Unconstrained Models

• One problem with models, even computational models, is that they are largely uncsontrained.
• That is, in essence they map inputs to outputs.
• How many functions are there to map a particular set of inputs to their corresponding outputs?
• As there are an infinite number of models, how can the models be constrained.
• One way is that the computational model can implement a theory.
• Another way is to have the model do more things. As it does more, each new task constrains the model.