liveBook · Manning

Metacognition

Metacognition is the capacity to monitor, evaluate, or regulate one's own cognitive processes.

How 3 Manning books explain it

Introduction to Generative AI, Second Edition

Metacognition is the capacity to recognize, monitor, or think about one's own mental processes.

How it is used
Frames consciousness and AI sentience through the question of whether a system can recognize its own mental processes.
How deep it goes
Defines metacognition through conscious awareness but stops short of providing a reliable measurement method.
Examples
Uses work-email planning, romantic uncertainty, breathing, knee-tap reflexes, and LaMDA’s self-description.
What you can do after
Evaluate claims about AI consciousness while distinguishing fluent responses from demonstrated self-awareness.

Read how Introduction to Generative AI, Second Edition explains metacognition →

AI Agents in Action, Second Edition

Metacognitive monitoring is the ongoing assessment of an agent's own knowledge and confidence while producing answers or taking actions.

How it is used
Uses metacognitive monitoring to help production agents respond to uncertainty, stagnation, and knowledge gaps.
How deep it goes
Defines three monitoring capabilities and illustrates knowledge-boundary assessment with workspace signals.
Examples
Illustrates stagnation through repeated unchanged findings and knowledge-boundary assessment through retrieval relevance, memory coverage, and confidence.
What you can do after
Evaluate agent behavior for overconfident guesses, stalled progress, and insufficient knowledge.

Read how AI Agents in Action, Second Edition explains metacognition →

Designing AI Agents

Meta-cognition is the ability of an agent to evaluate and improve its own cognitive activities or outputs.

How it is used
Adds an explicit feedback path that evaluates an agent’s artifact or outcome and can alter current work or future behavior.
How deep it goes
Defines reflection as review of first-pass outputs and reports coding-task results, but does not specify an implementation procedure.
Examples
Uses coding tasks as a reported case, where Reflexion-augmented agents improved from 50% to 90% success.
What you can do after
Evaluate first-pass agent outputs through an explicit review step that may revise current work or future behavior.

Read how Designing AI Agents explains metacognition →

What to know first

Related concepts

Questions about metacognition

What is metacognition?

Metacognition is the capacity to monitor, evaluate, or regulate one's own cognitive processes.

Which Manning books explain metacognition?

Introduction to Generative AI, Second Edition, AI Agents in Action, Second Edition and Designing AI Agents.

What should I know before learning metacognition?

Manning books that teach metacognition usually introduce language model first.

What are the parts of metacognition?

Metacognition includes model calibration, knowledge boundary detection, reflection and stagnation.

These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.