Open log

Experiments

This is the formal record of the laboratory's open experiments, written for academic peers. Each entry separates a proposed method from the materials and results actually linked on the site. It does not establish recruitment, data collection or protocol approval. Status is read from a single source shared by all three surfaces: the narrative version, in Spanish, is the laboratory log, and the experiments you can run yourself are in the laboratory app.

Editorial review of linked materials: . This is not a study or approval date. See the preprint policy and publication status.

EXP-01

How could vocational guidance relate to school retention?

A · Adaptive systems and psychometrics

A proposed study of vocational guidance and school retention. In Prisma, RIASEC frames interest exploration and ikigai inspires purpose-related questions; they are not presented as equivalent tests. This entry does not establish instrument validity or ongoing cohort collection.

Proposed method: A cohort comparison of guidance and school retention is proposed. Instruments, comparison, follow-up and analysis must be specified and reviewed before data collection.

Available: The log describes the question and proposed method; Prisma is presented as an experimental design.

Next steps: Publish the approved protocol, validate the instruments for the context, obtain participation authorization and open a verifiable collection record.

Results and limits: This entry presents no empirical results and does not demonstrate reduced school dropout.

EXP-02

Learning curves under an adaptive engine

A · Adaptive systems and psychometrics

A proposed comparison of practice under different difficulty rules. The browser demonstration produces reaction times; it does not adapt difficulty or validate a learning intervention.

Proposed method: Practice under different difficulty rules is proposed for comparison. Performance, retention and effort measures require their own definition and validation.

Available: A twelve-trial reaction-time demonstration runs in the browser. It does not adapt difficulty or apply a learning-study protocol.

Next steps: Publish the approved protocol, define comparison criteria and gather data showing whether SECA has an educational effect.

Results and limits: One session produces demonstration data; it does not establish learning, fatigue or a benefit from adaptation.

EXP-03

Axiom-TL: formal verification before the model answers

C · Neuro-symbolic AI and formal verification

Axiom-TL explores checking bounded domain rules before delivering a model output. This is a research hypothesis, not a production-ready verifier or a measured reduction in errors.

Proposed method: Model outputs would be checked against bounded domain rules, with evaluation cases specified before testing.

Available: The log presents the hypothesis of a formal-validation layer; it does not offer a production-ready verifier here.

Next steps: Publish a reproducible evaluation protocol, a public case set and comparative error measurements.

Results and limits: No measured error reduction is presented. Grammar validation alone does not establish that an answer is true or appropriate.