I hate zone training. It’s a pet peeve of mine. I hate the idea and the concept behind it. But so many coaches out there use it, and in actuality we all tend to classify workouts into different zones. I’ve wrestled with trying to explain my hatred for zone training and what I’d call the alternative method for a while, but it’s a difficult thing to wrap your head around. I knew I hated it, but I couldn’t effectively explain why to outsiders. I’d dance around the why’s but never had a satisfactory answer. Part of the reason was I didn’t have a simple way to explain the alternative. I could throw a bunch of information and examples to describe it, but there was no easy descriptor. Before getting into what the alternative is, lets look at what exactly the physiological model is and how we got there.
Defining a model:
The physiological model of training is one that relies on the premise that there are a few big physiological parameters that govern performance. The big ones mentioned in research and the literature are VO2max, Running Economy (RE), Lactate Threshold (LT), andsometimes lactate tolerance or anaerobic capacity/tolerance depending on publication. The idea is that these four things combine to create performance.
The central premise of the physiological model is that IF we improve one of these parameters, then performance improves.
But how does this relate to training? The model takes another step and says that there are certain intensities or zones that will improve X parameter.
So the full model really states that IF we do X training, Y parameter will improve and thus performance improves.
If this was a logic class it would look something like this: X->Y=↑ P
Sounds reasonable right? Well, hold your horses…
Before fully diving into what’s wrong with this model, it’s important to fully understand it so that it doesn’t look like I’m creating a straw man argument. As with anything there isn’t one strict concept of the physiological model, but instead many slight variations covering the spectrum. I’d like to give an example of that and show the limits of a normal physiological model.
Not meaning to pick on Dr. Jack Daniels, but his updates between his 1st and 2nd edition of the book Daniels’ Running Formula perfectly illustrate this. In his first edition, he outlined what have now become the classic zones used in many programs. What’s interesting, and what is typical of a strict physiological model system, is that in between these zones were what’s referred to as “no man’s land” training.
(Thanks to the blog http://sisyphusrunning.blogspot.com/2011/08/no-mans-land.html for this)
In his book he defines this no man’s land zones as:
“No man’s land of training. Training intensities that fall into ‘No man’s land,’ are either too easy or too hard to reap the benefits you want. You are not, as may sometimes be assumed, achieving the purpose of training the two systems on either side of the chosen intensity. What you are doing might be termed, “Quality-junk” training. At the least, it is training aimed at accomplishing an unidentifiable purpose. Always have a purpose for every training session; ask yourself the following questions: ‘What system do I hope to improve by doing this workout ‘and ‘What am I really trying to accomplish?”
This is what I’d call a strict adherence. If you train too fast or too slow of the zone it’s “junk” because it isn’t targeting the special zone which should target some parameter. This sounds good, but the issue is that training doesn’t occur in isolation and as we will soon find out those training zones don’t even attack those parameters terribly well. It’s missing the central point of training, that it is a stimulus. Yes different stimuli will result in different results, but everything is a stimulus to a degree. If you are too fast or too slow, it doesn’t negate the adaptations, it just changes them slightly. Secondly, you have to remember that this thinking is with the model behind it. There is no purpose for the “no man’s land” zone training IN THIS MODEL. How accurate that is depends on how well the model reflects what is actually going on. Because every model has holes and doesn’t perfectly reflect what is going on. In this case, the model fails to accurately reflect what’s going on as we shall soon see.
What is interesting (and very astute of Daniels) is that in his 2nd edition of the book he takes what I’d call a more loose interpretation of the model. Why? Because he expands his LT/tempo section of the book to say that unlike in the first edition, tempos can be slower and longer than the LT zone and still get great benefits.
Now that we get what the model is, what’s wrong with it?
(The above section might come across as harsh to Daniels but that isn’t my intention. Instead I think it’s cool that Daniels realized the potential of other zones of training andamended his thoughts. A smart coach!)
What’s wrong with the model?
As a quick reminder the physiological model relies on the following logic:
- There are certain parameters that define
- Training each one of these parameters improve
- To train these parameters you work at X
Hopefully you’ve noticed a few holes in this logic. Let’s go through them.
First, those big parameters don’t really do a wonderful job correlating with individual performance. Yes, some studies show strong correlation to a mixture of them, but you can’t really separate out good runners from great runners. Research shows that combining the big 3 of RE, VO2max, and LT explains around 70% of the variation in performance (Di Prampero, 1986). So, good, but not great? Vo2max in particular does a relatively poor job. You can check my fallacy of VO2max article if you want more details. For now I’ll briefly quote one relevant section:
“Showing the separation of VO2max and performance, the Vollaard et al. study found that the change in VO2max was not related to the change in time trial performance (2009). Studies demonstrate improved performances without changes in VO2max (Daniels et al. 1978). Also, studies show that VO2max can improve without changes in performance, which is seen in a study by Smith et al. that showed improvements in VO2max by 5.0% without an improvement in performance over either 3,000m or 5,000m (2003). In addition, in looking at long term changes in performance in elite athletes, changes in performance occur without subsequent changes in VO2max.”
As far as the other parameters, yes they all correlate to performance to a degree when looking at groups. So that is valid. The question is do changes in these parameters correlate with improvements in performance? The answer is it depends. In several studies you’ll see an increase in LT, RE, or VO2max as well as an increase in performance. In others, like those mentioned above with VO2max you’ll find no change despite an
improved parameter. Another quick example is a study by Vollaard et al (2009) that found that RE and VO2 changes didn’t explain performance improvements, or in their words: “This study demonstrates that improvements in high-intensity aerobic performance in humans are not related to altered maximal oxygen transport capacity”
Similarly, with lactate threshold. An increase in threshold is sometimes tied to a decrease in performance as demonstrated in the world by Jan Olbrecht. This is usually a result of the lactate threshold increasing while anaerobic ability is decreasing. For more on this concept see Jan’s work.
Which brings us to the point of training not occuring in isolation. We can’t independently try and improve these parameters in all runners and expect performance to increase. There’s an interaction between the training types. For instance, there’s a strong back and forth tug of war battle between LT work and the so called lactate tolerance work. This interaction effect is somewhat addressed by periodization, but when training is thought of in isolation such as do X and improve Y parameter, it’s often lost.
So, not surprisingly, we start with a somewhat decent premise and get a bit shakier as we move away from it. But where things really fall apart is in the next step, which is the most important because it translates lab work to the real world, which is what we are all concerned with anyways.
Training zones don’t really work?
Perhaps most importantly then, the training zones don’t really correspond with improving the parameters. And even if they do to a degree, there are multiple ways to improve the parameter. Let’s look at these zones individually to get an idea.
In this new quest to improve VO2max, it was first ASSUMED that training at VO2max was the best way to improve it. The logic was simple. Spend as much time at maximal oxygen consumption as possible and it has to increase right? Sounds logical enough…So the next step was figuring out what paces would elicit VO2max.
Why do we train at roughly 3k pace to improve VO2max? Well, we got to that point because a bunch of researchers took on the challenge to see what was the slowest speed which would maintain VO2max for a relatively prolonged time. Depending on what research article you look at, it generally came out to be something that lasts roughly 7-10 minutes. Or about the time it takes to run a 3k. Therefore, 3k pace became VO2max pace. Yes, running faster than that elicits VO2max too, but it’s not the slowest speed. We had the intensity. Onto the next step.
A bunch of researchers did more studies seeing different intervals that would allow you to spend the most time at VO2max during the workout. That’s where we get Billat’s famous 30-30 workout and a slew of others. After this we had our optimal interval speeds, lengths, and recovery.
So it was all set. Do these workouts at X speed with this recovery and you are set. The problem?
Well the research didn’t exactly back up the idea that to improve VO2max (even if it mattered much…) we should do that. I could go on and on but a nice little review of training to improve VO2max by Midgley (2006) sums it up nicely:
“Training intensities of 40–50% V˙ O2max can increase V˙ O2max substantially in untrained individuals. The minimum training intensity that elicits the enhancement of V˙ O2max is highly dependent on the initial V˙ O2max, however, and well trained distance runners probably need to train at relative high percentages of V˙ O2max to elicit further increments. Some authors have suggested that training at 70–80% V˙ O2max is optimal.
Many studies have investigated the maximum amount of time runners can maintain 95–100% V˙ O2max with the assertion that this intensity is optimal in enhancing V˙ O2max. Presently, there have been no well controlled training studies to support this premise.”
So there we go. We don’t know if that is the best intensityor not to increase VO2max. Tons of research, but no clue.
I also like one of their other statements:
“Synergistic and interference effects between optimised training protocols designed to target specific physiological performance determinants and the influence of individuality then need to be established before sports scientists can make recommendations to
runners and running coaches,with a high level of confidence, on components of an effective training programme.”
Individuality. Interaction. IMPORTANT.
So what we’re left with is a magical zone that we’re not sure if it even attacks what it’s supposed to attack (and if that even matters?)
What does this kind of thinking get us? Well, a quick example is certain studies came out a long time ago that said after 50-60mpw amount of mileage, VO2max doesn’t improve, so why would we run more mileage if that’s true since VO2max is the be all end all (Berg, 2003). This isolationist approach neglected the complexity of performance so we ended up with a bunch of coaches thinking you don’t need to run much mileage…whoops.
The lactate threshold is one of my favorite topics as I like to do lactate testing, but in the Jan Olbrecht inspired way, not the traditional way. I’ve covered that in my blog elsewhere, but there are some important lessons from that distinction. For instance, the actual threshold itself is influenced by factors besides “aerobic” ability. So the anaerobic side or lactate production side influences it too. So it’s entirely possible to get an improved LT curve while actually decreasing in aerobic capabilities (if the anaerobic capacity is decreased…). So once again interaction is key, but what about training at LT to improve LT, and thus performance.
Well if we look at the research, training at LT CAN improve LT. That’s good to know. But so does training at various other paces. Once again there is no magic training zone. In fact if we look at the one study on well trained runners, if they increased their training volume at near LT by 103%, it didn’t do anything to the actual threshold (Lehmann et al. 1991).
In other studies, with recreational runners, you saw big increases in LT after 2-3 months of training at threshold and then no further improvement in LT afterwards with continued training. What this tells me is if they wanted to keep improving the threshold, the stimulus had to change. They had to do some work above LT, some mixed intervals, aerobic intervals, or alternations. Whatever they chose, the stimulus had to change
In a review on training to improve threshold, Midgley et al. (2007) stated:
“In summary, we found only one study that investigated the effects of an increase in the volume of vLT or vLT training on the lactate threshold of distance runners. This study reported no significant increase in the lactate threshold. Several training studies have reported a significant increase in the lactate threshold of distance runners in response to the inclusion of supra-vLT training velocities,[15,29,39] althoughthese findings have not been consistent.[34,36]”
Once again, we’re looking at, ya training at LT improves it, but so does a lot of other stuff. And it depends on the type of athlete on how best it’s done.
But what about Running economy? Run fast and you get more economical? That’s the magic zone to improve RE, right? Well, sorta. If you look at research, again depending on the group studied, running fast, at VO2max, and at LT have all improved economy in various studies. Wait, so has weight lifting, plyometrics, whole body vibration, altitude, training in the heat and on and on (Berg 2003,Foster 2007). Obviously some improve it more than others, but the point is that like with all the other parameters, there are numerous ways to improve them. It’s not some magical zone that best targets it.
Ignoring whether or not RE is actually a good measure (that’s a topic for another time), it seems silly to assign one zone to improve this, when truthfully people are still arguing over what the heck improves it.
The point of this little tirade is to not say that training at these intensities is bad. It’s good. The point is that by sticking to this strict zoning concept we ignore both the individuality of the athlete and of the training response. In this system it assumes that an LT run does the same thing regardless of whether the athlete is a fast twitch runner or Slow twitch. While, we know that the effect can be vastly different because there physiology is different. Same goes with VO2max training or any of the other zones.
Nail in the coffin:
For my final nail in the coffin for why the model sucks, let’s look at a training study on sedentary people that had them all train at 70%VO2max. So they all trained at the same exact “zone”. What happened? Let’s check out the data (from Vollaard et al. 2009):
What happened was simple. A ton of different individual reactions to the same exact training load and training zone. All the way from big picture changes like performance , VO2max, and VE all the way down to enzymatic changes. There was a whole lot going on and a wide variety of responses. What this tells me is that stimulus matters. For some people the same training “zone” will give a different stimulus.
The same was found in a more recent study of untrained individuals. They followed them for a prolonged period and had them do all exercise at 60% HR reserve. What happened? Well a big variation in improvements in the big variables even though they were all training at the same intensity and volume (see full study here: http://onlinelibrary.wiley.com/doi/10.1111/j.1600-0838.2010.01139.x/full)
While you might say this was with untrained individuals so what’s the point, I’d counter that with untrained individuals you’d expect almost anything to work! I mean you’re going from nothing to something, so the fact that there’s a huge variation is interesting.
And finally, if we want to go all the way down to the genetic level, here’s what happens to people again all training at the same intensity. Look at the variation in high responders versus low responders for a whole slew of different genes: http://jap.physiology.org/content/110/1/46/T2.expansion.html
Why do we do this?
The answer is simple. In Daniel Kahnemann’s book Thinking Fast and Slow he describes essentially how human’s think. In there he talks about the concept of how when we get a difficult question, we often times replace it with an easier question and answer that. It’s subconscious so we don’t know we are doing that. But that’s essentially what we’ve done here.
We have a difficult question: How to improve the complex notion of performance?
And we’ve replaced it with “How do we improve these parameters?”
The second question is much easier to grasp our heads around then the first. It’s fine to simplify and reduce, but in this case we have to ask if the second question really answers the first. My contention is that it doesn’t.
How’d we get here? The story version:
Back in the day, coaching was almost purely trial and error. You learned from those who had come before you and tried your own manipulations. If something worked, it stuck for a while. The great coaches quickly were emulated until the next new idea or phase took over. The classic example is Arthur Lydiard experimenting on himself in terms of mileage before assigning training to his illustrious crew. Trial and error works. It just takes some time and occasionally a lot of error.
That got us most of the way towards modern training concepts, but something happened starting in the late 70’s, early 1980s and continued through the 1990’s. Exercise Science actually became a more defined field with actual research on performance. We began to be able to readily measure parameters like VO2max, lactate, etc. on well trained runners. Furthermore, we began to conceptualize what may effect performance. This was the rise of the VO2max paradigm. All those cool results that showed our best had really high VO2max started coming out and for those athletes who didn’t have crazy high VO2max values, the concept of RE was developed and explained the rest. Soon portable lactate analyzers came about and we could measure that evil fatigue component called lactate.
More testing, more variables.
Fast forward a little bit and we have the big 3 of VO2max, LT, and RE and then a few other variables that were occasionally thrown in to determine performance. This got translated to the intrepid coaches looking for an edge. It was the beginning of the age of Science. It was cool to make things sciency, and it offered a world of possibilities. Everyone was looking for the next edge, so science became that.
That’s when it got translated into the coaching world. With Coe and Martin’s book in the 1980’s, Daniels’ Oxygen power following, and many others it became the thing to do. The promise of explaining performance in a neat formula AND being able to make training almost a mathematical model was too tempting to pass up.
The model is thus part of our human notion to want to compartmentalize, our need for structure, and our seeing Science as the next ultimate step to sole all of our problems.
The models basic foundation and premise are broken though.
Done rambling, so what?
It’s not that the workouts are bad. It’s not that we shouldn’t run at threshold or 3k pace or faster. It’s that the model is bad. The model is broken. And if you strictly follow the model, you miss out on a bunch of sweet stimuli that need to be touched on. You also loose the ones thing that is essential to coaching, creativity. Instead, it becomes a plug and play system that discourages creativity and innovation.
Hopefully through this rambling mess you get a few things. If all fails, I hope it gets you to step back, think, and question.
But what the heck do we do about it, if I make this claim that the model is broken? After all I can’t just sit here and tear something down without suggesting an alternative.