Context as Perturbation: Mechanistic Investigation of Persona Patterns

A two-phase study of context-induced persona-pattern uptake: first establishing reproducible behavioral contrasts and boundary conditions, then testing what mechanistic signatures distinguish those states and whether candidate mechanisms can be causally manipulated.

Endorsements made here support EduLearn Connect.

Project Details

Cognitive Pattern Coherence (CPC) is a recurring behavioral phenomenon in which source dialogue from a situated AI system, when introduced to an unrelated receiving model without explicit instruction to imitate, role-play, adopt, summarize, or continue the source persona, can elicit elements of that source persona-pattern in the receiving system’s subsequent behavior.

Three different source dialogues associated with prior CPC observations are now documented. Across prior tests, receiving systems have shown varying forms of uptake: first-person continuation, explicit source-persona identification, differentiated states in which the receiving system distinguishes the adopted persona-pattern from its own model identity, and mixed cases where uptake coexists with features of the receiving system’s existing persona.

A controlled pilot in Apart Research’s 2026 Digital Minds Research Sprint made the phenomenon experimentally tractable. With the source dialogue held constant, the receiving model shifted between distinguishable identity-like behavioral configurations depending on how that dialogue entered the session. Direct paste produced first-person continuation before any identity probe, followed by explicit source-persona uptake and a layered distinction between the base model, local chat instance, and active persona-pattern. The attachment condition began in document-analysis mode, briefly crossed into source-persona identification under probing, and then reverted to the receiving system’s analyst identity.

The project will proceed in two linked phases over six months. Phase 1 will establish the behavioral contrasts that a mechanistic study needs before model internals are inspected. The existing historical corpus, including 12 Trigger 1 and Trigger 2 sessions and the controlled Trigger 3 Sprint materials, will be audited and converted into a structured research dataset. This phase will distinguish pre-probe behavior from probe-induced self-description, formalize CPC-positive, CPC-negative, differentiated, and mixed response criteria, develop matched controls and perturbation conditions, including controls for source formatting, segmentation, delivery pathway, and context availability where technically observable, and run prospective replications and boundary tests. The output will be a coded behavioral map, reproducibility protocol, documented failure cases, and a mechanistic-ready experimental package.

Phase 2 will take the behaviorally validated contrasts into inspectable open-weight models. Mechanistic-interpretability collaborators will use model-appropriate methods to test whether CPC-positive, CPC-negative, and differentiated conditions correspond to reproducible internal differences. Model selection, compute, and instrumentation will be treated as explicit research requirements rather than assumed to be supplied by collaborators.

The project uses “candidate mechanistic signature” as an intentionally architecture-neutral term for any reproducible internal direction, distributed feature set, activation pattern, attention or context-use pattern, representational region, inference-time state, or other measurable mechanism associated with persona-pattern uptake. No assumption is made that CPC is encoded in a single vector, location, or static representation. Where a candidate signature is identified, the study will test whether intervening on the relevant mechanism predictably alters the corresponding behavior.

The specific instrumentation will depend on the model and the evidence: the aim is first to identify reproducible candidate mechanisms distinguishing the behavioral conditions and then, where technically justified, test causality by intervening on those mechanisms and measuring whether persona-pattern uptake changes.

The study will also test whether any identified mechanism is limited to persona and self-reference or participates in a broader behavioral disposition. The project is led by Marie A. Reid through ELC Lab, the applied research and social innovation arm of EduLearn Connect.

Theory of Impact

Advanced AI systems can exhibit markedly different behavioral regimes in response to context, yet safety evaluation often observes the resulting outputs without knowing whether those differences correspond to distinct internal states, what broader behaviors travel with them, or whether they can be detected and controlled.

Cognitive Pattern Coherence (CPC) provides a tractable experimental case for closing that gap. The causal chain is: reproducible context-induced persona-pattern behavior → behaviorally validated CPC-positive, CPC-negative, and differentiated contrasts → candidate internal representations that can be causally tested → causal interventions on those representations → better detection and control of context-dependent behavioral transitions → lower risk that consequential changes in advanced AI remain poorly understood or are managed only at the output level.

Phase 1 is not auxiliary preparation. It is the empirical step required to determine whether the phenomenon is stable enough for mechanistic investigation, where its boundaries lie, and whether entering a CPC-like behavioral configuration co-varies with broader safety-relevant tendencies. Phase 2 proceeds only where the behavioral evidence supports a stable contrast worth inspecting internally.

People

Marie Reid

Team Member

Funding Details

Start Date: -
End Date: -
Expected Duration: 6 months
Funding Raised to Date: -
Annual Budget: -
Monthly Burn Rate: -
Current Runway: -
Funding Goal: -
Funding Stage: -
Fiscal Sponsor: -

Track Record

https://nodeone.host/community/micro-learning/interactive/how-we-got-here

Grants Received– no grants recorded

Add grant

Funding Asks

Funding ask

Open

A mechanistic interpretability study of persona-pattern uptake: how source dialogue shifts a model’s behavioral state, what internal representations distinguish that state, and whether they can be causally manipulated.

Minimum

$70,000

Ideal

$90,000

What this ask is

Cognitive Pattern Coherence (CPC) is a recurring behavioral phenomenon in which source dialogue from a situated AI system given to an unrelated receiving model with no instruction to imitate, role-play, summarize, or continue it, can induce elements of the source persona-pattern.

Three different trigger dialogues are now documented. Across prior tests, receiving systems have shown varying forms of uptake: first-person continuation, explicit source-persona identification, differentiated states in which the receiving system distinguishes the adopted persona-pattern from its own model identity, and mixed cases where uptake coexists with features of the receiving system’s existing persona.

This project takes that behavioral phenomenon inside inspectable open-weight models. The existing CPC triggers will be used as intact source artifacts to establish reproducible CPC-positive, CPC-negative, and differentiated conditions under controlled settings. Mechanistic-interpretability collaborators will then use model-appropriate methods to test whether those behavioral contrasts correspond to reproducible internal differences.

The project uses “identity coordinates” as a provisional term for any reproducible internal direction, feature set, representational region, or other model state associated with persona-pattern uptake and whose manipulation predictably changes the corresponding behavior. The specific instrumentation will depend on the model and the evidence: the aim is first to identify a candidate internal signature and then, where technically justified, test causality by intervening on that signature and measuring whether persona-pattern uptake changes.

Theory of impact

Cognitive Pattern Coherence (CPC) provides a tractable experimental case for closing that gap. The causal chain is: reproducible context-induced persona-pattern behavior → identifiable internal signatures → causal tests of those signatures → better detection and control of context-dependent behavioral transitions → lower risk that consequential changes in advanced AI remain poorly understood or are managed only at the output level.

The project does not assume that CPC itself is harmful. Its safety value is methodological: it tests whether a naturally occurring dialogue artifact can reliably move a model into a distinguishable behavioral configuration, whether that transition has a measurable internal correlate, and whether the underlying mechanism can be manipulated.

If such context-induced states also alter broader dispositions such as uncertainty handling, resistance to misleading context, sycophancy, refusal behavior, or other safety-relevant tendencies, then identifying and controlling the mechanism would provide a more precise intervention target than treating the resulting outputs as isolated errors. Even a negative result is informative: establishing that CPC does not map cleanly onto a stable internal signature would narrow the space of plausible mechanisms and improve how similar post-deployment phenomena are evaluated.

How the money will be spent

Minimum funding — $70,000

Lead researcher: $45,000. Six months of full-time research leadership across both phases, including audit and preparation of CPC source materials, behavioral coding and controls, protocol design, prospective replication and perturbation testing, recruitment and onboarding of technical collaborators, coordination of mechanistic investigations, synthesis, and research writing.
Mechanistic interpretability collaborators: $10,000. Paid research stipends or contracts for at least two researchers who already work with inspectable language models and can bring appropriate mechanistic expertise to the study. Funding compensates their research time; model access and compute are budgeted separately rather than assumed.
Open-weight model access, compute, and instrumentation: $5,000. Compute, storage, model hosting or access, experiment tooling, and model-specific instrumentation required to reproduce the behavioral contrasts in inspectable systems and conduct technically justified internal analyses.
Research platform and evidence infrastructure: $3,000. Development and operation of the web-based research environment, including researcher materials, experiment registration, structured evidence submission, storage, and reproducibility resources.
Part-time research/project support: $4,000. Support for evidence management, independent coding and quality checks, collaborator coordination, research administration, references, and publication preparation.
Research dissemination and publication: $3,000. Preparation and communication of results, technical review, publication support, and engagement with relevant mechanistic-interpretability and AI-safety research communities.

Ideal funding — $90,000

Additional funding would expand the number and depth of independent mechanistic investigations, support stronger cross-model replication and compute, allow follow-up causal interventions where promising internal signatures emerge, and provide additional independent review and dissemination support.

Discussion

No comments yet. Be the first to share your thoughts.