In addition, we computed an analysis of variance (ANOVA) for repeated measures in both time intervals on the average amplitudes, on eight frontal electrode sites (Fpz, AF1, AF2, Fz, F1, F2, F3, F4) for each item type. The ANOVA included factors of subsequent

memory performance (remembered and forgotten) and electrode sites. These electrodes were selected according to a priori Inhibitors,research,lifescience,medical expectations about a frontal distribution of the SME, as reported in the literature (cf. Otten et al. 2006, 2010). To assess the presence of an interaction between performance, condition (find more switch and stay) and time window (from −2 to −1 sec and from −1 to 0 sec) on the mean activity across the eight frontal electrodes, we Inhibitors,research,lifescience,medical have computed another ANOVA

for repeated measures with these three factors. Further analyses explored the SME in the stay condition and contrasted it with the switch condition and were based on methods that assess the significance of an ERP effect across the entire scalp. More precisely, we computed the amplitude differences in each condition and time window with the global field power (GFP) analyses Inhibitors,research,lifescience,medical that is a parametric assessment of map strength, computed as standard deviation of the momentary potential values and independent of topography (Lehmann and Skrandies 1980). The resulting amplitude differences indicate a different global strength in similar source distributions. In order to investigate the spatial distribution of the effects, we used TANOVAs (topographic analyses of variance)

applied to ERP data averaged across intervals and based on amplitude normalized maps. This was done to obtain a clear distinction Inhibitors,research,lifescience,medical between topographic effects and amplitude differences (e.g., Michel et al. 2009). A repeated measures TANOVA was performed in each condition and time window to analyze subsequent memory performance Inhibitors,research,lifescience,medical across the 64 electrodes sites. Based on randomization techniques, TANOVA is a powerful nonparametric test for the analysis of multichannel ERP data used to assess global dissimilarities between electric fields. This type of analysis corresponds to an ANOVA with all channels as repeated measures, but has the advantage that it considers all channels as a single entity avoiding a preselection of next electrodes, and does not require a correction for multiple testing across electrodes. Additionally, we have computed a post hoc TANOVA to assess the possible influence on the prestimulus SME of a third factor, instruction type (emotional and semantic) with the two factors already considered in the analyses namely conditions and performance. This factor was not considered in the main analyses for the lack of sufficient trials. Results Behavioral results At study, mean RTs were 1025 msec (SD = 157) for stay trials and 1078 msec (SD = 193) for switch trials. In line with the literature, RTs in hit trials were significantly shorter for stay than for switch trials (t(20) = −3.12, P < 0.