Cooccurrence Probability

Core statistical quantity in the paper's formal model: probability of two observations occurring within a time window

Neighborhood — ranked by edge-count

concept

Pointwise Mutual Information Kernel
associated_with
The kernel that contrastive learners converge to; similarity equals PMI between observations
World Model (statistical)
associated_with
The joint distribution over events in the world that generate observed data; the proposed endpoint of representational convergence

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.

Coherence Scoremethod0.741
GPT-4.1-mini-rated 0-100 score measuring response coherence; used to detect side effects of steering
Coherency Scoremethod0.740
GPT-4.1-mini based score (0-100) measuring clarity, absence of hallucinations, and lack of confusion in generated text
Coherence Score (C)concept0.736
GPT-4o scored 0-100 metric from Betley et al. measuring response coherence on open-ended evaluation prompts
Coherenceconcept0.731
A property that makes a segment of space stand out as a center; determined by symmetry, connectedness, convexity, etc.
Corrigibilityframework0.725
The property of an AI being safe to shut down or modify; discussed in context of GPT.
probabilityconcept0.722
Measure of expected sensory input, core to linking value and surprise.
Covariance Poolingmethod0.715
Novel aggregation technique replacing mean pooling; preserves joint activation structure (feature co-occurrence) in token embeddings.
Learning cooccurrence statistics in either vision or language domain recovers roughly the same perceptual color representationclaim0.713
Empirical validation that PMI convergence actually occurs on real data