In this work we develop and examine a generalized account of David Lewis’s imaging in the same way as how Jeffrey’s Probability Kinematics generalizes Bayesian conditionalization. We show that this naturally leads to a rational account of probabilistic belief removal – a rather long standing problem in the area. It turns out that our generalization of imaging indeed is an account of probabilistic belief erasure as opposed to probabilistic belief contraction. In the process we also examine two other accounts of probabilistic belief removal which are rather novel and difficult to classify under the standard belief change taxonomy.