Studies have revealed that neural substrates of social rejection and physical pain overlap in some brain areas, such as the dACC and AIC 6,10,11,14−16. In terms of emotional regulation and adaptive control perspective 17–19, this study was set up to examine if the overlapping areas also attend to the processing of positive emotions. At the behavioral level, our participants were in a normal status of positive and negative affect, as assessed using PANAS 29. Their rating data indicated that, first, they still felt hurt about the breakup event and felt delighted about the joyful event with their friend. Second, the breakup event elicited higher arousal than the joyful event did, promoting a highlight that the breakup event caused a higher salience level. When looking into the brain response data, in comparison with reliving a neutral event, reliving heartbreaking and joyful events elicited a wider distribution of neural activities. Especially, as is the aim of this study, we found that the AIC and dACC did show reactivity to the positive emotions elicited by reliving a joyful event (see Fig. 3b, 4c, and 4d).
Our brain does not react to stimuli in an invariant manner. Rather, it categorizes, abstracts, interprets, and evaluates incoming stimuli in light of current states and goals, as well as prior knowledge. There are theories that propose that emotions are constructed in a similar way and realized in a widely distributed group of brain areas 20,30−34. In our results, in addition to sensorimotor activity, the majority of the brain responses could be attributed to three networks, including one for emotional salience detection, one for emotional conceptualization, and one for craving and motivation. Our findings are in line with the framework proposed in the theories. Relevance of the findings is discussed as follows.
4.1 The salience network
The core structures of the salience network are the AIC and dACC. We found greater AIC and dACC activities with the breakup and joyful events than with the neutral one, indicating that this network could evaluate information not only for negative emotion but also positive emotion. These findings are consistent with research suggesting that the AIC and dACC are involved in processing both negative and positive emotions for detecting bodily changes and directing attention for action 14,20,31,35−40.
ROI analysis further indicated that the AIC and dACC showed a gradient pattern of activities for reliving the three events (see Fig. 4b): i.e., highest with the breakup, moderate with the joyful, and lowest with the neutral. This gradient pattern might reflect the fact that emotional salience of the three conditions differed. This is consistent with behavioral rating data—i.e., a higher arousal level for the ex-partner than for the friend. The bodily changes in response to the breakup event outperformed anther two conditions and caught the highest salience notice. On the other hand, reliving a joyful event also deserved some attention, though its salience was not that significant compared to the breakup event. This finding corroborates the feature that AIC and dACC are sensitive to social salience 41, and further demonstrates that activity amplitude of the network can denote salience degree.
4.2 The conceptualization network
The conceptualization network comprises the vmPFC and PCC, which can make sense of the internally and externally triggered affects through abstraction of the instances in context 20,32–34,42. In this study, we found that the vmPFC and PCC showed higher activities with the breakup and joyful events than with the neutral one. But there was no difference between the two emotional events (see Fig. 3a). The PCC activities positively correlated with the rating scores of the question (“How do you feel whenever this relationship comes to mind?”), meaning that the more they felt hurt when thinking about the breakup event, the higher their PCC activities were. This is consistent with the proposed function of the network—i.e., to make sense of the feeling associated with the breakup event.
It was important to note that the PCC activities of reliving the two emotional events were located substantially in the ventral PCC. Distinct from the dorsal regions, the ventral PCC is found to interact with the subgenual cortex—i.e., the vmPFC—to process self-relevant information for self-reflection 43–45. Research also posits that the vmPFC can integrate information from the world and information from the body, and works as a hub to guide affective behavioral and physiological responses 33. Therefore, together with episodic memory activities—i.e., the findings that the hippocampal and parahippocampal cortex also show higher activities for reliving the two emotional events—the conceptualization network serves to interpret the feelings elicited.
4.3 The network for craving and motivation
Relative to reliving a neutral event, we found higher striatal activities for both emotional events. In the ROI analysis, the caudate and putamen activities also showed a gradient response pattern, as the salience network did (see Fig. 4c). Previous studies have suggested that the ventral striatum has a role in supporting sustained salience and, therefore, can modulate activities of cortical areas associated with detection of changes, relevance, and novelty of sensory events 46–50. Similar response patterns of the salience network and the ventral striatum supported the proposed role of the ventral striatum—i.e., to sustain salience for emotion identification.
There are studies indicating that the mesolimbic reward system, including the VTA and ventral striatum, is involved in rumination of romantic love 51,52 and rejection 22. Other studies also have observed the involvement of the VTA and the striatum in cocaine-craving, which is considered a mediator of motivation and reward 53–55. Although there was a lack of VTA activity in our results, we did find robust activity in the caudate, putamen, and nucleus accumbens when reliving the breakup event. Therefore, in the context of reliving the breakup event, the striatal activities might reflect the fact that that our participants were still in a “wanting” state in regards to their ex-partner 56,57. In sum, with a breakup or joyful event, the salience maintained by the ventral striatum might serve as a base for the salience network to derive the relevance for other networks—for example, the conceptualization network.
4.4 The sensorimotor networks
In this study, we also found higher activities in sensorimotor areas, including the pre-SMA, thalamus, and occipital cortex, for both emotional events. First, the involvement of the pre-SMA seemed predictable because previous studies have observed that the pre-SMA not only works closely with the dACC for cognitive control and response selection 58–60 but also shares strong intrinsic connectivity with the dACC 61. On the other hand, in line with research that has integrated activities of the thalamus and hippocampus for processing of episodic memory 62,63, our findings of thalamic involvement in reliving both emotional events adds supportive evidence to this integrated view. Lastly, it has long been observed that episodic memory retrieval will rely on reactivation of cortical sensory cortex 64–66. Thus, our findings of higher visual cortex activities with emotional events are consistent with previous observations.