Recently, Buss & Spencer (2008) proposed a Dynamic Neural Field (DNF) model of the Dimensional Change Card Sort task (DCCS). This model is able to not only capture the details of 3- and 4-year-olds performance in the standard version, but also generalizes to account for performance in two other canonical variations. The distribution of features in space plays a central role in capturing these effects. To show that the model can generalize beyond space-based effects, we present preliminary simulations of DCCS variations reported by Fisher (2008) that examine the role of automatic and voluntary shifts of attention and randomize the spatial location of the target cards. Results show that the DNF model captures performance in these conditions as well.