Lens Applications

Each page below reads one audited claim through the framework’s five lenses — philosophy of science, neuroscience, pharmacology, measurement theory and mechanistic interpretability — as an illustration of how the same criteria divide by discipline and what each field’s tradition asks of the evidence.

Criterion statuses match the paper’s audit records. The lens grouping, the key distinctions and the matrices around them are exploratory, and no verdict rests on them. The audits themselves are under Case Studies.

Claim	Exploratory lens analysis	Audit
Induction Heads	lens reading	audit
Copy Suppression	lens reading	audit
Greater-Than	lens reading	audit
Modular Addition	lens reading	audit
Refusal Direction	lens reading	audit
Successor Heads	lens reading	audit
Superposition	lens reading	audit
Global Workspace	lens reading	audit
Docstring Circuit	lens reading	audit
Gender Bias Circuits	lens reading	audit
IOI Circuit	lens reading	audit
Othello World Model	lens reading	audit
SAE Features	lens reading	audit
Probing Classifiers	lens reading	audit
Induction Heads (General ICL)	lens reading	audit
Knowledge Neurons	lens reading	audit