All models are wrong, Some are useful

From nutrition to mental health, be skeptical of models.

Apr 26, 2023

I refer to Nassim Taleb’s work in my writing a lot because it provided me a pivotal change in my thinking. Before those ideas, I was an ardent proponent of Scientism. As in all that emerged from peer review science was valid, everything else wasn’t and that was it. I had memorized my logical fallacies & Carl Sagan’s Baloney detection and reveled in schadenfreude as Richard Dawkins snarked on the religiously inclined. This brought me at odds with my family a lot. An interesting battleground for these debates was traditional medicine, specifically Ayurveda. Interesting, because I had benefited from these in matters of my gut health consistently, even though clinical trials would continue to discredit the entire system. I would spare the reader details of the effectiveness that I witnessed with my gut, but I was convinced that it wasn’t placebo. However, I also couldn’t accept claims of elements in the body which different medicines supposedly balanced in order to cure ailments.

What “Skin in the Game” gave me was the perspective that these claims didn’t have to be true as long as they made some useful predictions. Also, their utility needn’t be studied in a clinical trial as their survival over centuries against the selection pressure of other ideas and of Time, was enough. This is not universal — Tobacco has been consumed for ages due to its utility as a stimulant but that doesn’t make it a “good” thing now — but the foundational ideas are sound.

This history of our collective understanding of things is that of evolving models.

All models are wrong. Some models are useful.

Think of the atom! Dalton came up with the idea of the indivisible units that make up stuff. It predicted that all atoms of the same element are identical and nothing more. Thompson’s model went one step ahead and said they’re positive spheres with embedded electrons. Now you could talk about charges on atoms. Rutherford then took it ahead and found how these charges were distributed — into a dense nucleus and electrons around it in empty space. But Rutherford’s atom would simply collapse unto itself until Neil Bohr rescued it with quantized orbits that could account for a lot of chemistry even if mostly for smaller atoms. Finally, Schrodinger's equation ushered in the Quantum Mechanics era which makes the most accurate predictions about all of matter.

neil bohr's model of the atom as pencil drawing

BUT, there are still competing interpretations of Quantum Mechanics that say different things about the nature of reality. And more importantly, this model of the atom doesn’t work with General Relativity which is the “ultimate” model in the physics of larger & faster things. Relativity itself displaced innocent Newton’s laws with fancy curves in space and time. Again, explaining vastly more things, yet staying incompatible with Quantum Physics.

This story tells us a few things about models being used in pretty much any field in science.

The model currently understood maybe incomplete or straight-up wrong about some things.
If modelling simple things like physics of particles is hard, complex systems like living things, economies, and societies are incredibly hard to model.
A lot of what a model can do is its predictions. Just because they’re incompatible, no one discards QM or Relativity. They are both celebrated for the vast testable predictions they make.

Viewed like this, I like to think of Ayurveda as Dalton’s atomic model. It is pretty effective for the time it was birthed in and makes plenty of useful predictions. Not all the predictions work, and not all are safe, but it IS a model of physiology with some utility and some risks

I have been musing about this more so because in my search for the friction I wrote about in my previous post, I got diagnosed with ADHD. It has been a dramatic and confusing week since then as I find relief in the predictions made by the condition that explain a host of frustrations that I have with myself; while scaring me into considering if I’m allowing myself to retrofit my observations into the mold of a condition to alleviate my pain through medically sanctioned kindness and clinical grade stimulants. Learnt voices of traditionalists in the family and on social media spoke through me to make me question if I was just an undisciplined screen-addicted kid refusing to get it together and finding excuses the West has come up with to justify weakness.

The good thing is, they’re both models! There’s the age old lack-of-discipline model and the ADHD model that try to explain the same set of behaviors. Looking inward, I found that while the lack-of-discipline model may explain some of my life choices, it’s predictions weren’t super useful to me. I spent the past 11 years trying to regiment myself to translate my intent into actions and failing miserably. Thus, the reason to discard that model was that its predictions were less than useless. Infact, in aspects of life where I was giving consistent output, my relentlessly optimistic motivation was driving results rather than any attempts to discipline myself.

a van gogh style depiction of an ADHD brain

The framework that decades of psychiatry and its application on millions of patients have developed, make for an effective standpoint for managing it both behaviorally and with drugs. Infact, it contains shades of the model that preceded it as many strategies effective in handling ADHD DO require discipline. Just that it is not a blunt instrument applied without context, rather wielded with precision to align with natural tendencies of the individual's psyche. While it is definitely the right choice to go with now, I caution myself because this could very well be the Rutherford model of the atom just paving the way for a more sophisticated one. And that’s not just a worry. This happened in the field of psychiatry itself as Serotonin Deficiency Model of Depression was challenged by a review study

causing a stir in the community. While I try to use the tools and framings of ADHD as it is currently understood, I do that with the knowledge that what is asserted as Gospel Truth today, might be upended tomorrow.

Going forward, I’m interested in seeing “what works” as long as it doesn’t pose major risks and discard what doesn’t!

While we may be tempted to look at models purely from their utility to make predictions, the terrifyingly prolific David Deutsch would disagree. In his book “The Fabric of Reality”, he refers to this view as Instrumentalism. He makes for a series of profound arguments as to why pursuing explanations is not only good but necessary for any meaningful scientific progress.

Imagine that an extraterrestrial scientist visited the Earth and gave us an ultra-high technology ‘Oracle’ which can predict the outcome of any possible experiment but provides no explanation. According to the instrumentalists, once we had that Oracle we should have no further use for scientific theories. But how would the Oracle be used in practice? In some sense it would contain the knowledge to build an interstellar spaceship. But how exactly would that help us to build one? In order to use it at all we must first know what experiments to ask it about. If we gave it the design of a spaceship and the details of the proposed test flight, it could tell us how the spaceship would perform. But it couldn’t design the spaceship for us in the first place. And if it predicted that the spaceship we had designed would explode on take-off, it could not tell us how to prevent such an explosion.
Prediction - even perfect universal prediction - is simply no substitute for explanation.
— The Fabric of Reality

It is a very tricky standard to hold our enquiry of reality to. In many situations, correlations are all we have to come up with theories. Explanations start out as theories and get tested by predictions they make. But they add a bit of logical cohesion to unify understanding. The fact that they will continue to be falsified and evolve doesn’t take away their power in the now as long as we are aware of the caveats.

By that standard too, ADHD holds its ground well so far! The idea that tonic levels of dopamine are low in patients causing them to scramble for stimulation to get motivated to do tasks is pretty cogent and explains a massive set of behaviors that cases exhibit. It is also what enables us to explain the effectiveness of drugs that ease symptoms of ADHD.

Still, both the predictive model and the explanations that come with it are being tested day in and day out in laboratories at universities and those of people’s lives. When deciding what path to take it is smart to expect the best of our current understanding to be proven inadequate at some point of the future.

For me, the biggest challenge in experimenting with alternative medicine is not that it may not work but it is that it may carry unknown risks. When a medical practitioner prescribes a psychiatric drug, the system (at least in principle) is invested in risk mitigation by regulating manufacture and sale. With alternative medicine, that doesn’t exist and supplements of all kinds from ayurvedic tonics to whey protein have been discovered containing heavy metals among other toxins. If the system were to study and mitigate risks, I’d be inclined to try a larger variety of medications just to see if they work.

Here’s a link to an article discussing this news.