Figure 3.

A graphical representation of why jittering the detection criterion leads to underestimation or effective lowering of detection sensitivity. Jittering the detection criterion is effectively identical to jittering the distributions, because with respect to detection behaviour, the scale of the horizontal axis is not critical. Therefore, given four conditions (left column) where different criteria are set, we can conceptualize them as equivalent to four conditions where the criterion is constant, but the distributions themselves move around (middle column). If we average these four conditions over a large number of trials, higher effective variances result for both distributions, thereby lowering detection sensitivity.