Context and candidate interpretation
Context is a first-class entity. Factors include expectation, voluntariness, self-generation, source trust and relationship, environment, attention, task, prior stimulus, repetition, habituation, predictability, novelty, internal state, social/cultural context and perceived intent. Each record declares its applicability. Not all factors apply to every organism or instrument.
A touch example
The mechanical-contact stimulus records dimensions for force, pressure, area, duration, onset description, vibration and temperature. Its physical values are unspecified. The pressure sensation has candidate links to contact, friendly touch and restraint.
expected-voluntary-contactdescribes an illustrative accepted contact context.friendly-touchlists it as possible context.unexpected-restricted-contactdescribes unexpected contact with restricted movement.restraintlists it as possible context.
These are editorial illustrations of representational capacity, not experimentally established mappings. The same measured contact does not reliably classify intent. Neither context guarantees the candidate interpretation, and no probability is supplied.
possible_context distinguishes a relevant factor from required_context. A required-context list is an asserted modeling condition, not executable inference logic. Temporal patterns and spatial frames remain separate: an anatomical site and a device coordinate frame cannot silently substitute for one another.
External appraisal
The restraint example contains openemotions:emotion:fear as a possible appraisal link, with explicit possible loss of control appraisal and conditional wording. This is a namespaced conceptual reference, not a foreign key or a live integration. It does not encode touch = fear.
Future processing API
A future POST /v1/process could accept stimulus, sensor, context and prior state and return candidate sensations/percepts with uncertainty and evidence. It is documentation only. v0.1 deliberately has no such endpoint or unsupported real-time interpretation engine.