People fixate on blank locations if relevant visual stimuli previously occupied that location; the so-called looking-at-nothing effect. While several theories have been proposed to explain potential reasons for the phenomenon, no theory has attempted to predict the stability of this effect with practice. We conducted an experiment in which participants listened to four different sentences. Each sentence was associated with one of four areas on the screen and was presented 12 times. After every presentation participants heard a statement probing one sentence, while the computer screen was blank. More fixations were found to be located in areas associated with the probed sentence than in other locations. Moreover, the more trials participants had completed, the less frequently they exhibited looking-at-nothing behavior. Fixations on blank locations seem to occur when an attempt is made to retrieve information associated with a spatial location as long as it is not strongly represented in memory.