Figure 3.

Heart-Rate and EEG Responses to Outcomes

$n=10$ subjects.

(A) Similarity between heart-rate responses recorded in different sessions following different types of outcomes. Similarity was computed as the average temporal (Pearson) correlation between heart-rate responses for six types of outcomes: reward and no-reward outcomes following choices of images associated with a 0.25, 0.50, or 0.75 reward probability. Similarity was computed separately for each subject and then averaged across subjects.

(B) Similarity between EEG responses recorded in different sessions following different types of outcomes. See Figure S3 for time courses of heart-rate and EEG responses for exemplar subjects.

(C) Reward prediction errors (RPEs) of the fast- and slow-learning processes were decoded from the physiological response to outcomes. The y axis denotes decoding accuracy, computed as the correlation between decoded and actual values. RPEs were derived using the model (see Figure 2) and decoded with cross-validated support vector regression (see STAR Methods).

(D) Correlation between actual and decoded outcomes and between actual and decoded expectations for the slow and fast processes. In (C) and (D), circles correspond to individual subjects. Shaded areas: SEM (dark) and SD (light).

^∗: $p_{\text{permutation}}<0.01$, NS: $p_{\text{permutation}}=0.2$.