Research Map · Operant Dyad™

How a working relationship accumulates, and what it raises

One sequence, understandable without reading the full paper. Wording follows the canonical paper's own description of how the relationship forms.

Canonical paper
  1. 01
    Repeated interaction & correction

    The human supplies goals, constraints, and defect reports across a task-bound working relationship with one AI system.

  2. 02
    Retained corrections (accumulated history)

    Corrections and context are retained rather than discarded after each exchange.

  3. 03
    Relationship State

    The paper defines this retained record, distinguishing declared state (what the person supplied) from inferred state (what the system derived).

  4. 04
    Later outputs, conditioned

    Retained corrections shape later outputs; the human inspects, accepts, rejects, or revises consequential results.

  5. 05
    The human-authority question

    Is authorship still preserved — or has the relationship drifted into Lock-In, Override, or Hidden Scaffolding?

Reading the map

Preserved human authorship is the governing condition for a healthy Operant Dyad, not a feature presumed to exist in current systems. This map does not assert that steps 1–4 necessarily produce a distinct, irreducible dyadic state; the wider research program is actively testing whether the effect at step 3 is better explained by human-side state, AI-side state, explicit memory, or shared artifacts. See the research brief for evidence status and open questions, or the canonical paper for the full argument.

Suggested citation
Preble, Micheal Charles. “Operant Dyad: Defining the Task-Bound Working Relationship in Human-AI Systems.” SSRN 7234238. Revised August 27, 2026. https://doi.org/10.2139/ssrn.7234238.