The study design (I’ll get straight to it) was to have 10 cyclists perform a maximal 5-second sprint, followed by a cycle to exhaustion at 80% of VO2max, and then at the moment they gave up, another 5-second sprint.
Easy design, with predictable results, because you all know intuitively what they’ll find. Let’s put it to numbers:
- During sprint 1, the average power output was 1075 ± 214 W
- During the ride to exhaustion, the cyclists averaged 242W, riding for approximately 12 minutes
- During sprint 2, supposedly after the maximal effort, the average power output was 731 ± 206 W
And that’s it. Nothing to it really, a fairly obvious result, because:
- The power output during the 5-second sprints declined from 1075 to 731 W as a result of the fatiguing ride in the middle. That’s a drop of 32%, pretty substantial but entirely expected.
- The riders were still able to increase the power output during that second 5-second sprint. They managed, for 5s only, a power output of 731W, when they’d been riding at 242W before. Again, if you’ve ever run, cycled, rowed, or even watched any of these sports, you’ll be well aware of how normal it is to increase power output at the end of exercise, when you know the finish line is in sight. How much more then when a researcher asks you to produce 5-seconds more of maximal effort?
So the study should have ended with those findings, supporting what has been found previously, and concluding that fatigue decreases the maximal sprint power output (by 32%), but that a reserve exists.
But no, this is science, where novelty and innovation are prerequisites (yes, I am frustrated with aspects of academia right now!). So rather than leaving a conclusion where it should be left, Marcora and co build a massive straw man, then burn it down while taking credit for a novel study that is in fact not novel, to present a “revolutionary” finding.
The piece by Fitzgerald contains this collection of pearls, which illustrated some of the problems with how our understanding fails to advance:
Late to the ‘show’, but don’t let the facts get in the way of a story
First, the study does little (not nothing, but little) to “completely shatter the concept of endurance fatigue”. It’s one of many studies or reviews that have document some kind of capacity to increase power output at the end of exercise, when a known finishing point is provided. Here are some:
Note that I have limited it to some studies published prior to 2010 – if you do a quick search on Pubmed for “fatigue perception exercise”, you’ll find dozens of others, all saying the same thing. If there ever was a concept of endurance fatigue, it was shattered long before 2010, and it wasn’t shattered by a single study.
So when Marcora & his co-author on this paper write “It is traditionally assumed that exhaustion during high-intensity aerobic exercise occurs because fatigued subjects are no longer able to generate the power output required by the task despite their maximal voluntary effort...”, they are in fact about a decade late to this particular party. It’s already been debunked. However, in the pursuit of novel theories and revolutions, I guess you cannot allow the facts to get in the way of the story, and so the straw man that fatigue is absolute is first built in order to burn.
The fact that they continue to write the following: “We have demonstrated for the first time that this is not the case…” betrays either that they’re new to the field, or choosing not to acknowledge the many studies that refute whatever ‘traditional’ idea they believe exists prior to their contribution.
Now, those studies, including the ones I list above, may not have overtly stated things to the extent that Marcora & co did (and do), but that’s because they’re so obvious and implicit. If there is an end-spurt, and increase in power output at the end of exercise, it can only be interpreted as an indication that a motor unit reserve existed prior to that point, and therefore a reduction in power output or voluntary exhaustion that happens BEFORE the end-spurt must be explained by something other than a direct effect of fatigue on the physiology.
Again, welcome to the discussion.
The A vs B fallacy of psychology OR physiology
The most glaringly frustrating falsehood in the Marcora paper, however, is the following:
“[I]f our subjects were able to voluntarily produce 731W for 5s immediately after exhaustion, they must have been physiologically able to produce 242W for much longer. The most likely explanation for the very high MVCP produced immediately after exhaustion is psychological. Subjects knew that the final MVCP test was going to last only 5s, and such knowledge motivated them to exert further effort after the time to exhaustion test which had a longer and unknown duration
First, this is so obvious it’s laughable to include. Of course the power output goes up when the cyclists are told they only have 5s remaining. We once did this study here in Cape Town, and make cyclists perform a maximal test, where they ride at ever-increasing power outputs until they can’t maintain the power output any longer, and then ask them for a 10-second sprint. Sure enough, the power output increases 3-fold in the sprint. So what? It shows the presence of a reserve, and asks some important questions of the underlying fatigue, but those had been asked by many people, many times.
Second, it introduces my absolute pet-hate in scientific discussions, and that’s a stupid polarization of a complex phenomenon. To write that “the most likely explanation for the the high power output…is psychological”, is to introduce an A vs B argument (A is physiology, B is psychology”. Once that is done, it becomes an A OR B scenario, which is totally unnecessary. Not only this, but the existence of A is taken to mean that B cannot exist, and that’s foolish.
Why, for instance, can fatigue not be the result of a combination of physiological and psychological factors? In fact, we don’t even need to guess at this, because in an outstanding series of studies, Markus Amann showed how central and peripheral factors influence the physiological contributions to fatigue, and numerous others have shown the psychological contribution.
For example, take the graphs below, from a study by Amann et al on pacing in various oxygen conditions. The cyclists performed 5km time-trials in four conditions: 15%, 21%, 30% and 100%
Unsurprisingly, the highest power output (bottom panel) comes in the 100% oxygen condition, the lowest in the 15%. Also, notice how the muscle activation levels (EMG, top panel) tracks power. Or rather, it happens the other way around – we activate muscle differently depending on the oxygen content, and that drives the differences in power output. Cause and effect.
The interesting aspect of this study came because every cyclist, before and after the 5km time-trial, had to perform what is called a supra-maximal magnetic femoral nerve stimulation. Basically, this is a technique where the quadriceps muscle is made to contract maximally using magnetic stimulation, and the force produced is assumed to represent the maximum capacity of the muscle. In other words, if the force drops when stimulated, then the muscle is less responsive to that stimulation. This was the finding:
So, you see that the force on stimulation drops by the same amount in every condition – around 34%, regardless of power output or muscle activation. The change in muscle performance, a measure of the degree of peripheral fatigue, is thus independent of the actual work done during exercise. All of which invites the very interesting possibility that perhaps the exercise bout is regulated, at least in part, specifically to prevent too large a peripheral ‘failure’?
This is not a novel theory – it’s the basis for the whole idea of anticipatory regulation, which was the subject of my PhD and which I began because others before me had already introduced the idea.
But what was novel is that Amann then took this into a follow-up study, this time cleverly using a drug called fentanyl to mess around with the normal regulation and communication between the muscles and brain. Fentanyl, you see, blocks the Type IV afferent nerves, and therefore prevents sensory information from ever making it to “central command”. So what happens then?
Two things. First, here’s the pacing strategy and performance during a 5km cycling time-trial:
So, as should be clear, the pacing strategy, the muscle activation (bottom panel) and the resultant power output (top panel) are totally disrupted. With fentanyl, the cyclists go out like a house on fire – hugely elevated muscle activation and power output until about halfway. That’s when the house, well, burns down, and the ‘price’ for those early efforts is paid.
So now, look at what happens to the muscle force production during stimulation. Remember, this is a method that allows us to quantify how much the capacity of the muscle has changed as a result of exercise:
Each green line there represents one individual, the blue is the group average. You can see how the control and placebo groups show a similar decrease to that measured in the hypoxia study I summarized above, but the Fentanyl trial shows a much larger decrease of 46%, which means that the disruption in pacing strategy caused by blocking those afferent nerves caused a “failure” or normal regulation. The result of this failure is a more severe loss of muscle capacity, and this does then suggest that the initial theory is correct – we pace ourselves specifically to prevent excessive changes in normal function. In this case, that normal function means muscle contractility or peripheral function, whereas in others, it may be related to high rates of heat storage and body temperature, perhaps energy depletion, oxygen delivery to the brain and so on.
The conclusion from the paper:
Central, peripheral and perceptual – but you knew that already
So, let’s relate this back to the nonsensical simplified conclusion from the Marcora paper. What these studies show, quite clearly, is that fatigue is a multifactorial phenomenon.
For instance, consider why, in the first Amann study in hyperoxia, the cyclists start out from the very beginnings of the trial, activating more muscle and achieving higher power outputs in 100% oxygen than 15% oxygen? That’s because the system is ‘clever’ enough to detect the environmental difference and then allow different regulation of pacing strategy. What Amann contributes to the field is that this may be at least in part able to regulate the degree to which peripheral systems (muscle in this case) are compromised by exercise.
Clearly, there is a physiological component to fatigue, which is why it is ridiculous and stupid to write that the increase in power output during a sprint, as reported by Marcora (and about a hundred others) is “most likely” to be psychological. It totally ignores the obvious physiology.
In fact, Marcora & co would do well to pay attention to studies like that, because they explain why the power output in the sprint went from 1075 W to 731 W. That’s a small matter of a 32% reduction, conveniently ignored by his “most likely” explanation for fatigue. In a world where perception is all there is, then the power output would return to a level a lot nearer the starting one. It should not escape your attention that the magnitude of that reduction is conveniently in the same range as the reduction measured by Amann. I suspect this is co-incidental, actually, but the point is that there are clearly changes that a “psychology-only” model does not account for.
The final point is that Marcora (and Fitzgerald, in his online piece) seem to be taking this finding as a revolutionary one. He writes, for instance: “As intuitively sensible as this explanation is, it is scientifically revolutionary, as it supposes that the true cause of endurance fatigue is perception of effort (i.e. psychological suffering), whereas perceptions are traditionally seen as having no causal force in exercise physiology.”
The idea that “perceptions are traditionally seen as having no force in exercise physiology” is preposterous and frustratingly ridiculous. In fact, consider this paper, published in British Journal of Sports Medicine in 2009:
They even have the words “perception” or “perceived” in the title... They too, must have escaped Marcora’s attention, as did the many preceding it. That review article, incidentally, was the culmination of my PhD thesis (in 2006), and it was built off the back of work done by others, not me. Scientists who, since the mid-90s, had recognized the value of perception and conducted work to show its importance in exercise and fatigue. I mean, we even have a number of rating scales measuring perception, so how anyone can write that perceptions have no force is beyond me.
Why it matters
Now, this may all seem very academic, but it’s an important debate, I believe, because it highlights the problem that holds science back.
When I did my PhD, I made the same mistakes of polarizing a complex issue and then offering MY simple explanation for it. I did the A vs B false-dichotomy error. I built straw men, and then burned them down with the glee that comes from driving after ego and recognition. I am embarrassed about that now, and I’ve realised since I finished my PhD how self-serving and ultimately destructive that method of science is. I am almost compelled to apologize at every scientific conference I go to for making the mistake of ignoring an integrated model of fatigue and over-simplifying complex realities into my desired simplicity, and Marcora and co are doing the very same thing right now.
One consequence of polarizing it into A vs B was that I allowed fatigue to become an “us vs them” argument. I did this, at the time, because I did not know better, and I was basically imitating what I saw as the best approach. Certainly this approach has merits, and I’m proud of most of the contributions, however small, made by my research into fatigue. However, I also overlooked research that would have made my understanding of fatigue better. I overlooked findings like those of Amann, because they didn’t ‘fit’ with the easy dichotomy of an unnecessary ‘war’.
Let’s not keep the same mistakes going, in the name of recognition and novelty, shall we? Let’s not simplify and offer up our work as novel, when frankly, it really isn’t. Let’s not simplify fatigue and do such an interesting concept such a great disservice, just because we’re trying to rise above the crowd and make ourselves look like revolutionaries.
Over and out.
Afterword. Of sorts
So Sam Marcora, unsurprisingly, has responded negatively to this article. I have subsequently conceded that questioning his motives, while plausible, should not be voiced publicly and have edited the original article above to remove insinuations that are unfair, with apologies to him. However, I cannot, and will not, back off the notion that he has unnecessarily simplified a complex theory to advance his own model and ideas, while consciously choosing to ignore that others before him have already addressed every aspect proposed by his so-called novel model, other than the application of brain training to performance (a contribution for which Marcora is novel and commendable). Why this has happened is speculative. That it has happened remains unquestionably true.
It is achieved in part by constructing straw men arguments, providing others with their own beliefs before criticizing them, and failing this, telling them they simply don’t understand, or to just shut up. One of the straw men that has emerged in the course of his responses is the notion that fatigue must be defined as being an absolute event. This is a constructed argument, though of course if you wish to create exercise conditions to satisfy this definition, it is relatively easy to do. The Marcora study I wrote about above does exactly this – it defines “exhaustion” and “fatigue” as the exact moment where the cyclist gives up in the 80% trial. Then because fatigue or exhaustion are discreet and absolute, then any increase in power output during the second 5-s sprint must be a revolutionary finding about fatigue.
Had fatigue been identified as having specific contexts (think exercise modality, situation, environment, duration), and being a failure to maintain a given force output (as has been defined, by neuroscientists and physiologists), then such a ‘miraculous’ revival would be obvious. All those papers I listed above recognize this, and that is why they don’t harp on about it. But this is the reason Marcora, on Twitter anyway, seems perplexed enough to keep asking for studies that have done what he did. They didn’t need to because they showed the presence of a reserve without having to rigidly define the parameters.
Besides, a study in 1961 had already showed that muscle force output, even when under maximal volitional control, could be increased by a gunshot going off. The reserve, in other words, was documented over 50 years ago, though it was not recognized entirely back then. And if that reserve exists during isolated muscle contraction during maximum tasks (shown by many, many studies), and if it exists when RPE is maximal in the final moments of self-paced exercise, then it is hardly unusual that it would exist at the end of 12 min of cycling at sub-max levels, when a new task, and thus new context, is introduce. St Clair Gibson described context specific fatigue years before.
I therefore concede on the issue of motives – who knows what they are, and I apologize to Marcora for inferring any. But the method by which the ideas are advanced, namely deliberate ignorance of previous arguments, claiming of novelty despite none present, the construction of straw men, and the method of putting words into the mouths of others – these are devices which do little for our advancement of understanding fatigue.