Table 4.

Clustering of embedded task-based networks.

	Within/Between	Std Dev	p-value
	t-SNE with 99% of the variance
2-Back vs. Rest	0.7263	0.0178	0.0009
1-Back vs. Rest	0.9662	0.0210	0.0186
2-Back vs. 1-Back	1.0055	0.0110	0.1613
	PCA with top 2 components
2-Back vs. Rest	0.6728	-	0.0043
1-Back vs. Rest	0.9361	-	0.0068
2-Back vs. 1-Back	1.0011	-	0.0992
	t-SNE with top 2 components
2-Back vs. Rest	0.6117	<0.0001	0.0043
1-Back vs. Rest	0.9266	0.0032	0.0072
2-Back vs. 1-Back	0.9937	<0.0001	0.0829

The ratio of the embedding distance within tasks vs. the embedding distance between tasks. For the t-SNE embedding, each measure is an average of over 50 unique replications. For PCA there is only one possible embedding so there is no variance associated with the measures. Significant clustering was identified using P-values that were determined for each embedding using a permutation test. The average p-values over the 50 embedding replications are shown for t-SNE.