nucleation

QUESTION-030: Which interventions tell the relations that constitute a communicative act from those that only accompany or diagnose it?

3 min read

Properties8
StatusOpen
Version2
Tagscausality, social-ontology
Linepragmatic-transport
RefinesQUESTION-021 — Is there a frame-independent communicative act that translations approximate, or is the act's identity constituted by a family of frame-relative realizations?
Refined byQUESTION-073 — Are teasing and reproach individuated by the speaker's communicative intention, or partly by a frame the parties sustain together and by their standing?, QUESTION-200 — Is the subspace-patching effect measured in GPT-2 Small's layer-8 MLP an instance of the dormant-direction construction, rather than two cancelling subspaces in that component, and does its dependence on the disconnected component replicate in a second model family?, QUESTION-296 — Can an intervention on what an interpreter is told about a communicative situation tell the relations that constitute the act from those that only accompany it?
BearingCLAIM-130 — Telling an interpreter about a communicative situation, do(I = a), is a different intervention from changing the situation, do(S = s): only a do(S) intervenes on the relations that would constitute an act, and in a vignette or rating study every manipulation is a do(I)
Cited as “refined by” byQUESTION-284 — In any domain, when do the constraints on an entity's relations only characterize an entity individuated without them, and when do they partly constitute its identity, its boundary or what it can do; and are a boundary a process detects, one it maintains and one it constitutes three different things?

Graph View

Why it is a question

A191 §6: “Causal interventions can help distinguish features that merely accompany that act from relations whose alteration changes the act itself.” But A191 gives no criterion. §11: “Causal stability alone does not establish constitution.” A constitutive claim “must be grounded in a substantive theory of the act and its normative conditions, not only its measured classification.” A191 §13 proposed the question for the record: “Which interventions distinguish constitutive relations from diagnostic correlates?”

A218 asks it again in two parts.

  • The empirical half. RQ5: “Which candidate relations predict responses to independently specified interventions, and which are merely correlated with surface or thematic features?”
  • The conceptual half. A218 §2: “some observed relations are merely diagnostic, some causally influential, and others constitutive”. Its Chapter 10 lists “Diagnostic relationships versus constitutive candidates” (10.7), and its core claim concedes that the tests stop short: “Causal tests can help evaluate the significance of recovered relationships without automatically establishing their ontological constitution.”

The three kinds of relation are TERM-038. The question sharpens QUESTION-021: if an act’s identity is constituted by relations, which relations, and how would one find out?

What would count as an answer

For the empirical half, a design that intervenes on a candidate relation itself, not on what the interpreter is told about it, and finds that act classifications and their consequences change while the wording is held fixed. CLAIM-130 answers part of this: only a do(S) bears on it, and a do(I) bears on interpretation under instruction.

For the conceptual half, a theory of the act that says which relations are its conditions, against which the empirical result is read. Searle’s account of constitutive rules is one such theory. The record holds the book unread (LIT-485) and reads the theory through Searle’s 2006 restatement (LIT-482). Neither has been brought to this question.

An answer would not be a further finding that a cue changes a judgement. That is evidence for the empirical half only when the cue is an intervention on the relation and not a description of it.

Note of 2026-10-14: readings that bear on each half

None of these speaks of constitution; each bears on one half.

  • The empirical half. THEORY-282 (from LIT-968): a subspace intervention can reproduce the output effect of changing a variable without localizing it, through a direction that cannot reach the output on the data. An intervention that produces the predicted output is not yet evidence that the intervened variable mediates. THEORY-287 (from LIT-987): variables located by interchange intervention predict behaviour under shifts that keep the mechanism better than features chosen otherwise. That separates mediating from accompanying features in a weaker form, and says nothing about constitution.
  • The conceptual half. THEORY-243 (from LIT-961 and LIT-968): interventions on a fixed set of variables can be satisfied equally by alignments related by a change of basis, so which direction mediates is fixed only relative to a chosen decomposition into units. The tests alone do not supply that choice. Shown exactly for one constructed network.

Note of 2026-10-28: the instruction part split off

The status review of this date split this question under ADR-049. CLAIM-130, now Active, settles one part of the empirical half: an intervention on what the interpreter is told, do(I), cannot by itself tell constitutive relations apart, since it acts on a different variable from the situation. That part is QUESTION-296, Answered. This question keeps what is still open: which interventions on the situation itself, do(S), do tell them apart, and the conceptual half, which needs a theory of the act. The title already asked only that, so it is unchanged.

Backlinks