There is a many-to-many mapping between neurons and concepts, meaning multiple high-level causal variables might be encoded in overlapping groups of neurons

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

• Polysemantic neurons are a major challenge for the circuits agenda, because N meanings in one neuron times M in another creates NxM effective connections that cannot be considered individually.claim0.804
  Precise characterization of why polysemanticity poses a combinatorial obstacle to circuit analysis
• DAS overcomes the localist limitation of prior causal abstraction by allowing individual neurons to play multiple roles via non-standard basesclaim0.794
  Central claim motivating DAS over prior methods.
• Each human is a collective intelligence; the parenthetical 'of neurons' can be dropped as a conceptual commitment.claim0.791
• The ability to represent relationships (e.g. correlations or associations) among variables, rather than a system of independent variables, is essential for any connectionist model of cognition.claim0.787
  Necessary condition for connectionist cognition.
• Superposition hypothesis: neural networks represent more features than dimensions using almost-orthogonal directions.hypothesis0.781
  Explanation for why dictionary learning can recover many more features than dimensions.
• The key question is not only the scale-up of a unitary cognitive capacity, but the many-into-one transition: the emergence of minds from collectives, working in spaces beyond those of their component cells.quote0.781
  Encapsulates the core challenge of the field.
• Neurons can correspond to interpretable functional roles but interpretations in terms of individual neurons are unlikely to be the most parsimoniousclaim0.777
  Claim from footnote 3, acknowledging neuron-level interpretability while arguing subcomponents are better.
• Investigating the causal substructure of neural representations is necessary to avoid misidentifying data structures of simpler representations as abstract conceptsclaim0.777
  Motivated by the finding that lexical entailment decomposes into word identities.