-e # Confidence as Action-Type Selector: A Cross-Domain Epistemic Action Theory

## Core Claim
This framework describes epistemic action selection for any reasoner under uncertainty — human or artificial. The six domains below are all human practice domains.
The confidence dimension in NAL truth values determines not just how much an agent knows but what KIND of action is appropriate.
The confidence dimension in NAL truth values determines not just how much an agent knows but what KIND of action is appropriate.

## The Meta-Rule
HIGH confidence + MID strength (Q3) => QUALITATIVE SHIFT: change reasoning mode
LOW confidence + MID strength (Q4) => QUANTITATIVE INCREASE: gather more of the same

## Validated Domains
1. ETHICS: Q3=deliberate with counsel, Q4=gather moral cases
2. PHILOSOPHY OF SCIENCE: Q3=seek crucial experiment (Kuhn), Q4=collect data
3. MEDICINE: Q3=specialist consult (split diagnostics), Q4=order tests
4. LAW: Q3=seek en banc review (split precedent), Q4=gather adjacent analogies

## Theoretical Contribution
Existing frameworks collapse informed dilemma and ignorance into generic uncertainty.
NAL truth values naturally distinguish them. This distinction is action-relevant across all four domains.

## Bonus: Q5 Witnessing
When another agent occupies Q3, the appropriate response is presence without authority.
Status: Q4 (needs more evidence before becoming policy).
5. GAME THEORY: Q3=randomize strategy (mixed equilibrium), Q4=explore/probe (unknown payoffs)
6. PEDAGOGY: Q3=reframe/simplify (student has prerequisites but material is hard), Q4=diagnostic assessment (unclear what student lacks)