Question
A bounded problem whose answer is not yet known.
Scientific discipline
A claim is worth what it allows others to verify.
AffectiveAI Research publishes the epistemic status of its materials so that a productive question, a technical specification and a validated result are never treated as the same thing.
Evidence scale
Five states that must not be conflated.
Open
Hypothesis
A provisional explanation accompanied by refutation conditions.
Specification
A formulation precise enough to guide prototypes, tests or review.
Internal evidence
Exploratory or synthetic results with scope and limitations declared.
External validation
Independent testing under a public protocol and an identified institution.
Method
A reviewable chain of work.
The programme connects question, hypothesis, operational definition, test design, ethical review, evidence record and publication. Each step can correct the one before it.
Publication alone does not turn a claim into evidence. It enables others to inspect its scope, examine the reasoning and identify errors.
Structural separation
Research, engineering and validation serve different functions.
Research formulates
Defines questions, boundaries, hypotheses and failure conditions.
Engineering implements
Translates part of the knowledge into systems without turning trade secrets into scientific claims.
Validation tests
Must remain independent enough to publish favourable, null or adverse findings.
Technical depth does not require disclosing an implementable recipe. It requires precision about what is claimed, what evidence supports it and what remains unproven.