Toward an Autonomous Background Cognition Layer for AI: A Taxonomy of Unconscious-Inspired Modules for Unprompted Insight Generation

This paper introduces the idea of an Autonomous Background Cognition Layer for AI: a system of background analytic modules that can monitor weak signals, unresolved problems, repeated omissions, behavioral patterns, and hidden decision conflicts over time. Inspired by unconscious human cognition, the framework does not claim that AI has a human unconscious. Instead, it proposes a practical architecture for unprompted insight generation: AI systems that help users notice relevant patterns they did not know to ask about and appear to them the same way solutions and inspiration comes to people randomly while walking or waking up when their brain has processed the neccesary information with the necessary data analysis methods that can be automated to produce the same results. The paper presents a taxonomy of unconscious-inspired analytic modules, a modular architecture for background inference, and an evaluation agenda focused on relevance, novelty, actionability, predictive value, user control, and ethical safety. Its central argument is that future AI should not only become better at answering explicit questions, but also better at detecting when the question itself is incomplete.