Domain-general and domain-preferential neural correlates underlying empathy towards physical pain, emotional situation and emotional faces: An ALE meta-analysis

Highlights

• A large-scale ALE meta-analysis including 103 empathy-related studies was conducted.

• Domain-general neural correlates underlying empathy were revealed in supplementary motor area.

• Activation convergence of empathy for physical pain, emotional situations and emotional faces was indicated to be different.

• Domain-preferential correlates were supported by functional characterization analysis using Neurosynth terms.

Abstract

Empathy is essential for social interactions and individual development. Through the empathy field, the domain could be divided into three subgroups according to the stimulus materials adopted in tasks: empathy towards physical pain (PhyE), empathy towards emotional situations (ESuE) and towards emotional faces (ExpE). By far, no empirical studies directly compared the neural correlates underlying three sub-domains. The current study, therefore, utilized ALE meta-analysis to identify the general and distinct neural correlates underlying three sub-domains of empathy. The results revealed the conjunction in bilateral supplementary motor areas which were generally activated across three sub-domains. Preferential correlates for PhyE were found in bilateral IPL, left middle cingulate cortex and left anterior insula, which were associated with pain, action and somatosensory functions. Left middle temporal gyrus was found to be preferentially engaged for ESuE. And the preferential activations for ExpE were identified in right amygdala and right dorsal lateral prefrontal cortex, the regions of which were statistically associated with functional Neurosynth terms “facial expression”, “face”, “emotion” and “social”. Through the current meta-analyses, we firstly indicated that the domain-general and domain-preferential neural correlates potentially exist to underlie the processing of empathy evoked by different types of stimuli.

Introduction

Empathy refers to the ability which we can use to consciously share and infer the cognitive and affective state of other individuals (de Vignemont and Singer, 2006; Hillis, 2014). Such ability enables us to share and understand others’ emotions and plays a key part in human social interaction, in which it probably encourages the prosocial behaviors (Eisenberg and Miller, 1987; Eisenberg and Morris, 2001; Li et al., 2013) and would affect the moral judgements (Bzdok et al., 2012; Decety and Cowell, 2014; Gleichgerrcht and Young, 2013).

The neural correlates underlying empathy have been attracting neuroscientists to explore, especially with the increasing applications of fMRI into neuroscience and psychology. Extensive brain regions have been approved to be engaged during empathy processing, including middle/anterior cingulate cortex (M/ACC), inferior frontal gyrus (IFG), anterior insula (AI), supplementary motor area (SMA), precuneus, temporoparietal junction (TPJ) etc. (Fan et al., 2011a; Lamm et al., 2011; Singer et al., 2004; Wicker et al., 2003). Such research field consists of several sub-domains which could be defined according to the stimulus materials adopted to evoke empathy-related responses. And the stimuli generally comprise three types: physical pain, emotional situations and facial expressions. Above empathy relevant regions were found activated in tasks employing specific stimulus material, whereas the domain-general correlates across empathy evoked by three stimuli still remain unclear.

Compared to the studies regarding empathy towards emotional situations and emotional faces (e.g. happy/sad/angry/etc.), the majority of fMRI experiments focusing on empathy employed painful situations or actions as the stimuli materials. However, it holds no doubt that empathy for different situations (e.g. physical pain vs. emotional situations) recruits distinct neural pathways, at least the distinct emotion-related structures should be noted. The study by Bruneau et al. (2015) compared the cerebral activations evoked by emotional pain and physical pain, and the result revealed that the amygdala was associated with the empathic regulation towards emotional pain, but not physical pain. Another study with the pain, happiness and anxiety involved also demonstrated the separated activation patterns regarding them. The contrasts pain minus happiness and anxiety both indicated the stronger activations in dorsal ACC, inferior parietal lobule (IPL) and IFG, and the reverse contrasts led to the activities in precuneus, medial prefrontal cortex (mPFC) and TPJ (Morelli et al., 2014), both of which received support from the subsequent study (Bruneau et al., 2015). Besides above two empathic situations, a fraction of studies was also concerned with the neural correlates underlying empathy towards facial expressions. Empathy-related response towards emotional faces seemed also to activate regions like ACC, IFG, SMA and TPJ (Schulte-Ruther et al., 2007), whereas the subsequent study confirmed the significant activations in insula, dorsal medial prefrontal cortex (dmPFC), amygdala and thalamus (Ernst et al., 2013). Moreover, the study focusing on painful expressions though revealed the activated regions in bilateral occipital and temporal cortices associated with face processing (Li et al., 2015). Until now, still no consistent findings were confirmed about the neural correlates underlying empathy towards emotional faces. Additionally, the fMRI study directly comparing the empathy towards emotional situations, emotional expressions and physical pain has barely been reported so far. And the domain-preferential correlates underlying three empathic situations still remain to be answered. Coordinate based meta-analysis (CBMA), such as activation likehood estimation (ALE) (Eickhoff et al., 2009, 2012, 2016; Turkeltaub et al., 2012), supplies us with reliable tool to address such problem. ALE can contrast the neural correlates of different psychological processing, so as to confirm the similarities and differences between the processes. Previous meta-analysis computed the neural convergences across experiments regarding empathy for pain and also compared the activations for evoked painful empathy in terms of stimulus materials (picture-based vs cue-based), which was the first attempt to differentiate the brain regions preferentially engaged in pain-evoked empathy by different materials and successfully revealed distinct activation patterns involved (Lamm et al., 2011). And the current study applied ALE to calculate the neural convergences involved in empathic responses evoked by emotional situations, emotional expressions and physical pain, with aiming at uncovering the domain-general as well as domain-preferential neural correlates underlying three empathic situations.

To distinguish the neural correlates involved in empathy towards physical pain (PhyE), emotional situation (ESuE) and emotional expressions (ExpE), three individual ALE meta-analyses and contrast analyses among them were carried out. Additionally, the conjunction across above three individual meta-analyses was computed. Through this we can to some extent locate the domain general neural correlates underlying empathic responses towards all three stimuli, which differed from the method adopted by Fan et al. (2011a) to define the core network. In their article, the core regions were computed by conjunction between experiments with and without explicit instructions to empathize. It is hypothesized that the empathy-related experiments would stably activate comprehensive regions in M/ACC, AI/IFG, SMA, IPL, TPJ and those regions related with emotional processing. And in our analyses, general as well as distinct neural correlates underlying three empathic domains would be defined.

To illustrate the functional associations of specific derived regions (for PhyE, ESuE and ExpE) and common regions across three individual meta-analyses, we calculated the functional associations of a given region with Neurosynth terms (Yarkoni et al., 2011), those of which were selected as biological relevant (Cheng et al., 2015). It is important to appreciate that such functional characterization approach aims at calculating relations between psychological processes and brain regions rather than claiming “a unique role” of the brain regions for any psychological processes.