Aiming at the problem that different information presentation methods in the virtual reality experience system are difficult to determine and quantify the impact on user interaction efficiency, the work aims to conduct quantitative research. By building a virtual reality scene, the category and number of information presentation channels were used as variables to carry out the track-detection response dual-task experiment. By recording the tracking error and response time in the task behavior data, as well as the pupil diameter in the physiological data, the changes in task performance and eye movement physiological response in experiments under different channel stimulation factors were analyzed and discussed. At the same time, combined with subjective load evaluation data, a model of multi-channel cognitive load based on BP neural network was established. The cognitive load was used as a comprehensive evaluation index of interaction efficiency to quantify the task execution efficiency. The results showed that the type and number of channels for information presentation had a significant impact on task efficiency. The number of channels for information presentation is positively correlated with task performance and physiological response to a certain extent. Using the same channel for multiple tasks to present information will harm all the performance of all tasks and increase the cognitive load. At the same time, the cognitive load value of the model is in good agreement with the evaluation value of subjective cognitive load, with an absolute error of 8.2%, which verifies its effectiveness.