Reading This Site

The site is organized around the two-layer structure of the framework. There is one page per validity type, one page per disciplinary mapping, one page per criterion, and one page per instrument. The navigation on the left mirrors that structure.

The reading order for newcomers

For a reader who has not used the framework before, the recommended order is:

1. Overview. What the framework is and why it has two layers.
2. The Five Validity Types. The upper-layer summary.
3. Construct → Internal → External → Measurement → Interpretive. Each in turn; the order is the order in which they gate one another.
4. Casebooks. The lower-layer summary.
5. Each casebook in turn. Neuroscience, Pharmacology, Philosophy of Science, Measurement Theory, Known Interpretive Failures.
6. Criteria by validity type. Where each individual criterion lives and what instruments contribute evidence to it.

This sequence takes about one hour to read through. It is the minimum the framework expects of a reader before applying the verdict language to their own results.

The reading order for readers with a specific question

A reader who already knows what they want to ask can enter the site at a more targeted point.

• Auditing a circuit claim. Start at the verdict checklist — it links to every criterion and every minimum-reporting rule.
• Choosing an instrument. Start at the Methods & Instruments index — each method page lists the criteria it contributes evidence to.
• Looking up a baseline. Start at the Published baselines table — every value is linked to its source.
• Understanding a partial-pass verdict. Each mapping page contains a partial-pass table that names the recommended verdict language for every pattern of evidence.

How pages link together

Every page on this site is annotated with the validity types it contributes to. A criterion page names its parent validity type and its parent disciplinary mapping. A method page names every criterion to which it contributes evidence, and through those criteria the validity types it touches. A case-study page reports its evidence under the framework’s verdict structure.

The cross-links are not decoration. A reader following them from any starting point should be able to reconstruct the full framework’s coverage of any given claim within a few clicks.

How to audit a claim

The framework’s primary use is auditing a circuit claim from a paper, a notebook, or a draft. The audit procedure is:

1. State the construct.
2. For each of the four validity types, ask which criteria are addressed by the evidence presented.
3. For each criterion that is addressed, ask whether the minimum-reporting rule is satisfied.
4. Report the verdict using the partial-pass language from the corresponding mapping page.

The audit is the framework’s central work product. A claim that scores 0.87 on faithfulness is not validated; a claim that satisfies internal validity’s necessity and sufficiency criteria with appropriate baselines and a held-out replication is validated for internal validity, and the audit makes the difference visible.

Conventions used throughout

• Criteria are written in lower-case (necessity, graded response). They are properties evidence is collected for.
• Instruments are written in lower-case (activation patching, resample ablation). They are procedures that produce evidence.
• Validity types are written in lower-case (internal validity). They are the abstract questions.
• Disciplinary mappings are written in title case (Neuroscience Mapping). They are the operational toolkits.
• Verdicts are written in italics and follow the partial-pass language from the relevant mapping page.

The vocabulary is fixed throughout the site, and the index pages list every term in current use.