Decoding Neural Signals of Memory Reinstatement and A↵ective State

Memory is thought to depend on the reinstatement of patterns of activity across the cortex that are similar to the patterns elicited by the original learning experience. According to this model, affective states such as stress might impair certain forms of memory by disrupting or delaying the reinstatement of these patterns. Characterizing the nature of cortical reinstatement under di↵erent a↵ective states is thus a critical initial step towards understanding the neural basis of memory. Here, we applied machine learning techniques to electroencephalography (EEG) data collected during a paired-associate retrieval experiment, when participants were under conditions of relative safety and stress. Specifically, we trained an algorithm on neural patterns representing categories of visual images, and then classified neural signals during memory retrievial to decode reinstatement of associate categories. Additionally, we classified di↵erent types of mnemonic status and a↵ective states, and we also find limited evidence that stress may impact memory-related neural activity.