metacognition

Cognition about one's own cognitive processes, comprising monitoring — assessing one's own knowledge, confidence, memory or errors, as in feeling-of-knowing and confidence judgments — and control, using those assessments to guide behavior by seeking information, checking work, abstaining or changing strategy. Psychologists measure it by how well confidence tracks accuracy (metacognitive sensitivity). In AI, analogous capacities include a model's calibrationcalibration, n.: How well a model's stated confidence matches reality: when a well-calibrated system says it is 80% sure, it turns out to be right about 80% of the time., its ability to predict whether it knows an answer, to abstain or ask for clarification, to detect errors in its own reasoning, and to report its internal states (introspectionintrospection, n.: Looking inward to observe your own thoughts and feelings. For AI, the question is whether a model can accurately report what is actually happening inside it, rather than guessing or making it up.); reasoning modelsreasoning model, n.: An AI model trained to work through a problem step by step in hidden 'thinking' text before it answers. Letting it think longer generally makes it better at math, coding and planning. that check and revise their chain of thoughtchain of thought, n.: Getting an AI to write out its reasoning step by step before giving an answer, the way a student shows their working. This often makes answers to math and logic problems more accurate. show a functional form of metacognitive control. Metacognition is central to higher-order theories of consciousnesshigher-order theories of consciousness, n.: Theories holding that a mental state is conscious when the mind, in effect, notices it — when some further thought or perception represents you as being in that state., but its presence in a system is not by itself evidence of experience.