Predictive processing is a leading theory of brain function. However, despite several conjectures it is unknown how predictive processing is implemented at the neural level. We will try to understand the exact role of the computations of the layer 5 pyramidal (L5p) cells within the framework of predictive processing. The experimental paradigm and the computational model used here are designed to answer, whether L5p cells carry predictions, as the theory of predictive processing suggests, or are they responsible for some other computation. To this end we will a) adapt experimental paradigms used with humans to study probabilistic computations in the mouse model, b) measure the dendritic input and somatic output of L5p cells in mice, c) capture this activity with computational models and d) manipulate the apical dendritic input and somatic output of L5p through noradrenaline antagonists. The project allows us to understand the mechanisms underlying predictions and their effect on perception.