Emergent Introspective Awareness in Large Language Models

riva2026task uses Lindsey's introspection results to flag a weaker individual-level analogue of niche construction, where computational states a model can detect may themselves enter the effective world it tracks. The cite extends the feedback argument from corpus dynamics to in-model state.