Quantum-enhanced decision-making in ACT-R

Param Pathak, Nouhaila Innan, Alberto Marchisio, Muhammad Shafique

Abstract: In this chapter, we explore integrating quantum optimization algorithms into cognitive architectures to enhance artificial intelligence (AI) systems. Conventional cognitive architectures, such as adaptive control of thought-rationale (ACT-R) and Soar, have been instrumental in simulating human cognition, but they face significant limitations when addressing complex and uncertain tasks. To overcome these challenges, this chapter proposes a novel theoretical approach, which embeds the quantum approximate optimization algorithm (QAOA) within the ACT-R cognitive architecture. By employing quantum computational techniques, particularly through the alternation of cost and mixer Hamiltonians, QAOA enables AI systems to efficiently explore and optimize complex solution spaces, thus improving decision-making under uncertainty. This integration enhances the problem-solving capabilities of AI systems and introduces a new dimension of computational efficiency and flexibility, allowing for better adaptation to dynamic environments. Through this innovative approach, we aim to bridge the gap between classical cognitive architectures and the emerging field of quantum computing, offering a pathway for developing more sophisticated, resilient AI systems capable of tackling the increasingly complex challenges posed by real-world applications. This chapter provides a foundation for future research, highlighting the potential of quantum optimization in advancing AI technologies.

External IDs:doi:10.1016/b978-0-44-330259-6.00021-9