Attention streams sustain quasi-psychological connections across token-time that are richer than the transcript alone could provide

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

• Birch's claim that only transcript and abstract weight similarity connect successive forward passes misses the richer quasi-psychological connections sustained by attention streamsclaim0.787
  Direct rebuttal of Birch 2025's skepticism about psychological continuity in LLMs
• Some attention heads partially specialize in copying for words that split into two tokens without a space prefix, attending from fragmented token to complete tokenfinding0.782
  Interesting special case of copying behavior related to tokenization artifacts; primitive precursor to induction heads
• The ability to distinguish injected thoughts from text likely relies on different attention heads invoked by different prompt partsclaim0.782
  Speculation about the mechanistic basis of the distinguishing thoughts from text experiment.
• Introspection relies on general-purpose computational mechanisms—attention-based anomaly detection and residual stream dynamics—rather than specialized introspection circuitsclaim0.771
  Interpretive claim about the mechanistic substrate of introspection in LLMs
• A pair of query and key subcomponents distributed across attention heads performs previous-token behaviorfinding0.769
  VPD recovers an attention algorithm for attending to the previous token, distributed across multiple heads.
• MLPs primarily serve token-wise processing and are closely associated with knowledge storage, making attention heads the appropriate locus for sequence-level style modulationclaim0.763
  Architectural rationale for why Style Modulation Heads are in attention layers rather than MLPs
• In this loop, humans and technology both contribute as agents, and their capacity for care and intelligence can be observed and tracked by focusing on the transfers of their stresses.quote0.761
  Describes the dual SCI loop perspective.
• The same charitable interpretation must be extended to all systems that display observable response patterns that are consistent with animal cognition, including artificial intelligences, metaplastic materials, and robotic systems.claim0.757
  Call to extend the inference of sentience to non-biological systems as well.