ABSTRACTThe generated mechanism of negative emotion involves the prefrontal cortex, hippocampus, and amygdala in the brain. Based on the biological mechanism, this paper proposes a memristive neural network circuit of negative emotion inhibition with self‐repair and memory. The proposed memristive neural network circuit consists of five modules: memory (hippocampus) module, inhibition (prefrontal cortex) module, damage detection module, repair module, and output (amygdala) module. The memory module does not respond to a small negative signal, but a large negative signal will cause the memory module to form memories. If the large negative signal repeats, the memory module will output a memory signal to the inhibition module. The inhibition module receives the negative signal and the memory signal and then generates an inhibition signal. When the negative signal is small, the inhibition module outputs normally. When the large negative signal is applied for the first time, the inhibition module becomes damaged and the output is abnormal. As the memristor in the inhibition module has exceeded its damaged threshold, the damage detection module will generate a damaged signal to the repair module. The repair module will restore the memristance of the memristor in the inhibition module after receiving the damaged signal. If the large negative signal is applied again, the memory signal from the memory module will help the inhibition module to output normally. The output module receives the negative signal and the inhibition signal and then generates the negative emotion signal appropriately. The PSPICE simulation results show that the hippocampus generates memories after learning and transmits them to the prefrontal cortex, and then the prefrontal cortex inhibits the amygdala to generate the negative emotion appropriately. The proposed memristive neural network circuit can be applied to the robots, which can flexibly alter the intensity of the negative emotion expressed by the robots.
Read full abstract