Earlier this year I wrote about the concept of physiological resilience—your body’s ability to resist deterioration in its fitness level over the course of multi-hour races like the marathon.
Since the concept of resilience was first proposed in a paper last year, there’s been a flurry of new research on it, confirming that it does seem to be an independent component of fitness, beyond the “usual suspects” of VO2max, running economy and steady-state max (SSmax).
Resilience is an extremely simple and elegant explanation for a common problem: people slow down at the end of the marathon because their fitness level is deteriorating.
A pace that used to be metabolically sustainable at the beginning of the race is no longer metabolically sustainable at the end. Moreover, the ability to resist this deterioration varies from person to person, and is amenable to training.
As I was putting together some writing for another project, I was reviewing some of the initial research on the experimental data on the rate of deterioration in running economy and VO2max.
I realized something remarkable: three key scientific papers that form the backbone for the arguments behind physiological resilience all came out in the same year—and that year was 1991, over 30 years ago. Resilience, as a conceptual model, was just sitting around waiting to be discovered for three decades!
I find it remarkable that, with the advantage of hindsight, this fact is so easy to see, and also remarkable that nobody at the time put the pieces together. It would not have required any new experimental data, just a thoughtful analysis of three papers published in the same year.
Today, there’s a constant deluge of new papers, so it’s functionally impossible to stay on top of everything in the field, but back in 1991 that was far less of a problem.
There were probably dozens, if not hundreds, of researchers and grad students who read all three papers in the next couple years, but didn’t have the right conceptual model that allowed them to put the pieces together.
It’s even got me wondering how many papers I’ve read and failed to see some key underlying insight!
So, with the benefit of hindsight, let’s go back and look at each of these three key papers and reason through how someone might have put the pieces together.
1. Joyner’s model of marathon performance
In 1991, Michael Joyner, a physician at the Mayo Clinic, published a short paper on modeling the limits of human performance in the marathon.
His argument was very simple: the marathon is long enough that you can only run it at an aerobically-sustainable speed—essentially, your speed at steady-state max (SSmax).[1]
Since speed at SSmax can be predicted by the mathematical product of your VO2max, running economy, and “fractional utilization”—which is just SSmax as a percentages of VO2max—you can estimate the fastest possible marathon achievable by a human as follows.
First, search the scientific literature for the highest possible values for VO2max, running economy, and fractional utilization. These limits, if they were to occur in the same athlete, would constitute the “best possible athlete.” Then, multiply these best-possible estimates for VO2max, running economy, and fractional utilization together and work out that pace for the marathon.
When Joyner worked out the numbers, the result was a prediction of 1:57 in the marathon. At the time, the world record was only 2:06, so it seemed outrageous. Still, Joyner’s model formed the basis of modern physiological profiling for marathon runners, and it heavily influenced the scientific testing done on the athletes in the Nike Breaking2 Project—one of which did end up breaking 2:00.
So, what was Joyner’s key contribution? Physiology predicts that marathon pace should equal VO2max × Running Economy × SSmax, but the result you get seems suspiciously fast.[2]
2. Brueckner et al.’s study of deterioration in running economy
The second key paper comes from J.C. Brueckner and a group of collaborators across Europe. Their experiment involved getting ten runners (all of whom had run 3:00 or better in the marathon) to complete six runs on an indoor track, while testing running economy both before and after each run. The runs, in order, were 15, 32, 42, 15, 32, and 42 kilometers in length.[3]
Three results are quite clear from Brueckner et al.’s results:
- Running economy deteriorates over the course of a long continuous run
- This deterioration happens in a linear fashion
- The rate of deterioration can be as high as 10% over the course of a marathon, to barely any change, depending on the runner
My reanalysis of the Brueckner et al. data below shows these phenomena quite clearly:
3. Dressendorfer’s study on deterioration in VO2max
Our third study comes from R.H Dressendorfer at New Mexico Highlands University.
The setup was quite similar to the Brueckner study: nine runners with marathon PRs ranging from 2:18 to 3:31 in the marathon completed a standard in-lab VO2max test either fresh and well-rested, or after a 13-mile outdoor run at PR marathon pace. Then, at least one week later, they switched to the other condition.
Dressendorfer’s results showed that 13mi at 100% MP caused a 6.2% drop in attainable VO2max, compared with running fresh, but no increase in maximal heart rate.[4] Since a lower VO2 for a given runner must indicate a lower cardiac output, Dressendorfer interpreted the results as evidence that stroke volume goes down after a long run at marathon pace.
In other words, the heart is beating just as fast, but less blood is being pumped per heartbeat. So, less oxygen gets delivered to the muscles.
Unfortunately, Dressendorfer did not quantify the variability in this change in VO2max, so it’s not clear if some runners were more resistant against this drop in VO2max compared with others.
In any case, Dressendorfer’s results make it clear that long efforts at marathon pace can decrease your achievable VO2max.
Putting the pieces together: resilience is your resistance against fitness deterioration
As early as 1991, there was plain experimental evidence that pointed to resilience as a new, independent component of fitness with significant explanatory power for the marathon and other multi-hour events. If some enterprising grad student or professor had realized this, their argument might’ve proceeded as follows:
- A runner’s fastest possible marathon pace should be the product (mathematically speaking) of their VO2max, running economy, and steady-state max (SSmax).
- When applied to real data, this model yields a marathon pace that’s about 10% too fast, even for elite runners.
- VO2max deteriorates by around 6% over the course of the marathon, and running economy likewise deteriorates by 2–10% by the end of the marathon—figures which vary across different athletes.
- Therefore, there must be a fourth component of fitness which represents the deterioration in VO2max, running economy, and possibly SSmax, over the course of the marathon.
- This fourth component of fitness (resilience) differs from person to person, and explains why elite marathoners run (in 1991) 2:06–2:07 instead of 1:57.[5]
The next question would be “is this new component of fitness amenable to training?” and as of 2024 we know the answer is yes, but had someone put together the pieces in 1991, it would’ve been an obvious next step.
So, what other physiological mysteries are waiting to be solved?
The other obvious next question is—what other puzzles in physiology are just waiting to be assembled by the right person? Clearly there must be more pieces of seemingly disparate evidence that, when put together the right way, snap into place and explain a physiological mystery.
Just off the cuff, here are some unsolved problems that spring to mind. Maybe the experimental data are already “out there” waiting to be assembled, or maybe there’s just one more simple experimental study that needs to be done:
- Why does performance vary from day to day? In other words, on my “good days,” is my VO2max measurably higher? My running economy? My steady-state max? None of the three?
- Why have performances in longer events (10k, marathon) improved so much since the early 20th century, compared to performances in shorter events (800m, 1500m) which have been more stable?
- Is there a “chemical signature” for fatigue at high-end aerobic speeds, analogous to the role of lactate at metabolically unsustainable speeds?
- Why is some fraction of your aerobic power output metabolically unsustainable? In other words, why does steady-state max occur at ~75-90% VO2max and not 100%, and when you exceed steady-state max, where is aerobically unsustainable metabolic power coming from?
- When a muscle becomes more resistant to running-induced damage (e.g. after doing eccentric training or runs through hills), what structural mechanism accounts for this improved damage resistance?
I hope in the next few years, we'll start learning the answers to at least some of these questions.
Learn more about the science of training
If you enjoyed this article, subscribe to my newsletter! It’s the best way to find out when I’ve got new content coming out. In fact, my newsletter subscribers got early access to my original resilience article and had the chance to read it over a month before it was posted here on my website! Sign up if you want to be the first to see what I'm working on.
I also have a book, Modern Training and Physiology for Middle and Long-Distance Runners, that focuses on the science of performance and training for events from 800m to the 10k. Check it out!
Footnotes
[1] Joyner used LT2, the second lactate threshold, as an estimate for SSmax.
[2] Technically if you express running economy in the traditional way, as mL/kg/km, you have to divide by running economy to get the units to work out right, since a “higher” running economy, i.e. a greater oxygen cost, is worse. But if you express running economy the way we express fuel economy for cars (e.g. “miles per gallon”) you can just multiply everything together.
[3] One of the clever aspects of the experimental design is that doing each distance twice allowed the researchers to control for the effects of “learning” to run more efficiently on the treadmill by statistically controlling for it. Indeed, in my reanalysis of the data using modern statistical methods, there is a strong learning effect. A more naive approach would have underestimated the deterioration in running economy because the learning effect would partially obscure it.
[4] Dressendorfer also did a steady-state test at marathon pace: the results showed a non-significant 1.4% increase in oxygen cost, but a 5% increase in heart rate. The heart rate increase might just be cardiac drift, though. The lack of significance in the oxygen cost might just be due to the small sample size: because Brueckner et al. used twelve repeated measurements (pre/post for two sets of 15, 32, 42 km) as opposed to only two in Dressendorfer’s study (control vs. experimental), the effective statistical power in the Brueckner et al. paper is much higher.
[5] In fact, if you make some reasonable “correction factors” from the Brueckner and Dressendorfer data, and apply them to Joyner’s calculations, it almost perfectly matches the world record at the time. This involves a bit of sleight-of-hand though, because you should expect elite athletes in the marathon to be better-than-average when it comes to rate of VO2max decay and rate of running economy decay.
Related articles
New web app: Predicting LT1 pace and Zone 2 pace from 5k time
Designing a plyometrics program for improving bone strength in young runners
Three theories of tissue damage accumulation during running
Lecture: The science behind modern marathon training
What assumptions are baked into your race prediction model?
Tendons do not store energy for free
In windy conditions, running at a constant effort is usually better than running at a constant speed
Loved this article, and your work in general. I probably disagree, however, that Unhjem provides evidence that resilience is "amenable to training?". He doesn't test the effect of training, nor establish causality (are people who become runners born with a better resilience? etc.) While I think the answer will be yes, someone needs to do that work.
I agree that some real experimental work is needed! I'm more convinced by a correlational analysis here, in part because all the other components of fitness (VO2max, SSmax, running economy) are also amenable to training, and it's a pretty plausible pathway from training --> better resilience
Really love reading this website and this article in particular. I'm a hobbyist runner who does science (not physiology) by day so thinking about these new ideas is real fun.
Riffing off your last section on open questions -- resilience is such new territory, and it doesn't seem like we understand the mechanisms behind it. Jones goes through a couple of possibilities in his paper but doesn't come to any firm conclusion.
I can imagine dividing the universe of possible mechanisms into two categories: centrally mediated vs peripherally mediated, so "leg muscle recruitment patterns" would be peripherally mediated but "heart gets tired / run out of glycogen in the liver" would be centrally mediated. Do you think we know which of these two categories is dominant? It seems like this would be pretty easy to test, even with N = 1 ironman participant, if you measure them fresh, after swim training where they mainly use their upper body, and then after run training to fatigue their lower body.
Again as a hobbyist who has been injured in the past, it seems like if a good model of resilience would help direct training. If it were central, maybe I would focus more on cross-training if my resilience were poor and too many miles would exacerbate injury risk. If it were peripheral maybe there is a short strength routine so you could go through to accelerate fatigue before running and reap some of the benefits of a long run without spending that much time. (Just spitballing here, never worked with a coach)
Finally RE: footnote 5 and the Joyner model, I came across this paper a few years back and found it fascinating. After reading your blog and understanding physiology better I find it even more fascinating. It's a different context but reading between the lines of their response, Joyner's team seems to be a little offended by an ad hoc "correction" as evidenced by their use of quotes. I imagine they went back and forth about exactly how to present this in a way that expresses their distaste but is still cordial.
https://www.nejm.org/doi/10.1056/NEJMc1906513
referencing their original article: https://pmc.ncbi.nlm.nih.gov/articles/PMC7315843/