Matt Perryman Matt Perryman

Fuzzy Science

[If you haven’t already, read Part I and Part II of this series before diving in here.]

It’s All So Fuzzy

Let’s take the question of muscle and what makes it grow. We can answer this question fairly well with some simple observations. Namely, you can go to any gym and you’ll notice that the people with well-developed muscles all tend to lift weights. That’s an anecdotal observation, but fortunately various research studies, both observational studies watching athletes and more direct interventions in the lab, have validated this “no kidding” conclusion.

We’ve got a lot of data that tells us yes, lifting weights makes your muscles grow. That’s an objective fact, in as much as we can ever define objective facts about exercise. If you want to get bigger, you pick up a barbell, or a dumbbell, or at least go to a cable station. You wouldn’t want to start running 10 miles a day. Extremes like this are easy to pick out — and they often tell us nothing interesting.

We turn to science in hopes of more detail. Case in point: it’s more or less true that all the people with big muscles lift weights, so we have that, but notice that little Jimmy also lifts weights and he doesn’t have big muscles. There’s obviously a relationship between “big muscles” and “lifts weights”, but the opposite — that all people who lift weights get big muscles — doesn’t hold true. We’re missing something.

Science is meant to find that relationship. Raw data, which is what scientists get when they notice things happening and write it down on a clipboard, and what you get when you go go the gym and notice that all the big jacked dudes “do this”, is like the static you hear when the radio isn’t set to a station. You might think you hear things in the hiss, but it’s really your mind imposing a pattern on the noise. Since our brains like stories, and we make up those stories from things which are near and immediate, we’ll hear what we expect to hear.

Sometimes we might be on to something, like when we noticed that all big muscular people lift weights. Our pattern-seeking nature means it’s hard to say for sure, though; we might just be drawing circles around the patterns that catch our eye. Even when we’re actually on to something, our mental blind-spots mean we’re usually leaving out something interesting.

Scientific research investigates the questions raised by observations, hopefully separating real cause-and-effect relationships from the patterns we impose on random noise. Without that, it’s all too easy to leave pieces of ourself in the thing we’re observing. This is a worthy goal, and it’s why I pay attention to research in the first place.

This is where it gets interesting, however, as biology is a few levels away from physics and even chemistry, where it’s easy to define general laws of behavior. The interesting things in biology happen in a way that doesn’t quite rely on simple “A causes B” relationships, and that puts some substantial limits on the scientific process in biological systems. I won’t say too much about this here as I’ve previously written about complexity, but it’s helpful to think of your body as more like the weather than a precision-engineered mechanical device. (I’ll also point you at these references for further reading on the philosophy of emergence and top-down causation in complex systems: [1] [2].

We can come up with rules that mostly get it right: lift weights and you’ll grow bigger muscles than when you started (the latter clause is key). When it comes to digging into the guts and trying to tease out more refined cause-and-effect relationships — say you want to improve a training routine by looking at cell-level physiology, or try to hammer out an “optimal” set/rep scheme — we’re chasing phantoms.

The relevant behavior happens at the “big” scale, when all the pieces act as a whole, and you don’t learn a whole lot by putting the details under a microscope. A water droplet from a hurricane is just ordinary water. It takes all the water drops acting together to produce interesting behavior.

The result is what Bruce McEwen calls “stability through change”. The only “optimal” in a biological system is ever-changing fuzziness. This is why you’ll find that seemingly “simple” ideas, like calorie balance, can be surprisingly complex (see here [3] for more on that); and more complex matters, like regulation of growth in a muscle fiber, often average out in aggregate, leaving you with a surprisingly simple real-world solution (which is why “go lift weights” is usually the right answer). We have to focus on the right level of detail.

This doesn’t jive with our intuitions. Since your average exerciser experiences a neurotic meltdown when faced with uncertainty, the placebo-effect-inducing rituals of the fitness industry step in as a substitute. Even if those supplements and magic programs and overly-detailed training and diet prescriptions taken from biochemical findings don’t “work”, strictly speaking, they provide guard rails that just might get the job done (unfortunately, too often they don’t).

My goal has always been an understanding of what we really need to do in order to see results, on the one hand, and things that people do anyway but don’t have any real effect. The question isn’t “does Mega Mass 10^15 work?” but “does it work better than things I’d normally do anyway?” The answer, for this question and most others that people will ask about workouts, supplements, and diets, is “no”.

In my thinking, close enough is as good as it gets. There are no equivalent to Newton’s laws for strength training or nutrition, and it’s unlikely there ever will be. It’s not easy for people to understand this, or to realize that most of the things they worry about are irrelevant to their success.

A History of Strength

You can see this fuzziness in play if you know a little about the history of strength training research. A lot of people seem to be under the impression that exercise science is on par with say pharmaceutical research or clinical research in medicine. Internet Bros thus love to dismiss science as being done by pencilnecked labcoats and holding no relevance to “real life”. Meanwhile the people who think “science” means “published in a peer-refereed journal” hold research up as the one and only source of perfect knowledge.

Let me cure you of those illusions.

Much of modern strength training knowledge is, in fact, based on “in the trenches” research and the practices of athletes training for their sport. Strength training has a long and varied history, dating back to at least the ancient Greeks and probably longer than that, although what we’d think of as science didn’t really begin until the early 20th century and only picked up a real pace after World War II. It’s been long known that picking up heavy things make you stronger and, when modern strength-science finally came about in the 1930s and 40s, it was verified in Western and Soviet labs — by researchers who watched and tinkered with athletes.

Much of the research done in the former-Soviet Union by names like Vorobyev, Medvedev, Ozolin, Verkhoshansky and Prilepin did exactly that. If you’ve ever used Prilepin’s Table to plan a workout, drawn on guidelines from a Russian manual, or done a Sheiko routine, then you’ve used “best fit” data drawn from competing Russian athletes (I’ll leave the bad news about “periodization” for another article).

Western science isn’t much better off. DeLorme, famous for the “three sets of ten” routine and the official creation of progressive resistance training, arrived at his program of “heavy” training — three sets pared down from an original 10 sets of 10 — by means that caused some question among his peers: he use trial and error in a group of weightlifters to discover the “optimum rate and duration at which strength increased most rapidly”. He edited his views according to what got the Bros stronger, not what “should have worked” according to his original (theoretical) ideas.

Even an old favorite like the 5×5 routine isn’t immune. As far back as the 1960s, Western researchers like Richard Berger had come to the conclusion that lifting 5RM weights for three to five or six sets of five to six reps, two to three times a week, yielded ideal strength gains. John Atha makes it clear that there was no hard consensus on strength-building weights and reps even in the 1960s and 70s. He writes that there “appears to be little to choose between a 1RM and a 10RM load in the strengthening effects produced. Both are held to produce inferior changes to the intermediate relative loads of 4RM, 6RM, and 8RM loads, while the 5RM to 6RM load is probably a reasonable estimate of the optimum.”

It was this understanding that led Bill Starr and Tommy Suggs to the magical “5×5” prescription, based on the idea that 4-6 sets of 4-6 reps, done three times a week, was right around the ideal for strength-building. Starr once wrote that the actual “5×5” numbers came from practicalities of coaching a group of athletes rather than any golden ideal derived from research. Even Starr’s method, which is about as solid as it gets, is based on “fuzzy” application of science.

This isn’t a criticism. The “natural experiment”, where the practices of real live athletes are quantified and analyzed, yields a wealth of data and has many advantages over the lab-style research most of us are used to. More than anything it reveals just how fluid and uncertain training is as a process, an issue that you are guaranteed to encounter in your own lifting.

But it’s also got the same problem mentioned above: without an intervention deliberately tested against a control group, it’s hard to nail down any definitive particulars. We know that this worked for this group — and that’s about all. It’s hard to gauge effectiveness relative to any alternatives, and even harder to prescribe universal principles (even though we can lay out lots of Good Ideas). Just as the guy dropping the book couldn’t make any definite statements about dropping things 10,000 miles away, it’s very hard for us to make absolute statements about specific training practices (a problem called inductive validity).

Not much has changed. Several years ago, Matt Rhea’s group at Arizona State and Mathias Wernbom’s group in Sweden independently published a set of meta-analyses of all this research which seem to provide some useful guidelines. A meta-analysis is limited to the body of literature available for analysis, and accordingly we run into some potentially severe methodological problems. For one, if the body of evidence you’re examining only ever tests limited groups, or limited types of interventions, it will only tell you what worked out of those methods for those people. What it won’t tell you is how these things compared to all the people or methods that weren’t tried. You know what you know, but you don’t know what you don’t know (to paraphrase Rumsfeld). All we can take away is what existing research has discovered, and perhaps pick out some trends worth future investigations.

These meta-analyses are brilliant insofar as they show you what the literature says, and they can even work fine as a starting point — much the same way as the Russian research — but we should be cautious in applying them as a universal “right answer” for how to train. There is no such thing. The inherent fuzziness of the process means that “it doesn’t matter” is, often enough, the only right answer.

It’s worth thinking about this for a moment. If a study were published tomorrow in the Journal of Strength & Conditioning that did exactly as the Russians did, and then came up with some reasonable prescriptions about training for X athletes in Y circumstances, the ensuing argument can be scripted to the last sentence. Here, let’s just get it out of the way:

“Oh those guys are elite Division I athletes which means they’ve got great genes and anyway everyone knows about the doping that goes on in those programs so you know the coach is giving them drugs which means this study is useless for any regular lifters.”

(The research might have been done in untrained subjects, in which case you sub in a different line of excuses, but the final conclusion remains the same.)

The same science that would be crucified by modern internet readers-of-research forms the basis of what we “know” and take for granted about effective strength training. Let that sink in for a minute, and then relish the consequences: armchair internet science has the unfortunate consequence of being anti-science even when it’s pretending to be pro-science.

I find that hilarious.

Observations of the training practices of athletes still tells us quite a bit, and those findings have helped us narrow things down to a range of training weights and set/rep combos which are good-enough for most instances, whether the goal is “get stronger” or “get bigger” (and whether you’re dealing with “advanced elite genetic wonders” or “average beginners”). But these are not “scientifically proven” absolutes in the same way as the laws of gravity or motion.

What we can’t see in the research about strength and hypertrophy training is, literally, what we can’t see in the research. This is a substantial blind-spot, and one rarely acknowledged by the Research First crowd, but for there to be evidence-based practice, there must be evidence.

This isn’t an insurmountable problem, of course, and even calling it a problem at all is a stretch; it’s only people who expect concrete rules that are troubled by fuzziness (or maybe it’s the other way around). What it should do is make you think twice before you brag that you get “all your information from Pubmed” or any such nonsense. Limiting yourself only to published papers is saying that you don’t care about other essential sources of information; like it or not, observation and anecdote do have to shape our ideas and, at the very least, guide our inquiry.

You should also have a hard think about your own convictions the next time you get the urge to argue with someone who does things differently than you. You might be combing the research and ignoring it’s weaknesses to justify your own biases. You might be treating biology as simplified clockwork with easy answers. You might not be as right as your brain leads you to believe.

Good practice means embracing the uncertainty, and realizing that trial and error are part of good scientific thinking.

[To be concluded.]