patients. My enquiring mind has for a number of years wondered exactly what a "dysfunctional cognition" or "maladaptive behaviour" might be, and how and why they originated. This was not adequately explained.

It seems that the "world" takes some extreme positions on Tavistock and Macy, and it is hard to elicit the facts. It is perhaps unfortunate that many of the groups involved seem to have been involved with the epidemic of LSD use. And they mostly seem to have had a surprisingly quiet war pushing paperclips.

Much more research is needed.

 

I see what you did there.

 

I had a good look over this. Awfully written, wasn't at all fun.

The main thrust of the article is whether some models of pathological fatigue can explain normal fatigue from exertion. So its not all that relevant to MECFS.

We’re bound to see more of this predictive coding stuff, though, so I thought a quick summary might be useful.

These are the general claims of the predictive coding approach (not specific to this article):

Sensation as Prediction. Researchers are increasingly thinking of sensory perception as not a passive process, where we "receive" signals and interpret them according to our knowledge and experience, but rather an active process, where we are constantly predicting what signal we will receive and what it will mean. For example, when a person hears a sentence like "The girl roasted a football”, there is a measurable burst in brain activity just after the word "football". That suggests we were actively predicting what we would hear next and reacted to the surprise when the word did not align with predictions.

We make active predictions at all kinds of levels, some we're aware of and can describe, others involve bodily systems like muscle groups, sensory processing networks, etc.

Prediction error and its role in learning. Prediction error (like the "football" surprise) is something we as organisms try to minimise. So when our predictions are disconfirmed, we work to adjust those predictions so that they will be more accurate next time. Consequently, we "learn" more from an error than from a correct prediction. Reducing prediction error is one of the major mechanisms by which we learn.

These are the claims that are specific to the content of this paper (to do with activity, adjustment and fatigue):

Predictive coding and physical activity. We can extend the idea of predictive coding to sensory signals from within the body, such as pain, heart rate, breathing, proprioception (information about where your body parts are located), as well as chemical signals indicating certain metabolic, autonomic, immunological and hormonal states. Healthy people are constantly predicting the signals they'll receive from their body during various activities. So an experienced jogger will anticipate the sensation of increasing breathlessness and faster heart rate as he sets out on his daily jog, not only at a conscious level but also at a bodily level.

Dealing with prediction error in physical activity. When it comes to sensations we experience during activity, we actually have two options for dealing with prediction error. One is by learning to adjust our predictions via the process described above. The other is by adjusting the functioning of various systems or changing our overt behaviour to bring it back into line with predictions. If our jogger suddenly feels a bit weird while jogging, he might (at an unconscious level) adjusting various bodily systems to improve performance, but when this fails, he will adjust his behaviour (he will slow down).

The experience of fatigue during exertion. Normal fatigue is the mental state experienced when continued adjustments fail to bring bodily sensations into line with predictions. The sensation is not all or nothing but increases with persisting failure to adjust and increasing mismatches between expectation and performance.

(edited to separate out the general claims of predictive coding models from the claims in the paper).

 

I just lost a long detailed reply to this thread, sigh.

I'll try to be more concise this time I guess.

The author seems to be basing his ideas on this: https://pubmed.ncbi.nlm.nih.gov/22641838/

Overall, the article is poorly focused and has a confused understanding of effort perception, fatigue and motivational/task failure. Task failure being confused with fatigue is an unfortunate, but common conceptual confusion introduced by Tim Noakes and colleagues.

Anyway, the authors do not seem to understand that effort perception during a task is not modified by afferent feedback. Effort aka the efference copy aka the corollary discharge plays a key role in proprioception. If there was an error in assessing effort, it would necessarily lead to motor coordination problems. As such, the authors (along with Edwards et al.) propose a feedback-loop which is not plausible to explain fatigue.

Fatigue IS an error in motor force/position prediction (undershoot of actual function), however this is a second order problem - issues of sensitivity/signal gain is easily corrected - the focus is on a reasonable level of accuracy, not precision. (note that fatigue is not experienced with muscles at rest - sensations of muscle stiffness is not fatigue)
It is the increased uncertainty tending toward force undershoot that signifies fatigue.
Hence the signal to the brain is that of reduced accuracy of output predictions. Metacognitively this is a warning that performance is not able to be maintained and one should expect reduced precision in performance predictions.

I therefore do not understand why the authors claim that this leads to increased "higher order" prediction in the brain, or the brain focusing on "increased precision" of afferent feedback (feedback that could be in error).
This ignores the fact that both hypothetical circumstances will lead to motor coordination issues if there was some sort of aberrant perceptual feedback loop. If there was a greater perceptual error, leading to greater perception of fatigue, feeding back into greater perceptual errors, then this would necessarily lead to motor coordination failure. If the motor units are not truly fatigued, then this will lead to overshoot of force/position - and fatigue will no longer be perceived (fatigue has a directional bias!) Likewise If the altered model is effective at predicting the output, then the perception of fatigue will be diminished (and thus there is no problem).
However, when there is a prediction error, the brain does not simply fall back to increased reliance on the primary muscle afferents as the authors suggest. Importantly, if the prediction error is of the second order as mentioned, the brain can modulate the primary afferent inputs to see if they are in error, and reduce the perceptual precision, with increased attention to alternative afferents, such as agonistic afferents, a modulation of gamma drive and increased reliance on other perceptual systems such as visual input. If this modulation doesn't increase accuracy, the system doesn't increase the weighting of afferent feedback, it simply tolerates the error and signals this as fatigue. Notably, this is contrary to the feedback loop proposed by the authors.

I certainly agree that meta-cognition is important in terms of motivation and task failure. But this is not fatigue. Meta-cognition can lead to an increase in baseline effort reporting, but we know from CPET results in CFS patients, that the rate of change of effort perception correctly predicts the anaerobic threshold and therefore is unlikely to be in error. In Greenhouse‐Tucknott's 2020 study "Interactions between perceptions of fatigue, effort, and affect decrease knee extensor endurance performance following upper body motor activity, independent of changes in neuromuscular function", Greenhouse‐Tucknott seems to be confusing reporting biases of effort perception on fatigue with effort itself. To avoid this bias requires some methodological finesse, namely realise that baseline effort reporting can be biased, which disappears as the task becomes more effortful (and the rate of change becomes more important).
(notably in Greenhouse‐Tucknott's 2020 study, the level of performance as well as the rate of change in effort perception was the same in the control group as well as the group that had been "fatigued" by a task using different muscles, so the initial higher reported level of effort perception was likely to be a reporting bias due to meta-cognition)

Not only is there no demonstrated process by which a generalised allostasis/metabolic is sensed, there is no need for some generalised notion of allostasis/metabolic homeostasis to be involved in fatigue perception. We know the metabolic situation within different muscles at the same time can be vastly different and fatigue is experienced at levels of exertion far below when central cardiorespiratory limits being reached. Hence any such notion of a body-wide-allostasis will be of limited use in predicting peripheral metabolic consequences anyway.

Lastly I mention two further points. I suspect the authors are unfamiliar with some of the literature regarding fatigue. Namely, that in (healthy) subjects, when fatigued, it is impossible to accurately estimate both force and position at the same time. This should give some clues about effort and proprioception when fatigued.

Secondly, patients with CFS, muscular dystrophies and peripheral neuropathies all have the exact same pattern of fatigue when tested using supramaximal twitch interpolation methodologies. Manipulation of type 3 and 4 afferents show that rather than some notion of central allostasis, metabolic and mechanical uncertainties are measured peripherally and mediated through spinal feedback upon central drive. The afferent measures of proprioception, namely muscle spindle, Golgi tendon of the respective muscle, the agonist and antagonist muscles and joint receptors are then compared with the predicted force and location given the efferent copy (effort). The notable consequence of this is that fatigue due to a decline in peripheral nerve function as well as central fatigue are perceived in exactly the same way.

Thus I think the ideas presented by Edwards et al. and Greenhouse‐Tucknott may be plausible for a hypothesis of psychogenic motor disorders, they cannot be plausible for a mechanism for chronic fatigue due to how fatigue is sensed and the motor coordination consequences of prediction error.

 

@Snow Leopard would you consider sending your critique to the authors?

People on this forum write excellent critiques and it seems a shame not to use them

 

The stuff you summarised is not controversial, and is considered a key part of our understanding of proprioception as explained:

https://en.wikipedia.org/wiki/Efference_copy

It is the notions of allostasis (borrowed from Noakes) and the weird attentional feedback loop that leads to an increased expectation (in the brain) of precision from primary muscle proprioceptive afferents (borrowed from Edwards et al.) that are used to "explain" how chronic fatigue could be caused by distorted meta-cognitions in patients.

Part of the problem is that the people being critiqued are unlikely to want to listen. Of course if any of the authors happen read this and disagree, then they can feel free to sign up and tell me I'm wrong!

 

More on this, one of the models is described here:
https://www.frontiersin.org/articles/10.3389/fnhum.2016.00550/full

The nonsense is revealed in figure 6 where "prior precision of the homeostatic belief is increased from 1 to 4. As a consequence, when a perturbation occurs at (6), this yields a considerably larger precision-weighted prediction error and hence greater interoceptive surprise"

Note that this "interoceptive surprise" is a effectively massive overshoot of position, quite distinct from the undershoot in the prior examples. Also realise that an overshoot would not be perceived as fatigue and would feedback to reduce the "prior precision of the homeostatic belief", because the increase in weighting was obviously in error. In reality, small modulations occur all the time as a control strategy and large jumps like this are unlikely and require a plausible physiological explanation, before we should be willing to consider this model to be realistic.

Also, feedback control does not work in real-time, which is why we have a predictive model of motor function and a sense of fatigue in the first place.

The authors don't explain why physiologically, some very high level of precision should be desired (in reality, rather than their model), rather than simply a reasonable threshold of accuracy, beyond which precision is inconsequential.

Hence I do not understand why this model is being claimed to model a chronic-fatigue like state, because it does not predict real-world motor function.

One key point by Greenhouse‐Tucknott that I do agree with, namely the differentiation of "acute tiredness" with "fatigue" by the above authors is not coherent. They then use this differentiation to make a fairly rubbish claim that this is why rest is not helpful in reducing the "interoceptive surprise", because in tiredness they claim, the problem is due to muscle metabolic balance disturbance, whereas "fatigue represents a metacognitive belief that arises from chronic experience of lack of mastery over bodily states" and "interoceptive surprise fails to decrease in the absence of actions, mastery cannot be experienced and the associated metacognitive beliefs cannot be adjusted."

Of course anyone who has stretched after waking up in the morning knows how their statement lacks empirical basis. Indeed, the "chronic experience of lack of mastery over bodily states" also necessitates motor coordination errors - the fact is, that the proprioceptive model is accurate most of the time contradicts the foundation of their model.

 

Okay, I'm a bit simple, but I don't get why cognitions, or even the brain, are involved, at all.

From single cell beasties - they must have had a way of monitoring 'fatigue' - the mitos release something, there may be by products produced - whatever, I cannot belive that a single cell organism has no way of determining it needs to rest or die.

Up to more complex life forms, surely cells in them must have had a way of load sharing, of communicating that one, or more, need rest, are fatigued.

Moving up to more complex life forms, why would the brain suddenly become the arbiter of this? Why wouldn't the presumably existing evolutionary methods of determining fatigue be used? Why would it be a matter for the brain to decide - granted the brain my notice, but the brain wouldn't produce fatigue in any way, other than locally.

So, why CBT when it's got sod all to do with the brain, the brain mearly passes on the sensation, it doesn't originate it.

Or am I being thick?

 

I don't have enough energy to break this down to understand the words and relationships here - and that is how it should be. The bps view is so superficial it is beloved by nonscientists but the body is complicated.

I think you are right in what you say. To my limited understanding they seem to be leaving out pathology or assuming that the pathology is all in the way the brain is analysing the signals when every part of the process can go wrong. It is good that there are people like yourself with an up to date knowledge of physiology as part of the forum.

I seriously doubt there are any truly psychogenic motor disorders. Movement involves so many systems that it could be thought of as an early warning of something wrong.

For instance, each muscular dystrophy is caused by repetitions of a certain gene. The number of repeats determines the severity of the disease and there is a threshold for a positive result. But if the number to be diagnosed is, say, 100 then at 98 you are "healthy" and any movement problems are psychogenic. But then even people who turn out to be in the disease category can be dismissed for years until they see an expert or a family member is diagnosed.

 

A lot of good points above. I find it hard to focus on the detailed theory because I have already concluded there are some major false premises at the start. I admire the detail of Snow Leopard's analysis.

Three things have struck me. One is that, as I think Snow Leopard is pointing out, the feedback theory put forward by Mark Edwards seems to be a*** over t*t. What should be too little is too much. The second is that although we have known about servo control mechanisms since Pat Merton's and others' work inn the 1960s the predictive coding bandwagon still largely depends on computer models that may bear little relation to reality. People who work with Friston are not always as gung-ho about it as one might expect.

The third, which relates to Wonko's remark in a way is that I see no reason to think that fatigue has much to do with precision quantitative control of motor output. I have just been sawing a six inch oak branch. When I get fatigued I continue to saw in exactly the same way but I know that pretty soon if I do not stop my arm will give up. I never get to the point where the action is actually unexpected or reduced.

One thing that interests me is that the whole predictive coding thing, and indeed the more traditional servo theory, depends on real time subtraction. Comparison is essentially subtraction ( '7 is the same as 3+4' essentially means that 7-(3+4) =0). What none of the neurobiologists talk about is the mechanism of high-precision subtraction. If neurons work by addition, which in general they are supposed to, then there is nothing to apply the computer models to. I have a paper coming out next year describing how I think cellular subtraction may work in brains but it is too complicate to go into now!

 

Neil Harrison is now at Cardiff Uni. https://www.cardiff.ac.uk/people/view/1458904-harrison-neil

He spoke at CMRC 2018, and at AfME AGM in Nov 2018. I got the chance to chat to him about scans showing increased "activity" in the basal ganglia, but he's not too sure why that was and why it only showed up on one side of the brain.

He seems genuinely interested in people with ME, but very closely associated with some of the BPS psychs. His latest retweet is:

Neil Harrison Retweeted




PNIRS

@PNIRS

Excellent news for @BrainBehavImm, the official journal of #PNIRS, with the new IF of 6.633, it is now in top 10% of psychiatry and neuroscience journals, and top 15% of immunology journals!



This journal is edited by Carmine Pariante. (PNIRS = Psychoneuroimmunology Research Society)