报告人：丁为

Title: Hippocampal State Distinguishes between the Formation and Utilisation of Relational Memory: Anterior-Posterior Gradient and Functional Connectivity Dissociation

Abstract: The hippocampus is a critical brain region for relational memory, but its specific role in memory formation and utilisation is not fully understood. In this study, we used functional magnetic resonance imaging (fMRI) during a sequential memory task to investigate the distinction between hippocampal formation and utilization states in relational memory. Our results showed regional and network level differences in representational patterns between the two states. In the formation state, the hippocampus shows higher correlation in the posterior part and functional connectivity with the precuneus cortex and the orbital frontal cortex, whereas in the utilisation state, the hippocampus shows higher correlation in the anterior part and functional connectivity with the superior parietal lobule and the middle frontal gyrus. These findings suggest that the hippocampal state switch involves an anterior-to-posterior gradient of coarse-to-fine representation in relational memory. The observed differences in functional connectivity patterns provide further evidence for a hippocampal state switch, with the hippocampus serving different functions during different stages of relational memory processing.

报告人 吴越

Title：Deep Generative Models Accounting for Developmental Cognitive Phenomenon

Abstract：Understanding the fundamental ability that leads to cognitive development is a key challenge in developmental psychology and cognitive neuroscience. In this work, we propose a novel approach that links deep generative models, specifically variational autoencoders, to the learning processes that occur in infants. We demonstrate that these models can capture essential features of human learning and behavior under various circumstances. We further dive into the connection between variational generative models and predictive coding theories.

报告人：殷子

Title: The Cognitive Basis of Intuitive Navigation Complexity Judgment

Abstract：Human intuition in judging the complexity of maps for navigation is a well-known phenomenon, but the underlying cognitive factors that drive this intuition are not fully understood. In this study, we hypothesize that the connectivity and path complexity between points on a map are key factors that influence human intuition in navigating maps. To test this hypothesis, we generated a series of maps and modeled them as directed graphs, manipulating their complexity based on the distribution of shortest paths between points. We found a strong correlation between the complexity measure of the maps and human intuition in judging their navigational difficulty. To investigate the development of this ability, we trained an agent using reinforcement learning to navigate the maps and calculated its success rate. The agent's success rate was highly correlated with both human intuition and map complexity measures, indicating that the ability to judge navigational complexity arises from interactions with the world. These findings shed light on the cognitive factors underlying human navigation intuition and have important implications for improving navigational aids and map design.