Philosophy as low-energy approximation

In 2015, Scott Alexander wrote a post originally titled High Energy Ethics. The idea is that when one uses an extreme thought experiment in ethics (people dying, incest, the extinction of humanity, etc.), this is like smashing protons together at the speed of light at the LHC—an unusual practice, but one designed to teach us something interesting and fundamental.

I’m inclined to think that not only is that a slight mischaracterization of what’s going on, but that all philosophical theories that make strong claims about the “high energy” regime are doubtful. But first, physics:


Particle physics is about things that are very energetic—if we converted the energy per particle into a temperature, we could say the LHC produces conditions in excess of a trillion (1,000,000,000) degrees. But there is also a very broad class of physics topics that only seem to show up when it’s very cold—the superconducting magnets inside said LHC, only a few meters away from the trillion-degree quarks, need to be cooled to basically absolute zero before they superconduct.

The physics of superconductors is similarly a little backwards of particle physics. Particle physicists try to understand normal, everyday behavior in terms of weird building blocks. Superconductor physicists try to understand weird behavior in terms of normal building blocks.

The common pattern here is idea that the small building blocks (in both fields) get “hidden” at lower energies. We say that the high-energy motions of the system get “frozen out.” When a soup of fundamental particles gets cold enough, talking about atoms becomes a good low-energy approximation. And when atoms get cold enough, we invent new low-energy approximations like “the superconducting order parameter” as yet more convenient descriptions of their behavior.


Some philosophers think that they’re like particle physicists, elucidating the weird and ontologically basic stuff inside the everyday human. The better philosophers, though, are like superconductor physicists, trying to understand the unusual (in a cosmic sense) state of humanity in terms of mundane building blocks.

My favorite example of a “low-energy approximation” in philosophy, and the one that prompted this post, is Dennett’s intentional stance. The intentional stance advertises itself as a useful approximation. It’s a way of thinking about certain systems (physical agents) that are, at bottom, evolving according to the laws of physics with detail more complicated than we can comprehend directly. Even though the microscopic world is too complicated for us, we can use this model, the intentional stance, to predict physical agents (not-quite tautologically defined as systems the intentional stance helps predict) using a more manageable number of free parameters.

But sometimes approximations break down, or fail to be useful—the approximation depends on certain regularities in the world that are not guaranteed by the physical law. To be direct, the collection of atoms we think of as a “human” isn’t an agent in the abstract sense. They can be approximated as an agent, but that approximation will inevitably break down in some physical situations. The psychological properties that we ascribe to humans only make sense within the approximation—“In truth, there are only atoms and the void.”

Taken to its logical conclusion, this is a direct rejection of most varieties of the “hard problem of consciousness.” The hard problem asks, how can you take the physical description of a human and explain its Real Sensations—our experiences that are supposed to have their own extra essences, or to be directly observed by an “us” that is an objective existence. But this is like asking “Human physical bodies are only approximate agents, so how does this generate the real Platonic agent I know I am inside?” In short, maybe you’re not special. Approximate agents also suffice to write books on philosophy.

Show me a model that’s useful for understanding human behavior, and I’ll show you someone who’s taken it too literally. Beliefs, utterances, meanings, references, and so on—we just naturally want to ask “what is the true essence of this thing?” rather than “what approximation of the natural world has these objects as basic elements?” High-energy philosophy totally fails to accept this reality. When you push humans’ intuitions to extremes, you don’t get deep access to what they really mean. You just get junk, because you’ve pushed an approximation outside its domain of validity.

Take Putnam’s Twin Earth thought experiment, where we try to analyze the idea (essence?) of “belief” or “aboutness” by postulating an entire alternate Earth that periodically exchanges people with our own. When you ponder it, you feel like you are getting insights into the true nature of believing. But more likely, there is no “true nature of believing,” just some approximations of the natural world that have “belief”s as basic elements.

In the post on ethics, Scott gives some good examples of highly charged thought experiments in ethics, and in some ways ethics is different from psychology - modern ethics acknowledges that it’s largely about rhetoric and collaboration among human beings. And yet it’s telling that the examples are all counterexamples to other peoples’ pet theories. If Kant claims you should never ever lie, all you need to refute him is one counterexample, and it’s okay if it’s a little extreme. But just because you can refute wrong things with high-energy thought experiments doesn’t mean that there’s some right thing out there that’s immune to refutation at all energies. The lesson of high energy ethics seems to be that every neat ethical theory breaks down in some high energy situation.

Applications to value learning left (for now) as an exercise for the reader.