The effect of adding to-be-ignored extra stimuli to a modified Stroop task is investigated. Adding extra stimuli of the same kind as the distractor causes a temporary improvement in performance (Stroop dilution), whereas adding extra stimuli of the same kind as the target causes an improvement in performance that is only detectable when the extra stimuli are removed (post-treatment). An attempt is made to explain these different outcomes in light of the existing theoretical accounts of the Stroop dilution effect. A computational model that accounts for the observed data is proposed. Results suggest that a top-down control mechanism compensates for lateral inhibition effects, particularly when they have a potentially disruptive influence on performance. This boost of control seems to last longer than needed, causing performance improvements in a post-treatment condition. A further implication of these results is that the top-down control function is trainable.