Post

Daily Learning Log - 2026-09-20

Records the workspace-wide correction that knowledge is a constructed, revisable world model.

Chinese version: /zh-hant/log/2026-09-20/

Conceptual correction

The learner requested that the workspace be corrected according to the premise: knowledge is the “world model” constructed in the brain.

The workspace now treats external notes, books, videos, data, measurements, and LLM responses as model resources, not automatically as the learner’s knowledge. Knowledge is the relatively dependable, appropriately supported, usable part of the learner’s internal and revisable world model. Facts, concepts, and procedures can be model components; understanding is demonstrated use and revision of a relevant model component; skill and mastery are levels of reliable performance using it.

Records updated

Evidence boundary

This session records the learner’s requested conceptual correction and updated workspace materials. It does not add evidence that the learner can independently distinguish a world event, external representation, internal model relation, prediction, and revision trigger in a new case.

Next action

Without notes, explain the distinction with one original everyday example. Name the world event, the external representation, the model relation in the learner’s mind, a prediction or action it supports, and what observation would revise it.

Learning-process correction

The learner further requested that learning be treated as destruction and reconstruction: a learner should identify the narrow or failed relation in an existing world model, absorb relevant information, integrate the update with relations that still hold, and test the reconstructed model.

The workspace now treats destruction narrowly as withdrawing, limiting, or replacing the specific assumption, variable, mechanism, or boundary condition that failed. It does not mean discarding the learner’s entire knowledge network. A new bilingual knowledge note, raw statement, and no-notes drill record this revision.

Additional records updated

The learner has not yet completed the new drill, so no independent evidence of model-revision performance has been added.

Six-question framework

The learner supplied a framework connecting: what knowledge is; what learning is; what ability is; what learning ability is; and how learning ability can be improved. The original text has been retained separately. A distilled bilingual note preserves the progression while correcting its shorthand:

  • external information is model resource, while knowledge is the learner’s usable, revisable world model;
  • learning is model testing, revision, and reconstruction;
  • ability is reliable performance in a defined task context;
  • learning ability is a task-dependent meta-level capacity to diagnose bottlenecks, select strategies, measure outcomes, and revise the learning plan.

First-principles questioning, plain-language teach-back, deliberate practice with feedback, metacognition, and sleep/spacing/retrieval are recorded as bounded tools rather than universal guarantees. Existing workspace sources support specific interventions under specified conditions.

Additional records updated

This adds no evidence that the learner can independently select a strategy or improve delayed performance. The next action is the new no-notes diagnostic drill.

This post is licensed under CC BY 4.0 by the author.