The method, developed by Washington State University professor, Richard Lamb, and colleagues is called "computational modelling," and involves a computer "learning" student behaviour and then "thinking" as students would.
    
Lamb said the process could revolutionise the way educational research is done.
    
"Traditionally, we'd be confined to a classroom to study student learning for virtually every potential theory we have about science education and curriculum implementation," Lamb said.
    
"But now, instead of taking a shotgun approach, we can test the initial interventions on a computer and see which ones make the most sense to then study in the classroom," he said.
    
"In the current model of research, we go into a classroom and spend months observing, giving tests and trying to see if changes to a specific model work and how to best implement them.
    
"It will still be necessary for researchers to go into the classroom; hopefully that never goes away. This just gives us more flexibility," Lamb said.
    
Lamb and his fellow researchers used an artificial neural network they named the Student Task and Cognition Model.
    
Students were given tasks to complete in an electronic game. The tasks were scientific in nature and required students to make a choice. The researchers used statistical techniques to track everything and assign each task as a success or failure.
    
"The computer is able to see what constitutes success, but it's also able to see how students approach science," Lamb said.
    
Because the computer is learning an approach to science, rather than just how to do a specific task, it will later try to solve a different problem the same way a student might.
    
Lamb said most entertainment video games have the same characteristics as educational videos games. So long as it asks a singular task of the students, any game would suffice - Halo, Call of Duty, Mario Kart and more, Lamb said.
    
"The computer is learning to solve novel or new problems, which means we can test different educational interventions before ever getting to a classroom," he said.
    
He said those initial tests will not only tell researchers if a specific educational model will work, but will give a specific percentage of success.
    
"For me to get 100,000 students, teachers to administer tests, professors doing research and all the rest, we could easily look at about USD 3.5 million," Lamb said.
    
"We can now get those 100,000 students for the cost of running software off a computer," Lamb added.