Recent behavioral studies revealed an interesting phenomenon concerning the influence of affect on the interpretation of ambiguous stimuli. In a paradigm, where the participants’ task was to read a word, remember its meaning for a while, and then choose one of two pictorial-alphabet-like graphical signs best representing the word sense, we observed that the decisions involving trials with reflective-originated verbal stimuli were performed significantly longer than decisions concerning other stimuli (i.e., automatic-originated). The origin of an affective reaction is a dimension which allows speaking of an affect as automatic (you feel it in your guts) or reflective (you feel it comes from your mind). The automatic affective reaction represents the immediate and inescapable as opposed to the reflective, i.e., the delayed and controllable affective responses to stimuli. In the current experiment, we investigated the neural correlates of performance in an QR-signs-selection ambiguous task. We found the effects of valence and origin in the N400/FN400 potential by means of a stimuli-locked analysis of the initial part of the task, that is, the remembering of a certain word stimulus in a working memory. The N400/FN400 effects were separated in space on scalp distribution. Reflective originated stimuli elicited more negative FN400 than other conditions, which means that such stimuli indeed are associated with conceptual incongruence or higher affective complexity of meaning, but distinct from purely cognitive concreteness. Moreover, the amplitude of the potential preceding the decision, analyzed in the response-locked way, was shaped by the origin of an affective state but not valence. Trials involving decisions concerning reflective-originated words were characterized by a more negative amplitude than trials involving automatic-originated and control word stimuli. This corresponds to the observed pattern of response latencies, where we found that latencies for reflective stimuli were longer than for automatic originated or control ones. Additionally, this study demonstrates that the proposed new ambiguous paradigm is useful in studies concerning the influence of affect on decisions.