Is Empathy for Pain Unique in Its Neural Correlates? A Meta-Analysis of Neuroimaging Studies of Empathy

Empathy is an essential component of our social lives, allowing us to understand and share other people's affective and sensory states, including pain. Evidence suggests a core neural network—including anterior insula (AI) and mid-cingulate cortex (MCC)—is involved in empathy for pain. However, a similar network is associated to empathy for non-pain affective states, raising the question whether empathy for pain is unique in its neural correlates. Furthermore, it is yet unclear whether neural correlates converge across different stimuli and paradigms that evoke pain-empathy. We performed a coordinate-based activation likelihood estimation (ALE) meta-analysis to identify neural correlates of empathy, assess commonalities and differences between empathy for pain and for non-pain negative affective states, and differences between pain-empathy evoking stimuli (i.e., facial pain expressions vs. acute pain inflictions) and paradigms (i.e., perceptual/affective vs. cognitive/evaluative paradigms). Following a systematic search, data from 128 functional brain imaging studies presenting whole-brain results of an empathy condition vs. baseline/neutral condition were extracted. Synthesizing neural correlates of empathy confirmed a core network comprising AI, MCC, postcentral gyrus, inferior parietal lobe, thalamus, amygdala, and brainstem. There was considerable overlap in networks for empathy for pain and empathy for non-pain negative affective states. Important differences also arose: empathy for pain uniquely activated bilateral mid-insula and more extensive MCC. Regarding stimuli, painful faces and acute pain inflictions both evoked the core empathy regions, although acute pain inflictions activated additional regions including medial frontal and parietal cortex. Regarding paradigms, both perceptual/affective and cognitive/evaluative paradigms recruited similar neural circuitry, although cognitive/evaluative paradigms activated more left MCC regions while perceptual/affective paradigms activated more right AI. Taken together, our findings reveal that empathy for pain and empathy for non-pain negative affective states share considerable neural correlates, particularly in core empathy regions AI and MCC. Beyond these regions, important differences emerged, limiting generalizability of findings across different affective/sensory states. Within pain-empathy studies, the core regions were recruited robustly irrespective of stimuli or instructions, allowing one to tailor designs according to specific needs to some extent, while ensuring activation of core regions.