Resting state connectivity predicts implicit sequence learning performance: a replication and extension

Abstract

PARTICIPANTS § 33 young adults ages 18-34 (M±SD = 20.7 ± 2.6) from Washington, DC metro area § 2 participants subsequently eliminated from all analyses due to image artifact or meeting study drug exclusion criteria (final sample: N = 31) fMRI DATA ACQUISITION & PREPROCESSING § Subjects scanned in resting state (eyes open) for 11 minutes in 3T Siemens Trio magnet § 264 whole brain images, gradient EPI acquisition (TR = 2500ms, TE = 30ms, 90oflip angle, FOV = 192 x192mm, voxel size = 3 mm isotropic) § Data preprocessing in SPM8 (Slice Time Correction/Realignment, Normalization to EPI template, Resliced to 3mm, Smoothing {8mm FWHM}, Band-pass filtering, and “scrubbing” (Power et al., 2012) with at least 200 volumes retained per participant BEHAVIORAL TESTING (see top middle panel) § Took place in scanner immediately following the resting scan § 3 sessions (720 trials) of the Triplets Learning Task (TLT; Howard, Howard, Dennis, & Kelly, 2008), modified for fMRI § Computer-based recognition task and interview to assess explicit awareness RESTING STATE FUNCTIONAL CONNECTIVITY CALCULATIONS § Bilateral dorsal caudate (DC) seed created in Marsbar as two spheres (R and L hemispheres) of radius 6mm, centered around coordinates taken from Di Martino et al. (2008) Seed based Voxelwise resting state connectivity: § For each subject, partial correlations computed between timecourse of DC seed and those of every other voxel in brain, controlling for motion (from realignment parameter timecourses) and physiological noise (approximated by timecourses of WM and CSF); resulting r-values converted to Z-scores before statistical analyses of correlation strengths § Produced brain map of intrinsic connectivity strength with the DC for each subject Correlations with SL performance: § Single-subjects’ whole brain connectivity maps entered as the DV in regression models (SPM8) testing for correlations with SL performance within a medial temporal lobe (MTL) region of interest, as well as across the whole brain § Mean framewise displacement included in models as a covariate of no interest § Whole Brain results corrected for multiple comparisons at P 54) Resting State Connectivity Predicts Implicit Sequence Learning Performance: A Replication and Extension