Embodied Interpretive Bidirectional Interface Convergence Architecture

Embodied Interpretive Bidirectional Interface Convergence Architecture (EIBICA) explores how wearable systems can interpret relationships among a person's signals, their environment, and their expressed intent. A system may respond, ask, defer, or remain silent. EIBICA keeps sensing, interpretation, authorization, action, correction, and evidence logging separate.

person signals + environmental observations + expressed intent
                              |
                              v
                    bounded interpretation
                              |
                              v
                         authorization
                              |
                              v
             respond | ask | defer | remain silent
                              |
                              v
                   correction + provenance

Each observation remains distinct from the interpretation built from it, and each action requires authorization. The person can approve, reject, correct, or stop a proposed response.

Current evidence boundary: this repository establishes theoretical, software, and record-contract foundations. It does not validate biometric inference, multimodal convergence, wearable benefit, participant outcomes, or adaptive-interface utility.

Current build focus

The planned first bench experiment uses a photoresistor, a bounded interpretation policy, an authorization button, and an LED. The system either pulses the LED or stays silent, while the laptop keeps the run record. Its evidence is limited to engineering behavior and record integrity. It cannot establish that EIBICA understands a person, infers a biometric state, or improves an experience.

Jahlon selected EIBICA as the current project and architecture name. The naming decision explains its scope and limits. Earlier EIA-* record IDs remain unchanged as stable historical identifiers.

Explore the build with an agent

Paste this repository's GitHub URL into an agent with this prompt:

Open this repository and begin with READER_GUIDE.md and agent-entry.json. Reconstruct the EIBICA build chronologically from its session, iteration, correction, disclosure, learning, and Git records. Distinguish Jahlon's directions and reflections from agent recommendations, identify changes and unresolved questions, preserve evidence boundaries, and do not infer events absent from the records.

The linked records let an agent explain how the work developed without inventing missing events. The repository does not contain raw private agent conversations. Public GPT Live learning transcripts, when added, are verbatim except for explicitly manifested sensitive-data redactions.

For a local deterministic view, run python skills/eibica-co-design-iteration/scripts/eibica_iteration.py journey --repo . --format markdown.

Evidence boundary

This foundation is theoretical infrastructure. It contains no simulation output and reports no empirical result. Equations and model sketches are proposals until explicitly tested and manifested.

Repository map

• paper.md: canonical manuscript
• model.md: definitions, assumptions, and candidate formalization
• references.md: citation ledger and verification status
• coordination/: agent-human build contract and operating boundary
• inquiry/: versioned scientific questions, constructs, and inquiry state
• benchmarks/: empty human-controlled review state and gate-design template; no benchmarks approved
• records/coordination/: ordered, human-gated build iteration records
• records/coordination/sessions/: programmatically opened agent-session snapshots linked to, but distinct from, build iterations
• records/learning/: per-session learning entries and GPT Live interview packets
• records/disclosure/events/: append-only disclosure events linked to sessions and iterations
• skills/eibica-co-design-iteration/: canonical source for the reusable EIBICA coordination skill and CLI
• simulation/: future implementation notes; no language selected
• results/: manifested outputs only
• DISCLOSURE.md: human and AI contribution record
• CORRECTIONS.md: append-only correction history

Coordination validation

Run python tools/validate_coordination.py . to verify that iteration records follow the seven-phase contract and preserve required human gates. A passing validator establishes record structure only; it does not prove that physical work occurred or that a scientific claim is true.

Run python tools/validate_scientific_records.py . to verify that the inquiry remains at Seed 0.1 and that no gate, metric, threshold, or expected result has been approved before the agent-human benchmark review.

Working coordination record

The live orientation and decision log is maintained in the EIBICA Stepwise Build & Decision Log, linked from the EIBICA Notion hub. Notion records status and decisions; versioned repository artifacts and manifested runs remain the evidence authority.

Jahlon Burruss accepts responsibility for research, acceptance, correction, and publication decisions.