(This article will be revised, as I add more to it. This is a living document.)
I recently read gwern's excellent "Why Correlation Usually ≠ Causation" notes, and, like any good reading, felt a profound sense of existential terror that caused me to write up a few half-formed thoughts on it. How exciting!
The article hasn't left my head in almost a week since I first read it, so I think it's time for me to start up a new labor of love around the topic. (It's around finals here, and I can't have the gnawing dread of the implications of the essay distract me from my "real" work, so this is just as much to assuage myself and say, "Don't worry, we have a plan on how to attack this, you can focus on what's actually important for now. You're good.")
I don't really feel like I understand the concepts at play here well enough to create a specific, well-formed question on it yet. So we'll go broad: What is the relationship between ontology and causality?
My basic hope is to read a lot of simple summaries on how different people have thought about these topics over time, and hopefully get to the point where I can at least sketch out the arguments of why they did so. (A little like learning the right way to think in order to generate a math proof without a vision problem, now that I think about it.) I think that most people who have done serious work on these topics are smart people, and the focal lens of history gives me at least a place to start thinking about them.
So: My reading list.
This section is probably going to get much bigger over the next few weeks/months as I get my bearings a little more in this world.
- These guys are awesome. The Stanford Encyclopedia of Philosophy is one of those "between Wikipedia and actual textbooks" kind of sites, much like nCatLab is, where they actually give you a taste of the details of a thing before you go into it. I'm actually going to make them my first stop for Causality, as well.
- Turns out there's no SEP entry on causality/causation proper, just a lot of useful links. In the absence of a central page, a decent first heuristic is to just read what comes up as links under the first page of the search for both terms.
- "Causality": https://plato.stanford.edu/entries/aristotle-causality/ https://plato.stanford.edu/entries/kant-hume-causality/ https://plato.stanford.edu/entries/causation-medieval/ https://plato.stanford.edu/entries/wesley-salmon/ https://plato.stanford.edu/entries/arabic-islamic-causation/ https://plato.stanford.edu/entries/al-ghazali/ https://plato.stanford.edu/entries/meister-eckhart/ https://plato.stanford.edu/entries/qm-retrocausality/ https://plato.stanford.edu/entries/occasionalism/ https://plato.stanford.edu/entries/reichenbach/
- "Causation": https://plato.stanford.edu/entries/causation-probabilistic/ https://plato.stanford.edu/entries/mental-causation/ https://plato.stanford.edu/entries/causation-backwards/ https://plato.stanford.edu/entries/causation-law/ https://plato.stanford.edu/entries/causation-metaphysics/ https://plato.stanford.edu/entries/causation-counterfactual/ https://plato.stanford.edu/entries/leibniz-causation/ https://plato.stanford.edu/entries/causation-mani/ https://plato.stanford.edu/entries/causation-medieval/ Almost entirely different! I wish search engines used word2vec or something to create results that were closer aligned when I just use synonyms than this. Oh, well.
- Let's do "Causal", just to round things out: https://plato.stanford.edu/entries/causal-models/ https://plato.stanford.edu/entries/determinism-causal/ https://plato.stanford.edu/entries/decision-causal/ https://plato.stanford.edu/entries/content-causal/ https://plato.stanford.edu/entries/causation-metaphysics/ https://plato.stanford.edu/entries/causation-mani/ https://plato.stanford.edu/entries/causation-probabilistic/ https://plato.stanford.edu/entries/wesley-salmon/ https://plato.stanford.edu/entries/paradox-simpson/ https://plato.stanford.edu/entries/mental-causation/
- On Hazard's rec, and on the excellent summary by Bayesian Investory of it, I'll throw in Judea Pearl's "Book of Why" as one of my first stops.
It may not be directly related, but I'd like to highlight that I just added a new section contextualizing the essay as a whole and explaining how it connects to the rest of my beliefs about correlation & causality: https://www.gwern.net/Causality#overview-the-current-situation
As for the broader question of where do our ontologies come from: I'd take a pragmatic point of view and point out that they must have evolved like the rest of us because thinking is for actions.
I would skip everything by Pearl except Causality: models, reasoning and inference itself. That book is the essential bit.
Best ontology book I've ever read was Li and Vitanyi's classic textbook on algorithmic information theory. My impression is that this is an area where traditional philosophers do a particularly bad job.