J-space: LLM's 'Recursive' Coordinates with Cross-Layer Consistency
Jack_W_Lindsey · x · 2026-07-07
Anthropic researcher Jack Lindsey detailed the core of his paper in a reply: 'J-space' consists of a set of vectors that persist across different contexts (not context-dependent); although LLMs lack true recursion, they can finitely simulate recursion in a single forward pass, and J-space provides a cross-layer consistent coordinate system, allowing network 'depth' to be viewed as a temporal coordinate. He also mentioned the discussion in Section 9.3 of the paper regarding the similarities and differences between 'global recurrent broadcast' and human cases. This is a high-quality technical interpretation in interpretability/consciousness research.
Related event: Anthropic Discovers Global Workspace Inside Claude(102 posts)→
More from Research
- New review: how prefrontal cortex dynamics turn world models into flexible behavior — AnnaCiaunica · 2026-10-27
- LIBERO-MAX: Robot Policies Drop 11–26 Points When the World Changes Mid-Task — drmapavone · 2026-10-27
- Kolab Benchmark Uses Pro Screen Recordings to Show AI Agents Overestimate Collaboration Quality — minnesotanlp · 2026-10-27
- One BBQ Example Pushes LLM Fairness Scores Past 90%, EMNLP Paper Questions Benchmarks — radamihalcea · 2026-10-27
- Real2Sem2Real turns one phone video into 2,000+ robot demos, matching 150 teleop demos — Scobleizer · 2026-10-27
- Understanding MuonH (Hyperball) through weight norm geometry of scale-invariant networks — jasondeanlee · 2026-10-27