V4 Planning Index

Status: consolidation map Created: 2026-05-08 Scope: V4 planning documents, phase boundaries, and documentation gaps

This document is the entry point for V4 planning. It does not replace the specification documents. Its job is to make the scattered V4 material navigable and to name the places where consolidation is still needed.

Canonical anchors

Role	Document	Use
V4 substrate anchor	v4_spec.md	Canonical list of V4 primitives and V4-bound claim cohorts.
Object/entity harness	v4_developmental_harness_spec.md	Object permanence, affordance schemas, intervention, and developmental object-learning tests.
V3 counterpart	substrate_roadmap.md	V3-tractable enrichments that should not be folded into V4 by accident.
Sleep/setpoint boundary	v3_v4_transition_boundary.md	V3 static setpoints that become V4 dynamic mechanisms; full sleep/offline consolidation scope.
Phase/TCL boundary	v3_v4_phase_substrate_boundary.md	V4-deferred phase, TCL, ephaptic, and V(t) substrate commitments.
Scaling frame	architecture_scaling_needs.md	Derived scaling versus deliberate intelligence scaling.
Failure-mode atlas	psychiatric_failure_modes.md	Architectural psychopathology mappings used as validation targets.
Deployment gate	../governance/deployment_gating.md	Policy that V3 remains sandbox-only and serious deployment is V4-gated.

Current V4 primitive stack

Primitive	Current anchor	Notes
V4-0 object/entity permanence	v4_spec.md, v4_developmental_harness_spec.md	Added as an explicit primitive because object schemas and persistent entity slots are prerequisites for social agents.
V4-1 multi-agent ecology	v4_spec.md	Requires separate agents with their own harm, drive, commitment, and policy state.
V4-2 self-model integration	v4_spec.md, sd_030_e2_self_forward_model.md, sd_031_e2_world_forward_model.md	DR-10..DR-14: z_self scoring, self-domain goals, E2 PE confidence, temporal self-depth, hedonic/proxy dissociation.
V4-3 long-horizon dynamics + persistent identity	v4_spec.md, v3_v4_transition_boundary.md	Lifetime-scale identity, sleep calibration, chronic dynamics, persistent damage and memory state.
V4-4 richer action repertoire	v4_spec.md, v4_developmental_harness_spec.md	Communication, manipulation, withdrawal, carry, push, place, and use actions.
V4 phase substrate	v3_v4_phase_substrate_boundary.md	Phase/TCL/ephaptic refinement of V3 synaptic stand-ins.
V4 clinical/failure-mode validation	psychiatric_failure_modes.md	Failure modes should be re-indexed against V4 primitives after V3 closeout.

Consolidation decisions

1. v4_spec.md is the canonical V4 substrate spec. Other V4 documents should reference it, not drift into independent scope definitions.
2. v4_developmental_harness_spec.md is the object/entity permanence harness, not a side note. It should be treated as V4-0.
3. substrate_roadmap.md remains the V3 enrichment boundary. If a feature can be tested honestly in V3, it belongs there unless governance explicitly authorises V4-only handling.
4. The two boundary documents remain binding: V3/V4 sleep-setpoint transition and V3/V4 phase-substrate deferral answer different questions.
5. Psychiatric failure modes should become a validation matrix, not just a narrative catalogue: each syndrome-like regime should map to substrate perturbation, REE metric signature, behavioural signature, human analogue, and falsifier.

Documentation gaps

Gap	Proposed landing place	Why it matters
V3 anatomical/functional concordance map	new V3 closeout doc	Needed before human-brain labels become too loose or inconsistent.
V4 object/entity permanence detail	v4_spec.md plus harness spec	Other-agent permanence depends on ordinary object permanence.
Failure-mode validation matrix	psychiatric_failure_modes.md or companion index	Makes psychopathological phenomenology experimentally actionable.
V4 experiment epoch contract	V4 runner/governance planning doc	V4 evidence needs a distinct architecture_epoch and script template.
V4 claim cohort refresh	claims.yaml plus planning docs	Current docs name claim cohorts, but claim registry changes need a governed pass.
V4 environment/training curriculum	new V4 curriculum doc	Environment expansion should train affective streams deliberately, not just add complexity.

Near-term documentation sequence

1. Finish V3 evidence closeout and freeze the final V3 substrate map.
2. Write a V3 brain-mapping/concordance document with evidence grades.
3. Convert the psychiatric failure-mode document into a matrix indexed by affected streams and control substrates.
4. Expand the V4 object/entity permanence section into a harness-to-substrate migration plan.
5. Draft the V4 environment and learning curriculum: object tasks first, affective stream training second, multi-agent ecology third.
6. Only after V3 full completion and governance authorisation, start V4 substrate implementation.