Modelling feature-integration in human vision with drift diffusion models