Expectations and Hypothesis Main hypothesis: We expect the difference in reaction times for congruent and incongruent trials under low load to be weaker or even vanish under high load (significant interaction effect in ANOVA), because high perceptual load of the primary task should impede the processing of unnoticed stimuli (e.g. Lavie 2005, 2006, 2014). Manipulation of a congruency effect affects performance in the primary task (manipulation check I): We expect mean reaction times to be higher for incongruent than for congruent trials under low load (it does not need to be significant under high load, because we explicitly expect it to be reduced under high load). Thus, if the unnoticed numbers belong to the same category as the target number (smaller than 5 or greater than 5) participants should respond more quickly to the target number compared to trials in which the unnoticed numbers belong to a different category. This should be the case even though these numbers are never consciously noticed. This finding would be a replication of the findings of Schnürch, Kreitz, Gibbons, and Memmert (2016) with different stimuli and task settings. Manipulation of perceptual load affects performance in the primary task (Manipulation Check II): We expect mean reaction times to be higher for high load than for low load trials in congruent trials. (This effect is probably but not necessarily significant in the incongruent condition, as high load inhibits the interference effect, which leads to slower reaction times in incongruent trials under high load, so it would be possible for those two effects (load and interference effect) to neutralize each other).