Neural mechanisms underlying memory, attention and cognition
The laboratory focuses on neural mechanisms underlying human memory and attention. We use multimodal neuroimaging methods and computational models to understand the cognitive processes, and tried to apply these mechanisms together with brain stimulation methods (TMS/tCs) to help special groups of people (ageing adults, schizophrenia patients, children with autism spectrum disorder etc.).
In the present study, we separately recorded fMRI and EEG data from healthy volunteers during short-term memory tasks to trace the dynamics of memory with deep learning algorithms. The established multilayer decoding model (MLP2-D) was efficient in predicting the contents of internal representations (chance level 50%) with the accuracy of above 85% in fMRI and of more than 65% in EEG, respectively, higher than other algorithms. This model could further be used to clarify the causes of patients’ impairment in memory and help people, both normal and abnormal, to improve their memory and cognitive functions.