Instructional design competency is essential for elementary education majors; however, the “knowing–doing” gap remains a common learning challenge. To strengthen the integration of educational analytics and intelligent sensing applications, we developed a computational cognitive-state modeling framework based on cognitive load (CL) theory and concepts relevant to future cognitive-state sensing applications. Using this framework, we investigated how CL affects perceived overall difficulty (CDP) through cognitive-emotional experience (CEE) and self-assessed design competence (DCS). A total of 630 elementary education majors participated in this study, and questionnaire-derived cognitive-state indicators were analyzed using structural equation modeling (SEM) to validate the relationships among the proposed latent constructs. Exploratory factor analysis identified a four-factor structure with a cumulative variance explanation rate of 74.23%, while confirmatory factor analysis demonstrated good model fit (comparative fit index = 0.967, root mean square error of approximation = 0.061). Path analysis revealed that CL significantly positively affected CEE (β = 0.890, p < 0.001), CEE significantly positively affected DCS (β = 0.512, p = 0.014), and DCS significantly positively affected CDP (β = 0.180, p = 0.010). Mediation analysis confirmed a significant serial indirect effect of CL on CDP through CEE and DCS. Latent profile analysis further identified three heterogeneous learner groups with different cognitive-load characteristics. The findings demonstrate that learning difficulty can be quantitatively analyzed through cognitive-state sensing and computational modeling approaches, providing potential applications for intelligent educational sensing systems and adaptive learning environments.