tags: #thought Researchers have been interested in [[Student model|student modelling]] and [[Simulated student|simulating students]] for decades, where the former can be used in the process of adapting instruction for personalisation, whilst the latter for evaluating EdTech. Whilst I am not a fan of simulated students given our existing struggles for naturalistic validity even with real students, I still entertain the thought here for the promise of what it could provide in the future. If we take education to be a design science, it could be possible to formulate student modelling and simulating students as two sides of the same coin. That is, [[Student model|student modelling]] induces properties by looking at the manner in which the student engages with the design, as well as the design itself. Meanwhile, [[Simulated student|simulated students]] creates some evolving representation of properties that is reasoned upon to determine how to perform student-like actions. So there are three components to be built: 1. The deduction of properties from student engagement 2. The format in which the student is modelled 3. The use of properties * Simulated student, to generate student-like responses * Adaptive systems, to adapt instruction Whilst (1), the deduction of properties from student engagement may not seem necessary for simulated students, it could be used to lighten the load of modelling the effects that the specific lesson may have on the student. That is, the student modelling is created to derive nuanced information about what has changed in the student through engaging in the design, and we can use this information to guide the next student-like generation. --- TODO * Write it in a way that actually makes sense to people * Turn into a neat diagram ![[Pasted image 20250814170906.webp]]