We present an axiomatic way of assigning probabilities to probabilistic models. In particular, we quantify an upper bound for probability of a model or in terms of information theory, a lower bound for amount of information that is assumed in a model. In our setup, maximizing probabilities of models is equivalent to removing assumptions or information stored in the model. Furthermore, we represent the problem of learning from an alternative view where the underlying probability space is considered directly. In this perspective both the true underlying model (Oracle) and the model at hand are events. subsequently, learning is presented in three perspectives: maximizing the likelihood of oracle given model, intersection of model and the oracle and symmetric difference complement of model and the oracle.