“Digital twins” is becoming a paradigm for research and applications in various disciplines. To avoid jumping on the bandwagon without a thorough understanding of what it means and what it takes to address the genuine needs in our investigations, this seminar targets the nature and necessity of developing digital twins for social science research. Starting from digital twins of cities, the seminar will develop arguments for constructing digital twins for social science investigations in general. It is argued that digital twins should go beyond the digital representations of real-world systems and should address social issues with a sound theoretical foundation and powerful information infrastructure. It should be a paradigm instrumental in bringing theories, particularly on dynamics and complexity, into the modeling, simulation, prediction, and practical management of our ever-evolving dynamical complex socioeconomic systems. The talk introduces a general framework for the learning and incorporation of dynamics and complexity into digital twins which themselves evolve throughout the life cycles of the corresponding physical systems. The framework lays the foundation for the construction of novel digital twins for the advancement of social science research and improvement of decision-making. The conceptual arguments will be substantiated by real-life phenomena and processes.