Section 118 of 440

APPENDIX AB: USING THE COCKPIT AND ENGINE

Cockpit and Engine User Guide

The short version

Run one bounded capability experiment at a time. Choose a real outcome, record a baseline, predeclare direct, retention and transfer tests, select the evidence mode that applies to every required result, and measure burden beside performance. Keep the Core profile and operational Capability records separate.

Seven candidates may be inspected but cannot be assessed, scored or added to a portfolio or Core result.

Fifteen-minute start

1. Complete the safety screen and record stop conditions, professional gates and access needs.
2. Choose one operational module connected to a real task.
3. Pick one worked exemplar or define an equivalent bounded task within the module scope.
4. Write the direct measure, quality floor and eligible-opportunity rules before practice.
5. Choose how the evidence will be verified; do not choose a stronger mode than applies to every required result.
6. Record a comparable baseline and predeclare the retention date and novel transfer task.

Where capabilities live

Choose an evidence mode

If the direct result is instrumented but the retention result is only self-recorded, the assessment mode is Self-recorded. Evidence mode does not replace the result, criterion, task provenance, burden or safety record.

The current workflow

1. Create the declaration before observing the result. Locking produces a declaration ID and local SHA-256 digest.
2. If the plan changes, amend it. The new declaration supersedes the prior one; neither is erased.
3. Save each assessment as a new event. Do not overwrite a prior result to make a trend look cleaner.
4. For a maintenance claim, link the retest to the prior assessment, confirm criterion/protocol comparability and stay inside the review window.
5. Log burden, near misses and adverse events as structured safety events so recurrence can be reviewed across modules.
6. Generate a coach or clinician handoff only after selecting the relevant modules, dates and fields. Review the preview before printing or sharing.

The local declaration digest detects a mismatch between the stored protected fields and stored digest. It cannot detect an edit followed by digest recomputation, and it is not a trusted timestamp. External declaration receipts and Ghostmark signing remain disabled until authority, key, time, privacy, rotation and revocation policy is approved.

Optional fixed headers in Titan Engine

The canonical workbook is the artifact-tool output and intentionally contains no serialized pane elements. The Engine bundle also includes a verified sibling workbook with 30 useful freeze panes and a hash-pinned patcher if you want to reproduce it. The patcher verifies the exact canonical input hash, refuses in-place or existing-output writes, and leaves the canonical workbook unchanged. Exact filenames are listed in the bundle README.

First forecast calibration cycle

Use the blank Titan Engine template or Cockpit declaration surface for one ordinary build cycle. This is an operational Forecasting assessment; Focus & Interruption remains an unscored candidate pilot.

1. Choose at least ten bounded tasks that will resolve within the cycle and define the start/stop rule for elapsed work time.
2. Before starting each task, record the point estimate, an interval or confidence range, task category and conditions.
3. Record actual duration without rewriting the estimate. Mark cancelled, merged or scope-changed tasks under the predeclared eligibility rule.
4. Calculate absolute percentage error or the declared alternative, signed bias and interval coverage on eligible tasks.
5. Review only after the declared sample resolves; then predeclare the next cycle’s calibration criterion, retention date and transfer category.
6. Do not enter a Titan level until the module-specific direct criterion and verification requirements are actually met.

The template is intentionally blank. A blank template proves nothing, and a reader’s completed record supports only the bounded claim permitted by the selected module and evidence mode.

One-cycle operating plan

1. Define

• State the task, context, population and desired real outcome.
• Choose the module whose direct measure best matches that outcome.
• Write what success will not mean; this is the claim boundary.

2. Route safety and access

• Apply the module guardrail, professional gate and local requirements before testing.
• Choose accessible cues, equipment, timing and support that preserve the intended function.
• Record support without treating it as failure or using it to infer evidence mode.

3. Predeclare the record

4. Baseline, practise and compare

• Use comparable conditions and record missing data honestly.
• Change one important feature at a time when practical.
• Do not improve the score by lowering quality, selecting easy opportunities or taking more risk.
• Review once a week: keep, adapt, pause or retire based on result and burden.

5. Earn only the supported level

Titan Engine and Cockpit workflow

1. Export a dated backup before migration or structural edits.
2. Choose only an operational module in the Universal Module Lab.
3. Lock a declaration first; then enter baseline, direct, retention, transfer, burden, support, structured safety events and evidence mode against its declaration ID and digest.
4. Use the module-specific anchor and let the level validator surface missing requirements.
5. Resolve legacy evidence-mode warnings by reviewing the original evidence; never infer a mode automatically.
6. Export a fresh backup after confirming results, provenance and record labels.

The workbook and Cockpit are two surfaces for the same discipline. Choose one as the working source of truth for a cycle, then reconcile deliberately. Duplicate live entry can create conflicting records.

Worked-exemplar catalogue

Perception, Spatial Cognition & Sensorimotor Navigation

• Sensory Discrimination & Substitution: legacy EX-ECHOLOCATION record retired from operational use. PWR-078 is held from explanatory publication and training, navigation, mobility, assessment, session, retest and handoff use.
• Sensory Discrimination & Substitution: Adult absolute-pitch naming. Intensive perceptual training can improve pitch naming in adults, and a minority reach high all-pitch accuracy; success, generalisation and durability vary.
• Spatial Simulation & Cognitive Maps: Sketch-map reconstruction and controlled novel route. Landmark, route and orientation representations can be practised and scored on defined maps and controlled routes; success does not establish safe independent navigation.
• Sensorimotor Intelligence & Fall Resilience: Screen-based driving hazard perception. Validated video tasks can measure latency to anticipate developing traffic hazards, and brief video-based training improved experienced drivers’ test performance with the benefit still evident after at least one week. This does not establish safer on-road driving, lower crash risk or fitness to drive.

Learning, Memory, Creativity & Metacognition

• Learning, Memory & Skill Transfer: Task-specific motor learning. A visible motor sequence can be acquired and measured without claiming broad cognitive enhancement.
• Metacognitive Calibration & Error Detection: Confidence-tagged quizzing. Recording confidence before feedback makes task-specific calibration and recurring error patterns measurable; it is not a general self-awareness score.
• Creative Cognition & Insight: Constraint-switch prototype sprint. Separating generation from selection and testing makes the diversity, novelty and usefulness of alternatives observable on a bounded problem; it does not establish a general genius mode.
• Prospective Memory & External Cues: Cue-anchored daily task. A future intention linked to one observable event cue and one fallback can be measured by timely completion and reminder burden; missed intentions are design data, not a moral score.

Quantitative, Decision & Causal Intelligence

• Quantitative Intelligence & Forecasting: Task-duration forecasting. Repeated estimates can be calibrated against actual task duration; the direction and size of bias must be measured rather than assumed.
• Decision, Causal & Experimental Intelligence: Low-risk N-of-1 time-of-day experiment. A randomized or counterbalanced within-person comparison can test whether performance on one safe task differs between two feasible time windows; it does not identify a universal chronotype or prove a medical cause.

Communication, Argument & Negotiation

• Argumentation & Dialectical Testing: Claim-and-counterargument decision memo. A short falsifiable memo can make evidence, assumptions, strongest opposition, cruxes and confidence updates auditable.
• Listening, Teaching & Negotiation: Teach-back with a delayed learner test. A learner’s immediate reconstruction and delayed application can test whether one bounded explanation produced usable understanding; fluency or agreement alone is not evidence of learning.

Collaboration & Collective Intelligence

• Collective Intelligence & Team Orchestration: Independent estimates before discussion. Preserving independent judgments before group discussion can expose useful disagreement and reduce premature convergence.

Making, Tool Use & Operations

• Making, Repair & Tool Intelligence: Dim-sum fold or culinary technique progression. A visible making skill can be improved through repeatable quality criteria, defect review and safe transfer to a related form.
• Operational Systems Design: Backup-and-restore fire drill. A backup is operationally useful only when a defined artifact can be restored and verified within the declared recovery target; one successful drill does not prove disaster readiness.

Human–Technology Augmentation

• Human–AI Complementarity & Reliance Calibration: Human-only versus AI-assisted brief. A human-AI workflow is useful only when it beats the better safe baseline on a declared quality-and-cost rubric.
• Assistive Technology & Person–Task Fit: Two-week assistive-technology fit trial. A time-bounded real-use trial can compare task benefit with setup, symptom, privacy and abandonment burden; availability or novelty alone does not establish person-task fit.

Older imported records

An imported record with no evidenceMode keeps its existing level and evidence text, displays verification mode not recorded, and is not silently downgraded. The next edit or revalidation requires an explicit mode. Unknown future values import as null with a warning rather than being guessed.

Signed and unsigned exports

Wave 1 acquisition and maturity fields

The Cockpit and Engine may store the following optional fields. They are metadata beside the operational level and never score inflation.

Engine validation rules

An incorporated status of Met requires a linked combined-cue, conflict or reliability-weighting result.

An automatic status of Met requires a linked dual-task result and declared acceptable cost.

A non-interfering status of Met requires a linked native-function retest.

A persistent status of Met requires a feedback-off, device-off or practice-free delayed result.

An adopted status of Met requires more than one real-use opportunity and a recorded opportunity denominator.

Maturity fields never raise the operational module level by themselves.

If direct, retention or transfer evidence expires, dependent maturity claims become stale rather than silently remaining current.

Research-governance and Candidate Lab boundary

Weekly review

1. What changed on the direct outcome?
2. How complete and comparable are the data?
3. How was each required result verified?
4. What support, burden, symptom change, near miss or adverse event occurred?
5. Did the result survive the declared retention and transfer tests?
6. Does the protocol earn its time, cost and attention?
7. Should the next action be keep, adapt, pause or retire?

Emergency boundary

Titan is not an emergency service. In a real emergency, contact the appropriate local emergency service and follow dispatcher and qualified-responder instructions. Do not delay help to open Titan or collect data.