It may come as a surprise but recovery from exercise is not scientifically well understood. As often, when understanding is poor, there is an overabundance of theory. And, also, bullshit. (Around 3.800 words, estimated reading time 19 min but only 13 min if you skip asides and footnotes.)
Perusing the science of recovering from exercise is a frustrating experience.
There is no broad scientific consensus on how human beings adapt to exercise (even with a rigorous definition of Adaptation) which entails that there cannot be any consensus on how they recover from it either. Of course, we know absurdly local things about some recovery processes. For instance, we know that contracting muscles damage capillaries and platelets repair them. But we don’t even know if the transient elevation of serum testosterone following strength training contributes to muscular anabolism. Adding to the adaptation problem is a methodological and epistemological issue:
- “Recovery” is universally understood as “recovery from stress” irrespective of the type of stress, as a stage of adaptation process to stress;
- “Stress” is a catch-all category and stuff that have been put in it may not have anything else in common than, well, to have been put in it and called “stress”.
Fortunately, stress-based theories have more fundamental problems than the definition of stress, and their best competitors are actually not stress-based, so we can spare the stress (pun intended) of looking at its definition (I love anaphora!).
Speaking of of theories, there are two families of theories of recovery from exercise: single-factor and dual-factor theories. The most prominent member of first family is supercompensation theory, where the only factor is “preparedness”. Strictly speaking, the second family has only one member, the fatigue-fitness theory, where the two factors are, well, fatigue and fitness (the word duh applies here), but it has clones. Both are masterfully exposed in Zatsiorsky & Kraemer’s Science and Practice of Strength Training (pp. 10-14) but it’s a textbook aiming at presentation rather than discussion, and they leave some fascinating details out, such as these:
- According to single-factor theories, recovering from exercise is recovering something that’s been spent, but identifying what is spent is an unsolved issue.
- According to dual-factor theories, nothing special is spent and thus nothing special recovered either and “recovery” is not even among the basic notions of those theories.
While there is a lot of bullshit (in the technical sense) about recovery, most of it is rather innocuous. So, instead of looking at specific examples, I’ll briefly conclude on why recovery is a breeding ground for bullshit. But that will come in Part 2, because I’ll split this post to keep it a manageable read. In Part 1, I’ll stick to single-factor theories which are quite the box of fun already.
Recovery ‘from’ as recovery ‘of’
The first general theory of adaptation to stress was formulated in the 1930s and was not co-opted for exercise until the 1960s.
In a short published in Nature in 1936 (vol. 138) and titled “A Syndrome Produced By Diverse Nocuous Agents”, Hungarian-born Canadian endocrinologist János Hugo Bruno Selye (1907–1982), better known as Hans Selye, described the physiological response of rats exposed to a variety of “nocuous agents” ranging from surgical injuries to environmental factors to drugs. According to Selye, in spite of the differences between those agents the rats’ physiological reaction was the same.
Extrapolating from his rat data, Selye proposed the existence of a general adaptation syndrome (GAS) in all living organisms, and lumped the nocuous stimuli together and called them ‘stress’ based on the supposedly common response. Selye would be nominated 17 times for the Nobel Prize of Physiology or Medicine for this from 1949 on (but receive no actual prize). Awarding him the Prize in 1949 might have saved the Nobel Prize Committee some later embarrassment, but that’s another story .
The GAS was a general theory, and a special application of the GAS theory to exercise was proposed in the early 1960s thanks chiefly to Australian Olympic coach and Olympian Forbes Carlile. So let’s start with Selye before moving to Forbes Carlile and, most importantly, differences in their approaches.
Selyes’ proposition of a general adaptation syndrome runs against mainstream thinking in physiology. One of Selyes’ underlying assumption that living organisms have a limited amount of “adaptation energy” to spend to adapt to whatever they have to. The GAS describes the successive phases of the spending of that energy, what happens when they go bankrupt. The flowchart below sums up Selye’s 3-phases GAS model.
The flowchart is complemented by the following diagram illustrates the magnitudes of the initial spending of adaptive energy (Alarm), its gradual recovery and build-up over baseline (Resistance Phases) and dip below baseline leading to bankruptcy and death (Exhaustion). The Alarm-Resistance-[Exhaustion] cycle and its associated curve are characteristic of the GAS. Pplease take a moment to commit it to memory, because Supercompensation Theory will alter that pattern in at least two ways.
[T]he conception of adaptation energy […] is quite contrary to common belief, since it is generally agreed that all vital processes are performed merely at the expense of the caloric energy of the ingested food.Hans Selye, 1938
Selye’s 1936 Nature note does not mention “adaptation energy”. But a 1938 follow-up published in the American Journal of Physiology and titled “Empirical evidence for the concept of “‘adaptive energy'” reveals that it is an underlying assumption of the 1936 model. Later proponents of the GAS would prefer to reduce “adaptive energy” to less mysterious chemical processes. This part of the GAS story had little consequences for its application to training. But “little” is not “no” so it’s worth an aside.
Russian GAS. Fast-forward to the 1950s on, Soviet physiologists are taking over the GAS concept and carrying extensive research. Of course, Selye’s vitalism is the polar opposite of Marx’s materialism, and the Soviets are very thorough in weeding vitalism out of the GAS model. This actually leads to sweeping changes throughout the 1960s and 1970s, following careful re-evaluation and extension of Selye’s data, in by a team lead by Dr. L. Garkavi. Whether an organism responds as predicted by Selye’s model depends of the magnitude of the disruption of homeostasis (the “stable” initial state of the organism) in the alarm phase. Selye’s curve is valid for cases of high disruption and lower disruptions elicit different response curves. Selye’s model is therefore not “general” and “stress” loses its operational definition (“that which elicit the GAS-like curve response”). You can learn more about this in Dr. Natalia Verkhoshansky’s incredibly detailed presentation of the GAS model, which to the best of my knowledge the only available online source in English about the physiological models developed from the GAS by Soviet-era scientists.
By the late 1950s-early 1960s, Selye’s GAS model has the attention of sports scientists in both the Eastern and Western block, but with very different appreciations. In Russia, Lev Matveev the father of periodization, dismisses GAS research as irrelevant, because Selye’s data pertains to pathological cases. Soviet GAS research seems to have found some applications in the 1980s after a thorough revision of the GAS model (again, my source is Dr. Natalia Verkhoshansky’s presentation of the GAS model, since most of the research she referenced is not accessible online).
West of the Iron Curtain, Dr. Ludwig Prokop (Austria) is first to suggest (in 1959) that adaptation to training followed the GAS model but publishing in German limits his impact. Soon after, and independently, a pair of popular articles published in Track Technique in September 1960 and September 1961 connect the GAS with training theory.
The 1960 article (“Stress and Training”) is penned by the journal’s then-editor Fred Wilt, and references both Selye and Forbes Carlile, Australian Olympic swimming coach, and 1952 Olympian (in Modern Pentathlon). Also, and perhaps more relevantly, Lecturer in physiology at Sydney University. The 1961 article is authored by Carlile himself, titled “The Athlete and Adaptation to Stress”, and postulates that the athlete’s adaptation to stress is described by ebbs and flow of a quasi-quantity which translates the athlete’s readiness to perform. This quasi-quantity will become preparedness in later incarnations of the model and cause all sorts of issues.
Preparedness is lost during the alarm phase, recovered during the resistance phase, and spent without counting during the exhaustion phase. But Carlile is ultimately interested in detecting and preventing overtraining and must modify the original GAS model to account for it. Carlile wants indeed to account for the difference between submitting oneself for a short period of time to a training dose that could not be sustained in the long run (‘overreaching’) and doing so until one can’t sustain it anymore (overtraining).
In Selye’s model, both overreaching and overtraining would belong to the exhaustion phase because both are states where “adaptive energy” is overspent. But overreaching is such that, if you stop just shy of spending all your preparedness, a miracle happens: you recover it with interests. You end up increasing the amount of preparedness you have to spend. Visually, this is represented by the flowchart below.
Complementing this flowchart is a canonical representation of the Supercompensation Theory curve (from Zatsiorski & Kraemer) showing the initial sharp drop in preparedness after training (Selye’s Alarm, re-named Training), followed by its gradual return to baseline (Resistance, becoming Recovery) and eventually by supercompensation (Exhaustion, not shown, would be the alternative: Overtraining). Notice that unlike adaptive energy in GAS, preparedness never exceeds baseline before the supercompensation phase, something that I’ll make a huge deal of in Part 2.
Interestingly, the differences between the GAS and Supercompensation Theory (ST) are typically not mentioned at all. That’s too bad, because they have to be justified empirically. For one, Selye’s data did not support supercompensation (then again, it did not support the GAS either, see the aside Russian GASabove). For two, no other data set supported ST either, whose differences with GAS come ultimately out broscience anecdotal data.
Lack of evidential support is an issue common to the original GAS, Carlile’s 4-Phase model, and its later incarnation in Supercompensation Theory.
As argued in a previous post of this series, when a theory’s predictions fail, one can always blame the data or the auxiliary assumptions. In the case of ST, one can blame the choice of observational correlate of “preparedness” and argue that the wrong indicator is responsible for the failed predictions. But it’s a crutch, and the theory is already limping.
I’ll have a look at two problems here: the timing of overtraining and the physiological correlate of preparedness. Together, they are literally a one-two punch that knocks ST out for the count and leaves the playing field open to the competition.
Punch one: the timing of overtraining
Endurance athletes (runners, cyclists, swimmers) who are overtrained may not recover from overtraining before months, and sometimes years of reduced load. Hence, it’s a serious and well-studied condition, and empirical data informs us that the drop in performance associated with overtraining often happens overnight. (The latter is also the case in strength-and-power sports but overtraining usually subsides much faster, and we’ll get a hint as to why in Part 2). On the surface, the effect of overtraining of endurance looks GAS-like post-exhaustion where no recovery occurs. But before that, the timing is off.
The consequence of performance drops happening overnight is that one must either reject ST or accept that performance and preparedness are not correlated. Assuming that ST is correct, preparedness is being spent at an alarming rate if training is too hard during the Recovery Phase, leading gradually to Overtraining rather than Supercompensation. But there is nothing gradual in the drop of performance. This, in turn, leads to two problems:
- there is a “performance inertia” that must be accounted for; and:
- preparedness must be re-identified with a physical quantity that varies according to the prediction of the model.
As we’re about to see, (1) is a more serious problem than (2).
Punch Two: The reduction of “preparedness”
There is evidence that many substances associated with effort metabolism do not supercompensate. Local ATP doesn’t, phosphocreatine doesn’t, and glycogen usually doesn’t either. . This is a source of major vexation for supporters of ST. And so, in the early 2000s, M.J. Hartman and L. Killgore tried to find an alternative, hypothesized that the serum testosterone/cortisol (T/C) ratio could serve as a physical correlate of preparedness.
Around 2001, M.J. Hartman and L. Kilgore teamed with G. Pendlay and followed 7 male elite-level weightlifters for 8 weeks. The research, published in 2004, lead to a 2008 follow-up, where Hartman, Killgore, and a bunch of others (but not Pendlay) followed 6 female elite-level American weightlifters. Both times, they managed to manipulate T/C ratio to match the 4-Phase Supercompensation Model by manipulating volume and intensity. The result for the 2008 experiment is shown below (the 2004 publication does not really count, see the aside Doctor Hartman, the boys, and the girls).
Didn’t they just validate ST? Well, not that fast.
First, Hartman’s manipulation is most likely of the same order as manipulation of training volume that leads to supercompensation of muscle glycogen (cf. footnote 3 above). Depletion, Recovery and Supercompensation interpreted as a sharp drop, increase, and over-baseline rebounds of the T/C ratio are likely the consequence of a particular training program. Supporting this, the 2004 study mentions a 6-week pilot experiment that had failed to induce supercompensation.
Second, Hartman et al. still had a blind spot of “performance inertia”. Well, maybe not so blind, since Matt Perryman reports in Squat Every Day a private conversation where Hartman still had it in its rear-view mirror: that some lifters had broke all-time personal 1RM records during the exhaustion phase of the program. As Perryman puts it diplomatically, “[a]re you really overtraining if you’re still hitting PR numbers?”
Of course, strength athletes seldom reach ‘true’ overtraining. But 2 weeks of build-up followed by two weeks of high-intensity, high-volume training, are still overreaching and thus, Exhaustion by either GAS light and likewise according to ST. Still, the 2004 and 2008 publications leave the misleading impression that new PRs occurred after supercompensation of T/C ratios while some occurred well before. This is borderline bullshitting (in the technical sense, again) since the publications seem to validate ST and correlate preparedness, T/C ratio and performance, while the data set does not.
Doctor Hartman, the boys, and the girls. The 2004 publication lists M.J. Hartman, G. Pendlay and L. Kilgore as authors, and appeared in a supplement to Medicine and Science in Sports & Exercise (vol 36, Supplement) collecting the poster presentations at an Annual Meeting of the American College of Sports Medicine. It is titled “Evaluation Of The Hormonal Control Model Of Competition Training In National-level Weightlifters” and being a poster there isn’t really much more to read than what fits in an abstract (and there’s no point summing up an abstract). The actual poster probably had some pictures and diagram, along with Hartman, Pendlay, and/or Kilgore standing next to it to explain them. The2008 follow-up study is an actual journal article, published by Hartman, Kilgore, and a long string of collaborators, in the Journal of Strength and Conditioning Research (vol. 22:2) titled “Force-Time Curve Characteristics and Hormonal Alterations During an Eleven-Week Training Period in Elite Women Weightlifters” which is accessible online (in HTML). The 2008 study does not reference the 2004 study. In the 2004 study, performance seems to have been evaluated through Clean & Jerk, Clean Pulls, Snatches, and Squats One Repetition Maximum (1RM). Those lifts were also those trained in the program. Finally, there is no indication of how intensity was measured (most likely percentages of 1RM). In the 2008 study, the Force-Time evolution was measured weekly with two accessory lifts that were not trained in the program, an isometric mid-thigh pull and a dynamic mid-thigh clean pull from high blocks. Weekly intensity was calculated as Total Volume Load/Total Repetitions and included all the exercises performed in the program. Neither study reports personal bests at the competition lifts, but the context of Perryman’s quote indicates that it pertains the 2004 study experiment. That being said, women have more endurance than men and are capable of training with higher loads for longer periods (no doubt a reason for the 11-week vs. 8-week programs). Hence, there is no reason to exclude that similar performances could have been observed.
Wrapping Up (For Now)
Supercompensation Theory builds on the simple idea that preparedness is correlated with some biochemical substance and that “recovery” is recovery of that substance.
For an athlete; “preparedness” would be readiness to perform at the top of her current abilities. Admitting that there was not a single biochemical substance characterizing preparedness but multiple candidates depending on what type of sport was considered was an important step towards empirical validation. Currently, only two cases (glycogen and T/C ratio) are known to conform to ST predictions and when they supercompensate, their rebound is correlated with improved performance.
But there are a hitch or two that proponents of ST definitely don’t want to scratch. First, glycogen and T/C ratios only supercompensate in response to some manipulations of training (and in the case of glycogen, food intake), but not all. Second, “performance inertia” still happens, and is still unexplained.
Interestingly, the original GAS curve shows something akin to supercompensation in the resistance phase. If the model had not been invalidated for other reasons, it would actually be a better fit with the data from sports performance. And fortunately, we do not need to look back to the GAS to find a better explanation.
But that is for Part 2, folks.
Interstingly enough, Selye’s first nomination occured in 1949, but the prize went instead jointly to Walter Rudolf Hess for his mapping of the brain area responsible for the control of organs, and António Caetano de Abreu Freire Egas Moniz, the inventor of prefrontal leucotomy (popularly known as “lobotomy”). Leucotomy was performed from the mid-1930s but by the time of Moniz Prize, it had already been invalidated for its clinical indications (in particular depression, which responded much better to electrotherapy). Still, it remained popular as a form of “surgical straitjacket”. And in fact, this had always been one of the intended application of the procedure, exemplified by the case of Rosemary Kennedy (1918-2005), sister to future President John F. and senators Robert F. and Ted Kennedy. Rosemary suffered from learning disabilities due to mood swings and a rebellious temperament (probably some form of bipolar disorder). In 1941, at age 23, her father Joseph P. Kennedy arranged for the procedure to be performed on her, hoping to contain her aggressiveness. The procedure left her unable to talk or walk. Her family had her committed in pa psych ward, never visited her, and claimed that she was “mentally retarded”. The Nobel institution’s recognition of Egas Moniz “for his discovery of the therapeutic value of leucotomy in certain psychoses” caused a surge of leucotomy in the US in the years that follow, often with little regard for medical ethics, and often to similar consequences. Ken Kesey’s 1959 novel One Flew Over the Cuckoo’s Nest testifies it was still common in the late 1950s (Kesey’s had a working knowledge of psychiatric facilities from his involvement in Project MKUltra). The Nobel Committee never admitted any responsibility in the enduring popularity of leucotomy and the Nobel’s website defense of the 1949 prize awarded to Egas Moniz whitewashes the decision with a counterfactual: Egas Moniz would have deserved the prize for his other (unrelated) research on encephalography. In retrospect, awarding the Prize to Selye would probably had been a better idea.
 Prokop proposed in particular that total training volume should be periodically decreased to avoid exhaustion. This suggestion appeared in a book co-authored with Rossner, Erfolg im Sport: Theorie und Praxis der Leistungssteigerung (Vien/Munich: Herbert St. Fürlinger, 1959) which would translate (according to Google, and with a modification of my own) “Success in Sport: the Theory and Practice of Performance Enhancement”. To the best of my knowledge, Prokop & Rossner’s book was never translated in English, and had no impact on Forbes Carlile.
 It is possible to induce glycogen supercompensation by combining training to depletion and carb loading. Then again, as noted by Zatsiorsky & Kraemer note (p. 12) it is a precompetition strategy, not a regular after-effect of training, and it’s to be used wisely.