A comprehensive literature search was conducted using MEDLINE (PubMed), PsycARTICLES (EBSCOhost), EMBASE (Ovid), and Web of Science, searching for combinations of keywords: “emotion regulation”, “affective regulation”, “distraction”, “selective attention”, “reappraisal”, “reinterpretation”, “cognitive change”, “fMRI”, “neuroimaging”, “functional magnetic resonance imaging”, and “functional MRI”. The search was limited to (a) the English language, (b) fMRI studies on reinterpretation and distraction in (c) healthy human samples (d) published until March 2020. In addition, several studies were selected by manual searches from review article reference lists. If the same neuroimaging data were used in more than one study, only the first reported study was included. If the data we needed were incomplete or absent, we contacted the corresponding authors for the complete data. The literature search resulted in 1741 articles after removing duplicates (Fig. 1).

Fig. 1.

PRISMA flow diagram. PRISMA, Preferred reporting items for systematic reviews and meta-analyses (http://www.prismastatement.org/).

We included studies comparing blood oxygen level-dependent (BOLD) responses during reinterpretation and distraction tasks among healthy volunteers. Specifically, participants were presented with negative images from the International Affective Picture System (IAPS, University of Florida, Gainesville, FL, USA, http://www.csea.phhp.ufl.edu), and were instructed to down-regulate negative affect [23]. The task comprises three conditions: maintain, reinterpretation, and distraction. We included studies in which the comparison of reinterpretation and distraction was conducted. Studies were considered only when the whole-brain analyses and standard anatomical reference space (Talairach/Tournoux; Montreal Neurological Institute (MNI)) were used.

Studies that conducted either functional-connectivity analyses or region-of-interest analyses were excluded. We also excluded studies in which whole-brain statistical results were not reported, or in which peak coordinates or statistical parametric maps (SPMs) could not be obtained. In addition, to ensure the reliability of our result, the studies were excluded if statistical thresholds varied in different conditions.

Four studies were included in our research (Table 1, Ref. [8, 11, 24, 25]). The number of subjects and peak coordinates were extracted from the original articles or supplementary material. The literature search, and decisions on inclusion and data extraction, were all conducted independently by two authors (Yuluo Liu & Xiangyu Wang).

All meta-analyses were performed using the revised ALE algorithm for coordinate-based analyses as implemented in Ginger ALE software (BrainMap, San Antonio, TX, USA, version 3.0.2, http://www.brainmap.org/ale/) [26, 27]. The ALE meta-analysis algorithm has four main steps: ALE scores; null distribution; thresholding; and cluster statics. Before performing meta-analysis, we classified and summarized the coordinate data of each experiment into four datasets: co-activation; co-deactivation; reinterpretation versus distraction; and distraction versus reinterpretation. Within each dataset, there were different foci groups according to different experiments. To describe the convergence of the assessed experiments for each dataset, a whole-brain modeled activation map (MA-map) was obtained by estimating the activation probabilities of each voxel in the brain [28]. The ALE image was a union of all of the MA-maps. Then, the MA-maps of each experiment were calculated to determine a voxel-wise ALE score. In the steps of null distribution, the p-values table of ALE scores was obtained by comparison to an empirical null distribution. The ALE image and the p-values table were used to create a 3D p-value image. In the steps of thresholding, a permutation test was adopted to identify the true convergence of foci across different experiments from random spatial association by comparing the ALE scores to an empirical null distribution [29, 30]. Histograms of the ALE scores obtained under the permutation distribution were then employed to assign p-values to compute the ALE map threshold [29]. The resulting ALE maps were determined at a cluster-level Family Wise Error (FWE) rate corrected threshold of p 0.05 (cluster-forming threshold at voxel-level p 0.001). The p-values were transformed into the appropriate quantiles (z) of the standard normal distribution. For illustration, the ALE maps were imported into MRIcroGL (University of South Carolina, SC, USA, version 6, http://www.mccauslandcenter.sc.edu/mricrogl/home) as an overlay on a standardized anatomical MNI-normalized template (Colin_27_T1).

To investigate the neural network of distraction and reinterpretation, we selected the activation areas of meta-analysis as volumes of interest (VOIs) for further analysis by using the meta-analytic connectivity modeling (MACM). Therefore, the left middle frontal gyrus/DLPFC, which was activated during both distraction and reinterpretation, was employed to uncover the common emotion regulation network. The left amygdala and the left inferior frontal gyrus/VLPFC, which were activated during the reinterpretation in the absence of distraction, were used to identify the neural network of reinterpretation. All VOIs were imported into the BrainMap database to search for the studies that reported activation within each VOI boundary [31]. The search criteria in the current study included: “context: normal mapping”; “behavioral domain matches: cognition attention or emotion”; “imaging modality: fMRI”; and “activations: activation only”.

We then downloaded whole-brain coordinates of activations from identified contrasts: left middle frontal gyrus/DLPFC = 502 subjects, 30 experiments, 494 foci; left amygdala = 764 subjects, 47 experiments, 821 foci; and left inferior frontal gyrus = 494 subjects, 29 experiments, 456 foci. These foci were then used to determine co-activated brain areas by ALE meta-analyses.

The pooled functional meta-analyses included 4 comparative studies with 110 subjects and 102 brain foci (Table 1). Brain clusters of the pooled analysis include the frontal gyrus, inferior parietal lobe, precuneus, insula, amygdala, and superior and middle temporal gyrus. Areas demonstrating common and different activation between reinterpretation and distraction are presented in Table 2 and illustrated in Fig. 2.

Fig. 2.

The common and different activation for reinterpretation and distraction. ALE meta-analysis of comparative neuroimage studies on reinterpretation and distraction. The co-activation of distraction and reinterpretation is represented by red clusters, whereas specific activation of reinterpretation is represented by green clusters. Results are displayed at a cluster-level Family Wise Error (FWE) rate corrected threshold of p 0.05 (cluster-forming threshold at voxel-level p 0.001).

The ALE analyses revealed activation of the left middle frontal gyrus/DLPFC during both reinterpretation and distraction (Table 2). The results also showed that no brain regions were co-deactivated during both reinterpretation and distraction. When compared to distraction, reinterpretation specifically recruited the left amygdala and the left inferior frontal gyrus (Table 2). In contrast, there were no specific brain structures activated in the condition of distraction when compared to reinterpretation.

MACM analysis for the left DLPFC revealed convergent co-activation in bilateral DLPFC, insula, IPL, DMPFC, left (pre)SMA, left middle temporal gyrus, and right superior temporal gyrus (see Table 3 and Fig. 3).

Fig. 3.

Connectivity maps corrected for multiple comparisons (FWE 0.05) for the left DLPFC, the left amygdala and the inferior frontal gyrus. The VOI of the left DLPFC, the left amygdala and the inferior frontal gyrus are shown in yellow. The red indicates coactivation patterns of the left DLPFC, green of the left amygdala, and blue of the inferior frontal gyrus.

Co-activation maps for the left amygdala included the bilateral amygdala, left lentiform nucleus, right superior temporal gyrus (see Table 4 and Fig. 3).

For the left inferior frontal gyrus, significant co-activation was observed in the bilateral VLPFC, the left amygdala and the left parahippocampal gyrus (see Table 5 and Fig. 3).

Humans employ multiple strategies to down-regulate negative emotion that include attention deployment (e.g., distraction), cognitive reappraisal (e.g., reinterpretation), and response modulation. Here, we compared the neural bases of distraction and reinterpretation using meta-analysis and data from four comparative fMRI studies of a total of 110 participants. Our study showed that there were significant commonalities in the neural correlates of the two tactics as well as differences. We found that the left middle frontal gyrus/DLPFC was the only brain area involved in both distraction and reinterpretation to rise above the statistical threshold. The activation of DLPFC during both distraction and reinterpretation was consistent with previous literature [32, 33, 34]. The DLPFC was hypothesized to play a critical role in the cognitive control process of emotion regulation [19, 35]. But the DLPFC may subserve different functions when adopting various emotion regulation strategies. In the process of reinterpretation, the DLPFC was reported to be associated with emotional information reprocessing and guiding the goals of regulation [16, 36, 37]. Although when combined with the OFC, the DLPFC is involved in selective attention during distraction [12, 14].

Regarding the differential activation of distraction and reinterpretation, we observed that the left amygdala and inferior frontal gyrus/VLPFC were specifically activated during reinterpretation. The well-established role of the amygdala is that it receives processed information from the prefrontal and parietal cortex and regulates emotion responses during the last stage of emotion regulation [38, 39]. Deactivation of the amygdala has been recognized as a feature of successful emotion regulation [40, 41]. However, our results showed that the left amygdala was activated during reinterpretation in the absence of distraction. By careful review of the included studies, we found that the bilateral amygdala was deactivation during both reinterpretation and distraction in three of four studies [8, 11, 24], although the bilateral amygdala was activated during reinterpretation in the absence of distraction in two of the three studies [8, 24]. Therefore, we suggest that this may result in the greater deactivation of the left amygdala during distraction. In other words, our results suggest that distraction is more effective in regulating negative emotion than reinterpretation. This can be confirmed by the self-reported negative affect in the included studies [24]. However, given the small sample size of our study, much more research is needed to confirm whether distraction is more effective than reinterpretation.

The activation of the left inferior frontal gyrus/VLPFC during reinterpretation was consistent with previous findings [19, 21]. A wealth of studies suggests that the left VLPFC is involved in various cognitive functions such as selection and retrieval processes, semantic and phonological processing, inner speech generation, and semantic generation [42, 43, 44, 45]. The VLPFC has been reported also to be involved in the selection process of appropriate reappraisals and in language processes [21, 46]. In view of the different psychological processes of reinterpretation and distraction, we prefer to connect the specific activation of the left VLPFC with the semantic processing of reinterpretation.

One of the striking results was that no brain region was involved in distraction to a degree that rose above the statistical threshold (compare to reinterpretation). Nevertheless, a previous meta-analysis showed a similar result [20]. Thus, the results indicate that distraction shares a common core with reinterpretation during emotion regulation, which includes the left DLPFC. In addition, reinterpretation also recruits activation of the left VLPFC during emotion regulation.

As the MACM analyses result showed, significant co-activation of the left DLPFC was mainly observed in bilateral DLPFC; DMPFC; inferior parietal lobule; insula; left (pre)SMA; left middle temporal gyrus; and right superior temporal gyrus. This indicates that distraction and reinterpretation share a common emotion regulation network. Moreover, our findings are similar to the results of previous meta-analyses on emotion regulation strategy [19, 20]. As outlined before, we suggest that the prefrontal cortex (DLPFC and DMPFC) may play a role in information reprocessing and selective attention during emotion regulation [36, 37]. The (pre)SMA is involved in the process of executive control [47]. Specifically, it plays a role in inhibiting emotion responses and switching tasks during emotion regulation. The anterior insula has been associated with representing emotional state and self-reported arousal [48, 49, 50]. The insula is also considered as a pivotal region for the detection of emotional stimulus during the emotion process [51]. Besides, it was reported play a role of the supervisory attentional control during the distraction [52, 53]. The functions of these brain regions play a fundamental role in the process of emotional regulation. We suggest that this emotion regulation network is the core network of reinterpretation and distraction. In addition, the MACM analysis results also indicate an emotion regulation network of the left VLPFC, encompassing the bilateral VLPFC, left amygdala, and parahippocampal gyrus. The VLPFC is mainly involved in the selection of appropriate reappraisals and language processes during reinterpretation [21, 46], and the activity between amygdala and parahippocampal gyrus increases during facial information processing [54]. Semantic processing and facial information processing play an important role in reinterpretation. We propose that the emotion regulation network of the left VLPFC may be a specific regulatory network for reinterpretation.

The previous meta-analyses study of emotion regulation strategies showed that the bilateral anterior insula, the left (pre)SMA, and the VLPFC were activated during both distraction and reinterpretation. The contrast analysis did not reveal a significant difference between distraction and reinterpretation [20]. In contrast to the previous study, we found that the left middle frontal gyrus/DLPFC was activated during both distraction and reinterpretation, whereas the left amygdala and inferior frontal gyrus/VLPFC were activated in the condition of reinterpretation in the absence of distraction. The difference may be due to two potential reasons. On the one hand, it may due to the fact that the current meta-analyses study was more direct, because only the comparative studies of distraction and reinterpretation are included. The previous meta-analyses were based on comparing the ALE results of the individual meta-analyses with each other. On the other hand, the sample of experiments in the current study was so small that makes it hard to avoid biased results. The current study may not have been comprehensive enough to investigate the common and different neural bases of distraction and reinterpretation. Further research is needed to investigate the difference between direct and indirect meta-analyses.

Although the present study maintains consistency in the emotion regulation goals, strategy, and stimulus material, it is not without limitations. First, to investigate the common and different neural bases of distraction and reinterpretation, our meta-analyses relied solely on the contrast of reinterpretation and distraction in healthy volunteers. Therefore, only four experiments meeting the inclusion criterion were incorporated into the current study. The results were not comprehensive enough to cover all of the common and different brain regions of distraction and reinterpretation. Second, another potential limitation stems from the gender imbalance of the participants. In the current study, 97 of 110 participants were female. It is generally assumed that women are more emotionally responsive to negative stimuli than are men [55, 56], and women are more likely to employ a frontal top-down network to down-regulate negative emotion, whereas men may redirect attention away from stimulus by using posterior regions of the ventral attention network [57, 58]. Therefore, the large number of female participants in current study may be one of the reasons for activation of the DLPFC. Third, the ALE analysis approach assigns weight to each experiment according to the number of participants, so the larger the sample, the stronger the weight in the meta-analyses. The results will deviate when there is a significant difference in the number of participants among the included experiments. In the current study, the n of Dörfel’s et al. [11] study is nearly twice as large as that of McRae’s et al. [24] study. We suggest this imbalance may lead to a bias in the results.