Intelligence, Learning, and Artificial Life Models: Evolution and Systematic Review

Acevedo D.A.V.M.; Mauricio D.; Rivera L. · IEEE Access (2026) · DOI: 10.1109/access.2026.3692536

Resumen

Artificial Life (ALife) seeks to understand and recreate the fundamental processes of living systems using synthetic models capable of robust, adaptive, and autonomous behavior. Despite significant advances, the field still lacks an integrated perspective that jointly examines modeling, intelligence, and learning, three key aspects for the study of ALife. This study presents a systematic review of scientific literature indexed in Scopus, Web of Science, and the ACM Digital Library, complemented by secondary sources, following the methodology proposed by Kitchenham and Charters. A total of 694 records were initially identified, from which 52 studies were selected through a rigorous multi-stage filtering process. The review identifies 45 ALife models classified according to Heudin's typology into Soft ALife (37), Hard ALife (7), and Wet ALife (1), highlighting the predominance of computational approaches. With respect to intelligence, 62 instances were identified across 44 articles, corresponding to limited-memory, cognitive, self-organizing, emergent, and evolutionary intelligence, revealing a growing tendency toward hybrid multi-intelligence architectures. In terms of learning, eight learning paradigms were identified, including supervised, unsupervised, reinforcement, self-organized, continuous, evolutionary, probabilistic, and static learning, corresponding to a total of 77 algorithms. Overall, the findings indicate that ALife models are becoming increasingly adaptive and cognitively sophisticated through the integration of diverse learning mechanisms and forms of artificial intelligence. This review constitutes the first systematic effort to jointly examine model typologies, intelligence categories, and learning mechanisms in ALife, offering an integrative and evolutionary perspective on the progressive enhancement of adaptive and cognitive capabilities in artificial living systems.

Instituciones: Universidad Nacional Mayor de San Marcos