To explore the mechanism by which the green space ratio in rural parks affects the psychological recovery of users, this study constructed three types of green space ratio scenarios ranging from 0% to 10%, 10% to 50%, and 50% to 80% using virtual reality (VR) technology. Physiological data were collected through electroencephalogram (EEG) and eye-tracking (ET) techniques, and a comprehensive evaluation was conducted using the Perceived Recovery Scale (PRS). Among them, the alpha waves (8 to 13 Hz) of EEG are closely related to relaxation and psychological recovery, and an increase in amplitude indicates a more relaxed state of the participants. The study found that the medium green space ratio (30% to 50%) scenario performed best in terms of security, fixation duration, and activation of EEG alpha waves (frontal F4 electrode), significantly outperforming the low green space ratio (0% to 10%) and high green space ratio (50% to 80%) scenarios. It can more effectively promote emotional regulation and thus reveals the non-linear relationship between green space ratio and psychological recovery. The low green space ratio scenario has a low security score, a shortened fixation duration to 200 ms, a decreased a wave amplitude, and the poorest perceptual recovery benefit. It is recommended to introduce dynamic landscapes to enhance the sense of depth. The high green space ratio scenario has good privacy but a single visual stimulus, a dispersed fixation duration (200~250 ms), and a lower recovery benefit. It should be moderately open to enhance transparency. The research conclusion indicates that the green space ratio of rural parks should be controlled within 10% to 50%, and combined with multi-level plants and dynamic landscape configuration, to achieve the best psychological recovery benefit, providing a scientific basis for the therapeutic design of rural parks.