Stimulus over-selectivity refers to behavior being controlled by one element of the environment at the expense of other equally salient aspects of the environment. Four experiments trained and tested non-clinical participants on a two-component trial-and-error discrimination task to explore the effects of different training regimes on over-selectivity. Experiments 1 and 2 revealed no differentiation between partial reinforcement (PR) and continuous reinforcement (CRF) on over-selectivity. Experiments 3 and 4 both found that a change in reinforcement (from CRF to PR in Experiment 3, and from PR to CRF in Experiment 4) did not reduce levels of over-selectivity, but rather continuing training with the same contingency (either CRF or PR) did reduce over-selectivity. The results support assumptions made by the comparator hypothesis, extending the growing body of literature explaining over-selectivity as a post-acquisition, rather than attention, failure. (Contains 7 figures.)