Cognitive Load from simple computations — LessWrong