Language And Learning

Claim Type: mechanism_hypothesis
Scope: Language and learning dynamics
Depends On: ARC-009, ARC-010
Status: candidate
Claim ID: MECH-011

Source: docs/processed/legacy_tree/architecture/language/language_and_learning.md

Elaborates Section 5 (Social Extension: Language) of REE_CORE.md.

Language and Learning

Language interacts with learning by:

biasing priors during model update,
accelerating generalisation of residue,
enabling counterfactual reasoning without direct experience,
transmitting “negative knowledge” (what not to do) socially.

However:

language-learned ethics is brittle without embodiment,
purely linguistic agents risk ethical drift and rationalisation loops.

REE therefore treats language as:

a multiplier of ethical learning,
not a substitute for harm sensing and residue continuity.

Open Questions

None noted in preserved sources.

MECH-011

References / Source Fragments

docs/processed/legacy_tree/architecture/language/language_and_learning.md