2. The Obvious Answer Creates Another Problem

Part I already found one way of organizing inquiry: convictions can be assessed under criteria that have become sufficiently stable to guide examination.

So to proceed well, maybe we should refine conviction by subjecting it to better criteria or standards.

We should demand consistency, evidence, reliability, corrigibility, empirical success, logical validity, Bayesian coherence, or causal adequacy.

These standards might help us distinguish disciplined conviction formation from mere persistence, prejudice, wishful thinking, or manipulation.

But Conviction Formation Theory cannot simply place them above the process it has been examining.

Consider the sentence: This is a better conviction.

If that sentence is to guide a finite agent's own assessment, it too must become sufficiently convincing.

We can work with a criterion without being convinced of it. We may follow it provisionally, because a procedure requires it, or because other people expect us to use it. But if we ourselves are to rely on it in judging one conviction better than another, its force cannot simply be supplied by calling it a criterion.

The same applies to "This is a better criterion", "This is better evidence", "This objection matters more", "This method is more reliable", or "This is the rational response."

Calling something a standard does not place it outside conviction formation.

Nor does a standard become convincing to an abstract agent. It has to convince someone with an already existing orientation, history, capacities, relationships, practices, and surroundings.

We can inspect some of this in ourselves. A standard may acquire force almost immediately or remain inert. Another may become convincing through training or repeated use, and later lose that force when we examine it or encounter cases in which its application becomes difficult to defend. What becomes compelling for me may remain unconvincing for you.

None of this is ours simply to choose.

A criterion can become sufficiently stable to guide inquiry.

Several people may also come to rely on the same criterion strongly enough to assess something together without reopening it at every step.

That does not make the criterion final. Part I already found that criteria can later become objects of inquiry themselves.

We can deliberately investigate a criterion, but we cannot command ourselves to remain convinced of it afterwards. We cannot simply command its convincing force to disappear either.

This changes the problem. What can finite agents do when convictions, criteria, and proposed methods of improvement all remain inside conviction formation?