hypothesis
active
hypothesis:if-network-representations-recapitulate-biological-population-coding-manifolds-will-be-represented-by-many-localized-features-tiling-the-geometry-rather-than-single-global-directionsIf network representations recapitulate biological population coding, manifolds will be represented by many localized features tiling the geometry rather than single global directions
Neuroscience-motivated prediction.
Source paper
extracted_from(2026) · Usha Bhalla · Thomas Fel · Can Rager · Sheridan Feucht +8
Neighborhood — ranked by edge-count
Papers (1)
paper
Findings (1)
finding
- Top SAE features on the years manifold show localized smooth tuning curves with overlapping supportassociated_withsupportsReminiscent of population coding.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- The paper's concluding summary statement asserting the deep interpretive significance of representation geometry.
- The paper's deepest interpretive claim, asserting that representation structure and behavioral structure are not coincidentally aligned but deeply connected.
- Neural representation geometry causally shapes behavior; interventions respecting that geometry will yield natural trajectories.hypothesis0.786Central hypothesis tested via manifold steering experiments across language models and video world models.
- Opening sentence framing the paper's core inquiry.
- Linear representation hypothesis: neural networks represent meaningful concepts as directions in their activation spaces.hypothesis0.784Foundation for interpreting features as linear directions.
- Developmental bioelectricity is proposed as a tractable entry point to understand the informational architecture of collective intelligence in morphogenesis.
- Manifold structure is present in representations and causally influences downstream behaviorclaim0.780Interpretation of steering results.