Neural Mechanisms of Implicit Rule Inference

Wang Shangyi (Wang Zuoren’s Lab), Graduate Student
CAS Center for Excellence in Brain Science and Intelligence Technology (CEBSIT)

Cognitive flexibility is a crucial executive function. A key limitation in current paradigms of cognitive flexibility research is the difficulty in disentangling the mechanisms of negative feedback from strategy shift. To overcome the limitation, we developed a novel behavioral paradigm that allows subjects to shift strategies without negative feedback, building on insights from creativity and cognitive flexibility studies.
In this study, internally-driven strategy shifts were compared with error-feedback-guided shifts in rats. Our findings demonstrate that the mPFC is specifically involved in internally-driven strategy shifts. By combinationally using optogenetics and electrophysiological recordings, we observed that while a subset of mPFC neurons encoded optimal strategies prior to behavioral shifts, network perturbation paradoxically accelerated behavioral switching. These results suggest that the prefrontal network serves dual roles: it not only encodes early indicators of new strategies but also actively maintains ongoing strategy stability during cognitive flexibility processes.

Reuse of Domain-Specific Representations: Neural Representation Studies of Lifelong Brain Flexible Learning

Tian Kaixi (Yu Shan’s Lab), Doctoral Student
Institute of Automation, Chinese Academy of Sciences

How does the brain maintain stable learned knowledge while flexibly adapting to new environments? This “stability–plasticity dilemma” is a major question in neuroscience. We hypothesize that the brain addresses this challenge by reusing domain-specific neural coding spaces (NCS), which constrain the space for a given function while maintaining relationships with other spaces.
In our study, we first verified the reuse of domain-specific representations in primate flexible learning tasks, showing that the brain achieves a balance between knowledge retention and flexibility through “functional separation” and “category encoding.” These findings provide important biological insights for designing intelligent systems capable of continual learning.

Date: June 13, Friday, 2025, 16:00–17:30
Venue:

• Shanghai A405 Lecture Hall
• Shenzhen Xili D602 Meeting Room
• Beijing Intelligent Building, 3rd floor, 3rd Meeting Room

Chaired by: Dr. Zhang Tielin