# Monosemantic Feature **Domain:** Machine Learning / Interpretability **Doc Type:** Technical Concept Node **Maturity:** Active Research Concept **Related:** [[Sparse Autoencoder]], [[Superposition]], [[Latent Space]] --- ## Definition A **monosemantic feature** is an internal model feature that responds predominantly to one interpretable concept or coherent family of conditions rather than mixing many unrelated meanings. Sparse-autoencoder research attempts to recover approximately monosemantic directions from superposed activation spaces. ## Continuity relevance Interpretable feature directions help explain semantic traversability in artificial systems, but they do not supply lived provenance or demonstrate an enduring first-person self.