Articles


Wittgenstein's Programming Lecture: Part II in Thinking About Thinking About Lifting

by Noah Milstein, SSC | December 25, 2019

You’ve finished your linear progression, now what? Texas Method, 5/3/1, Heavy Light Medium, Conjugate, Juggernaut, The Cube? There’s quite an array of options, each with their own battery of justifications. How do you decide? 

These program preferences are guided by an underlying rationale – either the mechanism of adaptation or mere personal preference. The former falls into at least one of four camps and the latter into two. 

1) Mechanistic rationales

  1. SRA cycle (stress, recovery, adaptation)
  2. Fitness-Fatigue model
  3. The intuition that it must be “harder” in some fashion
  4. Reference to a particular set of studies that recommends one pattern over another

2) Personal preference

  1. I like it
  2. It worked for X

There might be others but I am unaware of them. I’m concerned here with the first set. Let’s examine each in turn.

stress recovery adaptation cycle overview

SRA is a direct adaptation of Hans Selye’s General Adaptation Syndrome. An organism is exposed to a stress sufficient to disrupt homeostasis but insufficient to cause serious harm, the organism then marshals resources to adapt itself to this newly stressful environment, after which it has established a new homeostatic norm and is “adapted." 

The Fitness-Fatigue model suggests adaptation becomes evident or capable of expression after sufficient accumulated fatigue has dissipated. 

The “harder” intuition suggests that the stress of the current environment, the current training regimen, is insufficiently stressful to disrupt homeostasis, and thus insufficient to compel an adaptation. Thus the deduced need to expose oneself to a greater stress in order to elicit greater adaptation. 

How are we to understand the differences between these three (1a, 1b, 1c)? To figure that out, let’s make a digression into conceptual metaphor. This is a term coined by George Lakoff and Mark Johnson (Metaphors We Live By) that describes how ideas are understood through reference to other ideas. In this model all (or nearly all) ideas, all memes, are metaphors. Language is a metaphor of a metaphor of a metaphor. It’s metaphors all the way down. The regression stops where there is final reference to direct experience, either of the material or cognitive world. For this reason, this idea has also been called “grounded metaphor” and “universal metaphor.” This is why we can describe emotions in terms of temperature, value in terms of height or size, argument in terms of war, and why “Darmok” would never have posed any problem for The Federation because all language is metaphor. 

If you wish to communicate an idea, you must package it into a linguistic payload – a metaphor. The particular metaphor you choose is useful in so far as it may pump the intuitions of the message recipient in the desired direction, which is to say that it will accurately reconstruct the desired meaning in the mind of the recipient with minimal degradation and high fidelity. Think of this like the teleporter on Star Trek: Lt. Red-Shirt stands on the teleporter, he is deconstructed and sent as a beam of particles to some other location, where he will finally be reconstructed to appear in his original form. In this metaphor Lt. Red-Shirt is our meme, the beam of particles is the linguistic metaphor, the transmission of that beam to an endpoint is the communication process, the reconstruction at the endpoint is the consumption of that metaphor and it’s subsequent decoding and reconstruction in the mind of the receiver. If all goes well, then the receiver is able to reconstruct the lieutenant exactly as he stood on the teleporter pad. If not, then the lieutenant will suffer the unfortunate fate of so many Red-Shirts before him. 

An additional wrinkle in this process is that memes are more akin to molecules than atoms. That means that memes may be hard to accurately capture in a single metaphor, encapsulating instead only a segment of the meme-molecule. Some memes are harder still to understand when they are part of a large meme-complex and thus may greatly multiply the number or complexity of metaphors needed to properly translate the targeted idea from message transmitter to receiver. Worse still, the cognitive map, or schema, of each message recipient is unique, sometimes radically so, and is thus no guarantor of proper metaphor translation. The metaphor packaging of the message transmitter may be built off of a different mapping of the world than the mapping of the message recipient and thus, when translated, will produce a radically different message in the mind of the receiver. 

This can be seen in the difference between the first two of our adaptation models. SRA suggests discrete and discontinuous phases. First there is a stressor, you are now in the discrete stress/alarm phase, then there is a quantum leap into the recovery phase, finally you enter the adaptation phase. Notice that this is discontinuous – there is no overlap between phases. The quantum nature of the metaphor is purely a linguistic artifact and not necessarily an inherent attribute of the model itself. 

Contrariwise, the fitness-fatigue model implies a scalar continuity between phases. Stress is accumulated and continuously being recovered from/adapted to, but is simply not evident for display until a sufficient amount of the fatigue has dissipated. The constituent parts of the model are the same: one implicitly suggests phase continuity, the other discontinuity, all purely as a product of the linguistic packaging. 

What is the difference between “recovery” and “dissipation of fatigue”? Is this difference real? As in, do these two respective linguistic metaphors actually describe distinct real world phenomenon or are they merely slightly different metaphors pointing at a common material phenomenon? Like the proverb of the blind men and the elephant, each metaphor is an attempt to describe the same thing. 

This is what Wittgenstein called a “language game.” Wittgenstein’s ideas have since been expanded and brought to bear on the crucible of empiricism in the field of cognitive psychology, but the core idea is still present in his work, that being of the difference between the idea itself, the package/metaphor, and the unique mappings/schemata of the message recipient, causing parties to “talk past each other.” 

One of the more creative and enjoyable ways to navigate the impasse of a language game is to play “the taboo game.” That’s where you taboo certain words in the discussion, which forces the players to rethink their choice of metaphors. Try discussing the process of getting stronger without using the words “stress,” “adaptation,” “recovery,” or “fatigue” and see what that does to your intuition about what you’re actually discussing versus just the sounds you’re making. Usually your intuition about the topic has been rutted or rerouted by some metaphor that is preventing either accurate description of the desired phenomenon in question, or at least an accurate communication of the phenomenon. Making choice words taboo can force you to reevaluate your metaphors and thus the meme in question, which can help improve message fidelity and thus understanding. 

This seems to me to be the core difference between the first three rationales in our list. The first two are pointing at the same phenomenon but packaging them with slightly different metaphors that recipients are erroneously mapping as distinct real-world phenomena. The third rationale is merely the intuitive understanding of the same observed phenomenon without such explicit framing: 1c lacks an implicit intuition pump to suggest a planned pattern of stimulus in order to elicit adaptation, it reduces to simply “harder, bro.” At its core, both 1a and 1b presuppose 1c and as such all might be thought of as being the same. So, to where we first entered this long-winded digression, the difference is largely illusory. The lion’s share of disagreement emerges from the conflation of the metaphorical means of describing the thing with the thing itself, making it appear as though there is greater difference than in truth exists. 

The 4th rationale (1d) is probably an essay unto itself, being a set of distinct empirical claims. A rapid treatment of this would be to point out that ultimately whatever programming preference these distinct empirical claims produce, they would ultimately need to be tested in the lab of real world experience: the gym. The empiricism of the gym lab can be a death blow to some of the philosophical rationalism that emerges from heavy reliance on exercise science literature (which is typically subject to a raft of confounders). Empiricism requires inductive material evidence beyond the say-so of deductive abstraction and which Bayes theorem would require us to weigh more heavily while also reducing the weighing of the deductive claim in the updated prior probability (P), possibly sufficient to falsify the deductive claim entirely. 

That’s just a fancy way of saying something fairly simple. You have deduced that the sky is pink. You then test this hypothesis by observing the sky. The observed evidence suggests that the sky is blue. The probability of the truth of the claim “the sky is pink” is reduced and the probability of the truth of the claim “the sky is blue” is increased. That’s because observed evidence of the world carries greater probabilistic weight than does abstract deduction of the way the world is. That’s the shortest possible explanation of Bayes Theorem, which is basically just a way of quantifying the probability of a thing being true. The core idea is to take full account of all known, existing data (the prior probability) and rapidly updating those priors as new data becomes available. Simple enough. 

All evidence matters but not all evidence is created equal. All data must be weighed, but not all weighings are the same. Some data have a higher probability of truthfulness than others. In most cases, inductive evidence tends to merit a higher P than deductive evidence. It just so happens that much of the evidence – even though nominally inductive – in exercise science would produce a greatly reduced P because of the proliferation of methodological confounding that plagues the field. The specifics of that is worthy of a separate discussion. 

So, then, which program should you do after the linear progression runs out? I’m not sure, and it’s not clear that there’s only one right or best answer, but at the very least it should, in some measure, be harder.


Discuss in Forums




Starting Strength Weekly Report

Highlights from the StartingStrength Community. Browse archives.