A, Temporal generalization of distractor location classification. Results are presented as AUC. Significant clusters (p < 0.05) computed over the entire epochs (0–2.5 s) are marked by solid borders. Significant clusters (p < 0.05) computed over the presearch time window (0–1.5 s) are marked by dotted borders. B, Temporal generalization between presearch (0–1 s) and postsearch (1.5–2.5 s) time windows. The purple curve represents results obtained by training the decoder on the presearch time window and testing on the postsearch window (Fig. 3A, purple-dotted area). For this curve, graph origin reflects onset of a time interval beginning at onset of the placeholder display. The green curve represents results obtained by training the decoder on the postsearch window and testing on the presearch time window (Fig. 3A, green-dotted area). For this curve, graph origin reflects onset of a time interval beginning at the onset of the search display. Results are presented as AUC with error bars depicting 95% CI across subjects. Significant clusters (p < 0.05) are indicated by horizontal solid lines. Asterisks denote clusters with p < 0.10. Horizontal dashed line indicates the chance level. C, Raincloud plot of temporal generalization centered at 100 and 200 ms delay for presearch to postsearch (purple) and postsearch to presearch (green). Each condition displays the full data distribution, individual datapoints, and the mean with error bars indicating the 95% confidence interval. Horizontal dashed line indicates the chance level.