TRUTH ENGINE / BASELINE AND INTERPRETATION
An invitation to examine meaning.
The proposed Truth Engine studies relationships between declared responses, synchronised observations and context. Its outputs are estimates with uncertainty and a route for the participant’s correction.
A distinct research agreement.
The protocol proposes a fixed baseline task with obvious questions and participant-labelled truthful or deliberately false responses. Prompts, intended labels, context, timing, calibration and signal quality travel together. Participation and later use have explicit scopes.
Physiology does not settle a person’s truth.
Signals may change with posture, attention, speech, fatigue, stress or the meaning of a question. A probability names the outcome it estimates and the conditions under which it was evaluated. The National Research Council’s review provides context for the limitations of physiological deception inference.
Declarations, measurements and interpretation.
A declaration retains the original words. Signals retain their calibrated measurements. Analysis retains its method, uncertainty and version. Reflection adds the participant’s interpretation.
An unlock is an explicit contract.
A feature states the baseline, usable data, agreed task, evaluator and rule version it requires. A change in evidence or permission has a defined review route.
Evaluate on occasions the model has not seen.
Held-out sessions and tasks examine calibration, false positives, false negatives, context shifts and response to correction. Participant assessments and independent review contribute separate evidence. No validation results for an Aura Truth Engine are supplied by these design documents.