|
|
The EGI Framework: A Five-Layer Multi-Agent Cognitive Architecture
|
Abstract:
This paper describes the Extended General Intelligence (EGI) framework, a five-layer cognitive architecture that integrates semantic world modeling with persistent multi-agent reasoning. EGI organizes perception, semantic understanding, collaborative reasoning, orchestration, and meta-cognition into a unified structure, in which world-model information is represented as explicit semantic state variables shared across-agents. To illustrate how this architecture can be used in practice, we implement a multi-agent financial intelligence system and apply it to five years of central bank communications, macroeconomic indicators, risk factors, and market data (2020-2025). The system produces central bank tone indices, macro and risk-regime inferences, and market-reaction expectations. A comparative evaluation against a flat, single-prompt baseline suggests that the EGI-based design yields more stable semantic trajectories and more coherent cross-domain reasoning. The framework is intended as an architectural blueprint that may support further research on interpretable, long-horizon multi-agent intelligence.
| [1] | Ha, D. and Schmidhuber, J. (2018) World Models. https://arxiv.org/abs/1803.10122 |
| [2] | Hafner, D., Lillicrap, T., Norouzi, M. and Ba, J. (2021) Mastering Atari with Discrete World Models. https://arxiv.org/abs/2010.02193 |
| [3] | Lake, B.M., Ullman, T.D., Tenenbaum, J.B. and Gershman, S.J. (2016) Building Machines That Learn and Think Like People. Behavioral and Brain Sciences, 40, e253. https://doi.org/10.1017/s0140525x16001837 |
| [4] | Marcus, G. (2020) The Next Decade in AI: Hybrid Models and Reasoning. AI Magazine, 41, 22-35. |
| [5] | Baddeley, A. (2012) Working Memory: Theories, Models, and Controversies. Annual Review of Psychology, 63, 1-29. https://doi.org/10.1146/annurev-psych-120710-100422 |
| [6] | Hutchins, E. (1995) Cognition in the Wild. MIT Press. https://mitpress.mit.edu/9780262581462/cognition-in-the-wild/ |
| [7] | Clark, A. and Chalmers, D. (1998) The Extended Mind. Analysis, 58, 7-19. https://doi.org/10.1093/analys/58.1.7 |
| [8] | Doshi-Velez, F. and Kim, B. (2017) Towards a Rigorous Science of Interpretable Machine Learning. https://arxiv.org/abs/1702.08608 |
| [9] | Wooldridge, M. (2009) An Introduction to Multiagent Systems. Wiley. |
| [10] | Shanahan, M. (2022) World Models and Cognitive Maps. Trends in Cognitive Sciences, 26, 547-559. |
| [11] | Hafner, D., Lillicrap, T., Fischer, I., Villegas, R., Ha, D., Lee, H. and Davidson, J. (2019) Learning Latent Dynamics for Planning from Pixels. Proceedings of the 36th International Conference on Machine Learning (ICML). |
| [12] | Silver, D., Hubert, T., Schrittwieser, J., Antonoglou, I., Lai, M., Guez, A., et al. (2018) A General Reinforcement Learning Algorithm That Masters Chess, Shogi, and Go through Self-Play. Science, 362, 1140-1144. https://doi.org/10.1126/science.aar6404 |
| [13] | Gao, L., et al. (2023) LLaMA-Agents: Tool-Augmented Multi-Agent Systems. |
| [14] | Schick, T., Dwivedi-Yu, J., Dessì, R., et al. (2023) Toolformer: Language Models Can Teach Themselves to Use Tools. https://arxiv.org/abs/2302.04761 |
| [15] | Liang, P., et al. (2024) Structured Agents and Workflow-Supervised LLMs. |
| [16] | Park, J.S., O’Neill, D. and Zhu, Q. (2023) Social Simulacra: Building Memory-Augmented Agents. |
| [17] | Xu, Z., et al. (2024) Memory-Augmented Language Models. |
| [18] | Kahneman, D. (2011) Thinking, Fast and Slow. Farrar, Straus and Giroux. |
| [19] | OpenAI (2023) GPT-4 Technical Report. https://arxiv.org/abs/2303.08774 |
| [20] | Lewis, P., Perez, E., Piktus, A., et al. (2020) Retrieval-Augmented Generation for Knowledge-Intensive NLP. https://arxiv.org/abs/2005.11401 |
| [21] | Minsky, M. (1986) The Society of Mind. Simon & Schuster. |
