Table 1.

Proposed Analytic Permutations: 360 Total Modeling Combinations for MID task

First-level Pipeline Decisions	Options
A. Smoothing (FWHM)
1. 1.5x voxel	ON / OFF
2. 2x voxel	ON / OFF
3. 2.5x voxel	ON / OFF
4. 3x voxel	ON / OFF
5. 3.5x voxel	ON / OFF
B. Motion Correction
1. None	ON / OFF
2. Regress: Translation/Rotation (x,y,z) + Derivative (x,y,z)	ON / OFF
3. Regress: Regress: Translation/Rotation (x,y,z) + Derivative (x,y,z) + First 8 aCompCor Components	ON / OFF
4. Regress: Translation/Rotation (x,y,z) + Derivative (x,y,z) + First 8 aCompCor Components + Censor High Motion Volumes (FD ≥ .9)	ON / OFF
#5. Regress: Translation/Rotation (x,y,z) + Derivative (x,y,z) + First 8 aCompCor Components, Exclude mean FD ≥ .9	ON / OFF
#6. Regress: Translation/Rotation (x,y,z) + Derivative (x,y,z) + First 8 aCompCor Components + Censor High Motion Volumes, Exclude mean FD ≥ .9	ON / OFF
C. Task Modeling
1. MID: Cue Onset, Cue Duration only	ON / OFF
2. MID: Cue Onset, Cue + Fixation Duration	ON / OFF
3. MID: Fixation onset, Fixation Duration	ON / OFF
D. Task Contrasts
1. MID: Big Win > Neutral	ON / OFF
2. MID: Big Win > Implicit	ON / OFF
3. MID: Small Win > Neutral	ON / OFF
4. MID: Small Win > Implicit	ON / OFF

Blue text: Model hypothesized to produce the highest test-retest reliability; aCompCor: Anatomical Component Based Noise Correction; MID: Monetary Incentive Delay task; FD: Framewise displacement.

^#Due to the lack of low motion subjects (zero mean FD <.90 in 2/3 samples), this decision was not included in the Stage 2 analyses, resulting in 240 analytic models.