Cognitive Digital Twins: A Systematic Review of Definitions, Applications, and a Unified Definition

Cognitive Digital Twins (CDTs) are regarded as an evolved version of existing Digital Twin (DT) systems and are capable of certain cognitive abilities. However, the various introduced definitions and characteristics of CDTs, and different understandings of “cognition”, create conceptual ambiguity around CDTs. This paper critically reviews key definitions, application domains, capabilities, and proposed architectures of CDTs. Following PRISMA 2020 guidelines, a systematic review methodology is conducted across Scopus and Web of Science to map existing definitions, cognitive capabilities, and application domains of CDTs. Studies that explicitly implement or conceptualise a DT and explicitly mention cognitive, intelligent, autonomous, or AI-driven properties are included. Conversely, conference papers, book chapters, editorial pieces, review articles, and non-English publications are excluded from this review. The results of 59 reviewed studies present bibliometric metadata and a thematic analysis of early and recent definitions and applications of CDTs across various domains, such as manufacturing, which is the most studied discipline in terms of CDT implementation. Findings show that the understanding of cognitive enhancement has shifted toward the semantic enrichment of DT systems, with a significant emphasis on knowledge-driven approaches. The discussion focuses on identifying key differences between DTs and CDTs and synthesising existing definitions. The key contribution of this study is a unified definition of CDT, a mapping of cognitive capabilities and application domains, and a future research agenda. The review is not registered. The review is limited to journal articles, and the enabling CDT technologies, along with their implementations, are not addressed within this paper.