Magic, Memory, and Curiosity (MMC) fMRI Dataset

## Overview
<ul>
<li>The Magic, Memory, Curiosity (MMC) dataset contains data from 50 healthy human adults incidentally encoding 36 videos of magic tricks inside the MRI scanner across three runs.</li>
<li>Before and after incidental learning, a 10-min resting-state scan was acquired.</li>
<li>The MMC dataset includes contextual incentive manipulation, curiosity ratings for the magic tricks, as well as incidental memory performance tested a week later using a surprise cued recall and recognition test .</li>
<li>Working memory and constructs potentially relevant in the context of motivated learning (e.g., need for cognition, fear of failure) were additionally assessed.</li>
</ul>

## Stimuli
The stimuli used here were short videos of magic tricks taken from a validated stimulus set (MagicCATs, Ozono et al., 2021) specifically created for the usage in fMRI studies. All final stimuli are available upon request. The request procedure is outlined in the Open Science Framework repository associated with the MagicCATs stimulus set (https://osf.io/ad6uc/).

## Participant responses
Participants’ responses to demographic questions, questionnaires, and performance in the working memory assessment as well as both tasks are available in comma-separated value (CSV) files. Demographic (MMC_demographics.csv), raw questionnaire (MMC_raw_quest_data.csv) and other score data (MMC_scores.csv) as well as other information (MMC_other_information.csv) are structured as one line per participant with questions and/or scores as columns. Explicit wordings and naming of variables can be found in the supplementary information. Participant scan summaries (MMC_scan_subj_sum.csv) contain descriptives of brain coverage, TSNR, and framewise displacement (one row per participant) averaged first within acquisitions and then within participants. Participants’ responses and reaction times in the magic trick watching and memory task (MMC_experimental_data.csv) are stored as one row per trial per participant.

## Preprocessing

Data was preprocessed using the AFNI (version 21.2.03) software suite. As a first step, the EPI timeseries were distortion-corrected along the encoding axis (P>>A) using the phase difference map (‘epi_b0_correct.py’). The resulting distortion-corrected EPIs were then processed separately for each task, but scans from the same task were processed together. The same blocks were applied to both task and resting-state distortion-corrected EPI data using afni_proc.py (see below): despiking, slice-timing and head-motion correction, intrasubject alignment between anatomy and EPI, intersubject registration to MNI, masking, smoothing, scaling, and denoising. For more details, please refer to the data descriptor (**LINK**) or the Github repository (https://github.com/stefaniemeliss/MMC_dataset).

afni_proc.py -subj_id "${subjstr}" \
-blocks despike tshift align tlrc volreg mask blur scale regress \
-radial_correlate_blocks tcat volreg \
-copy_anat $derivindir/$anatSS \
-anat_has_skull no \
-anat_follower anat_w_skull anat $derivindir/$anatUAC \
-anat_follower_ROI aaseg anat $sswindir/$fsparc \
-anat_follower_ROI aeseg epi $sswindir/$fsparc \
-anat_follower_ROI FSvent epi $sswindir/$fsvent \
-anat_follower_ROI FSWMe epi $sswindir/$fswm \
-anat_follower_ROI FSGMe epi $sswindir/$fsgm \
-anat_follower_erode FSvent FSWMe \
-dsets $epi_dpattern \
-outlier_polort $POLORT \
-tcat_remove_first_trs 0 \
-tshift_opts_ts -tpattern altplus \
-align_opts_aea -cost lpc+ZZ -giant_move -check_flip \
-align_epi_strip_method 3dSkullStrip \
-tlrc_base MNI152_2009_template_SSW.nii.gz \
-tlrc_NL_warp \
-tlrc_NL_warped_dsets $sswindir/$anatQQ $sswindir/$matrix $sswindir/$warp \
-volreg_base_ind 1 $min_out_first_run \
-volreg_post_vr_allin yes \
-volreg_pvra_base_index MIN_OUTLIER \
-volreg_align_e2a \
-volreg_tlrc_warp \
-volreg_no_extent_mask \
-mask_dilate 8 \
-mask_epi_anat yes \
-blur_to_fwhm -blur_size 8 \
-regress_motion_per_run \
-regress_ROI_PC FSvent 3 \
-regress_ROI_PC_per_run FSvent \
-regress_make_corr_vols aeseg FSvent \
-regress_anaticor_fast \
-regress_anaticor_label FSWMe \
-regress_censor_motion 0.3 \
-regress_censor_outliers 0.1 \
-regress_apply_mot_types demean deriv \
-regress_est_blur_epits \
-regress_est_blur_errts \
-regress_run_clustsim no \
-regress_polort 2 \
-regress_bandpass 0.01 1 \
-html_review_style pythonic

## Derivatives
The **anat** folder contains derivatives associated with the anatomical scan. The skull-stripped image created using @SSwarper is available in original and ICBM 2009c Nonlinear Asymmetric Template space as sub-[group][ID]_space-[space]_desc-skullstripped_T1w.nii.gz together with the corresponding affine matrix (sub-[group][ID]_aff12.1D) and incremental warp (sub-[group][ID]_warp.nii.gz). Output generated using @SUMA_Make_Spec_FS (defaced anatomical image, whole brain and tissue masks, as well as FreeSurfer discrete segmentations based on the Desikan-Killiany cortical atlas and the Destrieux cortical atlas) are also available as sub-[group][ID]_space-orig_desc-surfvol_T1w.nii.gz, sub-[group][ID]_space-orig_label-[label]_mask.nii.gz, and sub-[group][ID]_space-orig_desc-[atlas]_dseg.nii.gz, respectively.
<br><br>
The **func** folder contains derivatives associated with the functional scans. To enhance re-usability, the fully preprocessed and denoised files are shared as sub-[group][ID]_task-[task]_desc-fullpreproc_bold.nii.gz. Additionally, partially preprocessed files (distortion corrected, despiked, slice-timing/head-motion corrected, aligned to anatomy and template space) are uploaded as sub-[group][ID]_task-[task]_run-[1-3]_desc-MNIaligned_bold.nii.gz together with slightly dilated brain mask in EPI resolution and template space where white matter and lateral ventricle were removed (sub-[group][ID]_task-[task]_space-MNI152NLin2009cAsym_label-dilatedGM_mask.nii.gz) as well as tissue masks in EPI resolution and template space (sub-[group][ID]_task-[task]_space-MNI152NLin2009cAsym_label-[tissue]_mask.nii.gz).
<br><br>
The **regressors** folder contains nuisance regressors stemming from the output of the full afni_proc.py preprocessing pipeline. They are provided as space-delimited text values where each row represents one volume concatenated across all runs for each task separately. Those estimates that are provided per run contain the data for the volumes of one run and zeros for the volumes of other runs. This allows them to be regressed out separately for each run. The motion estimates show rotation (degree counterclockwise) in roll, pitch, and yaw and displacement (mm) in superior, left, and posterior direction. In addition to the motion parameters with respect to the base volume (sub-[group][ID]_task-[task]_label-mot_regressor.1D), motion derivatives (sub-[group][ID]_task-[task]_run[1-3]_label-motderiv_regressor.1D) and demeaned motion parameters (sub-[group][ID]_task-[task]_run[1-3]_label-motdemean_regressor.1D) are also available for each run separately. The sub-[group][ID]_task-[task]_run[1-3]_label-ventriclePC_regressor.1D files contain time course of the first three PCs of the lateral ventricle per run. Additionally, outlier fractions for each volume are provided (sub-[group][ID]_task-[task]_label-outlierfrac_regressor.1D) and sub-[group][ID]_task-[task]_label-censorTRs_regressor.1D shows which volumes were censored because motion or outlier fraction exceeded the limits specified. The voxelwise time course of local WM regressors created using fast ANATICOR is shared as sub-[group][ID]_task-[task]_label-localWM_regressor.nii.gz.