In both paradigms, remembering faces was more difficult than remembering words (task main effect: RM, F[1,92] = 33.2, p < .0001; WM, F[1,109] = 494.7, p < .0001; Figure 1). Patients had poorer performance than controls (group main effect: RM, F[2,92] = 7.56, p = .0009, η2p = .141; WM, F[2,109] = 23.5, p < .0001, η2p = .301; between group contrasts: RM, HC vs. AH, F[1,92] = 11.5, p = .001, HC vs. NH, F[1,92] = 8.93, p = .004; WM, HC vs. AH, F[1,109] = 35.5, p < .0001, HC vs. NH, F[1,109] = 30.5, p < .0001), but hallucinators did not differ significantly from nonhallucinators (RM, F[1,92] < 1.0, n.s.; WM, F[1,109] = 2.57, p > .11). The difference in accuracy between the patient groups and controls was smaller for words than faces during the WM paradigm, which resulted in a significant group × task interaction, F(2,109) = 27.0, p < .0001. A ceiling performance of controls may have contributed to this effect, and no such interaction was observed during the RM paradigm. However, between group contrasts
