The Academic Salon of the Department of Psychology and Cognitive Sciences is held every Thursday afternoon. Welcome to all students and faculty members from every department!
Time: May 8 (Thursday) afternoon 13:00
Location: Room 915, 9th floor, Lv Dalong Building
Department of Psychology and Cognitive Sciences ZHANG Yisi's Research Group Academic Salon
Report One
Marmoset Family Development Tracking Study
Presenter: Bing Yuan
Content:
Humans' ability to form complex social cooperation networks and social cognitive abilities are inseparably linked to the biological evolutionary mechanism of long-term parental care. Long-term developmental tracking in natural environments is difficult to conduct in human families due to privacy and experimental cycle considerations. Marmosets share many commonalities with humans both genetically and in their relatively complex social behaviors, making them an ideal model for studying social cognition and exploring high-function biological mechanisms such as reproductive strategies. Traditional animal behavior analysis adopts manual evaluation methods, which have very limited throughput. With the development of artificial intelligence, computer vision and machine learning technologies have made large-scale behavior quantification increasingly feasible, and a series of animal automated tracking systems have been developed. However, existing systems are still difficult to directly apply to family development tracking research. On one hand, juveniles undergo body size changes as they grow, especially during infancy, when juveniles cannot be dye-marked and it is difficult to manually mark key body points, creating identification difficulties. On the other hand, migration across different family scenarios often requires large amounts of new training data to retrain models.
To address the above problems, this study introduces semantic segmentation and three-dimensional pose estimation to build a marmoset family automated tracking system. It eliminates breeding environments, segments juveniles and transfers learning of juvenile postures, quantifying and exploring marmoset family social interaction behaviors in complex scenarios. This provides effective technical support for studying the natural behaviors of marmosets, an important animal model widely used in social behavior research.
Report Two
Paradigm Exploration of Marmoset
Collective Problem Solving
Presenter: Haoxin Xu
Content:
Collective problem-solving ability is an important component of intelligence. To deeply explore the biological evolutionary mechanisms and developmental processes of collective problem solving, finding suitable animal models and developing corresponding experimental paradigms has become a primary issue. Previous research shows that marmosets have prosocial and cooperative breeding behaviors similar to humans, making them one of the few non-human primate models suitable for studying social cognition. However, there is currently a lack of flexible, variable, and easily quantifiable experimental paradigms for studying collective problem-solving processes. This study explores a marmoset collective problem-solving behavioral paradigm using tablets as task media. We first demonstrate that through two weeks of training, marmosets can skillfully click screens to obtain food rewards. Second, in social environments, marmosets' screen-touching behavior is influenced by companions, preliminarily proving that social interaction factors can be integrated into touchscreen-based task systems. In summary, we demonstrate that this behavioral paradigm can to some extent restore collective problem-solving processes and provide detailed quantification of learning and interaction.
