E.10.INT:4 - Solution
Recover the relied-on meaning, then continue the work that gives it value.
- Locate the claim that matters. Take the sentence together with the decision or activity it is meant to guide. Ask what would change if the object were judged interesting. If nothing depends on that judgement, keep the expression as it stands.
- Name the participant and relation. Identify who is engaged, affected, learning or choosing. Identify what draws attention, what might change, or what the participant stands to gain or lose. Use the distinctions in §4.1 only where they resolve the sentence.
- Preserve combined meanings. If a participant enjoys an experiment and the experiment could distinguish two models, say both. Let the corresponding motivation and inquiry methods supply their different contributions.
- Restore a needed comparison. When a score or optimization rule is proposed, name what it compares and under which model, experience or resources. Use §4.2 to resolve ambiguous appeals to surprise, progress or a Goldilocks region.
- Return to the action. Rewrite the claim so its recipient can choose, inquire, practise, retain a result or explain a concern. Continue once that meaning is sufficient. A shared decision can keep the clarification in its existing account; an ordinary sentence needs no additional form.
E.10.INT:4.1 - Recover the contribution being claimed
These are recurring uses of the wording. Select the ones needed for the sentence; several can apply together.
| What the wording is doing | What to recover | Example of the resulting claim |
|---|---|---|
| Expressing a participant’s stake | The participant, affected outcome and consequence | “The maintenance team wants the inspection window because it can then replace the worn coupling.” |
| Expressing attraction or engagement | The person or group, activity and response | “These listeners want to keep moving to this rhythm.” |
| Selecting information-seeking action | The agent’s uncertainty, available observation and use of the resulting information | “The controller samples input 1 to distinguish the two candidate relations before choosing its command.” |
| Describing learning progress | The learner or model, what changes and the comparison | “After training, the compressor describes the same material in fewer bits.” |
| Explaining active-inference policy selection | The generative model, candidate actions, information expected from them and preferences over outcomes | “The agent first observes the cue to learn which action is likely to produce its preferred outcome.” |
| Recognizing a promising continuation | The present clue, available variation or combination, and what is worth trying or retaining | “This construction permits a new family of transformations; keep it available for exploration.” |
An information-seeking policy can be attributed to a human, an AI agent or a collective whose members obtain and use the information. For a claim about a person’s experience, preserve that experiencer. A policy description alone leaves the question of subjective experience open.
A prospective continuation can be supported by a hunch, a newly available operation or a promising variation. State that basis at its actual strength. Where its future destination is unknown, the next action can be to explore or retain the construction. C.18 supplies retention in an exploration archive; C.19 helps choose which directions in an active search pool to continue, retain or stop; C.40 helps develop problems and ways together.
E.10.INT:4.2 - Recover the model behind the comparison
For surprise, first recover the quantity being used. The surprisal of an observation, -log p(observation), concerns its probability under a selected model. Bayesian surprise concerns the change from prior to posterior beliefs about chosen variables. Expected information gain evaluates a possible observation before it is obtained. A learning-progress signal compares successive predictive or constructive performance under a learning procedure. Those comparisons can disagree, as §5.1 shows.
In active inference, distinguish updating beliefs from selecting actions. Predictive coding uses prediction errors to update beliefs about hidden states and, in learning formulations, model parameters. Policy selection also depends on the information and outcomes expected from action. Recover the prior preferences when a source explains avoiding an outcome through its expected surprise. The source’s account may couple epistemic and pragmatic value. Preserve that coupling together with the quantities and assumptions that give it meaning.
For a Goldilocks region, recover what varies, for which agent and use, and over which range. It might concern an attainable learning challenge, a listener’s response to syncopation or a combination of rhythmic properties. Use the applicable account of the relation between those characteristics and the desired result. The useful region can move as the agent learns or as the task changes. In a search algorithm, a difficulty screen can select candidates for learning while novelty and transfer still govern which possibilities are retained.
When a quantitative answer is needed, C.16 supplies measurement and C.17 separates novelty, usefulness and sample surprise. Keep an adequate qualitative comparison when it already settles the action.
E.10.INT:4.3 - Combine contributions in a development project
A project can use information-seeking to distinguish alternatives, learning progress to choose a promising practice opportunity, and novelty or diversity to retain further possibilities. State the contribution of each rule to that project.
For example, a group can test the controller in §5.1 during a maintenance window. The researcher calls input 1 interesting because its response distinguishes two candidate relations. A colleague sees the query construction as material for other unknown interfaces. The maintenance team needs the controller returned to service before the window ends. Recover their claims as three instructions: “Query input 1 to choose the command”; “Keep this construction available for exploration of other interfaces”; and “Schedule controller work within the available window.” If the relation is supplied, the query becomes unnecessary. The potential use of the construction remains, and the window still constrains any remaining controller work.
C.11 compares formed options when a choice is needed. C.19 helps decide which directions in an active search pool to continue, retain or stop. A single agent’s information-seeking policy uses the applicable subject method; choosing among formed queries can use C.11. C.40 supports the development of problems and ways. The field’s methods supply the experiment, training activity or construction itself.