The effect of lineup size on discriminability is dependent on filler similarity and independent of encoding strength.

A photo lineup, which consists of one suspect and several physically similar fillers, is often used by the police to test an eyewitness's memory. To optimize memory performance, how similar should the fillers be to the suspect, and how many fillers should be included in the lineup? Recent work suggests that using fillers who match the basic characteristics of the perpetrator (e.g., same age, race, and gender) but who are otherwise maximally dissimilar to the suspect optimizes discriminability. However, the optimal lineup size has been found to vary with filler similarity, with larger lineup sizes increasing or decreasing discriminability depending on whether low-similarity or high-similarity fillers were used, respectively. Because manipulating filler similarity at retrieval affects overall performance, here we investigated whether encoding manipulations that affect overall performance also affect how lineup size influences discriminability. In three experiments, we first replicated prior findings (N = 502), then reduced encoding strength by making study images blurry when low-similarity fillers were used (N = 553), and finally increased encoding strength by repeating study images when high-similarity fillers were used (N = 501). We found that whether overall performance was low or high due these encoding manipulations, discriminability still increased as a function of lineup size when low-similarity fillers were used and decreased as a function of lineup size when high-similarity fillers were used. Thus, lineup size has opposing effects on discriminability when task difficulty is manipulated at retrieval, which narrows the theoretical explanations for why that effect is observed.