Experience Sampling Method (ESM)

Human psychology method for repeated in-situ self-report; methodological inspiration for the paper's approach

Neighborhood — ranked by edge-count

method

Numeric self-report
uses
Primary tool in human psychometrics for tracking latent internal states; adapted as the core measure in this paper for LLMs

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.

Experience Replaymethod0.741
RL technique using episodic memory to improve sample efficiency; used in some game-playing agents.
When LLMs produce experience claims under self-reference, is this sophisticated simulation or genuine self-representation, and how would we tell the difference?question0.729
The core interpretive question the paper narrows but cannot definitively answer
ESR Testing Pipelinemethod0.724
Three-step protocol: (1) object-level prompting, (2) SAE-latent steering, (3) judge model scoring of attempts
Does ESR reflect model scale, architecture, or training procedures?question0.724
Central unresolved question about the mechanism behind ESR's apparent size-dependence
TEM-t requires many fewer data samples than TEM to reach equivalent performance (sample efficiency improvement)finding0.724
Empirical performance comparison showing TEM-t is a more efficient learner than the original TEM.
Experiment 3: Semantic Clustering of Experience Reportsconcept0.721
Tests whether experience reports show systematic cross-model semantic structure via embedding analysis of five-adjective self-descriptions
Rejection samplingmethod0.721
A technique to filter model outputs; Redwood Research's project mentioned.
piecemeal evolutionary approachconcept0.720
Gradual, incremental introduction of morphogenetic sequences into the existing system, as opposed to revolutionary replacement.