Preprocessed fMRI, behavioral, and questionnaire data from self-other-touch, self-referential, & agency tasks

Description

Methods

Participants

Participants were recruited based on the criteria as a control sample for complicated grief and had to be 18 years of age or older and in good health. Participants were excluded if they had any psychiatric, neurological, or neurodevelopmental disorder, any other clinical diagnosis of significance, use of psychoactive medication, claustrophobia, and any other MRI contraindication. The included sample had a mean age of 37.7 ± 14.7 (age range: 18-74, females = 20). All participants provided consent prior to enrolment, and the study was performed in accordance with the declaration of Helsinki.

General Methodology

The study consisted of two parts, which could occur in any order on separate visits or consecutively on the same day. These parts were a neuroimaging and a behavioral session. During the imaging session participants completed a self-touch task and self-referential thinking task. During the behavioral session, participants completed an implicit binding agency task. Participants also filled in the generalized self-efficacy scale¹, GSES, and the self-concept clarity scale², SCCS.

fMRI Data
Procedure
During the imaging session, participants completed an earlier validated self-other touch task³ .  Participants stroked their own left forearm with their right arm (self-touch), stroked a pillow placed next to their forearm (object-touch; used as an offline control for movement), or had their left forearm stroked by the experimenter (other-touch). The current study expanded the task to also include simultaneously touching the object while being touched by the experimenter (TWT), which we used previously elsewhere⁴. Participants completed a single repetition of each condition and verbally rated pleasantness from zero to ten in the scanner prior to imaging. During imaging, each condition was repeated eight times in randomized order. A 3 s cue indicated the upcoming condition and then touch occurred for 12 s.

Participants also completed a self-referential thinking task⁵. Participants were instructed to consider if an adjective appropriately described a person or not. The person could be themselves (self), a close loved one of the participant’s choosing, the name of whom was used in the task (other-close), or the Swedish king (other-distant). As a control, participants had to consider whether adjectives contained exactly two syllables or not (syllable). The condition cue was displayed for 3 s, followed by 5 pseudorandom adjectives presented under the cue for 3 s each (total block duration 18 s). Answers were provided by button press. Each condition was repeated 6 times in a randomized order, and each block contained 5 pseudorandom adjectives.

Imaging

Imaging data were acquired using 3.0 Tesla scanner (Prisma, Siemens) with a 64-channel head coil. For anatomical images, a T1 weighted scan was collected using the following settings: repetition time = 2300 ms; echo time = 2.36; flip angle = 8°; field of view = 288 x 255 mm²; voxel resolution = 0.87 x 0.87 x 0.90 mm. Task images were acquired using echo planar imaging (EPI) with the following settings: repetition time = 1030 ms; echo time = 30 ms; slice thickness = 3 mm; matrix size = 64 x 64; field of view = 192 x 192 mm; in-plane voxel resolution = 3 mm²; flip angle = 8°.

Imaging data were processed and analyzed with MATLAB 2021b using SPM12 (Wellcome Department of Imaging Neuroscience). Functional images were realigned and resliced to the mean EPI image using rigid-body motion correction. The anatomical T1-weighted image was coregistered to the mean EPI image using normalized mutual information and segmented with SPM tissue probability maps. Spatial normalization to MNI space was estimated from the segmentation and applied to all functional volumes, resampling images to 2 mm isotropic voxels. Normalized functional images were then spatially smoothed with a 6-mm FWHM isotropic Gaussian kernel.

Behavioral data (Implicit binding task)

Procedure

The implicit binding task was identical to the one used by Imaizumi and Tanno⁹. Participants were seated in front of a screen with a fixation cross. The task consisted of two blocks, action and baseline. During the baseline block, 2 tones (first at 250hz, second at 440hz) were presented, with a random delay between them. Participants were instructed that the delay could vary freely from 0 to 1000 ms (i.e. 1 second). Participants were asked to verbally estimate the delay between the tones. In the action block, participants instead pressed spacebar. After a short random delay, a single tone was presented (250hz), and participants were asked to estimate the time between button press and tone presentation. As in the baseline block, participants were instructed that the delay could vary freely from 0 to 1000 ms (i.e. 1 second). Participants were also asked to rate how much they felt as though their button press caused the tone. This was rated on a 1-9 likert scale centered on 5. After answering, a fixation crossed was presented. The delays were fixed at 0.1, 0.3, 0.5, 0.7 or 0.9 seconds in both block types. The delays were presented in a random order. Each delay was repeated 5 times per block, and each block therefore consisted of 35 trials. Each block was repeated twice in an alternating order, counterbalanced between participants. Prior to starting the task, the participant completed a short training to ensure adequate understanding of the task.

Data processing

Data was processed on a trial-by-trial basis. Relative estimates (RE) were calculated by dividing the time estimate by the delay (i.e. RE >1 indicates overestimation of time, RE <1 indicates underestimation of time. Logarithmic scale). A linear regression was calculated on for each participant to predict RE by condition. This was done using MATLAB’s fitlm with the formula RE ~ condition, and the slope (i.e. beta-parameter of the linear regression), was used as a measure of the effect of condition, i.e. strength of implicit binding. For explicit agency, a similar approach was taken, however only considering the active block. Data was analyzed on a trial-by-trial basis within participants. A linear regression was calculated using MATLABs fitlm, where delay was used to predict agency ratings (Agency ~ Delay). This beta-parameter was used to quantify explicit agency. 

USAGE NOTES

Imaging data can be used for further imaging analysis including extracting BOLD signal activity or functional connectivity analyses. First and second level analyses can be directly performed on the swrEPI files using SPM (Wellcome Department of Imaging Neuroscience) in Matlab, or FSL or functional connectivity using software CONN (www.conn-toolbox.org). It suited to study the neural bases of self-processing, trait-judgment, and touch in healthy humans. We emphasize that that our dataset is representative of a specific population cohort, namely Swedish, without any previous loss of a close relative in the last 5 years of life and without any psychiatric self-processing disorders.

The log files include information about the timing of the events in both tasks . Experimental conditions can be reconstructed directly from the Presentation® task log files using the event codes recorded in the Code column.

For the touch task, condition onsets correspond to the events activeself_start (self-touch), activeobject_start (object-touch), passivestroke_start (other- touch), and twt_start (touch while being touched condition). The corresponding instruction periods are marked by cue_activeself, cue_activeobject, cue_passivestroke, and cue_twt, respectively. Event timings should be referenced to the first MRI scanner trigger (Pulse, code 99) and converted from the logged time units to seconds. In the original first-level analyses, cue events were modelled with a duration of 3 s and touch events with a duration of 12 s. In addition, a motion regressor was modelled as a 1 s event immediately following the active conditions (touch_self, touch_object, and touch_twt) to account for movement-related effects associated with task performance.

For the self-referential thinking task, condition onsets correspond to the events Själv (Self judgments), Kungen (judgments about the Swedish king), CLOSE_OTHER (judgments about a personally familiar individual; original names were replaced for anonymity), and Stavelser (syllable-processing control condition). Event timings should be referenced to the first MRI scanner trigger (Pulse, code 99) and converted from the logged time units to seconds. In the original first-level analyses, cue events were modelled with a duration of 3 s. Reflection periods began 3 s after cue onset and were modelled as 15 s blocks corresponding to the conditions think_self, think_king, think_love (CLOSE_OTHER), and think_syll.

All raw agency files are provided. Each file corresponds to one subject. Within the files, each row corresponds to a trial, with the condition (action/baseline) in the first column, the delay (in seconds) in the second column, the subjective agency in the third column (if action block. If baseline block, this field contains NaN), and the guess (in ms) in the fourth column.

The demographic file contains the subject age, gender, their processed implicit and explicit agency scores, the total scores of the questionnaires (GSES & SCCS), as well as subjective pleasantness ratings for the touch conditions during the fMRI procedure (as described above). 

Notes