In section 2 we describe the mixture models considered. Firstly, we describe the discrete classification mixture models. We then describe how we can approximate these discrete classification mixture models using the continuous weights mixture models to allow adaptive spatial regularisation. In section 3 we describe the class distributions. We then describe how we infer on the model using Markov Chain Monte Carlo techniques in section 4. In section 5 we examine the behaviour of this model when applied to artificial data with different spatial characteristics, and finally in section 6 we apply it to FMRI statistical parametric maps.