In order to balance the relationship between preschool children's cognitive ability and icon modeling design, the work aims to explore the interface icon modeling design of the science popularization APP based on preschool children's cognitive load. Firstly, based on existing literature research, three different types of icon modeling and three modeling evaluation dimensions were extracted. Secondly, 6 representative eye movement indicators and 3 EEG indicators were screened out, and the data were collected through eye movement-EEG experiment. Thirdly, the improved Pass scale was used to obtain subjective evaluation data of cognitive load. Then, the stepwise regression method was used to screen out 3 eye movement indicators and 2 EEG indicators that had the greatest impact on cognitive load. Finally, the relationship among eye movement indicators, EEG indicators and cognitive load was explored, and a multiple regression model was established. It was found that there were significant differences in the effects of different types of icon modeling on cognitive load. By integrating eye movement and EEG indicators and establishing the regression relationship model with cognitive load, the icon modeling affecting children's cognitive load is better analyzed and a quantitative analysis of visual aesthetics and cognitive load is formed, providing a theoretical basis for the design of icon modeling for preschool children's science popularization APP interfaces and program evaluation.