Our intention is provide a credible demonstration of the emotional impact that a researcher might expect when using a sample size that is typical of human fMRI studies (Szucs and Ioannidis, 2017). Twenty-four members of the University of Münster community participated in the fMRI experiment. All participants gave written informed consent, as approved by the University of Münster Human Subjects Review Board. One participant’s data was excluded due to excessive head motion, leaving 23 participants in the final sample (11 female, mean age 26.9 years, SD 3.8 years). All participants reported no neurological abnormalities and had normal or corrected-to-normal vision. Participants received 50 Euros compensation.

In brief, the functional imaging protocol involves a passive scene viewing period of 5 min in the scanner. Participants are asked to maintain central fixation on a projection screen while the scene stimuli are presented in a mixed, event-related design.

Participants were fitted with earplugs, headphones, and given a patient-alarm squeeze ball prior to entering the bore of the scanner. Padding inside the head coil and explicit verbal instruction were used to limit head motion. After a 5-min structural volume was collected, a 5-min functional scan was collected. Participants were instructed to attend to the series of 20 emotional and neutral scenes, while maintaining fixation on a red fixation dot at the center of the screen. The scenes were rear-projected onto a screen visible through a coil-mounted mirror. Following the 20 scene series, participants left the scanner room and reviewed each scene on a laptop computer, while providing their ratings of pleasantness and emotional arousal using the Self-Assessment Manikin (SAM; Bradley and Lang, 1994). Participants then received transcranial direct current stimulation, and viewed two additional scene series while functional imaging data was collected (described in Junghöfer et al., 2017).

Participants viewed an event related series of 20 natural scene photographs presented in 256 levels of grayscale, at 800 × 600 resolution, over a 30° horizontal field of view. The stimuli depict 8 pleasant (happy children and families, erotica), 4 neutral (people in daily activities, land, and cityscapes), and 8 unpleasant scenes (threatening people and animals, bodily injuries). All scene stimuli included a central fixation point and were balanced across emotional and neutral content to be statistically equivalent in luminosity and complexity as measured by the 90% quality Joint Photographic Experts Group (JPEG) file size using GIMP 2.8.¹

The scene series began with a 2 s checkerboard acclimation image, followed by the 20 experimental scenes, presented for 2 s each, in pseudorandom order, separated by interstimulus intervals of 10 or 12 s. The total acquisition time was 4 min and 42 s, including 5 dummy acquisitions (10 s) prior to the first trial to allow MR signal stabilization.

In the original experiment, the scenes were presented using the Psychophysics Toolbox (Brainard, 1997) which runs in Matlab (Mathworks, Inc.). To enable researchers to avoid a dependence on a Matlab license, the presentation paradigm was rewritten using open-source Psychopy software (Peirce et al., 2019). The code was created with the “builder” graphic user interface such that it can be simple to use and adjust for individual research applications. The code includes a generic trigger wait loop to enable synchronization with the first functional image acquisition, which will need to be set appropriately for each scanner and chosen image repeat time (the default is a 2 s TR). In addition, the scene presentation order is randomized for each run, with a requirement that a specific scene content (e.g., pleasant, neutral, unpleasant) is not repeated twice in succession. This process occurs before the experiment is fully loaded. There is also a means of tracking and correcting the slight time variation (∼1–2 ms) that occurs as each scene is loaded into video memory. This variation is corrected in the following inter trial interval, such that each trial has the same duration, and the total presentation series ends on time. The scene presentation order and onset/offset times are recorded in a log file for each run.

Using a Siemens Prisma 3T MR scanner (Siemens Healthcare, Erlangen, Germany), a T1-weighted structural volume was collected consisting of 192 sagittal slices with a 256 × 256 matrix and 1 mm³ isotropic voxels. The functional prescription covered the whole brain with 30 interleaved axial slices with 3.5 mm³ isotropic voxels (64 × 64 gradient echo EPI, 224 mm FOV, 3.5 mm thickness, no gap, 90° flip angle, 30 ms TE, 2000 ms TR). The functional time series was slice-time corrected using cubic spline interpolation, spatially smoothed across 2 voxels (7 mm full width at half maximum), linearly detrended and high-pass filtered at 0.02 Hz. Structural and co-registered functional data were resampled into 1 mm³ voxels and transformed into standardized Talairach coordinate space, using BrainVoyager (Brain Innovation).²

A repeated measures ANOVA using a 2-gamma hemodynamic response function (Friston et al., 1998) convolved with scene presentations assessed the effect of emotional scene content on brain activation. The resulting group-averaged BOLD signal map was thresholded at a false discovery rate of p < 0.01 (Genovese et al., 2002) and a minimum cluster size of 250 mm³. Our goal in this paper was to enhance the reliability of our BOLD activation results, thus we chose a relatively large ROI volume to enable the averaging of roughly 70 voxels at raw acquisition size. From each participant’s data, 5 regions of interest (ROI) were sampled, the locations of which were derived from prior studies employing this paradigm (Frank and Sabatinelli, 2014, 2019; Sambuco et al., 2020) as well as 3 large meta-analyses of visual emotional processing (Kober et al., 2008; Liu et al., 2011; Sabatinelli et al., 2011). These ROIs are not intended to represent an complete list of all brain regions that show emotional sensitivity, but instead are used to demonstrate the effects sizes of BOLD signal differences that may be expected when employing this paradigm and protocol. These regions include bilateral amygdala (±22, −7, −12), fusiform gyrus (FG; ±44, −50, −14), lateral occipital cortex (LOC; ±45, −70, 2), inferior frontal gyrus/anterior insula (IFG; ±34, 19, 6), and the medial dorsal nucleus of the thalamus (MDN; 0, −19, 5). A 100 mm³ sample of BOLD signal was extracted from each ROI location for each participant. From these functional time series, BOLD signal change scores were calculated using the average percent signal change 4–8 s following scene onset, deviated from the 2 s pre-stimulus baseline.

Analysis of scene content effects (Figure 1) were assessed with repeated measures ANOVA across pleasant, neutral and unpleasant scenes, applying corrections for violations of sphericity where necessary. Effect sizes will be listed in generalized η² (where 0.01 indicates a small effect, 0.06 indicates a medium effect, and 0.14 indicates a large effect) and in Cohen’s d for ease of interpretation in Figure 2.

As expected, scene content significantly modulated ratings of valence (F(2,44) = 168.71, p < 0.001, η_G² = 0.818) and ratings of arousal (F(2,44) = 99.86, p < 0.001, η_G² = 0.741). On scales of 1–9 (standard error), the sample rated pleasant scenes at 7.12 (0.181), followed by neutral scenes at 6.70 (0.184), followed by unpleasant scenes at 2.99 (0.175). On a scale of 1–9, the sample rated unpleasant scenes as most arousing at 6.90 (0.196) followed by pleasant scenes at 5.53 (0.203) and neutral scenes at 2.87 (0.235).

Shown in Figures 2, 3, emotional scene perception significantly modulated BOLD signal in amygdala (F(2,44) = 6.34, p < 0.01, η_G² = 0.138), fusiform gyrus (F(2,44) = 12.94, p < 0.001, η_G² = 0.129), lateral occipital cortex (F(2,44) = 30.10, p < 0.001, η_G² = 0.337), inferior frontal gyrus/anterior insula (F(2,44) = 5.47, p < 0.01, η_G² = 0.127), and medial dorsal nucleus of the thalamus (F(2,44) = 5.83, p < 0.01, η_G² = 0.172).

Consistent with much prior work, both pleasant and unpleasant scenes elevated BOLD signal compared to neutral scene perception in all regions. To provide an estimate of the strength of emotional reactivity in each ROI, Cohen’s d effect size was calculated for BOLD signal enhancement by emotional (pleasant and unpleasant) relative to neutral scene perception, as shown in Figure 2. These effect sizes were quite strong, ranging from medium to large in FG (0.66), large in IFG (0.78), large in amygdala (0.82), large in MDN (1.04), and very large in LOC (1.47).