Towards a Computational Model of Why Some Students Learn Faster than Others