Computational cognitive architectures aim to model the human mind by integrating various psychological functionalities within a unified framework. Traditionally, these models have faced limitations predominantly due to the computational tools and techniques employed in their formulation. However, recent advancements in large language models (LLMs) have demonstrated their superior computational capabilities, prompting researchers to explore their incorporation into cognitive architectures to address real-world complexities and ensure psychological realism.
The article presents a case study on merging the Clarion cognitive architecture with LLMs, capitalizing on the implicit-explicit dichotomy intrinsic to Clarion. This innovative approach seeks to marry the robust computational power of LLMs with the nuanced psychological modeling offered by Clarion. The implications of this research could be profound, enabling cognitive architectures to better simulate human thought processes and serve as more effective tools in both academic and practical applications, while simultaneously presenting risks related to the interpretability and alignment of these integrated systems with human cognitive principles.
👉 Pročitaj original: arXiv AI Papers
