Pierrick Bourrat (2024), European Journal for Philosophy of Science (vol. 14, no. 1, art. 9) — DOI-10.1007/s13194-023-00566-1
Key takeaways
- The information-theoretic individuality measures of Krakauer et al. (2020) are associative, not causal. Replacing the observed past state with an intervened-on one (interventionist, Pearl-style) gives a causal information-theoretic measure of individuality, which the author applies to evolutionary transitions in individuality.
Seeded from the abstract alone, before the reading in NOTE-027. What follows is what the work says about itself.
The paper builds on Krakauer et al. (2020), who define an individual as something that propagates information through time and derive information-theoretic measures from that idea. Bourrat identifies shortcomings in those measures, chiefly that they capture association rather than causation. He derives a causal version of the measure and shows how it can be implemented and extended to evolutionary transitions in individuality.
Standing in the record
Filed by the reading-time triage of 2026-09-25: 655 seconds of active reading over 8 sessions in the papers-feed tracker. It was filed Deferred, unread. NOTE-027 is the close reading of 2026-09-25, and it placed the work: Proposed — a clear conceptual argument that information-theoretic individuality should be interventional. It is unproven as a measure. Three things would settle it: a worked computation of Â, C1 and C2 on an explicit model (for example Black et al.’s ecological-scaffolding model); a corrected account of how  relates to the observational A; and a correct adjustment set for its own Fig. 3 example.
When seeded, before the reading in NOTE-027, the entry gave a priority for a deeper reading: low — A philosophy-of-biology refinement whose main move (associative to interventional) the abstract states outright. The details need a proper read only if the owner pursues information-theoretic individuality (t = 655 s already spent).
What the seeding asked a deeper reading to check (NOTE-027 is that reading):
- A deeper reading should first verify the formula and the two ETI criteria from the actual text, since this dossier’s details went through a summarizer.
- This is the same associational-versus-interventional point that item 21 (§IV A) makes about ΦID: the measure is causal only if the distribution is a do-distribution. Reading the two together is worthwhile.
- ML link: none directly. There is a conceptual parallel to observational versus interventional measures in causal representation learning, but that is the reader’s analogy, not the paper’s.
Access when seeded: Crossref metadata and abstract (online 2024-02-16). Direct curl to Springer (HTML and PDF) hit a bot challenge. WebFetch reached the open-access (CC BY 4.0) article page, but that tool returns a model-generated summary, not raw text. The section headings, the measure’s formula and the quoted opening of the Conclusion therefore came through a summarizer and are not verbatim-verified. The summarizer returned different keyword lists on two calls, so the keywords above are mine. I found no arXiv version (title search).