Compute ROI-by-ROI Representational Connectivity from Feature RSA Predictions

Forms an ROI x ROI similarity matrix by correlating lower-triangle predicted RDM vectors across ROIs. Sparsification, when requested, is applied only to the final ROI x ROI matrix and never to the per-ROI RDM vectors themselves.

Usage

feature_rsa_connectivity(
  x,
  method = c("spearman", "pearson"),
  keep = 1,
  absolute = FALSE,
  use = "pairwise.complete.obs",
  verbose = FALSE
)

Arguments

x

Either a regional_mvpa_result produced by feature_rsa_model(..., return_rdm_vectors=TRUE) or the tibble returned by feature_rsa_rdm_vectors(). Regional results may store RDM vectors either in-memory or in file-backed batches written by run_regional(..., save_rdm_vectors_dir = ...).

method

Correlation method used across ROI RDM vectors, one of "spearman" or "pearson".

keep

Proportion of ROI-ROI edges to retain after optional sparsification. keep = 1 disables sparsification. For example, keep = 0.1 retains the top 10% of finite off-diagonal edges.

absolute

Logical; when TRUE, rank edges by absolute magnitude during sparsification. Defaults to FALSE.

use

Missing-value handling passed to cor.

verbose

Logical; if TRUE, emit block-level progress messages while connectivity is being computed.

Value

A symmetric numeric matrix with ROIs in rows/columns.

Examples

# Usually `vecs` comes from feature_rsa_rdm_vectors(run_regional(...));
# here a toy table of per-ROI lower-triangle RDM vectors is used.
set.seed(1)
vecs <- tibble::tibble(
  roinum = 1:4,
  rdm_vec = replicate(4, as.numeric(dist(matrix(rnorm(30), 10, 3))),
                      simplify = FALSE)
)
conn <- feature_rsa_connectivity(vecs, method = "spearman", keep = 0.5)
round(conn, 2)
#>      1     2     3     4
#> 1 1.00  0.00  0.14  0.00
#> 2 0.00  1.00 -0.02 -0.02
#> 3 0.14 -0.02  1.00  0.00
#> 4 0.00 -0.02  0.00  1.00