Larger hidden representations create more random structure that DAS can search through, allowing manipulation of counterfactual behavior even in randomly initialized networks

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

• Patching h[1] with a divergent representation can activate distinct, hidden pathways that result in misleadingly confirmatory behavior and/or undetected behavior.quote0.781
  Load-bearing description of the core pernicious divergence mechanism illustrated in Figure 1
• Direct probes over learned activations in standard basis may fail to reveal the actual causal role of representations because they are highly distributedclaim0.778
  Supported by the finding that non-trivial rotations are required to find aligned representations.
• There are fewer representations competent for N tasks than M<N tasks, so training more general models should yield fewer possible solutionshypothesis0.777
  Selective pressure toward convergence via task generality
• DAS overcomes the localist limitation of prior causal abstraction by allowing individual neurons to play multiple roles via non-standard basesclaim0.774
  Central claim motivating DAS over prior methods.
• Deep representations have a special significance in recurrent networks, allowing coordinated behaviour without losing sensitivity to new inputs.claim0.772
  Importance of hierarchical structure for flexible coordination.
• probably helps not only with faithful reconstruction but also creates interference patterns that encode nuanced information about the deltas and convergences between states.quote0.769
  Key quote connecting path redundancy to interferometric information encoding.
• Shallow interaction structures cannot compute non-linearly separable functions; depth (hidden layers) is necessary for ETI-relevant individualityclaim0.765
  Assertion that deep organization is mandatory, based on connectionist theory
• We hypothesize that representation geometry drives model behavior — the geometric structure of internal representations causally shapes what models do externally.hypothesis0.764
  The causal hypothesis motivating the use of causality (intervention) as the lens connecting representation and behavior geometry.