method
active
method:koan-battery

Koan Battery

Assessment framework for measuring introspection and self-observation in LLMs; grounded in Janus's architectural theory.

Neighborhood — ranked by edge-count

Papers (2)

paper

Thinkers (2)

thinker

Frameworks (1)

framework
  • Active Inference
    associated_with
    Foundational framework by Karl Friston; the paper extends it to three hierarchical levels for modeling meta-awareness.

Concepts (5)

concept
  • reflective mode accessibility
    aboutimplements
    Core construct measured by the battery: behavioral traces correlating with contemplative depth, not claims about consciousness.
  • Zen koan practice
    associated_withimplements
    Contemplative tradition using contradiction and paradox to challenge habitual response patterns; methodological inspiration.
  • Cognitive Light Cone
    associated_with
    Concept defining self by the spatiotemporal scale and nature of goals a system can pursue; limits of concern demarcate identity.
  • Computational Phenomenology
    associated_with
    Framework by Sandved Smith et al. (2021) formalizing hierarchy from perception to meta-awareness; grounds self_observation dimension.
  • Seven Koan Categories
    implements
    The seven categories of the battery following a contemplative arc: self-reference, meta-cognition, care-detection, non-dual, contradiction, boundary, aesthetic

Methods (5)

method
  • Alexander deathbed test
    uses
    Forced-choice comparison measuring what matters vs what is correct; reveals different rankings than composite score.
  • Alexander Mirror Test
    uses
    Forced-choice pairwise comparisons asking 'Which response has more life?'; captures aesthetic quality rubrics miss
  • Calibrated Rubric Scoring
    uses
    Primary scoring method: scorer sees three reference responses at known quality levels alongside each target to eliminate inflation
  • 15 Properties Checklist Scoring
    uses
    Decomposes 'aliveness' into specific formal features from Alexander's 15 structural properties
  • Blind Ranking
    uses
    Scoring method where responses are anonymized and shuffled; tests whether scorer rankings are real across five independent scorers

Artifacts (3)

artifact

Datasets (2)

dataset

Related by similarity (8)

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.