question

pending-review

question:does-the-geometric-structure-of-activation-space-causally-shape-neural-network-behavior

Does the geometric structure of activation space causally shape neural network behavior?

wurgaft-2026-manifold.md

Frontmatter (9 fields)

{
  "doc": "wurgaft-2026-manifold.md",
  "context": "Central research question driving the work.",
  "norm_label": "Does the geometric structure of activation space causally shape neural network behavior?",
  "graphify_id": "geometry_shapes_behavior_hypothesis",
  "source_file": "wurgaft-2026-manifold.md",
  "imported_from": "/Users/antonborzov/Documents/Research.nosync/papers/extract_typed_out/wurgaft-2026-manifold/graph.json",
  "extracted_type": "question",
  "source_location": "Abstract",
  "graphify_file_type": "question"
}

Outgoing (2)

answered_by (1)

Interventions along activation manifold M_h yield behavioral trajectories following behavior manifold M_y, and vice versa — bidirectional relationship demonstrated across language models and video world models.(finding)

gates (1)

Geometry in neural representation is not merely incidental, but is in fact the proper object for enabling principled control via intervention on internals.(claim)

Incoming (0)

None.

Mentions (1)

papers-typed
wurgaft-2026-manifold.md