01, two-way ANOVA). For robustness, we also repeated all analyses using bootstrap methods for bin-wise significance (Figures S3D and S3E) (Paz et al., 2006, 2009). We conclude that there is a double dissociation in the roles of the amygdala and the dACC in mediating acquired aversive responses for partial and continuous reinforcement schedules. Selleck KRX 0401 We next turned to examine the

correlations between simultaneously recorded pairs of amygdala and dACC neurons (n = 483 pairs) and tested the hypothesis that altered correlation patterns at acquisition underlie the delayed extinction. Correlation matrices at all delays (in 25 ms bins) were computed for each of ten trials from the habituation and the acquisition phases and significance was assessed per bin (Figures 4 and S4) (Paz et al., 2006). The mean number of significant bins was then normalized by the distribution of significant bins in the shuffled data of the corresponding pair, producing a correlation-factor index. This did not buy PD0332991 differ during habituation between ConS and ParS (p > 0.3, t test). During acquisition, we found that ConS induced an early sharp increase in correlations that diminished to habituation levels in parallel to reaching plateau level of behavior. In contrast, ParS induced an increase that remained high throughout the whole acquisition stage,

even much after behavioral plateau was achieved (Figures 5A and 5B), and this dynamic was evident in the correlation-factor index (Figure 5C, p < 0.001; interaction in two-way ANOVA). When the correlation factor was inspected separately above and below the diagonal, the same effects were observed (Figure 5C, insets, p < 0.05, interaction in two-way ANOVA), suggesting that it is an overall increase in reciprocal interactions between the two regions. For robustness, we repeated the analyses using more conventional statistical tests to assess bin significance (Pearson statistics) and revealed the those same dynamics (Figure S5A, p < 0.001 in two-way ANOVA).

Notice that analyses are performed only on reinforced trials; hence, the number of preceding reinforced trials is identical for each data point in the comparison between ParS and ConS. Finally, in accordance with the distribution of projections from dACC (area 24) to the amygdala (Stefanacci and Amaral, 2002), we found more significant correlations at medial penetrations to the amygdala than at lateral ones (Figure S5B; p < 0.01, main effect in two-way ANOVA). If the sustained correlations at the end of the acquisition indeed support a more resistant memory, then their magnitude should somehow predict the time it takes to extinguish. To test this, we compared the mean correlation-factor index with the memory-persistence index (as in Figure 1G) on an individual session basis.