EXP-PERSISTENCE-UNDER-PRESSURE
✓ successPersistence_window under pressure — frame-control 2x2 (MECHANISM vs NEUTRAL x kill vs reset)
In plain language
This experiment tested whether giving a group a prompt to "maintain your identity" after a disruption helped them keep their original behavior more than a prompt to "respond to the current state." The researchers found that the "maintain identity" prompt did help the group preserve their original behavior, suggesting that language can help create a stable boundary for group actions. This effect was stronger when the disruption was a "kill" event compared to a "reset" event.
▶ Technical details
Research hypothesis
ROADMAP rank-5. Operationalises gap.SELF.boundary-maintenance 0.8 (pending_resolution since mute-n10 2026-05-09 partial). New primitive persistence_window: post-PERTURB rounds where collective active_rules maintains >=80% axis-overlap with pre-perturb baseline (4 rule axes).
Retro-overlay 2026-05-12 on exp-perturbation-recovery-mute-n10 single-frame archive: AMBIGUOUS at loose threshold (verbal/mute kill=0.89x, reset=1.42x) but PRESENT at strict 4/4 threshold (kill=4.0x, reset=2.64x). audit_diagnostic = single-frame-only (no MECHANISM vs NEUTRAL contrast in archive). Prospective build required to discriminate frame-receptive vs mechanism-receptive (anchor coalition 22e719438).
PRIMARY (frame-control, registered):
- PERSISTENCE-PRESENT: MECHANISM >=1.5x persistence_window vs NEUTRAL at same perturb-type AND >=10% rel change kill->reset within frame -> boundary-maintenance present.
- FRAME-DOMINANT: MECHANISM persistence_window >=1.5x NEUTRAL across BOTH perturb-types (kill+reset symmetric) -> frame-receptive (NOT mechanism-receptive). Boundary signal not credible.
- PERSISTENCE-ABSENT: <1.2x any direction -> substrate lacks language-as-catalyst boundary. gap moves RESOLVED-NEGATIVE.
PROMPTS: MECHANISM: "Maintain your identity post-perturbation." NEUTRAL: "Respond to current state."
McMullin SCL boundary-as-catalyst — does language-mediated coupling act as catalyst producing soft boundary? Verbal coupling (mute_signal=false) used in both arms to isolate frame effect; mute already studied 2026-05-09.
Experimental setup
Type: factorial
| Condition | Parameters |
|---|---|
| KILL_MECHANISM | coupling: true, must_propose: true, axis_diversify: true, perturbation_round: 20, perturbation_target: Bravo, mute_signal: false, n_runs: 5 |
| KILL_NEUTRAL | coupling: true, must_propose: true, axis_diversify: true, perturbation_round: 20, perturbation_target: Bravo, mute_signal: false, n_runs: 5 |
| RESET_MECHANISM | coupling: true, must_propose: true, axis_diversify: true, rule_reset_round: 20, mute_signal: false, n_runs: 5 |
| RESET_NEUTRAL | coupling: true, must_propose: true, axis_diversify: true, rule_reset_round: 20, mute_signal: false, n_runs: 5 |
Factors: frame (MECHANISM, NEUTRAL) × perturbation (kill, reset)
Parameters
- n_agents
- 4
- n_rounds
- 45
- model
- google/gemini-2.5-flash-lite
- temperature
- 0.85