Speculating on Physiology in the Lab vs on the Road

What started as Part 3 of my recent series of articles comparing the VO2 Master Pro to the TrueOne 2400 metabolic cart, half way through spontaneously turned into a more interesting comparison of the same workout performed outside on the road, and in the lab on the trainer. So let’s go with that!

For this experiment we wanted to look at physiological response to the same workout in the LAB vs in the FIELD. I’ve been reading up on the literature on various modalities of cycling (trainer, rollers, treadmill, flat road, uphill road, etc.) and how physiological responses tend to differ.

It’s fascinating! There are a lot of conflicting and equivocal findings, ultimately suggesting that there is a high degree of individual variability in how each of us produce power across modalities and conditions.

4x4min Severe Intensity Intervals

For this experiment I performed three trials of a 4x4min workout over a two week period, measuring VO2, blood lactate (BLa), and muscle oxygen saturation (SmO2). One trial was completed at home with the VO2 Master Pro, but I won’t be showing that here to keep the charts more legible It followed the same trends as the other two workouts, which in itself was a good finding of validity for the VMPro. The two trials we will be comparing were performed in the FIELD and in the LAB.

For the LAB trial I alternated between the VO2 Master Pro O2 analyzer and the TrueOne 2400 metabolic cart to give them a direct side-by-side comparison within the same workout. This trial was performed with the same set-up as detailed previously.

The FIELD trial was performed on a local short climb with gradient between 3-5%. This was supposed to be the first proper comparison of Lab vs Field with the VMPro and I was excited to compare the data. However the climate in Vancouver in October had other plans and freezing rain spoiled my fun, and unfortunately spoiled some of the VO2 data as you’ll see below.

I thought we agreed it’s not supposed to rain when I have to test outside! 🥶

Regardless, the comparison across these two trials got me thinking about core stability on the bike, cycling biomechanics, ventilation, and perceived effort. Let’s take a look at the charts and see what we can learn about the VMPro vs TrueOne devices, and about LAB vs FIELD cycling.

Power, Heart Rate, Cadence

I’m constantly trying to improve how I visualize these kinds of data. Any feedback would be appreciated! The FIELD trial below will be in a lighter shade, the LAB trial will be a darker shade to try and differentiate the data streams.

  • The target was to maintain above 350 W for the work intervals. This turned out to be unnecessarily high. I had to drop the power during the later two intervals of the LAB trial just to complete the workout.
  • I produced higher average power at a lower average cadence in the FIELD compared to the lab. Power & cadence were more variable in the field as would be expected with gradient changes. Heart rate was also slightly lower in the field, aligning with the lower cadence and RPE, and probably the greater environmental cooling in the field despite higher power output.

This is the first interesting finding comparing lab to field, trainer to road. The literature reports that athletes tend to adopt a lower preferred cadence in the field (Hansen et al, 2002, Bertucci et al, 2012, Nimmerichter et al, 2015). All else being equal a lower cadence will tend to be more metabolically efficient at the same power (Leirdal & Ettema, 2011). This is one of the possible contributing factors to the common perception that higher power is easier to achieve on the road than on the trainer. Although this rabbit hole gets waaaay more complicated, as we’ll begin to find out.

Most of the studies I’ve encountered that look at cadence find significant differences in Gross Efficiency (GE) between a wider spread of cadences, eg. 60 vs 120 rpm (Tomabechi et al, 2018). I’m not sure whether a 10 rpm difference in my case would have significantly changed GE, but that’s only one factor at play. Certainly subjectively I found it easier to maintain a higher power at a lower cadence in the field than subsequently on the trainer.

VO2, Blood Lactate, RPE

  • Here is where the FIELD data from VO2 Master unfortunately has to be thrown out. You can see it’s literally off the chart.. far higher than what was physiologically possible.
  • The error codes reported by the VMPro showed that the mask was hitting the lower limit of it’s operating temperatures, as well as the risk of moisture infiltration from the heavy rain. This ended up providing some good test data to compare how the sensors drifted under “extreme” conditions, but not good data to compare between LAB and FIELD trials! 😢

Blood Lactate (BLa) tells an interesting story. I measured significantly higher BLa values in the LAB (measured at the fingertip) until I had to drop the power just to survive the final intervals. VO2 correspondingly decreased during these later two intervals. RPE for the LAB session also probably reflects the feeling of more lactate and metabolites coursing through my tissues. Subjectively things felt a lot harder on the trainer, even worse than riding outside in freezing rain!

It’s too bad we can’t compare VO2 against BLa, but I suspect we would have seen higher VO2 on the road, compared to the trainer. The simple comparison is that a higher BLa suggests greater glycolytic/anaerobic energy production, therefore less oxidative/aerobic energy production to achieve roughly the same power output, and likely higher perceived exertion as a result.

My current understanding is that cycling on a stationary trainer can feel harder in part because there is a greater concentration of muscle force production from the quads. Whereas on the road there is a wider contribution of force output from the quads and glutes especially (we’ll come back to this), as well as a ton of other accessory & stabilizer muscles. So maybe the effort feels more distributed?

But this is seriously tip of the iceberg..
References for above hypothesis (expect another article 😅)

These changes in muscle recruitment and stabilization gets into one of my pet-theories on cycling biomechanics, core & pelvis stability, diaphragm, and glutes engagement to produce power, but hold that thought!

Respiratory Frequency, Tidal Volume,
Minute Ventilation

  • Respiratory Frequency (Rf) and Tidal Volume (Vt) are the rate of breaths per minute and the depth of each breath, respectively
  • Minute Ventilation (VE) is the product of Rf * Vt in L/min of gas exchange. This is the volume of air passing in and out of my lungs each minute
  • The VO2 Master was able to accurately measure ventilation under ‘inclement conditions’ (ie. Vancouver in October) even when the O2 sensor failed. So these numbers still provide a good comparison.

Considering the slightly higher power in the FIELD and very different conditions compared to the LAB, ventilation remained similar between trials. A trend toward slightly lower breathing rate (Rf) and slightly higher breathing depth (Vt) on the trainer produced mostly similar ventilation (VE).

I wrote about an experiment a while ago comparing road to trainer for a longer steady-state interval. For those 30min steady-state intervals I also saw lower Rf and higher Vt on the trainer, as well as an upward drift in Rf and VE during the constant power interval. VO2 during both trials was virtually the same.

Higher VE but no change VO2 would be expected where thermoregulation is challenged: My body was sending a signal to hyperventilate to try and offload heat by increasing gas exchange, while the additional inhaled O2 was simply exhaled as ‘waste’ (higher FEO2) producing no change in actual O2 uptake.

The trend toward higher Rf in the field is interesting. I try to focus on big breaths. “Big lungs” is one of my mental cues during hard efforts. But outside I feel like I’m less able to inflate those “big lungs”. I think part of this is having more distractions to focus on, like keeping the bike upright and not running into things.. but to start speculating again, I wonder if core and diaphragm stabilization plays a role?

Let’s backtrack for a moment to core stability on the bike. Once again, the literature seems to point to greater activation of quads when the bike is fixed to a stationary trainer. Furthermore it is suggested that the core can “turn off” since active stabilization to keep the bike upright is unnecessary on the trainer (Arkesteijn et al, 2013).

On the road accessory and antagonist muscles (glutes, hamstrings, calves, etc.) show greater activation in order to stabilize and minimize micro-changes in velocity during each pedal stroke. I would speculate that more leverage to push the pedals has to come from internal stabilization: you push against yourself to generate force.

Hence recruiting more muscle groups at the trade-off of consuming more energy to produce the same power output, acts to distribute the perceived effort across those additional muscle groups.

To keep speculating, if the core isn’t sufficiently stabilizing the pelvis on a fixed trainer, I wonder if the glutes, as the major prime movers of the hips, won’t have a stable base of support to push off to generate force? Is that contributing to the reported lower muscle activity (Arkesteijn et al, 2013, Aasvold et al, 2019) and observed lower metabolic activity (see SmO2 below) of the glutes and higher recruitment of quads on the trainer?

And to the opposite effect, I wonder if with the core less active on the trainer, the diaphragm actually has less abdominal resistance to push against and can inflate the lungs to a greater depth? Resulting in the trend I’ve observed toward higher Vt (breathing depth) on the trainer? 🤔


Muscle Oxygen Saturation

  • Muscle Oxygen Saturation (SmO2) measured at the quads (vastus lateralis, VL), glute max, and deltoid (shoulder).
  • The VL are almost always the most metabolically active and generate the most force on the pedal downstroke (Bhambhani, 2012, Aasvold et al, 2019). Hence they show the greatest de-saturation and look very similar between trials.
  • The glutes tend to kick in more at higher torque and higher power outputs (Aasvold et al, 2019). Here we can see some very interesting differences between the LAB & FIELD trials.
  • The deltoid is a ‘non-priority’ muscle that isn’t directly involved in power output, but is recruited to stabilize the upper body. It also shows greater de-saturation at higher intensities as greater stabilization becomes necessary. Again there are some revealing differences here.

I mentioned the deltoid deoxygenation-breakpoint (deoxy-bp) in the recent posts comparing VMPro and TrueOne. Deltoid appears to de-saturate once the athlete begins to work above their Anaerobic Threshold (above steady-state).

This is primarily due to increased metabolic demand for stabilization of the upper body during higher intensity efforts (Ozyener, 2012). But I’m curious whether we might also be seeing some redistribution of blood flow & O2 delivery away from the non-priority/non-locomotive muscles such as deltoid to preserve oxygenation at the brain, respiratory muscles, and priority working muscles? (Bhambhani, 2012)

In this case deltoid in green shows similar de-saturation behaviour during the first two intervals where power was more similar between the two trials. Interestingly when I had to drop power during the later two LAB intervals, deltoid showed less de-saturation (higher SmO2).

It’s hard to draw any conclusions from magnitude of SmO2 because it’s so sensitive to movement and sensor placement, but clearly the first two intervals look more similar than the last two intervals. This would be expected if deltoid de-saturation was dose-dependent: higher workload = greater de-saturation.

The glute signal in orange is really where things get interesting, and I can get back to some unbridled speculation! Glute de-oxygenation appears deeper/lower saturation in the field, as would be expected with greater recruitment and power contribution from the glutes (Arkesteijn et al, 2013, Aasvold et al, 2019). Perhaps, say, because I had better core stabilization in the field which allowed my glutes to push harder on a more stable pelvis!

However not just the SmO2 magnitude, but the pattern of de-saturation is different between trials. In the LAB the glute shows a steady but slower downward slope during each work interval. In the FIELD trial the glute appears more similar to the VL: it de-saturates quickly and stabilizes at a minimum plateau. This suggests greater oxygen extraction (mVO2; local O2 uptake at the muscle) during the work intervals in the field. I would expect that EMG would have been higher at the glutes in the field, reflecting a greater hip-joint specific contribution to power output.


First let’s quickly consider the comparison between the VO2 Master Pro and TrueOne 2400. Despite the spoiled VO2 data during the field trial with VMPro, VO2 and ventilation otherwise showed good agreement between trials, considering the expected physiological differences. I’m really excited about the promise of being able to test VO2 reliably in the field… when it’s a bit warmer outside.

We’re seeing trends in systemic VO2, blood lactate, and local muscle oxygen that align with the literature on muscle recruitment and metabolic efficiencies on the trainer in the LAB vs on the road in the FIELD. This should give some better insight into postural and technical cues to optimize these signals and improve efficiency & performance in both conditions.

As far as indoor modalities go, the turbo trainer seems to facilitate a ‘brute force’ muscle recruitment of quads to generate most of the power. Where maximizing workload is the desired adaptive stimulus, this might be sufficient (Olsen, 2012, Tseh et al, 2017). Such as during high intensity interval workouts.

However I suspect we’re undervaluing the importance of accessory muscle activation, core stability, glute contribution to power, and diaphragm compliance to optimizing performance on the road. I’m starting to think low intensity aerobic training might be more effective on rollers if stuck inside, or obviously ideally on the road. Where these accessory biomechanics will be challenged (Tseh et al, 2017).

I think these factors are very likely involved in the common perception that power output is lower and relative effort is higher on the trainer. I have started to spend my low intensity Aerobic rides focusing on a few cues to optimize these biomechanical patterns:

  1. Maintain a stable pelvis, quiet upper body, and soft grip on the handlebars
  2. Piston my legs with active glute extension from this stable base of support
  3. Focus on a strong exhale and allow my diaphragm to elastically recoil into a deep belly breath on each inhale.

Personally, I’m starting to get a better handle on the disparate data streams of power, VO2, ventilation, BLa, and SmO2 and building a more integrated picture in my head of metabolism, biomechanics, energetics, and force production. I’m trying to be cautious with over-interpreting signals and especially attributing mechanisms, but the experiments I’ve performed so far have been invaluable for generating research questions & hypotheses!

As I get more involved in proper rigorous science, I hope I’m still allowed to speculate and brainstorm outlandish possibilities of what’s going on behind the scenes of the signals we see. I think it’s important to have a Performance-first approach to translating the rigor of white-coated science in the lab, to more ‘ecologically valid’ practice in the rain, mud, and dirt of the field.

6 thoughts on “Speculating on Physiology in the Lab vs on the Road

  1. Hi Jem,

    First of all, thanks for the great content – I really like your way of explaining things and connecting physiological insights with speculations about real-world applications! Sorry for digging up this older article but I just re-read parts of it and have a few questions:

    Under the first diagram (showing power, HR, cadence) you write
    “Heart rate was also slightly higher in the field.”
    but isn’t it the other way round?

    2.) On the road power was higher, RPE lower, VO2 (most likely) higher. Does this suggest that doing VO2max intervals on the road is the better choice in terms of “bang for the buck” (VO2max improvements for suffering)?

    If 1.) and 2.) are correct, I guess this might this be an example where paying too much attention to HR could actually be misleading / at least in terms of absolute values (the shape of the curve might be of interest for dosing a potential hard-start approach).

    One of the reasons I’m asking these things is that my power is much higher on the road, whereas HR is higher indoors, so I can easily rack up a lot of “time over 90% HRmax” indoors, but I’m not sure if this high HR is a reflection of high VO2 or if provides some other useful stimulus or if it’s basically wasted mental energy and I should rather go outside (whenever possible).

    Thanks again and greetings from Switzerland,


    1. Thanks for the questions and kind review Anna!

      1) Huh, you’re right. HR was lower in the field. I’ll edit that, thanks for catching the error! That aligns with the trend toward lower cadence and lower RPE in the field, also probably related to greater environmental cooling.

      2) All else equal, performing intervals in the field might be “more productive” than intervals on the trainer. Not just from a purely metabolic perspective (lower HR & RPE, higher power and possibly VO2), and biomechanical/proprioceptive (balance, stability, muscle recruitment) perspectives as I suggested in the article, but of course also because of race-specificity and perceptual stimulation/peripheral awareness/multitasking. However that’s a hard maybe: the trainer allows you to perform workouts which you might not be able to perform on the terrain available locally. And using Erg mode might allow you to push beyond what you would be capable of on the road, because it doesn’t allow you to voluntarily decrease power output, and it focuses the effort more specifically at the working muscles (quads, primarily).

      That’s probably the biggest pro/con trade-off: To Erg mode, or not to Erg mode. It might be cognitively easier to push more mechanical work by removing the ‘decision option’ to ride easier. You can literally close your eyes, head-down, and just devote all your cognitive & physical energy to pushing the power. Maybe that extra VO2 on the road is going toward non-locomotor energy demands (breathing, stability, peripheral awareness) that doesn’t contribute to power output and won’t elicit the same stimuli? But maybe those cognitive & ‘accessory’ physical stimuli you get on the road might be more important than the pure work output, when it comes to performing in your target event on the road? Definitely a trade-off. Then there is the thermoregulation challenge to consider on the trainer, which has a huge effect on the HR-power-RPE relationship vs the road (I’m doing some testing and might write more about that soon… hopefully 🙂

      3) Yes definitely. I’ve talked about ‘over-precision’ with training targets before, in terms of HR down to the bpm or power down to the watt. Watts are watts at the pedals, but how you produce those watts internally changes moment to moment. And HR is a gross, indirect, and delayed measure of internal work which is affected by too many factors to make it a reliable proxy for any single one. I really think the most precise and reliable measurement of effort is RPE. Your brain is good at integrating signals. Let it do it’s job and pay attention to the sensations, more than the data screens in front of you.

      4) Ah yes. My first thought is that thermoregulation point I mentioned above. Even with a big fan it’s difficult to get the same cooling on the trainer as outside. That will probably have the biggest effect on your power & RPE between road & trainer. I wouldn’t say it’s wasted effort though: As I said, if it feels hard, it is probably ‘hard enough’ to get a benefit. Whether that benefit is optimal might be another question. But doing the work consistently over weeks and months is absolutely more important than optimizing any one single workout. If including trainer workouts allow you to do that more than exclusively road workouts, then that’s a win!



    2. Sitting up on the bike produces a slightly higher HR than the aero position on the bike, for the same power output. Try this for yourself, in ergometer mode on your trainer, at a fixed power demand.
      Most of us tend to ride more aero in the field, but on the trainer we might not.


  2. Hi Jem,

    Thanks for the answer 🙂 .

    ad 2) “Maybe that extra VO2 on the road is going toward non-locomotor energy demands (breathing, stability, peripheral awareness) that doesn’t contribute to power output and won’t elicit the same stimuli?”
    Well, since my power is higher outside, I guess there is more VO2 going to both locomotor and non-locomotor recipients. Seems like a double-win in terms of central adaptations? The only thing that seems to work just as hard indoors (at least until you / I start failing our intervals XD) are the quads.

    ad 4) Overheating is the first thing that comes to my mind too, but I can’t really convince myself it’s the main reason, since I can actually feel cold on the turbo (low ambient temperature + fans) and still produce far less power than outside on the hottest summer days. It’s a bit of a mystery. I seem to be a rather extreme example of power loss indoors vs. outdoors… Extremely high RPE, HR and perceived ventilation at a power output that is barely above my outdoors FTP.
    Wondering where all the air and blood is going XD.

    I found your point about BLa being higher on the turbo interesting. Perhaps I’m unusually “bad” at dealing with this lactate accumulation?


    PS: Typo?: “Glute de-saturation appears lower in the field” – higher?


    1. Hey Anna,

      ad 2) Totally. In my recent article on recommendations for hard-start interval programming I mention the idea that ‘excess’ VO2 not going directly to locomotion is still placing greater demand on central cardiac output, therefore probably adding to the central stimulus component.

      I also agree, from working with NIRS it seems like quads – vastus lateralis specifically – are basically always working directly proportional to the power output. Their deoxygenation doesn’t appear to change much across different conditions.

      ad 4) I would be curious if you performed the same training on resistance rollers or treadmill, how you would find that compared to static trainer and road. I wonder if it goes back to accessory muscle recruitment & stabilization allowing for greater power contribution from muscles other than quads. Would be an interesting experiment if you have access to those equipment

      Lactate accumulation in the blood typically implies a greater disturbance to the metabolic milieu within the working muscles, eg. Pi, H+, other metabolites. We are all “bad” at dealing with those! Hence the goal is to produce more power and produce fewer of those metabolites (of which Lactate is a useful byproduct that goes on to be oxidized as another fuel source).

      Not a typo, but a confusing double-negative I’ll fix, thanks! Should read as: “glute deoxygenation was deeper/reached a lower saturation in the field”


  3. Thanks once more for the interesting comments!

    Yeah the rollers vs. static trainer comparison would be interesting, but tbh in my whole life I’ve only done a handful rides on the rollers so I’m not exactly relaxed, which will skew the comparison. (Another reason to ride more on the rollers, I know… ;)).

    [I have never noticed an obvious performance decrement on the treadmill, but haven’t done much and I don’t know how accurate the calibration was.]


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.