Visual Legend
The visual elements represent the evidential target, perceived balance, assigned credence, skill depth, uncertainty, and irrationality.
Markers and Lines
The objective evidence marker indicates the evidential target. The perceived evidence marker indicates what the agent takes the evidence to show. The assigned credence line shows the final confidence commitment.
The left-to-right semicircle is the epistemic gradient: the 0% to 100% scale on which Objective Evidence (E0), Perceived Evidence (EP), and Assigned Credence (CA) are positioned. Assigned Credence (CA) is the epistemic agent's chosen location on that gradient.
When these elements cluster together, the state is aligned. When they separate, the model asks which kind of separation has occurred.
Skill and Uncertainty
The skill-depth marker reflects Deep Rationality. As skill rises, warranted uncertainty narrows. As skill falls, the agent is entitled to a wider range of uncertainty because their tools are less precise.
The concentric semicircles are Deep Rationality (SD) percentage markers. The center is low skill depth; each ring farther outward represents a higher percentage, with 100% at the outer rim.
The green uncertainty region is not a defect. It represents the range in which variation in credence may remain reasonable given the agent limitations.
The Red Region
The red region appears when Assigned Credence diverges from Perceived Evidence. It visualizes core irrationality: not merely being wrong, but adopting a confidence that departs from what the agent takes the evidence to warrant.
The stronger and more stable that red divergence becomes, the more the model moves from ordinary bias toward epistemic delusion.