nucleation

LIT-1834: Causal feature learning: an overview

1 min read

Properties7
StatusActive — read 2026-10-29 (NOTE-1415) in full, from the publisher's PDF as posted on Frederick Eberhardt's Caltech publications page (https://www.its.caltech.edu/~fehardt/papers/CEP2017.pdf); worth reading as the plainest statement of causal feature learning: a macro cause is a class of micro states with the same P(effect | man(state)), a macro intervention is any micro manipulation into the class, and the causal coarsening theorem says that, except on a measure-zero set, the causal partition is a coarsening of the observational one, so one experiment per observational class finds it. Read it knowing that its example is a one-dimensional toy with four hue bands; that the theorem is a genericity result, not a test, and holds only for discrete macrovariables and an acyclic micro system; that the effect classes are defined by equal likelihoods while the merging step normalizes them; that two of its displayed equations carry slips (NOTE-1415); and that it gives no measure of how much causal information a coarsening keeps.
Version2
Tagscausality, complex-systems, probabilistic-modeling, anthology-candidate
ExtendsLIT-1833 — Multi-Level Cause-Effect Systems
Cited as “grounds” byCLAIM-497 — A system's causes are found at the coarsest description whose interventions are unambiguous: a macro description is causally proportionate where it merges only micro states that intervention cannot tell apart, and no coarse-graining carries more causal information than the micro description it supervenes on
Cited as “paper” byNOTE-1415 — Chalupka, Eberhardt and Perona — Causal feature learning: an overview
Cited as “source” byTHEORY-1169 — For a discrete mechanism scored with one full-support intervention distribution over micro causes at both levels, a coarse-graining of cause and effect keeps all of the micro effective information if and only if it merges only causes with identical rows P(Y | do(x)) and only effects with proportional columns, whatever the distribution; so the coarsest coarse-graining that ties the micro is the causal partition of causal feature learning, with effects grouped by normalized likelihood, and any merge of causes with different effects loses information under every such distribution

Graph View

Chalupka, K., Eberhardt, F. & Perona, P. (2017). Causal feature learning: an overview. Behaviormetrika, 44(1), 137–164. doi:10.1007/s41237-016-0008-2

Read in full on 2026-10-29 (NOTE-1415). One account was drawn from it and from NOTE-1420: THEORY-1169.

Why this is here

Eberhardt and Lee (LIT-1839, NOTE-1413) recommend causal feature learning as the account of macro variables that avoids the failures of effective information they document. QUESTION-204’s first half is answered (THEORY-1167); its second half asks whether a coarse-graining chosen by a causal criterion agrees with an independent interventionist test of which level the effects track. CFL’s causal partition is such a test: it merges micro states only when their interventional effects are the same. This is the framework’s overview; the two-sided theorem it restates is from LIT-1833.

Backlinks