Abstract:Cognitive computing and granular computing are effective approaches for simulating the cognitive mechanisms of the human brain from different perspectives to construct intelligent computing models. Multigranularity cognitive intelligent computing integrates the advantages of these two computing paradigms, aiming to integrate the seemingly contradiction mechanisms between the processes of human cognitive and information processing, thereby constructing computing models with robust, efficient, and interpretable intelligent. It first summarizes classical multigranularity cognitive computing models about addressing uncertainty issues and their latest research advancements. Furthermore, the current research progress on multigranularity cognitive intelligent computing theory and its applications are elaborated from the perspectives of granulation and cognition. Finally, important issues that urgently need to be studied in multigranularity intelligent computing are analyzed and discussed.