Announcing MCog Core: A Foundation for Metacognitive AI

We’re excited to announce the release of MCog Core, a new domain-agnostic ontology for representing metacognition.

As many of you know, our internal R&D focuses heavily on crafting intelligent systems for strategic decision-making and AI-augmented consulting.

A core part of that has involved developing systems that can reason about their own reasoning. This is where metacognition comes in – the ability to think about thinking.

Metacognition is a crucial aspect of human intelligence, allowing us to monitor our cognitive processes, detect errors, and adapt our strategies. It's also becoming increasingly important for artificial intelligence, especially as AI systems become more complex and autonomous. By equipping AI with metacognitive capabilities, we can create systems that are more robust, reliable, and adaptable.

MCog Core and its key features

MCog Core is an ontology designed to capture the fundamental concepts of metacognition in a structured and reusable way. It provides a framework for representing:

• Different types of reasoning processes

• Heuristics (mental shortcuts)

• Hypotheses (testable assumptions)

• Reflection (introspection on reasoning)

• Biases (systematic errors in judgment)

• Confidence assessments

• Feedback loops

• Learning processes (including double and triple-loop learning)

Benefits of using MCog Core

One of the key features of MCog Core is its domain agnosticism.

It's not tied to any specific field or application, making it reusable across a wide range of domains.

This initial release is a "lite" version, focusing on core concepts to foster broad adoption and gather feedback.

It's also designed to be modular and extensible , so you can easily adapt it to your specific needs or integrate it with other ontologies.

Motivation for creating MCog Core

The primary motivation for developing MCog Core was to support our work in AI-augmented consulting. We needed a way to represent the metacognitive processes of AI agents within the context of consulting engagements.

However, I quickly realized that a general-purpose metacognitive ontology could be valuable to a much wider audience, including researchers and developers working on AI, cognitive science, decision support systems, and educational technologies.

You can find the MCog Core ontology, along with detailed documentation, on GitHub. The ontology is released under the Creative Commons Attribution 4.0 International (CC BY 4.0) license, so you're free to use, share, and adapt it as long as you provide appropriate attribution.

Feedback and contributions

If you have any suggestions, comments, or ideas for extensions, please share them on the GitHub repository. I believe that this ontology can be a valuable resource for anyone working on metacognitive systems, and I'm excited to see how it evolves with your input.

In the coming months, we may expand MCog Core with more detailed representations of biases, heuristics, and uncertainty. I'm also exploring how to integrate it with existing cognitive architectures. Stay tuned for updates!

Thanks for reading, and I look forward to your feedback and contributions!