More questions than answers

An extract from the 1st chapter of The Nature of Training

MORE QUESTIONS THAN ANSWERS

People with less knowledge in the world of training often think that there is a magic recipe, a way of training that is suitable in the light of what training science has discovered in recent years—and many coaches contribute to spreading this myth, which is highly convenient for us.

However, we are at the antipodes of knowing the most appropriate type of training for an individual based on a number of variables. Sports science is evolving, but it is still anchored in tacit assumptions and beliefs of very questionable validity and, sometimes, looking for answers to exactly the same questions as a century ago: How do we improve aerobic fitness? How do we improve the anaerobic threshold? The lack of critical thinking has hindered and continues to hinder progress in sport science.

Seventy per cent of scientists surveyed by the prestigious journal Nature admit to having failed in their attempts to reproduce the results of other researchers, and some scientists said that, in publishing this lack of reproducibility, editors had required them to minimise comparisons with the original study they were trying to reproduce (Baker, 2016). This demonstrates the extent to which the defence of the traditional and the prevalence of the status quo dominates over constructive criticism and scientific evolution (Pol, 2021).

This prevalence of the status quo can be interpreted with the “sunk cost bias” (Kahneman, 2013). If we have to decide between several options, we tend to opt for those in which we have already invested a lot of effort. If I had spent my whole life learning about one way of interpreting the human organism, I would have a hard time changing my view because it would mean acknowledging that I was wrong and thinking that I had wasted my time—even though, in reality, it was not time wasted, but time invested.

THE INDIVIDUALITY OF TRAINING

The Briton John Kiely is one of the first sport scientists to say loud and clear that the emperor has no clothes. I share his view that we must face up to the uncomfortable truth that the foundations of what we think we know about training are unstable.

In his famous 2018 article “Periodization Theory: Confronting an Inconvenient Truth”, Kiely leaves us with a review of intervention studies on types of training. He compiles the results of the different trials and concludes that when adaptations are analysed at the group level, we see how participants following similar training programmes develop very different training adaptations. Furthermore, we find very similar training adaptations in athletes following very different training programmes.

In Kiely’s words:

These findings highlight the futility of arguments spanning much of the history of periodisation and sports training, whereby advocates of predetermined training models claim their superiority over other models. . . . This claim would only be possible if humans responded to training stress in a predictable way, over generalised time frames and according to a stable dose-response model. However, contemporary evidence clearly shows that this position is no longer logically tenable.

Moreover, responses to methodological interventions are not only variable in magnitude, but also in form. In the face of similar improvements in performance—say a 3 per cent improvement in a time trial—each subject achieves it by improving on different parameters (Pickering, 2019):

It is not that there are responders and non-responders, but people who adapt in one way and others who adapt in another way or who need a different type of stimulus to achieve the same adaptation.

Moreover, the larger the sample size of the studies and the more general the research methodology—meta-analysis—the more difficult it is to understand how and why this variability occurs and how and why each intervention works for each athlete (Pol, 2021).

The extrapolation of average study results to actual application in athletes is, according to Rose (2016), “worse than useless, because it creates the illusion of knowledge, when in reality the average hides the most important thing about an individual”.

THE TRAINING LOAD

Impulse-response models of training, which assign a hardness score to sessions in an attempt to monitor loads, fatigue, adaptations and athlete performance, are the foundations underpinning traditional training methodologies, and some modern ones focused on the analysis of some potentiometer data (TSS [see glossary], CTL [see glossary], etc.).

However, as Passfield and Murias show, these models have never been validated. It seems that we are again faced with another of these training myths that have been passed on from person to person without anyone questioning the original source—or, rather, silencing those who questioned it.

In their first paper, Banister and colleagues (1975) proposed a model that predicted changes in a swimmer’s performance as a consequence of his 105-day training programme, showing both the fatigue (ATL [see glossary]) and fitness (CTL [see glossary]) induced by training. To calculate the effects of training on fitness and fatigue, their swimming and weight training were quantified in TRIMP [see glossary]. The resulting performance predictions were presented in a very simple form along with the actual performances, without any statistical evaluation.

In subsequent papers, they repeated this training model, presenting the findings in a similar way to the original paper. The work of Banister and his team, in developing TRIMPs, did not include any kind of validation, not even a formal statistical analysis (Passfield, 2022).

Subsequently, other researchers proposed modifications to the way TRIMP was calculated for different sports, either based on heart rate zones (LTRIMP), perceived exertion (sRPE) or power (CTL-ATL). However, these modifications of the original formula have been tried to be validated against the original TRIMPs—which were not validated—and some of them have not been validated at all.

Recent research shows that these models would be limited at best. The effects of training load are non-linear (Fullerton, 2021), and how this amount of training is achieved is much more important than its overall magnitude. Most training load metrics do not show agreement between observed attrition and the score generated (Passfield, 2022).

FATIGUE AND OVERTRAINING

If the yan of training is adaptation, the yin is fatigue. It is the toll paid for doing a job. We might think we understand this sensation that we have all felt, but the reality is that we have generally been far from understanding it.

Scientists and trainers have not stopped looking for the culprit of fatigue. If we could unmask it, we could work on reducing it. However, no physiological parameter has been found that marks the point at which an athlete reaches exhaustion and is unable to keep up the pace. Neither lactate, oxygen deprivation, glycogen depletion, dehydration or increased body temperature, among other factors, have been consistently linked to the point at which an athlete is unable to maintain pace (Noakes, 2004). While all of these factors favour the onset of fatigue, none of them is the ultimate killer—we will explain this better in chapter 6.

Moreover, we have been seeing how mental fatigue, motivation and mindset affect performance as much—or sometimes more—than purely physiological factors (Marcora, 2009; Ven Horst, 2018). The new central governor theories (Noakes, 2003), the psychobiological model (Marcora, 2009) and the three-dimensional model (Ven Horst, 2018) integrate the aforementioned aspects, but still do not provide a clear answer to the question of why we slow down.

And, although it seems to be clear lately that the best athletes in stress tests are not necessarily the best in competition, and that fatigue is an extremely complex process, the athlete continues to be treated with mechanical and reductionist models, focused on working exclusively on these mechanical factors. What is the point of knowing that fatigue is a complex process if reductionist training plans are then proposed?

For example, acute fatigue and overtraining have so far been treated as two independent concepts. The impulse-response models still in use today (CTL-ATL) differentiate between chronic load (adaptations) and acute load (fatigue) so that performance improvement could be a potentially unlimited event were it not for accumulated fatigue, which prevents us from actually executing those capacities that we would potentially have.

However, we now know that increasing the training load does not always lead to greater adaptations, but rather reduces them above certain thresholds. Overtraining causes processes such as increased catabolism, which leads to loss of haemoglobin mass and lower levels of testosterone and growth hormone (San Millán, 2019). Increasing the training load not only does not further increase the number and functionality of mitochondria, but damages them, worsening the health and performance of the athlete (Flockhart, 2021). The relationship between training and performance-health therefore follows an inverted V-shaped curve.

Overtraining remains an unknown for scientists and coaches. Even the definition of overtraining is complicated. The European College of Sport Science distinguishes between three phases of overtraining: functional overtraining (overreaching), non-functional overtraining and overtraining syndrome.

Thus, overtraining syndrome is an extrapolation in time of the fatigue process. Current impulse-response models cannot capture this transition, which is why overtraining has so far been seen as a different state, even though its symptoms are diffuse and more theoretical than real.

Until now, overtraining was thought of as something purely physical, as distinct from burnout. Thus, it could not be explained that people with relatively low training loads would experience fatigue related to overtraining (Meuseen, 2013).

However, the appearance of syndromes of this type in people with a low training load, and the similarities in their symptomatology with occupational burnout, indicate that we could be facing a similar problem. Furthermore, the separation between body and mind in 2022 seems absurd, and it has been shown that perceptions of fatigue and the presence of negative psychological states coincide with overtraining in endurance athletes (Meuseen, 2013). Emotions have the power to affect our organism and even provoke physiological responses, while alterations in our physiology provoke changes at the psychological level (Damasio, 1998; Sterling, 2012).

Different hypotheses (Cheng, 2020) have been proposed for this overtraining process: glycogen deficiency hypothesis, central fatigue hypothesis, glutamine hypothesis, oxidative stress hypothesis, autonomic nervous system hypothesis, hypothalamic hypothesis and cytokine hypothesis. However, on a practical level, so far the only marker that has been linked to overtraining is a prolonged decrease in performance. No biochemical marker or any other kind of parameter has been established as a reliable indicator of the state of overtraining (Grandou, 2020). The same symptoms that show fatigue would reflect overtraining.

Although all of the above hypotheses may be partly true, it seems clear that fatigue and overtraining are two complex phenomena that cannot be fully understood from a reductionist view. We will see how complexity science offers us new tools to understand these processes.