Fig. 1: Classification analyses based on neural activations of the ‘threat circuit’. | Nature Communications

Fig. 1: Classification analyses based on neural activations of the ‘threat circuit’.

From: Distributed neural representations of conditioned threat in the human brain

Fig. 1

A Cross-validation performance on the discovery dataset (n = 425). We divided each experiment phase into 4 trial-blocks, and separately did cross-validation on each trial-block. The vertical dash lines separate trial-blocks of different phases. The accuracies are based on forced-choice test. The top of the gray-shaded area indicates the threshold of significant accuracy based on the permutation test (p < 0.05, one-sided). B, C Generalization performance on the validation dataset 1 (n = 98) and the validation dataset 2 (n = 127). The classifiers were trained based on the discovery dataset, and directly applied to the validation datasets. Error bars in (AC) are SEM for accuracies. D Predictive patterns of the nodes of the ‘threat circuit’. A voxel with red/blue color indicates that this voxel is more activated to CS + /CS-. For the extinction learning and extinction memory recall phase, only predictive patterns of the first trial-block were shown because the other trial-blocks did not show robust classification performance. For the memory recall phase, CS + E vs. CS- and CS + U vs. CS- were shown separately. Cond. Conditioning, Ext. Extinction learning, Rec. Extinction memory recall.

Back to article page