Cognitive reasoning, through analogy to milestone events in human development, focuses on demonstrating "small data, big task" human-like learning abilities from raw signals (such as vision, hearing, touch, or multimodal inputs) to achieve precise modeling of complex physical and social interactions, thereby driving the advancement of the next generation of strong artificial intelligence.

Cognitive reasoning is a highly interdisciplinary frontier that integrates cutting-edge discoveries and experimental approaches from computer science, statistics, developmental psychology, anthropology, linguistics, neuroscience, and other fields. The main research topics include affordance, functionality, intuitive physics, causal reasoning, tool use and imitation, nonverbal communication, intention understanding, animacy perception, theory of mind, abstract reasoning, utility, explainable AI, and communication.

Led by Prof. Songchun Zhu, this field has a global perspective and extensive international collaborations. The team has led multiple basic research projects funded by agencies like DARPA and ONR MURI in the United States, focusing on complex cognitive reasoning problems. The team's achievements have been published in prestigious international journals, including Science Robotics, and featured as headlines on the homepage of Science. The research outcomes have also received widespread media coverage in both domestic and international outlets, including interviews, People's Daily, Xinhua News Agency, China Daily, MIT Technology Review, IEEE Spectrum, and Smithsonian.