Table 3.

Source variables used for prediction analyses

Variables	Range	Description
behavioral response frequencies	$o_{i1},...,o_{i5}\in [0,1]$ $n_{i1},...,n_{i5}\in [0,1]$	proportion of old items replied to with 1, …, 5 and proportion of new items replied to with 1, …, 5
fMRI summary statistics	$y_{i1},...,y_{i4}\in \mathrm{ℝ}$	two scores (FADE-classic, FADE-SAME) computed from two fMRI contrasts (novelty processing, subsequent memory)
fMRI contrast images	$Y_i\in {\mathrm{ℝ}}^v$	voxel-wise fMRI contrasts computed in SPM, representing activations related to novelty processing (novel images – master images) or subsequent memory (PM with memory response)
resting-state fMRI maps	$Y_i\in {\mathrm{ℝ}}^v$	voxel-wise PerAF (mPerAF) computed using the REST toolbox, based on fMRI signals measured during a resting-state session
structural MRI maps	$Y_i\in {\mathrm{ℝ}}^v$	voxel-wise gray matter volumes computed in CAT12, based on each subject’s T1 image

FADE = functional activity deviation during encoding, SAME = similarities of activations during memory encoding, $\mathrm{ℝ}$ = real numbers, $v$ = number of (in-mask) voxels.