Smartphone users touch the screen of the phone thousands of times per day. The inter-touch intervals follow a power-law distribution. We propose a multi-scale statistical model for those smartphone touches. At short-time scale, the model is governed by refractory effects, while at longer time scales, the touching rate is governed by the priority difference between smartphone tasks and other tasks. We show that both the statistics of the short intervals as well as the longer intervals are well captured by the model. The model is described in continuous time and the inter-touch interval distribution can be computed analytically.