The N2pc (PCN) component of the event-related potential (ERP) waveform provides a useful tool for directly assessing the locus of spatial attention in visual search. It is still unclear whether the amplitude of the N2pc/PCN relates to the deployment of attentional resources. A key issue is the lack of evidence that top-down allocation of attention affects the N2pc/PCN amplitude. Previous findings could be explained if manipulating different expectancy strategies changes participants’ search mode, causing them to redefine the target’s features. In this study, we explored the relationship between N2pc/PCN amplitude and top-down attention allocation by manipulating the discriminative difficulty (differences in the response-defining feature) but leaving the search difficulty (target’s saliency) unchanged. Using the same sets of stimuli, in a blocked condition, participants showed the expected higher amplitude of N2pc/PCN in the hardest condition, compared to easier discrimination conditions. Importantly, there was no difference in the N2pc/PCN when the exact same stimulus sets were presented in a randomly interleaved mixed set. At a behavioral level, in both conditions performance was significantly slower for the hardest condition. This finding indicates that the N2pc/PCN component is modulated by the predictability of discriminative difficulty, which reflects the modulation of top-down attentional deployment.