Change detection, the ability to notice changes in a scene, is a notoriously difficult task for humans. Change detection is commonly measured using a disrupted, or flicker, model in which a scene is shown, quickly replaced by a blank buffer screen, and followed by an altered version of the original scene. Slow changes, those that occur gradually in the scene, are even harder to detect than those that occur abruptly. This study investigates gradual change detection. Facial expressions of famous and non-famous people are changed slowly over the course of a 12-second video and participants are asked to identify the change and rate their confidence in their change detection abilities. It is anticipated that changes to famous faces will be more readily detected than changes to unfamiliar faces because of participants' greater familiarity with them, thus more attention is available to be used for noticing changes occurring. Preliminary data show results that are consistent with this prediction.