Fig. 2: Brain states represent coactivation within and between resting state functional networks.

a Brain states defined as the centroids of clusters identified using an unsupervised machine-learning algorithm applied to rest and n-back task fMRI data. Brain states are labeled based on cosine similarity with a priori resting state functional networks (RSNs)43. The top label corresponds to the RSN with the most overall similarity, and the bottom two labels separated by a forward slash reflect the RSNs with the most similarity to the positive and negative components of each state, respectively. b Cosine similarity between positive (black) and negative (red) components of each state with binary state vectors corresponding to a priori definitions of RSNs43. Larger radial values correspond to higher cosine similarity. DAT dorsal attention network, DMN default mode network, FPN frontoparietal network, LIM limbic network, SOM somatomotor network, VAT ventral attention network, and VIS visual network.