DAS overcomes the localist limitation of prior causal abstraction by allowing individual neurons to play multiple roles via non-standard bases

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

• DAS achieves substantial causal effect even on arbitrary input-output mappings where no causal mechanism should existfinding0.820
  Replication of Wu et al. 2023 finding; DAS expressivity concern validated in CausalGym setup
• There is a many-to-many mapping between neurons and concepts, meaning multiple high-level causal variables might be encoded in overlapping groups of neuronsclaim0.794
  Fundamental theoretical claim motivating DAS, attributed to Smolensky/Rumelhart/McClelland.
• Early causal abstraction methods (Geiger et al. 2021) implicitly rely on the privileged bases hypothesis, while recent methods (Geiger et al. 2024b) rely on the linear representation hypothesisclaim0.790
  Historical framing of how representation assumptions have evolved in causal interpretability
• The discovery of perfect abstract equality representations that cannot be decomposed into entity representations is a foundational result informing our understanding of how symbolic and connectionist architectures coexistclaim0.786
  Concluding claim about theoretical significance of the hierarchical equality finding.
• Smolensky (1986) proposes that viewing a neural representation under a basis that is not aligned with individual neurons can reveal the interpretable distributed structure of the neural representations.quote0.780
  Load-bearing theoretical claim providing the conceptual foundation for DAS.
• causal abstraction implicitly relies on strong assumptions about how features are encoded in deep neural networks (DNNs), and becomes trivial without such assumptionsquote0.779
  Load-bearing formulation of the paper's central argument
• You can only get the profound multiple structure of centers by unfolding each bit from the previous state, allowing the next layer of structure to appear from the previously established layers.claim0.777
  Explains why time and sequence are essential for generated complexity.
• Causal abstraction implicitly relies on strong assumptions about feature encoding in DNNs, and becomes trivial without such assumptionsclaim0.777
  Authors' interpretation connecting their proof to practical interpretability methodology