For example, when a person gets lost, he / she predicts a scene that appears first while moving, and updates the position prediction by comparing the prediction with the newly obtained observation.The "anticipation" at this time is the "virtual reality" created in the brain.Also, self-confidence in expectations seems to influence the expression of this virtual reality.

　In this research, we investigated whether such virtual reality in the brain and self-confidence that people think introspectively can be extracted from brain activity by artificial intelligence (AI), and whether virtual reality can be reproduced well when there is self-confidence. ..

　Therefore, we conducted a space movement game to explore the inside of the VR maze where the rooms are lined up in a grid pattern for the experiment participants who memorized the map of the maze in advance.We predicted the scene of the room to move to before opening the door, and asked them to answer their confidence (confidence) in the prediction, and measured their brain activity.

　Then, using AI technology, we investigated whether the prediction scene and the certainty of the prediction could be deciphered from the brain activity.As a result, it was shown that the prediction scene can be decoded from the superior parietal lobule / inferior parietal lobule and the dorsal premotor cortex, and the certainty of the prediction can be further decoded from the anterior prefrontal cortex.It was also found that the decoding performance of the scene prediction changes depending on the confidence level for the prediction.

　According to the results of this research, new applications such as technology for directly connecting the brain and machines and control methods for connecting the brain and AI in space movement such as drones and vehicles are expected in the future.

Paper information:[Communications Biology] Confidence modulates the decodability of scene prediction during partially-observable maze exploration in humans