The Brain-Machine Interface (BMI) allows us to extend the human body and sensing by communication between external devices and brain. A study in monkeys (Nicolelis et al., 2000) showed that effective motor control signals to decode arm movements appeared not only in the motor cortex but also in the frontal and parietal cortices. This result suggests that the accuracy of the BMI can be improved by expanding areas of interest where neural activity is monitored to decode the motor planning. However, current BMI systems that use electrodes to record neural activity are limited in the flexibility of the region of interest. In this study, we aim to develop an optical BMI system based on activity from multiple brain regions. We designed a BMI system in which mice obtain rewards according to a specific pattern of neural activity. We established a lever-press behavioral task in which head-fixed mice under a fluorescence macroscope gained rewards in response to auditory cues. We recorded the neural activity from the cerebral cortex to identify specific activity patterns during the task. We are constructing a real-time closed-loop system that detects the pattern and triggers a reward. This study will contribute to the development of new therapies to control and restore neural function.