Skip to contents

MS-ReVE: build temporal nuisance RDMs from a spec

Source: R/temporal_rdms.R
msreve_temporal_confounds.Rd

Convenience helper that produces a named list of condition-level temporal nuisance RDMs for MS-ReVE analyses given an mvpa_design, a time index (trial onsets or order), and a compact specification. Each entry can be a kernel-based nuisance (delegates to temporal_nuisance_for_msreve) or an HRF-overlap-based nuisance built trial-wise then reduced to condition level.

Usage

msreve_temporal_confounds(
  mvpa_design,
  time_idx,
  spec,
  reduce = c("min", "mean", "median", "nn_min"),
  within_blocks_only = TRUE
)

Arguments

mvpa_design

an mvpa_design object

time_idx

numeric/integer time index per trial (seconds, TRs, or ordinal)

spec

a named list of entries. For kernel entries: include kernel and optional width,power,lambda,sigma,units,TR,metric. For HRF entries, set kind="hrf" and include TR plus optional durations,hrf,oversampling,length_s,similarity,metric.

reduce

reduction method for condition-level aggregation ("min","mean","median","nn_min")

within_blocks_only

logical; if TRUE ignore across-run pairs

Value

a named list of K x K matrices aligned to levels(mvpa_design$Y)

Examples

ds <- gen_sample_dataset(D = c(4, 4, 4), nobs = 24, nlevels = 4, blocks = 3)
spec <- list(lag = list(kernel = "exp", lambda = 3),
             hrf = list(kind = "hrf", TR = 2))
conf <- msreve_temporal_confounds(ds$design,
                                  time_idx = seq(0, by = 2, length.out = 24),
                                  spec = spec)
lapply(conf, dim)
#> $lag
#> [1] 4 4
#> 
#> $hrf
#> [1] 4 4
#>