Most theories of how decisions are made assume that the accumulation of evidence from the environment is a noisy process. Recently, models have been proposed which do not have this micro-variability, and as a result are simple in the sense of being analytically tractable. We use a global model analysis method called landscaping to show that in terms of flexibility, simply removing micro variability does not necessarily make a model more simple. Our landscaping also highlights an experimental design which might be helpful in discriminating between different response models.