Change Detection Test

Change Detection estimates visual working-memory capacity. On each trial, an array of 2, 4, 6 or 8 colored squares appears for 100 ms. After a 900 ms blank, the whole array appears again in the same places. On half of the trials one square has a new color. Participants press the left arrow key if nothing changed and the right arrow key if one square changed.

Colors come from a set of nine and never repeat within an array, and a changed square takes a color that was not in the sample. Squares are at least 80 pixels apart and away from the central fixation cross. Change and no-change trials are split evenly within each set size. Each display switches on a screen refresh, and the measured sample and blank durations are saved with every trial. Participants have 3 seconds to answer. Eight practice trials at set sizes 2 and 4 give feedback.

Scores include accuracy, hit and false-alarm rates, d-prime, and Cowan's K = N x (H - FA) at each set size (Cowan, 2001). The capacity estimate is the mean K at set sizes 4, 6 and 8, because K at set size 2 cannot exceed 2. Because the test shows the whole array, Pashler's K = N x (H - FA) / (1 - FA) is also reported, the estimate Rouder et al. (2011) recommend for this design.