How to Read the Model

Start With the Target

Begin by locating Objective Evidence (E0). This is the model estimate of where the evidence really points. In real life, E0 is often uncertain, but in the lab it gives the visual system a target.

Then compare Objective Evidence (E0) to Perceived Evidence (EP). If they are close, the agent is reading the evidential situation well. If they are far apart, something has gone wrong in access, attention, interpretation, method, or environment.

Then Read the Commitment

After Perceived Evidence (EP), look at Assigned Credence (CA). This is where the agent finally places belief. When Assigned Credence (CA) tracks Perceived Evidence (EP), the agent is internally consistent even if they are mistaken about the world. When Assigned Credence (CA) diverges from Perceived Evidence (EP), the agent is believing against their own evidential perception.

The outer semicircle from 0% on the left to 100% on the right is the epistemic gradient. It is the confidence scale on which the epistemic agent applies a credence. A point near 0% means the agent treats the claim as very unlikely; a point near 100% means the agent treats it as very likely.

Use Deep Rationality as Resolution

Deep Rationality (SD) functions like resolution. High SD means the agent has better tools for narrowing the distance between the world and their perception. Low SD means more fog and wider warranted uncertainty.

On the chart, Deep Rationality (SD) increases as the yellow skill marker moves farther away from the center point and toward the outer semicircle. This is intentionally radial: greater depth is farther out, even though people may intuitively expect “deeper” to mean lower.

Do not read low Deep Rationality (SD) as failure of integrity by itself. A novice may lack tools and still be honest. The model becomes diagnostically interesting when you compare Deep Rationality (SD) with the Perceived Evidence to Assigned Credence gap (EP-CA).