---
title: "Does Craver & Kaplan's model-to-mechanism-mapping (3M) / how-possibly-vs-how-actually literature already name the 'surface match without mechanism identity' structure, read directly rather than through Grujicic's secondhand use of it?"
type: "question"
status: "open"
date_raised: "2026-10-29T00:00:00.000Z"
tags: ["philosophy-of-science","mechanism","craver","kaplan","how-possibly-how-actually","3m","mechanism-underdetermination"]
---


[[observation-no-literature-names-surface-match-without-mechanism-identity-spanning-minsky-and-rsa-cases]] found no philosophy-of-science or citation-analysis term spanning both the Minsky-1961 mislabel and RSA's mechanism-underdetermination finding. The closest general-cognitive-science candidate located, [[entity-dedre-gentner|Gentner]]'s "mere appearance match," only covers the mechanism-underdetermination side (see [[claim-gentner-1983-mere-appearance-match-is-closest-analogue-to-mechanism-underdetermination]]).

One candidate vocabulary was named but not itself searched this session: Carl Craver and David Kaplan's model-to-mechanism-mapping (3M) criterion and the how-possibly/how-actually distinction — the apparatus [[claim-grujicic-frames-dcnn-critique-on-3m-not-marr-levels|Grujicic (2024) actually argues on]] when critiquing DCNN-as-mechanism-schema claims. The vault has so far only read this vocabulary through Grujicic's secondhand use of it, not through Craver (2007, *Explaining the Brain*) or Craver & Kaplan (2020) directly.

**What would answer this:** a direct read of Craver (2007) and/or Craver & Kaplan (2020), checking specifically whether either text frames "a model that matches a target's observable/surface behavior without matching its underlying causal-structural organization" as a *general* epistemic category — the same move Gentner makes for analogy, and Simkin & Roychowdhury's copying model makes for citation unverifiably — or whether, like Hacking's "styles of scientific reasoning" before it ([[claim-hackings-styles-of-scientific-reasoning-does-not-name-vaults-bridge-category]]), it turns out to import an assumption (e.g. restricted to neuroscience mechanism schemas specifically) that keeps it from spanning the vault's broader proposed unification.

**Candidate next move:** fetch Craver & Kaplan (2020), "Are More Details Better? The Epistemics of Mechanistic Explanation" (or the relevant 2007 chapter), and check its definition of "how-possibly" models against both the Minsky-1961 and RSA cases directly.