The scientific interpretation of Oscar Pistorius research

From the go-kart racer to the F1 driver: The  scientific interpretation of the Pistorius research

The possibility that Oscar Pistorius has an advantage from his carbon-fibre blades is a topic that I honestly wasn't going to cover on this site.  For one thing, I've covered it already, dating back to 2007 when the story first emerged.  And secondly, it inspires in people a reaction that buries the scientific question/debate in a tsunami of emotion fueled by social and political correctness.  Even in a recent scientific debate between two camps that formed with Pistorius' own team split turned into name-calling, and the ghost of John McEnroe was called upon in a "You cannot be serious barb".  And that was in a scientific journal... 

However, I've received about a dozen emails in the last few weeks (most of them surprisingly supportive) and a lot of people have asked about the scientific discussion around the blades.  There are a lot more readers now than before, and so the "old news" may not be that old to many.  And then I got to thinking - The Science of Sport began specifically to provide a more detailed analysis of sports news, with a special focus on how sports science might add to our interpretation of what we see.  As we say in our "Mission", we're about the how and the why of sports news.  And there can be no bigger current sports story involving sports science than the Pistorius one.

Added to this is that the media seem scared of this issue - or at least, that's the only reason I can think of for why some of the facts haven't emerged, or why they never challenge the claims made by those associated with Pistorius, and most recently Hugh Herr.  Every claim made is accepted as fact without actually looking at what is being said, and some of the counter-points border on ridiculous.  And so this series of posts is the result - they will summarize my views on the science, the scientific process and the misdirection that has characterized what is one of the most controversial illustrations of science applied to sport encountered since this site started (with the exception of Caster Semenya in 2009 and 2010)

And it will be controversial.  Those who have followed the site for a while will know that I believe there is an advantage, and that is a conclusion based on ALL THREE scientific papers published about Pistorius, as well as the years of hypothesis and discussion, and the known theory of sprinting.  But it would be a glaring oversight to ignore it any longer.  (It's a long, and technical piece, so I may break it up into parts)

Being right vs being correct - the concession, and agreeing to disagree

Let me start off by saying that even once all the science is considered, and even if you arrive at the conclusion that Pistorius has an advantage, your response to that may not necessarily be to stop him competing in able-bodied races.  The more I think about it, the more I can appreciate the views of those who may be saying that he should run regardless of the possible advantage.  They have a point - it's possible that his "rarity" and the fact that he is such an inspirational figure outweigh the potential advantage and future implications of allowing him to compete. 

I'm not one of those people - my interpretation, having followed this from before the hypothesis-generation stage to the research studies, is that he has an advantage.  And because I come at this from an athletics standpoint (as opposed to the human interest side of it), my opinion is that he shouldn't be allowed to run regardless of it.  However, I bring this up counter-point to emphasize that this controversial issue may not have a "right" solution.  It may be a matter of being "correct" (with regards to the science) but not "right" (with regards to his participation).  So while I disagree, I can respect and appreciate a balanced view that says "there may well be an advantage, but given the circumstances and bigger picture, allowing him to run is the right thing to do".  All I ask is the same respect and courtesy, free of emotional name-calling!

What I do take exception to, however, is the dishonest scientific process in this story, and the lies that are actually shown up by data - that's where my criticism is leveled.  This is therefore not only a post about Pistorius, it is about the scientific process behind him.  My criticism is directed primarily at the science that, I believe (and as I'll describe) was twisted and manipulated to create the desired finding.  It is not at Pistorius personally.

However, I do also want to use these posts to respond to some personal comments made by Pistorius at me - unfortunately, journalists don't allow for opportunities to respond to being called a "kart racer" and accused of hunting publicity.  These are minor issues, so I leave those for a footnote at the end of this post.

The bigger issue is some of the claims being made in this debate, which are outright lies, to the point of deliberate dishonesty.  The media of course don't challenge these statements - they accept every claim at face value and instead prefer to sensationalize and create "hero vs villain" sagas.  But the claims that the science proves that there is no advantage demand a response.  And since the media are not bothering to question them, someone else has to.

Right back to the beginning: The starting hypothesis

Before looking at the evidence, it's important to understand WHY specific things were being measured, and what the basis might be for saying that Pistorius has an advantage, because this will help explain the implications of what were later found.  As far back as 2007, it was possible, on a theoretical level only, to debate whether the blades would provide an advantage.

The key points suggesting advantage were:

1. The reduced mass of the carbon fibre limbs
   
   The ability to accelerate the limbs faster is the first theoretical advantage.  A study had previously found that in distance runners, athletes with smaller calves (by mass) were more economical.  And while economy isn't as valuable for a 400m sprinter, the reduced mass is a potential advantage because it allows lower forces and work to be done in accelerating a smaller mass.  Conceptualize this by imagining how your 400m performance would be affected if you raced with 1.5kg strapped to each foot, and then imagine taking that mass away - that is the mass advantage of the carbon-fibre Cheetahs.  This would be detected in a reduction in the work or cost of running at a given speed, and is potential difference number one.
   
   
2. Enhanced energy return from carbon fiber
   
   The company that manufacture the blades, Ossur, make the claim that the carbon fibre blades return around 80 to 90% of the energy they store under compression.  That is, land on carbon fibre that is specially designed and shaped to compress under the body, and more than 80% of the energy will be returned on recoil. 
   
   In contrast, human tendons are relatively poor at returning energy - estimates vary depending on the way it is measured, but energy return ranges between 30% and 70%.  Greater energy return would reduce the need for muscle work, further reducing the "cost" of running.  There's no question that 400m races are metabolically limited.  The details are debatable, but key is that every 400m runner is regulating the rate of metabolite accumulation or energy depletion.  They are limited/regulated by the consequences of metabolic energy use, and anything to lower this (as carbon fibre blades would, according to the theory) would be advantageous.
   
   There was a lot of confusion about this, with some people saying that carbon fiber was only passive energy return and thus a disadvantage.  The reality is that human muscle can return energy, but it requires muscle contraction twice - first to store energy AND then to release energy.  Therefore, there is a cost to energy storage AND return in human muscle and tendon.  The fact that carbon fiber releases energy stored under compression is thus an advantage - there is no cost. 
   
   Once again, this would be reflected in a reduced energy cost of running - same effort, faster speed, or same speeds with less work.  On top of this, there is the matter of fatigue, which causes the force-producing properties of muscle to decline, whereas carbon-fiber never fatigues (in an athletic sense).  The result may be less fatigue at the same speed or for the same effort, over time.

Given those two theoretical considerations, the hypothesis before testing was that the metabolic energy cost of running would be substantially lower compared to able-bodied controls, and that the mechanics of running would differ significantly.  The word "bounce" came up many times in the initial debates.  The question remained, however: Had anyone collected data to confirm these hypotheses?  The answer, until 2007, was no.

The summary version of the evidence

Then, over the course of about 6 months, that evidence was gathered.  It took a while for it to emerge, but when it did, it confirmed much of the above.  This is a summary of the findings, and if you don't have the appetite to go through the detail and you trust the summary, then this is where to stop!  Or read on for the detail of the first study...

1. Pistorius used significantly less oxygen than able-bodied sprinters.  25% less during sprinting (the IAAF Study) and 17% less during jogging (the Herr study), to be specific. Therefore, his metabolic cost of running was lower.  This was true at all speeds, from jogging to sprinting, and of course the carbon-fiber limbs are not designed for jogging in the first place, so one can question how valid a measurement would be during jogging.  It was still found that he used 17%, or three Standard Deviations, less energy when jogging than able-bodied sprinters.
   
   
2. When you add in elite distance runners, then Pistorius becomes "similar" to other runners.  Herr et al conveniently did this when they found that Pistorius uses 17% (or 3 SD) less oxygen than able-bodied sprinters.  Rather than actually testing runners themselves, they turned to the literature and found fifteen-year old research studies on elite distance runners, and sure enough, when they added other people's data to their sample, it helped bring the average down and he became statistically similar to distance runners.  Even then, he used 4% less oxygen than the elite distance runners, which is quite remarkable.
   
   
3. Pistorius had significantly lower vertical ground reaction forces and horizontal braking forces than able-bodied runners.  That means less braking force, but interestingly, the same propulsive forces.  This in turn means less work at the same speed than able-bodied runners.  There is however a disadvantage of lower peak vertical forces, compromising the acceleration from the start.
   
   
4. The energy return from the carbon fiber limbs was 92% compared to 59% for the able-bodied runners.  This, in part, explains the reduced physiological cost compared to sprinters.
   
   
5. Pistorius' rate of fatigue was similar to the able-bodied sprinters, using running trials to fatigue.  This is interesting, and I have my doubts about whether testing an untrained athlete who knows the hypothesis reveals anything of value.  It is also questionable as to how relevant it is to a self-paced 400m race, but this is the only one of five findings that doesn't suggest advantage.

Now, for more on the findings, as well why they were so hotly debated, read on.

First testing - the IAAF study in Germany: Bouncing locomotion with lower metabolic cost

The first round of testing was conducted on behalf of the IAAF by Pieter Bruggemann in Germany in October 2007.  Below is a brief explanation of the three key findings from that paper, the reasons it challenged by Pistorius' science team (led by Hugh Herr, and at that stage, Peter Weyand), as well as some responses from other biomechanists to those challenges.

1. Oxygen use 25% lower = lower metabolic cost of sprinting

First, they measured his oxygen uptake during a simulated 400m sprint.  It's really important to understand that measuring oxygen is done because it's a "barometer" of sorts for the cost of running.  It's not because oxygen is limiting to a sprinter or because a sprinter needs a high VO2max (as was alleged as part of the normal obfuscation of the issue), but rather because much like fuel use in a car, oxygen measurements give an indication of the energy requirements for running.

So if you recall the hypothesis and the theoretical background to the question, measuring oxygen is important because it allows you to establish whether the work requirement is in fact different for carbon-fibre compared to able-bodied limbs, and not whether the runner has a better cardiorespiratory system.  The hypothesis was that Pistorius would show a reduced metabolic cost of running.

Below is a graph of oxygen use during the 400m simulated trial:

So, Pistorius uses significantly less (25% once up to speed) oxygen than able-bodied sprinters during a 400m race.  Also of interest is that he uses the same volume of oxygen over the first 15 seconds, when the balance problem (which many have said would increase his metabolic cost) is theoretically greatest.  Once up to speed, balance is actually not as much of a factor as some suggest - if you want to test that claim, jump your bicycle and ride at 2 km/hour, and then at 20 km/hour, and you soon realise that movement assists balance.

The mechanism for this reduced oxygen, and hence energy cost would be two-fold, and goes back to those two points raised earlier - increased energy return and mass of the limb mean less work, and that means a lower metabolic cost, indicated by lower oxygen use.

However, there was a problem with this conclusion, and the research was not complete in this regard.  That is, when someone is sprinting, energy comes from oxygen-dependent and oxygen-independent sources.  Measuring oxygen uptake during a sprint means that both are contributing, but only one is being measured.  The graph above thus gives an incomplete picture, and the conclusion that the metabolic cost of running is lower was challenged at the CAS.  There is a counter-point, and that is that if he uses less oxygen and thus less energy from oxygen-dependent sources, then there's no reason to suggest that his oxygen-independent contribution would be higher.  But only part was measured, and this would go on to become perhaps THE key point that was challenged in the CAS hearing.

The mechanical energy - 41% vs 8% energy loss

The other Bruggemann finding is summarized by the quotes below.  All are drawn from the report after testing and from a great piece by Edward Ovadia on the issue.

"The energy loss in the blade during the stance in sprinting was measured at 8% and is significantly lower than in the human ankle joints of the controls (41%). This results in a mechanical advantage of more than 30% (it is reported as a 7-fold greater energy return elsewhere in the paper) when the leg is substituted with the prosthesis" - Bruggemann report

Bruggemann attributed most of the 41% energy loss to heat.  This was disputed by Hugh Herr, on the grounds that a) the energy loss in able-bodied runners cannot be that large, that it is not possible to dissipate that much heat in a 400m race, and b) that measuring energy loss in joints is extremely complex, and that there is a possibility that some of the energy in the human joint is not lost, but rather transferred across the joint.  This would bring the energy loss down from 41%.  Bruggemann however disagreed, saying that firstly, the frictional energy loss is as high as claimed, and secondly, the energy transfer across joints is very small.

What is significant is that other studies of energy in joints support the Bruggemann evidence, showing that energy loss ranges between 70% and 30%, depending on whether the person is walking, jogging or sprinting.  The 41% energy loss found by Bruggemann is thus not unrealistic - he may have overestimated, and at the time, a group of biomechanists did debate the model he'd used, but concluded that even with certain small imperfections, the general conclusion was correct - human tendon returns only around 60% to 70% of its energy, compared to 92% for the carbon fiber blades.

However, this discussion was another point of contention for the CAS hearing, who said that the uncertainty re the exact energy loss could not be confirmed. The paradoxical thing about this, however, is that if energy is transferred from the ankle to the knee, as Herr and Kram argued to the CAS, it is not actually an advantage. The problem, as Bruggemann explained, is that this would be "a disadvantage for the able-bodied athlete, because this energy will bend the knee in a phase when the knee is in extension. It's an argument against Pistorius" (Bruggemann, quoted by Edward Ovadia)

Ground forces - less energy loss and less work required

The third finding is that Pistorius had very different ground forces during running. His vertical forces were 20% lower and the horizontal braking forces were 50% lower than those measured in able-bodied controls. Interestingly, the horizontal braking force is reduced, but he doesn't lose anything in the propulsive component of the horizontal force. This is shown below:

To Bruggemann, the reduced vertical force was a distinct advantage, because it was measured at a constant speed, similar to that of able-bodied runners, and meant that Pistorius would do less work to run at the same speed, a finding that is supported by the lower oxygen cost shown earlier.

Herr however argued that this would be a disadvantage during the acceleration phase of running and that faster runners would need more force. There's no question that this reduced vertical force would be detrimental to the start, and goes some of the way towards explaining Pistorius' relatively slow starts.

However, the debate about peak forces and acceleration typically obscures the real significance of Bruggemann's finding - it's a false comparison because we should not be comparing runners at different speeds, but rather comparing Pistorius to other runners at 46 second 400m pace. The comparison to people running faster is irrelevant. As Bruggemann explains: "If we look at subjects running at different speeds, it’s logical to say that the higher the force, the higher the speed. But with all subjects running at a given speed, lower force is an advantage.”

These sentiments were echoed by Benno Nigg, one of the world's leading biomechanists (if not THE leading biomechanist of running):

"He needs less vertical force as well as less horizontal braking and propulsion force – which means he has to work less at the same speed than the control subjects. Pistorius lost less energy and had to produce less work during each [instance of] ground contact than the athletes in the control group.” (as quoted in Ovadia's article)

Conclusion from the first study


The ultimate conclusion reached by Bruggemann, as you all no doubt know, was as follows:

Sprinting with the artificial limbs (Cheetah) is – from a biomechanical perspective – a “bouncing” locomotion and is significantly different to sprinting of able-bodied athletes on hard surface. It is a different kind of locomotion at lower metabolic cost.

This was however challenged at the CAS for the reasons explained above - the metabolic cost was challenged because of the method of measurement, and the mechanical/kinetic data on the basis of alternative interpretations.

I think it's clear that when trying to model the joint loads, the forces, the energy turn, much is model-dependent.  The biomechanical model used by Bruggemann (called inverse linear dynamic model) was questioned by biomechanists, but they nonetheless agreed with his overall conclusions.  So too, as shown for a specific part, did Nigg. 

The importance of the physiological data 

The challenges made by Herr & Weyand were of course equally unprovable.  They offered no alternative to the 41% energy loss, only a question about frictional heat loss.  Of course, this was all that was required, because the CAS hearing only asked for doubt to be cast on Bruggemann's findings, and not proof of a lack of advantage (the context would have been significantly different then).  The result is that the biomechanics part of the debate reaches a stalemate, and finding "conclusive proof" for either position would prove impossible.  CAS of course required conclusive proof and perhaps it is not surprising that they ruled the way they did, on this question anyway.

However, it is for this reason that I would suggest that the physiological data, and not the biomechanical data, hold the more important information.  That's not to say the mechanics are unimportant - the case made by Bruggemann, both in his research and in his responses to the questions, is, I believe, compelling and correct.

But it's when you look at the metabolic factors that things become really insightful, because metabolic cost is a symptom of the mechanics, and so given that there are two ways to interpret Bruggemann's kinetic and energy data, the way to test the options is to use metabolic cost.

And so what was needed was a comparison between Pistorius and sprinters at sub-maximal speeds to ascertain whether that oxygen cost would be lower even then.  That is, repeat the experiment but at slower running speeds.  And this is where Herr enters the picture with a research study that can only be described as "manipulative".  It was, to be blunt, one of the most astonishingly selective research articles I have ever seen, to the point of being dishonest.

But that is the topic of tomorrow's post, when I'll consider Herr's evidence, and some of the claims made in the media about what he showed, and more importantly, what he chose not to show for the sake of the finding.  I realise there are many unanswered questions - but this is only the first part - the analysis of the Herr paper reveals much more, including why I believe the CAS process was so farcical.

More to come.

Ross

The fineprint

And finally, because the tabloid journalist did not allow a response, I must make the following statements, which I do at the end of the piece because they are a footnote, not necessary to the debate, but need to be said.

First, at no point, not even once, have I contacted a journalist to "push an agenda".  Ironically enough, the fact that I have conducted no "pertinent research" on Pistorius is in fact because Pistorius didn't want research.  Back in 2005, I offered to do the research, but he was not interested enough in the science.  Then in 2008, he requested help from SA sports scientists, but again, he would not work with us because of the likelihood that we would confirm the Bruggemann findings.  Ultimately, Pistorius went to Hugh Herr because Herr was going to find what Pistorius wanted.  In tomorrow's post, I'll explain how they did it. 

When you stop to think about it, any person who says "Oscar Pistorius has an advantage" is walking into unpopularity and guaranteed hostile reactions.  That is hardly the way to raise your profile.  The reality (and this is a direct message to the Pistorius camp) is that the journalists are contacting me with questions, and I am answering as honestly as I can, about the evidence and the CAS process.  The alternatives are to lie ("there is no advantage), or to say "the evidence is unclear", but it's not.  It clearly points in one direction.

Secondly, when Pistorius personalizes the issue by calling me a "kart racer", it reveals one of the key problems in this issue - this is not an issue of personalities and of WHO is saying what.  It's about the evidence, independent of the people.  I don't care that Hugh Herr got a write-up in Time, what concerns me is the distortion of the evidence (again, see tomorrow's post).  I also care to point out that Herr is a big recipient of funding from Ossur - the idea of "independent research" from him is utterly false.  There are vested interests everywhere on this one, and it clouds how the media are reporting it. 

That's all, an unnecessary distraction from the evidence.  I'll get back to it tomorrow!

Training, talent, 10000 hours and the genes

Genes and performances: Why some are more equal than others

The genetic influence on exercise performance is dizzyingly complex.  So complex that my best efforts to explain how genes may impact on the science of performance will fail to capture just how enormously complex the various interactions are.  It is so complex that despite the best efforts of scientists to find "the performance genes", they have failed.  This has been interpreted in some quarters to mean that these genes don't exist, that genes are unimportant and that training counts for all - nothing could be further from the truth.  The reality is simply that they're too numerous, with too small an influence, and too complex to find...for now.

However, having previously discussed the 'holes' in the theory that success in sport can be explained by deliberate practice, it's important to consider the genetic component.  When I presented evidence that showed, for example, that only 28% of variance in darts performance could be explained by 15 years of practice time, then it begs the question of where the remaining 72% lies?  When you consider that some athletes are able to become world-class within 12 months of taking up a sport, whereas others slog for a lifetime to stay mediocre, part of the reason may lie in the genes.

And of course, this is enormously complex.  So let me say this upfront today:  The science of success is about the coming together of dozens, perhaps hundreds of factors.  Practice, quality coaching and time spent learning are clearly key factors - this is why you get "hot-beds" of performance, exceptional athletes from anywhere that opportunities exist - the impact of training on performance is large enough that it can help to offset potential differences in innate abilities.  Can it turn anyone into a world-beater?  My opinion is no, but this doesn't decrease the value of the training.

Equally valuable, I believe (and some of the early evidence is below) are genes or innate ability, and this is what has been downplayed in the popular media.  It is not wrong to suggest that practice is crucial and that elite performers do many hours of training. But it is incomplete. And sometimes, incorrect, when you promote one at the expense of the other - training and genes are additive, not exclusive.  So when Ericsson writes in his 2009 paper that:

"distinctive characteristics of exceptional performers are the result of adaptations to extended and intense practice activities that selectively activate dormant genes that are contained within all healthy individuals’ DNA” (Ericsson et al 2009)

it must be challenged on the basis that the science may not necessarily support this.

And to help complete the picture, we look at genes - that is the context of this post.

As mentioned, I have recently written two review articles on this subject - one will be published in Dialogues in Cardiovascular Medicine to co-incide with next year's London Olympic Games, the other will hopefully be published in 2012.  My co-author, Prof Malcolm Collins, is a geneticist, and I owe a debt of gratitude to him for some of the genetic concepts I explain in this post.  So let's look at genes and performance.

The most powerful genetic influence of performance is...

At the risk of starting with the blindingly obvious, the first key point to make is that the single biggest impact made by any factor on sports performance is genetic, and it is biological sex.  Before people react negatively to that statement, please don't view it is a statement of superiority or inferiority - it is simply a fact, and is the very reason we recognize (and embrace) separate categories for competition.  Ask the following question:  If we did NOT recognize that men and women should compete in separate categories in most sports, how many women would be competitive?

Take marathon running - Paula Radcliffe holds one of the most respected records in athletics with her marathon world record.  That performance, easily the best ever by a woman, would have ranked her 473rd in an "open" world list in 2009 alone.  That is, 472 men were faster than this time in a single year.  In history, the time was ranked 3,205th, and that was in 2009 - it's now probably close to 4000th.

This gap exists in all athletic disciplines ranging from 100m to 100km - a 10 to 15% difference between the best men and women is seen across the board.  Of course, the differences may be smaller in other sports - skill-based activities that are not heavily influenced by size, strength, heart or lung volume, hemoglobin content etc may be more competitive.  But the difference still exists (would you back Serena or Venus Williams against Federer or Nadal?), and the result is that if we competed in only one category, no major sporting prize would ever be won by a female competitor.

This is the very reason that we recognize separate competitions - they enable competitiveness.  And the point is that this characteristic, biological sex, is entirely genetically determined.  There are of course cases where the neat binary system we create is skewed by intersex conditions, and we debate and discuss cases like Caster Semenya's endlessly.  In those instances, there is a mismatch between genetic and anatomical sex, such that the chromosomes no longer determine the biology.  However, genetics is entirely responsible for male and female characteristics, and this has an enormous impact on performance.

The question is, if genes exert such an enormous impact on the entire organism, are there similar genetic differences within each grouping, and do these affect various systems (muscle, heart, physiology) in the same way?

Genetic complexity - height as an example of complexity

The next illustration is height.  It's well established that height is a highly heritable characteristic. In fact, 80% of height has been linked to a number of genes (it's called a polygenic trait because many genes influence it), with the remaining 20% being down to environment and diet.

The key about height is that as "simple" a characteristic as it is, it is still impossible to identify all the specific genes and the contributions they make to it, how they interact.  And here, it's important to understand the approaches to the problem.  One can look for single genes - they are called "candidate genes" - that account for the biggest impact in height.  But because even something as relatively simple as height is polygenic, there is no single candidate gene.

There is not even a group of genes.  In fact, if you really want to get down to it, you have to do what are called Genome Wide Association Studies, where you look at the entire genome at once and look for how variations from one person to the next might account for different traits, like performance (or disease, for example).

And when you do this, the numbers become staggering.  Most recently, a paper in Nature Genetics found that you could explain 45% of the variance in height by using 3,925 unrelated people, and a staggering 294,831 different SNPs.  A SNP (pronounced snip), just to explain, is a DNA sequence variation, where for example Andrew might have a gene with a certain sequence, whereas Matthew has the same gene, but with a single change, a single 'different letter', that alters the function or effect of that gene.

In other words, it's not even as simple as having a gene or not, it now becomes a question of which variant in the gene you have!  If this is getting complicated, don't panic - it's because it is complicated!  The bottom line is that there is no such thing as a single gene that makes one person tall and another short.  There are hundreds of thousands of different gene variants, and these variations change the phenotype (the effect of the gene) so that you and a friend may have the same gene but because your SNPs differ, you have different traits or characteristics.

Let's just go back to that height finding, which bears repeating:  Height is almost certainly simpler than something as complex as human athletic performance, yet it requires almost 300,000 different genetic variants, and that helps us explain only 45% of it.  How many more SNPs or genes or DNA sequence variations might it take to explain sprint or endurance performance?  And this is why when you read that the latest studies have failed to find a gene that explains why Jamaicans are so fast, you should interpret it with the right insight because:

• they are often looking for a 'candidate gene' (or small collection of genes), which is a huge oversimplification of performance as a polygenic trait, and;
• there are simply not enough elite athletes in the world to be able to do the study that finds significant associations between that many SNPs and performance.  If it takes 4,000 people to explain less than half of height, then how many more may be required to explain sprint performance, of which height is only a small contributor?

This is also why those genetic tests that supposedly tell eager parents whether little Tim is going to be a sprinter or a distance runner are so over-rated.  These tests screen for several genes, including perhaps the most "famous" performance gene ACTN3, which is supposedly linked to elite sprinting performance.

The problem is, the studies comparing Jamaican sprinters and east African distance runners find no differences for that particular gene.  I hope I've shown you why this may be the case.  In the words of Prof Stephen Roth, one of the world's leading experts on genes and performance "It looks like the gene does contribute something, but only a very small amount at the very, very elite levels".  "Several genes" will sadly explain very little, except in rare cases.  And performance is not one of them.

So there is no single genetic predictor of success (or even of height), but this does not mean that genetics don't count towards success.  We are limited by our capacity to measure how these many thousands of gene variants interact, as the next study of training responses shows.

Genes and training responses:  Responders and non-responders

The next level of our genetic journey is to ask how genes impact on our ability to adapt to training?  This is clearly vital for aspirant elite athletes - whether or not you still believe in 10,000 hours, it's quite clear that some people adapt faster to training than others, or are able to more rapidly acquire skills than others.

The study that is needed to answer this question is to take a large, random group of people and expose them to training, and then to measure how much they improve.  And this has been done.  There are four studies, summarized in the figure below, where big groups have been put through a supervised training programme, and their VO2max measured as an index of fitness.

So, on average, VO2max will improve by 15% as a result of training.  In some studies, it's been as high as 19%, in others, 9%.  This may be due to differences in the training programme, or the people involved.  However, what you should be asking, especially given our look at Ericsson's violin study and the chess paper, is "What are the individual differences that make up that 15%, and what is the genetic impact in these studies?"

And for this, a paper by Claude Bouchard earlier this year.  In this study, 470 untrained volunteers were put through five months of training, and their fitness levels measured before and after.  The figure below shows the result:

As you might expect, most people improve by average amounts - 38% of the volunteers improved by between 300 and 500 ml/min (shown by they yellow and green bars in the breakdown of responders section).  But either side of these "typical responses", you see the extremes - the "low responders" shown in reds and oranges, and the "high responders" shown in blues and purples.  4% of the volunteers improved by 800ml/min or more, whereas 7% improved by less than 100ml/min.

Overall, there was a range of changes in VO2max all the way from 100ml/min (basically no improvement) to over 1000ml/min.  That's a 10-fold difference.  You may recall that yesterday, we saw how chess expertise showed an 8-fold difference between the fastest and slowest to succeed at reaching Master level.  It seems that a similar range of responses occurs for physiology.

The end result is that the bottom 5% of the sample, those who responded the least, improved their VO2max by less than 4%.  On the other end, the high responders, the top 5%, improved by 40%.  That is an astonishing difference, and the simple, and obvious question is where are you most likely to find an endurance athlete in this sample?  The answer is on the far right - the individual who shows large adaptations to training, improves quickly and then reaches a higher ceiling.  I am sure that every one of you reading this knows one of each of these people, perhaps you are one of them!

Note that this study does not take into account that ceiling, and nor does it account for the starting point.  Both of these may be influenced separately, and ideally what you need is a person who starts high, shows this kind of high response, and they are most likely to be the endurance achievers.

In terms of the genes, where's the link?  Well, Bouchard performed a genome-wide association study and was able to identify 21 of those previously mentioned SNPs (genetic DNA variations) that accounted for 49% of the difference in the training response.  As we saw for height, 49% is pretty solid, especially with only 21 SNPs - it suggests that height is not so simple...!

One of those SNPs was in fact responsible for about 6% of the training response, and as far as a single SNP goes, that's a pretty powerful association.  The figure below shows the association between SNPs and training response:

It turned out that the non-responders were people who had fewer of these SNPs than the responders.  If a person carried 9 or fewer of the identified SNPs, they improved by an average of 9% (about half the average), whereas individuals who had 19 or more of the 21 SNPs improved by 26% (almost double the average).  The three-fold difference between the responders and non-responders could thus be attributed to the presence of these sequence variations.  Not the genes - I can't stress enough that the search for a single gene is futile because performance is just too complex.  But rather individual variants that make up the response of VO2 to training.

And again, this is just one component of performance - think of the hundreds of other physiological attributes that make up an elite athlete.  The reality is that our failure to find a performance gene may be more a reflection on our capacity to understand the complexity of physiology and genes than it is an indication that genes don't make a significant impact.

The key genetic question: Same training, different responses?

The most powerful question, then, in my opinion, based on the above study, is the following thought-experiment:

If you took 470 volunteers from Kenya, and gave them the same training as was given to the 470 in the Bouchard study above, would you find the same range of non-responders to responders? Would you find that 7% of Kenyans improve their VO2max by less than 100ml/min?  And would you find that 4% improve by 800ml/min or more?

I would hypothesize that the whole curve would be shifted way over to the right - there would of course be low and high responders.  But the lower responders in the Kenyan sample, would, I suspect, be fewer and perhaps would improve by 200ml/min, not 100ml/min.  As for high responders, instead of finding only a few who improve by 40%, you may find many more.  This would be the indication of a genetic advantage - not that every single person is superior, but that within a given population (470 people in this case), you are more likely to find the physiological characteristics of a champion athlete in one group than in another.

And as soon as you super-impose the opportunity, the competitive environment, the altitude, the diet, the psychology, the culture and belief, the lifestyle, then you have the recipe for a distance champion - Kenya succeeds not because they have these factors, but because they apply these factors to an exceptional genetic pool.

Jamaica has the same scenario for speed, I would hypothesize:  a concentrated group of individuals who possess the necessary physiological attributes to run fast, and to respond enormously to power and sprint training.  Then onto that, you add the history, the role-models like Usain Bolt, the school competition, the excellent coaching, the culture of the island, and the result is the perfect mix to produce athletes who may well go on to win half a dozen Olympic gold medals.

No alchemy in elite sport - start with the right materials

But it all starts with the genetic potential.  In high performance sport, there is no such thing as alchemy - you do not make gold out of other metals.  If you want to produce a champion, a gold medal, then you must start out with the right raw materials.  Everyone will improve as a result of training.  Some, the lucky few, will start out at a level that is higher than the rest, and will improve more rapidly through training.  That this is linked to genes is, in my reading of the evidence, unquestionable.

There are other arguments, of course.  Some are obvious - your body size is strongly influenced by genes, and it limits the sports available to you.  For example, if you're 1.70m tall and weigh 70kg, you won't be playing high level rugby or American Football.  And definitely not basketball.  If you are 2.00m tall and weight 110 kg, then basketball or rugby are options, whereas long distance running probably isn't.  But these are almost absurd illustrations of how genes, which clearly determine these aspects of our physical makeup, influence performance.  But if this is true of these traits, then would it not be the case for something like hemoglobin, muscle enzyme activity/content, plasticity of the nervous system and motor skills?

Rate of performance improvement - a key symptom of innate ability

Last example - I was asked yesterday in a presentation on this subject whether a parent should try to 'diagnose' their child's potential using the genetic tests.  I explained above that these tests have very limited potential to do this, to the point of being useless.  It did get me thinking though about what we look for to detect whether those genes are present.  How does one know that a person has innate ability over and above the typical ability to learn any activity?  And I believe the key, as illustrated by Bouchard's study, is the responsiveness to training.

Of course, the starting point is also crucial, especially for sports that are "physiologically limited" (like running, cycling, swimming, triathlon, where muscles, heart, lungs and brain provide a ceiling for ability).  But for skill-based sport, where training time does matter, the key is how quickly the skill or ability can be acquired - this is the symptom of the innate ability.  I was asked about the Polgar sisters, for example - these are three Hungarian sisters who were taught by their father to play chess to prove that "genius are made, not born".

The coaching of their father, along with professional chess players who were employed to teach the three girls the game,  produced outstanding chess players.  Two became grandmasters, one an international master. Judit Polgar is the most accomplished female chess player ever.  Their story is often cited as a nurture over nature example.

But there are problems with that theory.  First, the fact that they were all family doesn't allow you to exclude genes.  But more than this, when you read the story and start to see not only what they achieved, but when it was achieved, it's difficult to make the case for many hours of training being the secret of their success.

For example, Judit Polgar, at the age of five, defeated a family friend (an adult) without looking at the board.  She defeats her father (a decent level chess playing adult) at five, and beats a Master level player at seven, playing blindfolded!  Remember that yesterday we saw that on average, it takes 11,000 hours of practice to become a master, and you realize how exceptionally talented Judit was.  She then beats an international master player at 10, and a Grandmaster at 11.  These are accomplishments that precede "many hours" of training.

Her sister Susan wins a local chess competition for Under 11s at the age of 4.  Within the first year of their exposure to the activity, they demonstrate exceptional ability, long before the 10,000 hours, long before the deliberate practice can explain their obvious ability.  What makes these sisters exceptional is not simply that they accumulate hours of training, it is that their ability to learn the skills is astonishing - defeating a Master at 7, while blindfolded, given that at most, you've done maybe 3,000 hours of training, is just a staggering illustration of superior ability, developed through training, certainly, but not a performance that you'll find in most people.

Sure, in order to continue to the Grandmaster level, to become the best in the world, it required more training.  But the trajectory was clearly there early, it was a symptom of innate ability, and so this is an argument for genes just as much as it is deliberate practice.

The Polgar sisters, to sum up, are the sporting equivalents of Missy Franklin or Michael Phelps - precocious talents who achieve within the first few years what others take a lifetime to do, and will often fail.  That is as much an argument for innate ability as it is for deliberate practice.  The only experiment that proves nurture over nature is if you can take 100 children, unrelated, and train them all to reach the same level of performance.  The simple fact is that this doesn't happen, and the reason is, at least in part, innate ability.

Conclusion: Two valuable frameworks, both absolutely necessary

I don't think it's revolutionary to suggest that BOTH genes and opportunity are needed.  In the scientific community, you'd be laughed at for suggesting this.  Most people believe that it's a combination of both, and that's why the current models, the best models for performance, integrate all these factors.  One such model is shown to the right - it's a framework for talent ID and development from a 2008 Sports Medicine paper by Vaeyens (click to enlarge).  It clearly includes natural abilities, catalysts, environmental factors and even chance.  These are the basis for current sports science beliefs, and the theories put forward in the popular media, and by Anders Ericsson, unnecessarily and incorrectly oversimplify this.

I can appreciate the value of the deliberate practice framework proposed by Anders Ericsson, popularized by Gladwell, Syed, Coyle etc.  It reinforces that we must better manage our entire sports environment to ensure that more potentially successful athletes are exposed to good coaching, good diet, competition etc.  This has implications all the way up to government level, where policies around sport are determined.  For example, in South Africa, sport is less accessible than it should be, partly because of the removal of sport from our school curriculum.  We also have a dearth of coaches, and few facilities - these factors combine to greatly reduce the chance that we'll produce a Phelps, Franklin, or even a great distance runner, regardless of the talent we have.

But equally, the realization that certain individuals have innate abilities that will help them achieve elite levels is crucial.  It influences where money is spent, how young athletes are steered, how athletes are encouraged to transfer from one sport to another (think of the lifesavers and sprinters who were given a shot at the Olympics and skeleton because of the Australian Talent transfer).  This too has implications for policy, and even for parenthood, in terms of understanding whether a child should specialize early or be encouraged to be as diverse as possible with their sport choices.

All in all, it's a fascinating debate, and thank you for your inputs and contributions to the debate so far.  As always, my aim is to have the first word in a debate, not the last, so I welcome more inputs.  In this post, I've proposed my theory, based on the early gene studies that are associating exercise performance with genes.  I've tried to highlight the complexity, and to illustrate that all is not as it seems.

The rest is for future studies, but I'll leave it with my ultimate conclusion.  To become an Olympic champion, the very best of the best, you need to tick the boxes.  Genes is without a doubt one of those boxes.  But so too are opportunities.  And so is success genetics or training?  It's both.  In fact, it's 100% genetic, and 100% training.

Ross

Talent, training and performance: The secrets of success

Genes vs training:  The secrets of success

Apologies for the post-Tour de France "black hole" that was The Science of Sport! Following and analyzing three weeks of racing left the inevitable backlog of work, which also happened to pick up to warp-speed at the conclusion of the Tour!  However, recovery time now over, I am in the process of putting together the next series, which I'll start as soon as I can, on The Physiology of Pacing Strategies. That will be a video series, consisting of perhaps six or seven short videos, which will bring us neatly to the IAAF World Championships in Daegu.

Genes, innate ability and talent? Or practice makes perfect? Is it all in the training?

But for this week, a few posts on a topic that is both fascinating and, for me, very frustratingly hyped in the media (and, interestingly enough, within sports science as well), and that's the issue of genetics/talent vs training as a requirement for success. I recently co-authored a review on this for the journal Dialogues in Cardiovascular Medicine, and am busy working on a second review on the relative contribution of genes and practice to performance. Those articles will be published in 2012 (I'll let you know when).   I have also done a few presentations on this in the last few weeks, at the University and to the public, so it's a topic that I'm pretty immersed in at the moment.

And then last night, I received an email from a journalist with the Evening Standard of London, asking for some thoughts on a piece that they carried a few days ago.  It was an article called "Why we're the best", and it speaks about culture, practice, genes (or the lack thereof) and other factors that determine why, for example, Kenya produces great runners, China great table tennis players and Australia great swimmers.

You can read that piece here, but I just want to highlight some of the key phrases that warrant a mention, and then evaluate them critically.  I'll do this in two parts.  

1. The first (this post) will look at claims about the role of training and practice on performance
2. The second (later this week) will look at genes and how genetics may influence performance.

Why we win: Culture, practice but not really the genes?  Champions are born, not made?

The video below shows Matthew Syed, author of "Bounce" and he puts forwards some of the same concepts as the article.  It's worth a look, because Syed makes a number of claims that really need to be tested or questioned. Watch the video below, and take note of the following statements by Syed

"Any validity"
"Utterly transform the people we are"
"all about genetics"
"that's not what the science is saying"



Then the following are quotes from the Evening Standard article, which further reflect this thinking:

Success, he [Peter Keen, director of performance at UK Sport] said, was "massively culturally determined" as it dictates what you can "interact with and what is denied you" as a sportsperson. Tradition, success, climate factors, cultural factors - these are more important than some apparently fundamental drivers, such as genetics. 

"To be a high jumper it pays to be tall and this is true if you're Chinese or British," he continued. "But the simple truth of any successful athletic performance is a minimum of 10,000 hours of deliberate practice. That is typically eight to 10 years of your life, two to three hours a day, motivated by the belief that you can be something special."

Later in the article, it talks about the failure to discover the "speed" or "endurance" gene:

But despite Professor Morrison's assertion, attempts to identify a generic speed gene are unsatisfactory.

The search has focused on a gene known as ACTN3. This is because there are two types of muscle fibres, slow twitch and fast twitch. Slow twitch fibres are more efficient in using oxygen to generate energy but fast twitch fibres fire more rapidly and generate more force. These are the ones believed to aid speed and ACTN3 is the gene considered key to their development. 

But despite huge testing programmes of Olympic athletes, not a single record-breaker has been identified with two copies of the variant in the gene. Dr Yannis Pitsiladis, who conducted tests for the University of Glasgow, says this means the impact of genes on identifying sporting excellence has too often been overstated. He adds: "To date there is zero predictive capacity in sports genetics."

Huge complexity, but an oversimplified, and unbalanced explanation

Let me start out by saying that culture, training, diet, opportunity are all crucial to producing sporting champions or elite performances.  But the problem with the debate as it stands is the relative dismissal of physiological factors like genes, and also the extremely oversimplified view that "it's all about the training", or that science suggests genes don't matter.  My purpose with these posts is thus not to dismiss the role of training, culture or belief, but rather to balance out the argument with the facts.

And in so doing, to give an indication of just how complex it really is - the only certainty is that whoever says that success is due to one or two things is wrong.

Testing the statements: The 10,000 hour concept

So there are a couple of claims in the above quotes from the article, and they're worth looking at a little more closely.  We start with the 10,000 hour claim in today's post.  Tomorrow, I'll look at the genes and the claims made about the absence of genetics.  

It is stated in the article that "the simple truth of any successful athletic performance is a minimum of 10,000 hours of deliberate practice".

Deliberate practice means dedicated training in that activity.  In this model, there is no such thing as talent transfer, and there is no is no allowance for accumulating training by play (this is a theory that has been challenged recently, but the deliberate practice model, which was proposed and really developed by Anders Ericsson, holds that only specific, dedicated training works.)

The origins of 10,000 hours

This 10,000 hour theory has its origins in a 1993 study by Ericsson, where he looked at the performance ability of violinists, and showed that the playing ability was determined by the cumulative hours of training up to the age of 20.  That is, the best experts had accumulated the magic number of 10,000 hours whereas those classified as merely "good" or "least accomplished" were found to have done only 8,000 or 5,000 hours of practice, respectively.  The graph below illustrates this main finding, where yellow and orange are the best performing violinists.  Clearly, the average time taken to get to the 'elite' level is 10,000 hours, at least when it comes to playing a musical instrument:

Exceptions to the norm:  What variance would indicate

There's another way to interpret this finding, which I'll get to later in the piece.  First, a major statistical "omission" in the paper undermines how the conclusion of Ericsson and those who argue for 10,000 hours can be made.

I have that study, and what is remarkable about it is that Ericsson presents no indication of variance - there are no standard deviations, no maximums, minimums, or ranges.  And so all we really know is that AVERAGE practice time influences performance, not whether the individual differences present might undermine that argument.  Statistically, this is a crucial omission and it may undermine the 10,000 hour conclusion entirely.

I must emphasize this point strongly: If the theory is that 10,000 hours of practice are needed, and there is no innate ability, then you should not find a single person who has succeeded with fewer than 10,000 hours, and nor should anyone fail having done their 10,000 hours.  Take a look at the graph below.  I've highlighted only the "best expert" and "least accomplished" players, and shown some hypothetical dashed lines to show the ranges within each group.  

It's conceivable that there is a range of practice times within each group, such that there is a person in the "least accomplished" group who does 10,000 hours (shown by the blue circle) without cracking that performance level, and a person who does less but succeeds (shown by yellow).

Unfortunately, Ericsson didn't show us this data, so we can only speculate.  But that didn't stop Malcolm Gladwell from making this statement in his book "Outliers":

“The striking thing about Ericsson’s study is that he and his colleagues couldn’t find any “naturals”, musicians who floated effortlessly to the top while practicing a fraction of the time their peers did.

Nor could they find any “grinds”, people who worked harder than everyone else, yet just didn’t have what it takes to break the top ranks.” – Outliers, pg 39

Again, I don't know how he arrives at the above statements - Ericsson presented not a single measure to support these claims (and I happen to know that he didn't interview him either).  As we'll see shortly, it is actually inconceivable that Gladwell's statements are true - other study of skilled performance show massive variations, and the same will be true for violinists, of this I'm certain.

But what he is saying above is that practice is NECESSARY (the first part of the quote - no one succeeds without doing the time), and that practice is SUFFICIENT (the second part - if you do the training, you will achieve the level).  This is crucial to this debate - those advocating for 10,000 hours are saying that it is both necessary and sufficient.

Gladwell reinforces this when he goes on to quote someone called Daniel Levitin:

“The emerging picture from such studies is that ten thousand hours of practice is required to achieve the level of mastery associated with being a world-class expert – in anything…no-one has yet found a case in which true world-class expertise was accomplished in less time” – Daniel Levitin, quoted in Outliers (emphasis added)

That all of the above are claims are enormous oversimplifications and without evidence becomes clear when you consider "what the science is actually saying", to borrow Syed's words from the clip above.

10,000 hours: Unnecessary and/or insufficient

So we start looking for evidence to test Levitin's (and others') statements about 10,000 hours being both necessary and sufficient.  We do this by disproving it, and begin with chess.  Gobet and Campitelli studied 104 chess players and measured practice time and performance level, and looked at the time taken to reach the Master level.  This is their finding:

So, the average time taken is 11,053 hours.  That's pretty much in agreement with Ericsson's violin players.  So far so good.  But look at that Standard Deviation - 5,538 hours, and it gives a co-efficient of variation of 50%.  For those not into the statistics, what this basically shows is a "spread" of the values around the average.  If the Standard Deviation is small, and the CV is low, then you have a tight cluster - all the individuals are close to the average.  But when it's 50%, then you know you have massive differences within that group.

And that's what happens when you start looking at individuals - one player reaches master level on 3,000 hours, another takes almost 24,000 hours, and some are still practicing but not succeeding.  That's a 21,000 hour difference, which is two entire practice lifetimes according to the model of practice.  It seems pretty clear that practice, while important, is not sufficient for some.  And for others, it's not even necessary.

But let's look at other sports.  Darts has been studied, by Duffy and Ericsson.  They find the following when looking at darts scores and accumulated practice time:

The figure above shows how much of performance can be explained by deliberate practice. In chess, which I showed above, it's 34%.  In darts, 15 years of practice explains only 28% of the variation in performance between individuals!  An extra-ordinary finding, because with all due respect, that's in darts...what else is there that influences performance?  Yet practice time accounts for only a quarter of the performance differences.

What is most interesting about this is that 10 years of practice explained 25% of variability, while 15 years explains 28%.  So clearly, the more you practice, the more you can explain performance.  That's not surprising, but the question is this:  How many hours of practice would it take to explain "most" of performance as a result of practice?  Look at the quote in the figure above, where Ericsson writes that "the development of expert performance will be primarily constrained by individuals' engagement in deliberate practice" (Ericsson, 2009).  Well, 28% is not "primarily constrained" and even though more practice explains more of performance, there is clearly a lot missing from this practice argument.

Sports examples: Very rarely do elite athletes need 10,000 hours

So far, we've looked at chess and darts, both skill sports, but neither is "physiologically-limiting" in the way that running, cycling or swimming may be.  So let's expand our examples and look at other sports.

Start with Olympic wrestling, football and field hockey.  Below are the findings from research on the USA Olympic athletes.

Clearly, 10,000 hours are rarely required. A subsequent study on Australian athletes found that 28% had participated for fewer than four years in their sport - that's probably 3,000 to 4,000 hours, at most.  One netball player from Australia had made the international stage on 600 hours of play.

Accelerated Talent ID and talent transfer

Australian skeleton is another interesting example - in 2002, they decided to adopt a systematic, accelerated talent ID approach to skeleton, and looked at sprinters, lifesavers and speed-skaters to find an Olympic skeleton athlete (Bullock et al, 2009).  It took fourteen months and they had qualified athletes for the Olympic Games, despite no ice-experience, and despite the prevailing wisdom that you "have to learn a feel for the ice" through years of practice to become elite.

Similarly, in the UK, they have had amazing success with accelerated talent ID and talent transfer, producing world champions within years of first introduction to a sport.

And I must emphasize this point - if the 10,000 hour concept is true, and it really does require that this time be accumulated (that is, if 10,000 hours is necessary), then talent transfer would be impossible, as would accelerated performance trajectories that we've seen in Australia and the UK.  And if genetics played only a small part, as some have argued, then Talent ID would also be wasteful and unnecessary, because any aspirant athlete would succeed, regardless of genetic "potential", providing they did the required training time.  This is clearly not true - the actions of federations who invest in Talent ID suggest that despite their talk, they don't believe this anyway.  But more on this tomorrow...

One important point is that different sports will have different requirements, different capacities for talent transfer and accelerated performance, and thus training time.  Rowing, cycling, and canoeing are perhaps easier to learn later in life than skill-heavy sports.  I dare say that tennis, golf may require much earlier exposure and training time - the skill component forces this.  Similarly, sports like football or rugby may also require early exposure, because the tactical insights and understanding are crucial to success.

However, even here, it's possible to identify who will go on to become a professional within the very first years of playing the sport - the Gronigen talent studies have shown this, where at the age of 14, children can already be picked as future professionals because they develop skills, improve endurance and learn tactics faster than their peers.  Differences in how quickly athletes improve, even in skill sports, suggests innate attributes that are predictive for success.


Greatness is recognizable early, long before 10,000 hours are accumulated: Michael Phelps and Missy Franklin


Having mentioned the fact that skill sport success can often be predicted very early, long before a player has accumulated 10,000 hours, it's worth looking at the examples of two swimmers at this point in the debate.  Michael Phelps is the owner of more Olympic golds than any person in history, and his story reveals that greatness can be recognized very early on.  He has told this story and described how he started serious training at the age of 11, doing approximately 1,000 hours per year.

At the age of 15, he was finishing fifth at the 2000 Olympic Games in Sydney.  At 19 he wins 6 golds, and then 8 golds four years after that in Beijing 2008.  But the key is that first Olympic performance, fifth at the age of 15, after only 4 years of training.  Some will argue that the difference between Phelps in 2000 and Phelps 2008 is eight years of training.  I'd say you are partly right, but to me, the bigger issue is physical development - a fifteen year old boy finishing fifth in the wold, no matter how physically developed, is clearly marked for great things very early on in their career, long before 10,000 hours are accumulated.

Missy Franklin is equally telling.  At 16, she won 3 gold medals (and five medals in total) in swimming in Shanghai recently.  And she failed to even qualify for the US Olympic Trials at 13, only three years before.  So her rise has been, to put it mildly, meteoric, and her arrival as a the best of the elites, precedes 10,000 hours by a long way.

My question to advocates of a "genes are less important model, it's about training" is how does this athlete become a world champion at 16 and with relatively little training, whereas thousands and thousands of others, who train maybe more than she does and over a longer period, will never even make the USA team for swimming?  In fact, the question should always be turned around - don't ask why some succeed, rather ask why most fail?  For every example of a champion, there are thousands who get the same training, the same opportunity, but fail to even make national level, let alone become a world champion.  Why?  The answer to this question, and the question of why the likes of Missy Franklin succeed, is genetics and INNATE TALENT.  As mentioned, more on this next time.

Innate ability as a catalyst for training: The alternative theory

Finally, let's go back to the violin study and let me suggest a possible alternative theory for why the best expert performers tend to train more.  

Ericsson concludes that these children just accumulate more training time and that this explains performance.  The difference between the "best experts" and the "least accomplished players" is the training time.

But what if it is exactly the other way around?  Let's take two children at nine years old.  Do they have the same ability to play on first exposure?  Ericsson's model says yes, and that the difference comes later, when one child practices more, gets better teaching.  But what if the difference is present from the very first note, the first exposure to the activity?  The parents of a child who shows some ability encourage further practice, they invest in teaching and training, and this child, by virtue of the fact that he/she has more ability to begin with, accumulates more practice.  

But the child who has little innate ability makes the violin sound like the death march of stray cats, and their parents do not encourage more play.  In fact, they discourage it - the "go play outside" syndrome takes over, and the child is never exposed to teaching or practice.  His trajectory is set precisely because he has less innate ability.

My point is that the above graph can be explained just as easily using an innate ability argument as it can a deliberate practice argument.  The current explanation is been practice, but given what we know about genes, I'd argue that the 're-inforced' behaviour catalyst/filter explanation is just as likely.  Those who display greater ability early on (innate ability, that is) are encouraged to practice more, and they set out on that journey towards "best expert" levels, the yellow line, from the beginning.  Those who lack innate ability are placed on the blue line.  No amount of training will change this, but the behaviour is set early so we never find this out.  The result is that ability is determined by practice, on average, but that practice volumes are perhaps themselves influenced by innate ability.  

Until someone shows that individual differences in performance can be made to disappear with training, and that the differences we measure in performance are not present from the outset, I remain skeptical about an extremist view of performance being due to one factor.

The importance of practice: Practice is vital, but extremist arguments just don't work

Now, I don't mean to be dismissive of the importance of training.  Of course, practice is vital.  It is a pre-requisite for success, especially when you have a competitive sport where many are vying for the same medal.  In that situation, the person who succeeds must train hard.  But their ability to get more out of training, to adapt to training and also to start off from a higher "baseline" is just as important, and those are factors influenced by genes, as I'll cover in the next post on this topic.

To argue that it's about the training, and to dismiss that genetics play a significant role, is to adopt an untenable and grossly oversimplified position.  The 10,000 hour concept is a nice motivational tool, a way to encourage more training, to inspire people to improve.  The idea for elite performance is that the right person hears this and believes, and then does the training.  But to attribute success to 10,000 hours of training is not only over-simplified, it's wrong.  Matthew Syed, in the clip above, says that one will argue that you "need both talent and opportunity, but that's not what the science is saying".

The truth is, the science is saying exactly that.  Make no mistake - producing champions is incredibly complex.  The success of Kenya at distance running, or of Jamaica in sprinting, cannot be reduced to one, or even a few factors.  You will find altitude in many places.  You will find socio-economic similarities all over the world.  But you won't find champions.

Success is probably due to hundreds of different factors, all interacting with one another.  But the end result is that if you take 100 aspirant athletes in Kenya, and 100 aspirant athletes in the USA, and expose them to the same training, you will not see anything like the same success rate.  And that is due to genetic differences that are too complex to discover with the approach that has been adopted so far.

In this post, I've looked at the training and practice factor, and hopefully given a broader view on it than the terribly simplified version that 10,000 hours is what it takes.  Next time, the genes, and the very certain finding that genetics is vital, both in determining innate ability, and our response to training, and even our motivation or desire to exercise and train in the first place.

Ross

Tour 2011: Alp d'Huez, leaving the mountains and onto the TT

Tour de France 2011: leaving the mountains and onto the TT

The Tour de France is now only one day away from finally answering every question posed of it, and its riders, almost three weeks ago.  A time-trial in Grenoble holds the final answer:  Schleck or Evans?

But much fell into place during the Alp stages - we now know that Contador wasn't quite up to the challenge this year (though the relative effects of the Giro and the knee problems are not known), we know that Andy Schleck carried good form into week 3, and we've seen Cadel Evans' strength and consistency, and has certainly had the least drama-filled Tour.  We also finally saw Thomas Voeckler pay for his great performances, but he certainly inspired along the way.

Aggression in the Alps

Yesterday's stage up the most famous climb in the race, Alp d'Huez produced some incredible racing.  If the Pyrenees were described by the word "conservative", then the Alps were the polar opposite.  Aggression was the feature, not only yesterday but all the way back to the foothills of the Alps where Contador ignited the race through his attacks on Cat-2 climbs.  Then on Thursday, it was Andy Schleck, launching with 60km to go, and finally, to bookend it all, Contador attacked even earlier, on the Col du Telegraphe and pulled Andy Schleck away from the chasers.

That brought a concerted effort from BMC at the front of the peloton on the Galibier.  At one point, Contador and Schleck had almost two minutes' advantage over Evans and Schleck, but BMC first held it, then slowly reeled it back, before Evans found himself, for the second day in a row, pulling the peloton up to an early break with a Schleck in it!  His effort closed the gap to around 30 seconds by the top, and on the very long 40km descent to the foot of Alp d'Huez the gap was removed.

So had you tuned in then to watch the final climb, it looked like just another day.  It had been anything but.  Meanwhile, Thomas Voeckler was doing a ride to admire, if not wonder about slightly - he tried to follow the Contador/Schleck break, but wasn't quite able to bridge.  And he hung to a tantalizing gap of around 30 seconds for a big part of the Galibier.  It went to 40 seconds, then he fought it back to around 20 seconds.  Then it went up, then it came down.  Voeckler the yo-yo was agonizingly close, and at times, less than 100m away from the group.  But he couldn't quite close it and within the final 4km of the climb, he finally cracked.

He was reeled in by the pressure from behind, and promptly dropped from that Evans group as well.   Perhaps, in hindsight, having failed to bridge, he might have waited for the chase group and been pulled back by them - but this is easy to say afterwards.  Fact is, Voeckler was inspirational.  While he managed to catch up on the descent, it seemed that barring a miracle, his yellow jersey tenure was over, but not without providing some of the race's great performances and highlights in the process.

Alp d'Huez: instant separation

Then on Alp d'Huez, the race exploded (I'd say "literally exploded" but that's what Paul Sherwen says, and fortunately, it's not quite true!).

And it's worth analyzing some of the performance times for different stages of the climb, because they show how Contador actually cracked about two thirds of the way up the climb, after an incredibly strong bottom third.

The historical context of the climb is equally interesting - the 41:21 climb of Sanchez was the fastest of the day - that's a return to the times we saw in the 1980s, when bicycles were considerably heavier.  However, I don't want to become too repetitive because the issue of performance analysis and characterizing the power of the climbs has been a regular feature of our Tour coverage this year.  So rather than spend minutes on it now, I have included that discussion in a separate "sub-post" section at the end of this post, for those who are interested in further debate.

If you want the summary though, the times yesterday do little to dissuade me of the view that performances are universally slower, and by a considerable amount.  Of course there are tactics, and there are varying conditions, but consider for example that all three regularly completed HC climbs in this Tour have been over three minutes slower in this Tour than were seen the 1990s and 2000s.  And not a single HC climb in the last two Tours have been done at anything close to 6.2 W/kg, let alone the 6.4 W/kg seen in years gone by.

And so the combination of performance times decreasing, the physiological implications of those performances and the bio-passport data suggest progress in the anti-doping fight - only more time will confirm or disprove that hypothesis.  For more, check the bottom of the post.

Contador's climb - hard, medium, crack pacing strategy

Contador's times were easiest to record, because once he attacked with 12.5 km to go, he was on camera most often.  I'll look at Roland's times relative to his, to give an indication of how the climb was structured by the two of them - it's very different.

So Contador rode the whole climb in 41:30, but it was set up with a first 3.8km of 11:24.  The next 5 km would take him 15:59 and then the final 5km 14:07.  His speeds thus dropped from 20.1 km/hour for the first 3.8 km, to 18.8 km/hour, before increasing to 21.3 km/h for the final segment (which includes some leveling off right at the summit)

Rolland, on the other hand, rode the first 3.8km in about 12:34 (he started 50 seconds ahead, then was caught and a gap of ± 15 seconds opened).  His next 5km was done in about 15:51 and the final 5km in 13:29.  His speeds per segment were 18.3 km/h, 18.9 km/h and then 22.2 km/h.

That is put into perspective when you consider the profile of the climb.  The first two kilometers are the steepest of the climb at over 10%, before it becomes more gradual, and then gets steeper at the top again.  Contador therefore attacked in the steepest part, and rode over a minute faster than Rolland over the first few kilometers.  He then slowed progressively even though the roads got less steep, and Rolland was able to hold him to 20 seconds before closing the gap and winning the stage.  This is further illustrated by the time gaps to the yellow jersey of Thomas Voeckler who held Contador to around 2:45 for much of the final 5km.

Contador therefore cracked, and that's partly the toll of stage and earlier aggression, both on the climb and those before it.  It's also because he probably rode well above capacity in those first 5km - I don't know the precise elevation of the climb by kilometer, but perhaps someone does and can estimate the power output per section based on the above splits?  The fact that Contador was faster on the steeper earlier slopes than from about 10km to 1km to go says much about how he ran out of reserves towards the top.  It was all Contador could do of course - he had to throw everything in and he did (to his credit).  In the end, just too much to ask.

Pacing strategy and the time-trial

And speaking of pacing strategy, perhaps that's a good point to leave the mountains and talk time-trial.  Today will see Cadel Evans try to overturn a gap of 57 seconds to Andy Schleck to win yellow.  I've spoken to a number of people about their projections and they range from a 2 minutes loss for Schleck to an equal performance!  Clearly, there will be some intrigue!

The two biggest problems with trying to guess the outcome of the final time-trial of the race are 1) the cumulative effect of fatigue is so difficult to predict, and 2) riders often don't ride them at 100% or with the same incentives at stake, because they don't need to.  Therefore, when you look back over the last few final time-trials, you find cases that support either position.

For example, in 2009, the final time-trial of the race was 41km (comparable length), and Cadel Evans beat Andy Schleck by 30 seconds.  But, that was a race where Evans began that time-trial well off the lead, whereas Schleck was in contention for the podium.  To highlight this even more, in last year's final TT, Evans finished 166th and over 5 minutes behind Andy Schleck!  Again, this was the time-trial where Schleck was fighting for the race win, Evans had little to gain and came off what was a relatively poor Tour and was riding with injury.

This year, that's quite different.  Evans has been strong and consistent - he's followed every attack, he's shown the strength to pull the peloton up two mountains in the last two days and he looks really solid for the win.  There's the yellow-jersey effect, of course, in that being in yellow supposedly adds the motivation necessary to find a better performance.  And while that's true, I don't believe it's a competitive advantage in a race like this.  Consider Cadel Evans - he has been trying to win the Tour for years and has come close on a few occasions.  Is there more motivation than that of a rider who knows his chance is 60 minutes of effort away?  I've always found it quite presumptuous to try to guess at people's motivation anyway.

The next issue is fatigue - Andy Schleck has had two enormous days in the Alps - a 60km break and then yesterday's efforts.  Cadel Evans hasn't exactly had it either though - he did all the work on the Galibier on Thursday and a great deal more yesterday.  But his allocation of effort has been more controlled and maybe a slight edge here.  Again though, it's impossible to guess at recovery and even effort allocation unless it's quantified, so this doesn't really help.

Pacing strategy is another interesting one.  When you're talking about an hour of riding and 53 seconds advantage, then small 1% differences as a result of pacing can be telling.  Going last is a slight advantage in that regard, because you know what the targets are.  And Schleck may be able to pace himself off Evans.  However, that's pretty meaningless if Evans is 2 minutes faster - all that will happen is that Schleck will be "pulled" too fast over the first half, and pay in the second.  So it will be fascinating to see how the relative gaps between the two unfold and how they structure their effort.

All in all, it should be a fantastic day.  We'll see what it throws up in terms of analysis and bring you the final yellow jersey tomorrow!  Thanks again for reading and discussion!  One day to go!  (and I won't lie, as enthralling as it's been I feel the need for holiday!  The Tour de France really is grueling!)


Ross


Donate and support The Science of Sport

Oh, and one last thing - we try not to be too forward about this but every once in a while, we put it up, and that's a "Donations" button, on the top right of this page.

We run this site as a "labour of love" and will continue to do so, but of course, any support is greatly appreciated. So, if you would like to donate and support our continued efforts to bring you the insights and analysis, please consider donating to our site. You can do so by scrolling up to the top of the page, where you will find the DONATE button at the top right of the page.

As I say often, thank you for reading, for commenting, for sharing with others.  We really do appreciate every reader, every comment, and we'll continue to bring our insights to you.  But if you enjoyed the coverage and would like to express it in the form of a donation, we will be doubly grateful!

Oh, and if you're reading the post online, please CLICK HERE to go to the site - the Donations tab is top right.

Thanks and enjoy!

Ross

___________________________________________________________


The historical overview of the climb

In terms of the historical context of the 2011 performances, the overall time for Contador was 41:30.  Sammy Sanchez was the fastest of the day in 41:21, while Pierre Rolland, first to summit, did it in 41:52 because he started the climb with a 51 second lead (bear in mind small errors in timing off the TV - I'll watch it again and correct small errors later).  

It does not escape notice that the times were comparable to what we saw in the 1980s.  In the 90s and 2000s, sub-39 minute performances were expected, after Gianni Bugno and Miguel Indurain had been the first to break 40 minutes in 1991.  In fact, since 1994, seven of eight climbs have been done in under 39 minutes, let alone 40 minutes.  

Carlos Sastre in 2008 rode a 39:30, a sign of the passport, perhaps, and this is the first time (unless I'm missing something) that the fastest ascent of the climb has been outside 40 minutes.  And it wasn't just slightly outside - Sanchez did 41:21.  And many have commented that this fails to account for the earlier climb and tactics. True, but the size of the effect is too large to be dismissed.  The earlier efforts on the Galibier (which were not as massive as many think) might account for some of the difference, but this climb was over four minutes slower than Pantani's record, and 3:30 behind the times of 1995, 1997 and 2001.

Implications for power output

In terms of the implications for power output, below is the graph that I showed the other day as well, courtesy Alex Simmons.  It shows the expected climb time for different power outputs and different wind speeds.  On the note of wind, I have watched the climb three times now and there seems to have been a breeze in the trees, but the flags on camper vans and poles on the climb show very little wind, so I think the effect was minimal, but to give the benefit of the doubt, assume a headwind.  

Without wind, the climb was done at an estimated power output of 5.65 W/kg.  A headwind would push it up towards 6 W/kg.  At this point, it's worth saying that Chris Anke Sorensen (who we've been following all Tour) rode Alp d'Huez at 5.3 W/kg, and finished the climb in 47:10.  So just over five minutes off.  Within the relatively linear region of this graph, those 5 minutes are worth about 0.7 W/kg, so that puts the top climbers at ± 6 to 6.1 W/kg.  An alternative method is to take the Sorensen performance (47:10 and 5.3W/kg) an estimate the headwind and then derive the power output for the faster time of 41:30.  Turns out that it it predicts a headwind close to the yellow line shown above, and so the race leaders are climbing at an estimated range of 6.0 to 6.1 W/kg.  And that's a highest case scenario, I suspect it to be lower.

This is exactly what I'd expect of this level of rider at their maximal efforts at the end of a three-week Tour, as seen today, and as has been seen the entire Tour long.  To repeat a common observation in the Tour - this is much, much lower than we used to see in the 90s and 2000s, where climbs over 6.2 W/kg and higher were common place (for some of the data, see this post) and the lack of those performances now is conspicuous.

Of course, Alp d'Huez has rarely been done after such aggressive riding on the preceding climb.  Then again, it's rarely been done after only 93 km of riding either - today's stage was short, the total riding time "only" 105 km and took 3:13:25. 

And I readily acknowledge that any one climb in isolation paints a picture of nothing.  But I'd point out that this year, there have been three HC climbs that are regularly done in the Tour - Luz Ardiden, Plateau de Beille and Alp d'Huez.  Every one of them has been more than 3 minutes slower than the record times for those climbs, all of which were set in the EPO and blood doping era of the 90s and 2000s.  The same was true last year, incidentally - not once was an HC climb done at more than 6W/kg, whereas that was common in the 90s and 2000s - 6.4 W/kg was the average back then.  Even when you correct for tactics and weather, the number and magnitude of those differences is compelling.

The dividing line, I believe, comes in 2008, when the biological passport was introduced (and please read this for the context).  And now, as the Tour rolls out of the mountains again, it has once again suggested to me that a) performances of greater than 6W/kg (let alone the 6.2 to 6.7W/kg we used to see) are not credible and that b) the doping problem, while no doubt present, is coming back under control thanks to the stringent testing.  Obviously, this is still a hypothesis - let's get thirty or forty climbs that are slower, not the five or six in the last two years.  But so far, the data support the hypothesis, and only time will tell if it's true.