In the present study, we aimed to evaluate whether the number of training trials performed bythe participants during the baseline protocol in equivalence class experiments could modulatethe N400 evoked component. Two groups of 15 participants each followed a matching-tosampleprotocol to train on the conditional relations between four sets of abstract stimuli.Participants in the Extended Group performed approximately twice as many trials as those inthe Reduced Group. After having achieved the learning criteria in the equivalence tests,participants’ neural activity was recorded during semantic judgment tasks that includedstimulus pairs of both the same (related pairs) and different (unrelated pairs) classes. Ourfindings indicate that participants in the Extended Group had similar N400 components forrelated and unrelated stimulus pairs. Conversely, participants in the Reduced Group had morenegative waveforms for unrelated stimulus pairs compared to the ones for related stimuluspairs. We discuss the necessity of a more careful analysis of the choice of the number ofbaseline training trials in experiments on equivalence relations and N400 component, anddraw parallels between our findings and the N400 effect previously described for high and low frequency words in the participants' native language.