ABSTRACT
Mental models perform simulations from their very nature. This can be seen by considering the character of mutors. Mutors are active things, which transform inputs into outputs. Mutors are small neural networks, with inputs from some effectors, and outputs to others, depending on the specific mutor. Every simulation will require a certain allowance of mental resources: mutors activated, values of effectors stored, etc. Thus, the amount of mental power that a simulation uses could be measured, and compared between simulations. Many of the differences between simulation and propositional inference arise from the difference between mutors and logical rules. It is clear that the basic notion in assessing the quality of mental models must be the quality of the simulations that they allow. The chapter is also concerned with one sense of 'understanding': the sense in which we understand things or people.
