Abstract: Spatial ability refers to the ability of individuals to recognize, encode, store, represent, decompose, combine and abstract objects or spatial figures in their minds, which is the cognitive foundation for understanding one's environment and solving problems. Building an accurate, convenient and effective assessment system of spatial ability is of great significance to the enhancement of STEM education and the quality of talent cultivation. Due to the complex, multi-dimensional and implicit nature of spatial ability, it is difficult to evaluate spatial ability via computer-based assessments. This study aims to accurately, effectively, and massively evaluate spatial ability by using multimodal learning analytics methods to explore the characteristic behavioral expressions of learners' spatial cognition, and by developing key technologies and tools for spatial ability stealth assessment based on video game environments.
This study further develops spatial ability assessment methods based on video games, building upon previous research. Unlike prior game-based assessments that primarily focus on post-hoc analysis of backend game data, this study innovates by deeply integrating gamified assessment with multimodal learning analytics to provide process-oriented evaluation and a comprehensive understanding of spatial abilities. Firstly, in the self-developed video game assessment tool, an evidence-centered design framework combined with Bayesian networks is creatively applied to identify and aggregate multimodal behavioral data that can infer learners' spatial ability levels. This approach leverages existing research and expert knowledge, enhancing the identifiability and interpretability of the assessment model. Secondly, a novel paradigm for spatial ability assessment driven by multi-source data is proposed, incorporating learner behavior data (observational data), game backend data (interaction data), and physiological signals during gameplay (contextual data). These data are aligned over time and incorporated into the inference system according to evidential rules, updating probabilistic predictions of spatial abilities to achieve automated assessment. Additionally, the cognitive mechanisms underlying the impact of video games on spatial abilities are explored using 3D puzzle games as a virtual environment for training and assessing spatial abilities. The xAPI standard is employed to automatically collect learner interaction data, enabling stealth assessment of spatial abilities and effectively mitigating issues such as test anxiety and the Hawthorne effect, while ensuring both internal and ecological validity. This lays a technical foundation for the future application and dissemination of gamified spatial ability assessment tools. Lastly, the study aims to achieve a holistic understanding of spatial abilities, covering different scales and cognitive processes through the assessment results, thereby enriching and expanding the concept and scope of spatial abilities, providing instrumental support for theoretical research in this field. Using an evidence-centered design paradigm, an original assessment tool is constructed to deepen our understanding of spatial abilities. Building on theoretical research and leveraging cutting-edge multimodal analysis techniques, the tool collects diverse physiological signals such as EEG and eye-tracking data during gameplay, along with operational logs and behavioral indicators from the game backend. Following a "proficiency-evidence-task" framework, variables at various levels and categories of spatial ability are summarized within a Bayesian network model, facilitating unobtrusive assessment and a comprehensive understanding of spatial abilities. The gamified assessment tool created in this study not only addresses limitations commonly found in traditional psychological tests, such as the lack of process data and susceptibility to social desirability biases, but also cleverly harnesses the appeal of video games as a mass medium, offering low-cost and wide-reaching advantages. Leveraging the convenient dissemination properties of social networks and internet platforms, gamified assessments can stimulate broad participation among players, thus collecting massive amounts of data, with sample sizes reaching thousands, tens of thousands, or even millions. Such large-scale data samples present unprecedented opportunities for group-level studies of cognitive abilities. In summary, the video game-based spatial ability assessment designed in this study holds promise for overcoming longstanding challenges in quantifying and evaluating spatial abilities, with significant implications for large-scale talent selection and development efforts, and the potential to bring about new breakthroughs in the field of spatial ability research.

Key words: game-based assessment, spatial ability, multimodal learning analysis, game-based learning, stealth assessment