Introduction

In general, emotional contagion refers to emotional transmission from one person to another (Belkin, 2009). There are two types of emotional contagion. One is intentional emotional contagion, a phenomenon of emotional convergence through intentional association and engagement (Barsade, 2002; van Kleef & Côté, 2022; Zuo et al., 2014). The other type is primary emotional contagion, which happens automatically and unconsciously between individuals (Hatfield et al., 1993; van Kleef & Côté, 2022; Zuo et al., 2014).

Some researchers view primary emotional contagion as spontaneous copying of others’ emotional states (Nakahashi & Ohtsuki, 2015). Meanwhile, the results of intentional emotional contagion are similar to those of primary emotional contagion. Considering that intentional emotional contagion is characterized by agency (Liu et al., 2009) and the relative distinction of self and other (Huang & Su, 2012), people use more cognitive resources and are more consciously engaged when intentional emotional contagion occurs. However, primary emotional contagion mainly depends on the automatic association of others’ emotional expressions and one’s own emotional experience (Decety & Lamm, 2006).

Moreover, compared with primary emotional contagion, intentional emotional contagion can improve the efficiency of emotional information transmission between interacting individuals. Specifically, the perceivers are asked to feel the sender’s emotion from the sender’s perspective. Therefore, the perceivers pay attention to the sender’s expression and try to image a situation that matches the sender’s emotion (Neumann & Strack, 2000). Hence, the perceiver’s intentional engagement can facilitate the efficiency of information processing. Besides, the senders can improve the suitability of the emotion and situation by regulating their expression consciously (van Kleef & Côté, 2022). Consequently, they can transmit their emotion to the perceivers better (Anders et al., 2011; Kinoshita et al., 2019), and thus improve communication efficiency (van Kleef & Côté, 2022). Therefore, research on intentional emotional contagion has a much stronger application value.

In addition, storytelling is considered an important part of human’s daily life. It is a powerful means to share emotions with others (Komulainen et al., 2021). In general, listening to others’ emotional stories, especially happy stories, is a fascinating experience. For instance, listeners who immerse themselves in stories during social interactions may evoke strong emotional reactions (Komulainen et al., 2021). They may share the narrator’s feelings without going through the actual experience. Hence, individuals’ autobiographical memory was selected as a suitable carrier of intentional emotional contagion (Smirnov et al., 2019). What’s more, few researchers have explored intentional emotional contagion using individuals’ autobiographical memory (Smirnov et al., 2019).

Besides, the corresponding mental mechanism of emotional contagion is not sufficiently understood (Anders et al., 2011; Kinoshita et al., 2019; Smirnov et al., 2019). Meanwhile, the self–other overlap theory (Aron et al., 1991, 1992) can be considered the mental mechanism of intentional emotional contagion in social interaction. Factors such as the communication object, communication content, and communication mode in social interaction (Zhang & Liu, 2018) can influence the degree of overlap between two interacting individuals. When people put themselves in others’ shoes and consciously try to feel others’ emotion, this results in greater integration with others. Then, with the deepening of interaction, individuals develop closeness, and their behavior and opinion become increasingly similar (Zhong et al., 2015). This means that higher levels of self–other overlap can promote people’s empathy for others’ emotions, enabling them to understand others’ intentions and emotional states quickly and accurately (Decety & Meyer, 2008). In other words, higher levels of self–other overlap can promote the effects of emotional contagion between two interacting individuals.

Furthermore, according to previous studies (Anders et al., 2011; Kinoshita et al., 2019; Smirnov et al., 2019), the brain mechanism of intentional emotional contagion under social interaction mainly involved two neural system: mirror neuron system (MNS) and mentalizing system (MS) (Kingsbury & Hong, 2020; Wang et al., 2018). MNS is activated when we observe others’ expressions. MNS includes inferior frontal gyrus (IFG), inferior parietal lobule (IPL) and superior temporal gyrus (STG). Shamay-Tsoory et al (2009) found that IFG was involved with primary emotional contagion. MS, inconcluding temporoparietal junction (TPJ) and dorsomedial prefrontal cortex (DMPFC), is involved when we try to understand others’ intentions or emotions by their gestures, behaviors and facial expressions. The PFC, likes a commander, is responsible for the planning, regulation, integrating of information, and other high cognitive functions (Wang et al., 2018). TPJ is more engaged when participants make allocentric choices in a communicative context (Zhen et al., 2021). That is to say, MNS and MS is the important brain region of intentional emotional contagion in social interactions.

In conclusion, naturalistic and complex emotional stimuli, especially individuals’ autobiographical memory, pervade human’s daily life. In the emotional contagion literature, few studies have assessed the neural responses to individuals’ autobiographical memory (Smirnov et al., 2019). Importantly, research on intentional emotional contagion arguably has a strong application value. This study focuses on the intentional emotional contagion in people’s daily conversation using a functional near-infrared spectroscopy (fNIRS) technique because this technique has a high tolerance of participants’ movement artifacts (Wang et al., 2018). When telling a story, certain body movements, such as head and hand movements, appear involuntarily. Consequently, an fNIRS technique is a better choice. Specifically, we want to explore the listener’s behavior and brain activity when processing others’ different emotional stories. We also want to explore the corresponding mental mechanism behind the occurrence of intentional emotional contagion (Anders et al., 2011; Kinoshita et al., 2019; Smirnov et al., 2019).

Methods

Experimental design

This study used an approach consistent with that of previous studies (Jospe et al., 2020; Smirnov et al., 2019), namely, a one-factor (emotion type: neutral or happy emotion), within-subject design comprising two phases. The first phase involved the speakers’ video production. In the second phase, these videos were presented to the listeners while we obtained listeners’ brain activities when viewing these videos. All the participants had normal or corrected-to-normal vision and had no history of brain injury or mental disorder. They all signed an informed consent form before joining the experiment and received some compensation after the experiment. This study was approved by the Ethics Committee of the department of Psychology, Renmin University of China.

Phase one: speakers’ video production

In this study, individuals’ autobiographical emotional story was selected as the carrier of intentional emotional contagion. We asked 20 people for suggestions on the topics for storytelling. These topics had to meet certain requirements. Specifically, the topics must be familiar to college students (Gandolphe et al., 2018), and they must be based on the individuals’ autobiographical events. Moreover, these events should be recent (D’Argembeau & van der Linden, 2004).

Finally, the happy topics were about receiving a gift, joining a party, and being praised, while the neutral topics were about one’s daily life, cleaning, and self-study. Each speaker was asked to narrate a story for each topic, meaning that each speaker had to narrate three happy stories and three neutral stories based on their autobiographical memories. The duration of each story was 2 min (El Haj et al., 2021). The description of each story should also be specific (i.e., when and where the story happened, who was present, what the individual was doing at that time, and what the individual was feeling; El Haj et al., 2021). In addition, the speakers must be engaging in their storytelling. For instance, facial expressions play an important role during social communication (Song et al., 2019). In this regard, the speakers must be good at conveying their emotions through facial expressions and vocal language as well as body language. According to these requirements, we started to recruit suitable speakers and train them before they came to record the corresponding stories. Finally, we chose four speakers and had them record their stories. We planned to pick two suitable speakers from the four speakers.

At the recording stage, four speakers were seated in front of the camera. They told neutral and happy stories in separate blocks. The sequence of the stories was customized. The speakers held the topic cards and were allowed to determine the order of the stories to help reduce their anxiety. The speakers tried to calm themselves before telling a story. Pressing the space bar produced the “Ding” sound, which served as a cue to start the storytelling. A countdown was done to remind them of the time and help improve their performance. The resting time between two stories was at least 30 s. The overall rhythm of storytelling was self-controlled. The speakers’ brain activities were recorded while they were sharing their stories. Then, the recording was edited in Adobe Premiere 2020 software to eliminate background noise and resize it to an appropriate size. All the stories were edited into a two-minute video clip (El Haj et al., 2021).

Subsequently, 16 female raters (20.44 ± 1.71 years (M ± SD)) evaluated all the stories narrated by the four speakers. The selection criteria comprised six items—Q1: Describe the speaker’s emotion; Q2: Evaluate the speaker’s emotional intensity; Q3: Evaluate your own emotional intensity after listening the story; Q4: Evaluate the attractiveness of the speaker’s appearance; Q5: Evaluate the speaker’s infectivity; and Q6: Evaluate the naturalness of the speaker’s narration. Two female speakers (21.00 ± 2.83 years (M ± SD)) scored well, and they were selected as the final speakers (see supplementary files regarding the evaluation results).

Phase two: collection of listeners’ data

To prevent the confounding effect of gender, we selected females as our participants. Thirty-two female participants took part in the experiment. Three pieces of the brain data were discarded during the fNIRS recording. One was dropped because the collection software malfunctioned, and two were interrupted because of the participants’ uneasiness. Moreover, two pieces of behavioral data were deleted. One was mistakenly deleted and one was interrupted during collection because of the participant’s uneasiness. Consequently, brain data of 29 participants were retained (20.62 ± 2.19 years (M ± SD)), while behavioral data of 30 participants were retained (20.73 ± 2.30 years (M ± SD)).

Like the procedure for the speakers, listeners were asked to calm themselves before listening to a story. They rated their own emotional experience (valence) using a nine-point scale. Subsequently, the preparation cue would present at the screen’s center unless they pressed the space bar. Each video took 2 min. Simultaneously, the brain activity was recorded using the LABNIRS system (Shimadzu Corporation, Kyoto, Japan). While viewing the videos, the listeners were instructed to try to feel the speaker’s emotion from the speaker’s perspective. This manipulation can help listeners internalize the speaker’s emotions to facilitate the occurrence of emotional contagion (Hawk et al., 2011). After viewing the videos, the listeners evaluated their own emotional experience (valence) using a nine-point scale as well as the degree of self–other overlap between the speaker and the listener using a six-point scale (Aron et al., 1991, 1992; Peng et al., 2021). The blocks of happy stories and neutral stories were counterbalanced among the participants, and the sequence of stories was randomized within each block. The whole experiment was recorded by camera.

fNIRS data acquisition

As participants viewed the videos, data were continuously recorded using the LABNIRS system (Shimadzu Corporation, Kyoto, Japan; Fig. 1). Two mainly nervous systems: mirror neuron system and mentalizing system (Kingsbury & Hong, 2020; Wang et al., 2018) may be involved in the occurrence of intentional emotional contagion in social interactions; thus, three sets of optode probes covered frontal, temporal, and parietal cortices, with 16 laser sources and 16 laser detectors, corresponding to a 43-channel montage. The distance between two optode probes was 3 cm, which took measurements approximately 15–25 mm beneath the scalp (Hoshi et al., 2005; Okada & Delpy, 2003). The distance between the listener and the screen was approximately 70 cm.

Fig. 1
figure 1

Setup of the experiment

Full size image

In addition, a SIEMENS TRIO 3-Tesla scanner was used to collect the anatomical images from two typical female participants according to the method of Long et al. (2021, 2022). A high-resolution T1-weighted magnetization-prepared rapid gradient echo sequence was used (TR = 4,200 ms; TE = min full; FA = 7◦; slice thickness = 1.33 mm; FOV = 256 mm × 256 mm; scan matrix = 1 × 1; scan layer = 144 slices). The positions of the channels were further confirmed and adjusted based on the functional magnetic resonance imaging scan of the two typical female participants. For each participant, three sets of optode probes covered the frontal, temporal, and parietal cortices (Fig. 2). In the preparation phase of the experiment, the optode probe sets were examined to ensure that the positions were consistent among participants.

Fig. 2
figure 2

The optode probe set (placed on the bilateral frontal, temporal, and parietal cortices)

Full size image

Three wavelengths of 780, 805, and 830 nm were used to measure the hemodynamic changes in oxyhemoglobin (HbO) and deoxyhemoglobin (HbR) blood. The signal was recorded at a sample rate of 9.5238 Hz. Signal quality was adjusted and calibrated on the LABNIRS system before starting the experiment.

fNIRS data preprocessing

The fNIRS data were pre-processed using the functions in MATLAB2013b software. The oxyhemoglobin, deoxyhemoglobin, and total hemoglobin concentration changes (HbO, HbR, and HbT) were assessed based on the modified Beer-Lambert law. In this study, only HbO concentration changes were used because they have the highest sensitivity to changes in the regional cerebral blood flow and signal-to-noise ratio (Hoshi, 2007). For every participant, general linear model (GLM) analyses were performed for each of the 43 channels. The hemodynamic response function filter and a wavelet-minimum description length detrending algorithm were used to remove physical noise and artifacts (Jang et al., 2009). For each condition, the beta-coefficient values of the GLM from each participant’s different trials were extracted as weights to account for brain activity (Xie et al., 2018).

Statistical analysis

IBM SPSS Statistics 26 was used to analyze the behavioral results and the single-brain activation results. One-way repeated-measures analysis of variance (ANOVA) was performed based on the self-assessment scores and the individual-level HbO beta-coefficient values for each subject, condition, and channel. Additionally, for the single-brain activation results, false discovery rate (FDR) correction was applied across 43 channels (Benjamini & Hochberg, 1995) to account for multiple comparisons and obtain a corrected p-value for each channel. Only FDR corrected p-value was reported as significant.

Results

Behavioral results

Manipulation of story valence

A pair t-test of the self-assessment concerning the emotional state was conducted to examine the validity of the manipulation. Neutral stories did not significantly evoke mood changes in the listener (t = 0.85, df = 29, p > 0.05; M ± SD: before ~ 5.00 ± 0.09, after ~ 5.09 ± 0.57). However, happy stories significantly evoked mood changes in the listener (t = 13.69, df = 29, p < 0.001; M ± SD: before ~ 5.11 ± 0.33, after ~ 7.08 ± 0.78). These results meant that our manipulation of the story valence worked.

Emotional contagion induced by the emotional stories

One-way repeated-measures ANOVA was conducted to show emotional contagion that was induced by the emotional stories. The independent variable was the emotion type (neutral or happy stories). The dependent variable was the change in self-assessment about the emotional state (self-assessment score after viewing minus self-assessment score before viewing). A significant main effect of the emotion type was observed (F (1,29) = 148.47, p < 0.001, η2p = 0.84). The degree of emotional contagion induced by happy stories (M ± SD: 1.97 ± 0.79) was larger than that by the neutral stories (M ± SD: 0.08 ± 0.54; Fig. 3).

Fig. 3
figure 3

Emotional contagion induced by the emotional stories. Note: *** means p < 0.001

Full size image

Self–other overlap between the speaker and the listener

One-way repeated-measures ANOVA was conducted to demonstrate the self–other overlap between the speaker and the listener, as felt by the listener after listening to the emotional stories. The independent variable was also the emotion type (neutral or happy stories). The dependent variable was the self–other overlap scores after the listeners heard each story. A significant main effect of the emotion type was observed (F (1,29) = 16.29, p < 0.001, η2p = 0.36). The degree of overlap induced by the happy stories was bigger than that by the neutral stories (M ± SD: 3.53 ± 1.21 for happy stories, 4.27 ± 1.09 for neutral stories; Fig. 4).

Fig. 4
figure 4

Self–other overlap between the speaker and the listener. Note: *** means p < 0.001

Full size image

Single-brain activation

One-way repeated-measures ANOVA was conducted to illustrate the brain activity induced by the emotional stories. The independent variable was also the emotion type (neutral or happy stories). The dependent variable was the individual-level HbO beta-coefficient values for each subject, condition, and channel. We found a significant main effect of the emotion type in channel 11 (F (1,28) = 14.73, p = 0.001, η2p = 0.35) and channel 15 (F (1,28) = 20.73, p < 0.001, η2p = 0.43). The activation pattern in channel 11 and channel 15 was similar. Specifically, neutral stories induced greater activation than happy stories (M ± SD: -0.006 ± 0.005 for neutral stories, -0.009 ± 0.004 for happy stories in channel 11; -0.004 ± 0.007 for neutral stories, -0.008 ± 0.006 for happy stories in channel 15; Fig. 5).

Fig. 5
figure 5

Significant main effect of the emotion type in channel 11 and channel 15. Note: *** means p ≤ 0.001

Full size image

Discussion

Social interaction is a part of human’s daily life (Dumas, 2011; Redcay & Schilbach, 2019), while emotional contagion is the foundation of human’s social interaction (Wei et al., 2023). In addition, storytelling in conversation is one of the most typically social interactions. Moreover, intentional emotional contagion during storytelling is like a fantasy; it allows people to share others’ feelings. In this study, we investigated listeners’ behavior and brain response when processing others’ different emotional stories. We also investigated the corresponding mental mechanism in the occurrence of intentional emotional contagion (Anders et al., 2011; Kinoshita et al., 2019; Smirnov et al., 2019). In the following, we discuss the results and suggest some future research directions on the topic of intentional emotional contagion from the perspective of social interactions.

To be a happy girl is quite easy

Face-to-face conversation consists of verbal and nonverbal components. It is impossible to act and adapt to circumstances without mentalizing, that is, thinking about others’ thoughts and mental states to predict their intentions and actions (Suda et al., 2010). Hence, face-to-face conversation requires social cognitive function.

The behavioral results showed that happy stories can be more contagious than neutral stories, suggesting that people are more easily affected by happy stories. The brain activation result demonstrated that compared with the happy stories, the activation degree of the neutral stories was larger. PFC plays a prevailing role in emotional processing (Balconi et al., 2019). Perhaps listening to others’ happy stories did not require acts of intentional communication, as obtaining pertinent information from the emotional states of others may on occasions require little effort (Puścian et al., 2022). Hynes et al. (2006) found that the frontal pole was recruited when slower, conscious, cognitive processing was required to make sense of the social encounter. Other researchers also found that orbitofrontal cortex functioning is critical for social cognition processes, moral decisions, and emotion control. It is a brain region associated with computing and evaluating predictions of other persons’ actions and the comparison of these predictions with subjective states across both affective and non-affective situations (Brink et al., 2011).

In conclusion, both the behavioral and brain activation results provide evidence that emotional scenarios require fewer or less difficult processing (Hynes et al., 2006). In other words, to be a happy girl is quite easy.

Sharing happiness leads to greater closeness

Happiness is regarded as one of the most fundamental human goals (Matsunaga et al., 2017). Our behavioral results demonstrated that compared with neutral stories, happy stories allow greater closeness between individuals and speakers. This finding suggests that when individuals are happy, they become less self-focused (Yogan, 2020) and then feel more intimate with others. Therefore, sharing happiness could strengthen interpersonal bonding. People who use and respond to emotional expressions appropriately may develop better social networks, receive more social support, and lead more successful social lives (van Kleef, 2016, p. 15).

Future directions

Social interactions are at the root of happiness and subjective well-being (Yogan, 2020). Meanwhile, the topic of intentional emotional contagion is relatively underexplored, especially intentional emotional contagion in social interaction. Below are possible directions for future research.

First, the ecological validity of the research on this topic needs improvement. Importantly, with the new research methods and techniques, the study of social cognitive neuroscience has been questioned by researchers in terms of ecological validity (Sonkusare et al., 2019). Moreover, the new hyperscanning technique has provided a novel perspective—from a “single-brain” to a “multi-brain” perspective—and improved the ecological validity of the respective research (Li et al., 2018). In other words, the hyperscanning technique breaks the traditional one-way communication mode in the field of emotional contagion and expands research into natural contexts. In addition, this study was not conducted under natural settings. Hence, future research needs to pay attention to the inter-brain neural synchronization of emotional contagion in natural social interactions using the hyperscanning technique.

Second, natural situation paradigms can be used to explore the inter-brain neural synchronization, for instance, through an index of interaction quality in people’s daily conversation. Future research may also explore whether the enhancement and blocking effects of intentional emotional contagion are not just the result of single-brain activity but are more related to inter-brain activity.

Third, our study was conducted in female strangers. In the future, research should explore the interaction in friends, teacher-student, romantic relationships. And gender of the dyads could also be included in the sample to see if gender-related effects occur (Balconi et al., 2019).

Fourth, in the process of social interaction, the input and output of emotional information are often accompanied by facial expressions, changes in physiological activities such as electrocardiogram and skin electricity, and changes in dyads’ neural activities. Hence, we may build a multimodal hyperscanning platform for intentional emotional contagion and explore the influence of different social factors on intentional emotional contagion. Furthermore, when building a multimodal hyperscanning platform, it is necessary to consider compatibility issues between technologies and whether the wearing experience of multimodal devices interferes with participants' natural emotional responses (Wei et al., 2023).

Additionally, we could examine the subsequent prosocial behaviors after sharing happy experience. Some researchers found that positive emotional contagion interventions may be performed in people with subthreshold depression (Yamashita & Yamamoto, 2021). Therefore, some practical research is also needed.