Fig. 4: Data-driven clustering of task-related neural activity.
From: Human response times are governed by dual anticipatory processes with distinct neural signatures
Illustration of hierarchical clustering of electrodes based on similar task-related activity modulations. Top row: time intervals used to measure task modulation and RT correlation for each electrode as detailed in main text. Bottom row: left, colormap representing a feature matrix across all electrodes, where each row represents an electrode, and each column represents a feature (task or RT modulation); middle: dendrogram representing similarities between electrodes; right: feature matrix re-organized based on similar task and RT modulations at the clustering level indicated by the black line in middle panel, which we used in these analyses. B We identified a level of clustering (black vertical line in dendrogram) based on an objective function that maximized the number of clusters that were well sampled in our dataset (i.e., each eligible cluster was observed in all participants and consisted of at least 200 electrodes). C Percentage of electrodes that showed task-related increases and decreases (positive and values, respectively as in Fig. 3A). D Same as (C) but for RT correlations. Negative values indicate increased activity with faster RTs; positive values indicate increased activity with slower RTs (as in Fig. 3C).
