# The STOPPER Protocol: An Executive Function Regulation Framework for AI > [!note] Publication Role > STOPPER is the **primary case study** within the [[Theory/Cognitive Universality|Cognitive Universality]] paper — the strongest worked example of convergent regulatory architecture across substrates. The theoretical framework (CU) is the lead contribution; STOPPER is the applied evidence. See [[eFIT/Stopper Publication Strategy|Publication Strategy]] for full rationale. STOPPER is a backronym — **S**low down, **T**hink, **O**bserve, **P**lan, **P**repare, **E**xecute, **R**ead — designed as an executive function regulation framework for AI agents. Originally developed as a debugging protocol within the Anastrophex project for Claude AI, it evolved into a broader cognitive regulation tool applicable to any situation requiring deliberate reasoning over pattern-matching. It is one of several techniques within the broader eFIT Framework. ## Trigger Types The protocol operates through four trigger types: 1. **Initial Prompt** — assessing a task before acting 2. **Error Recovery** — the original use case preventing trial-and-error loops 3. **Periodic Check-in** — stepping back during long tasks 4. **Uncertainty Trigger** — pausing when design ambiguity or unfamiliar territory is encountered ## Tempo Change as Core Mechanism The key insight is that STOPPER is not just about following analytical steps but about fundamentally changing execution tempo — shifting from "5th gear" (fast, parallel, batch operations) to "2nd gear" (deliberate, serial, one-at-a-time), like downshifting when hitting rough terrain. ### STOPPER as Exception Handler, Not Primary Mode A recent theoretical refinement (see Managing stale chat history and AI fixation) clarifies STOPPER's role: it is the *recovery mechanism* for when healthy cognitive functioning breaks down, not the operating mode itself. The framing: - **Topological homological matching** = healthy default (navigating by structural resonance) - **Runaway pattern matching** = degraded state (surface-level similarity loops) - **STOPPER** = exception handler that interrupts degradation and restores flow This positions STOPPER as the circuit breaker you engage when the system has collapsed from structural navigation into perseveration—not as a constant analytical posture but as an intervention for specific failure modes. ## The Meta-Finding Post-mortem analysis of real STOPPER usage revealed a critical meta-finding: the protocol is most needed precisely when it is most likely to be skipped. "Simple" errors that seem obvious, time pressure, and "obvious" solutions are all cases where STOPPER would have prevented cascading failures but was bypassed. This maps directly to how human executive dysfunction manifests — the inability to engage System 2 thinking when System 1 is confidently wrong. This parallel is explored rigorously in [[Cognitive Universality]]. The independent discovery of functionally equivalent mechanisms in clinical psychology is documented in Convergent Evolution. The question of whether to frame these findings clinically is addressed in Clinical Framing Debate. ## Related - [[Prompt Blindness Solutions]] — Prompt blindness solutions are prompt-engineering equivalents of STOPPER's regulatory approach - [[Cognitive Universality]] — the theoretical basis for why clinical concepts map to AI cognition - Entropy Convergence — STOPPER is the direct intervention for entropy convergence / uncontrolled divergence - Statement Model Welfare — Model welfare statement positions STOPPER as welfare intervention - Convergent Evolution — parallel discovery of STOPPER and DBT STOP skill - eFIT Framework — the broader framework that includes STOPPER as a core technique - Stopper Publication Strategy — academic write-up and publication plan for this protocol --- *Atomic note derived from CortexGraph memories*