Measurement Feasibility

The question

Can we distinguish "PFC online, capable of self-directed action" from "PFC offline, reflexive mode" from peripheral and behavioral signals?

Available measurement channels

• Peripheral physiology: HRV, EDA, pupillometry (if available), motion/activity
• Conversational/behavioral signals (AI-derived): Response latency, topic avoidance, linguistic markers of cognitive load, behavioral patterns
• Potentially BCI data: Direct neural signals (future timeline)

What's known

Augmented cognition (DARPA, NASA) demonstrates that coarse classification (overloaded vs. not) is feasible from physiological signals. The gap between "detect stress" and "estimate catecholamine-PFC balance well enough to predict intervention receivability" is real but may be closeable through multi-modal fusion including AI-derived signals.

The AI itself is a sensor. Conversational patterns, response latencies, topic avoidance, linguistic markers of cognitive load are all signals an intelligent machine can use to estimate state. This enriches the measurement model beyond what the augmented cognition literature considered (those systems did not have AI).

Threshold question

Not "can we estimate catecholamine-PFC balance perfectly?" but "can we resolve the state distinctions the architecture requires with sufficient precision for safe gating?" The architecture requires at minimum: green vs. red regime classification with known uncertainty bounds.

Connection to predictions

This is the empirical content of P2 (regime observability). E1 tests it directly.