Prosocial versus instrumental motives for interpersonal emotion regulation
Abstract
Why do people try to influence the way others feel? Previous research offers two competing accounts of people’s motives for attempting to regulate others’ emotions. The instrumental account holds that people use interpersonal emotion regulation to benefit their own goal pursuit. Conversely, the prosocial account holds that people use interpersonal emotion regulation to benefit others’ goals. This article juxtaposes these accounts across two studies. Study 1 demonstrates that when given the chance to benefit themselves through their interpersonal emotion regulation, people choose to do so, even when this involves making a friend feel unpleasant. Yet when given the chance to benefit a friend through interpersonal emotion regulation, with no personal gains, people also choose to do so. Study 2 reveals no overall tendencies toward either motive when people can choose between benefitting themselves or a friend through their interpersonal emotion regulation. However, people’s motives can be reliably predicted by their values: individuals with high values of care and concern for others show a greater tendency to regulate a friend’s emotions prosocially and a lower tendency toward instrumentality.
1 INTRODUCTION
People try to regulate others’ feelings in a variety of social contexts. Friends try to cheer each other up when one is feeling sad, romantic partners try to lessen each other’s anxieties, sports coaches try to enthuse their team members, and rival coworkers try to make each other feel envious. But what drives people to regulate others’ emotions? In recent years, two competing accounts have been put forward to describe the motives that underpin this process. Netzer, Van Kleef, and Tamir (2015) suggest that interpersonal emotion regulation is engaged under instrumental concerns of benefitting one’s own goals. In contrast, López-Pérez, Howells, and Gummerum (2017) argue that interpersonal emotion regulation is engaged under prosocial concerns of benefitting others’ goals. The aim of the present paper is to examine the relative support for these competing accounts, in order to provide more nuanced insights into why people try to shape the feelings of those around them.
1.1 Interpersonal emotion regulation
Emotion regulation is “the process of initiating, maintaining, modulating, or changing the occurrence, intensity, or duration of internal feeling states” (Eisenberg, Fabes, Guthrie, & Reiser, 2000, p. 137). Traditionally, research on emotion regulation concentrated on intrapersonal regulatory processes, in which people attempt to shape their own feelings. More recently, the focus has shifted toward interpersonal emotion regulation, during which people try to manage feelings of others (Gross, 2013; Niven, 2017; Zaki & Williams, 2013). Research on interpersonal emotion regulation highlights that people attempt to influence the feelings of those around them within work, peer, familial, and romantic relationships (e.g., Little, Kluemper, Nelson, & Ward, 2013; Niven, Macdonald, & Holman, 2012; Vangelisti, Daly, & Rudnick, 1991). While the consequences of these attempts have been relatively well established (e.g., Little, Gooty, & Williams, 2016; Martínez-Íñigo, Poerio, & Totterdell, 2013; Niven, Holman, & Totterdell, 2012; Niven, Totterdell, Holman, & Headley, 2012), the underlying motives are less clear.
A popular account of why people regulate their own emotions is the instrumental account. Contrary to the traditional assumption that emotion regulation is simply a matter of hedonism (such that people always want to feel pleasant emotion, for example, Westen, 1994), the instrumental account suggests that emotion self-regulation is undertaken primarily as a means of fulfilling one’s goals (Tamir, 2009; Tamir & Ford, 2012). Since different emotions are useful for achieving different goals, this account predicts that people will improve or worsen their feelings, depending on which state will be most beneficial. The emotional state of happiness may contribute to the goal of affiliating and collaborating with other people, as it promotes sociability, friendliness, and cooperation (e.g., Forgas, 1998; Van Dijk, Van Kleef, Steinel, & van Beest, 2008). Conversely, anger may facilitate goal pursuit in confrontational or aggressive situations, as it conveys power and promotes assertiveness and competition (Van Dijk et al., 2008; Van Kleef, De Dreu, & Manstead, 2004).
Recently, it has been proposed that the instrumental account may extend to explaining interpersonal emotion regulation (Netzer et al., 2015). According to this proposition, people may regulate others’ feelings for their own benefit, such that others’ emotions are simply a means to achieving personal goals. For example, a manager might try to make his or her team members feel enthusiastic to get them to accept changes to work practices, or make the team members feel anxious in order to meet an important deadline. Similarly, a person might try to calm his or her romantic partner so that the partner listens to his or her point during an argument, or make the partner feel guilty in order to get the partner to complete outstanding chores. Across three studies, Netzer and colleagues (2015) provided support for this suggestion, showing that participants were motivated to induce emotions in others that they expected to personally benefit from, even when this involved worsening the feelings of partners or improving the feelings of rivals.
However, another possibility is that interpersonal emotion regulation is prosocially motivated (Niven, 2016). In other words, people might try to regulate others’ emotions in order to help those others achieve their goals. Thus, interpersonal emotion regulation might be engaged in order to benefit the target rather than the regulator. This idea is consistent with various theories which posit that people act to benefit others’ goal pursuit in a prosocial or altruistic manner, because this facilitates the formation and maintenance of social relationships and promotes cooperation (e.g., Batson, 1995; Finkel & Rusbult, 2008; Grant, 2007). In line with this possibility, a recent study by López-Pérez et al. (2017) demonstrated evidence that when primed to take the perspective of others, people selected stimuli intended to induce emotions in others that would benefit those others’ performance.
1.2 The present research
Previous research provides preliminary support for two very distinctive motives for interpersonal emotion regulation, but remains mute on the relative strength of prosocial versus instrumental interpersonal emotion regulatory tendencies. In the present research, we aimed to provide insight into this issue. We conducted two studies in which we contrasted instrumental versus prosocial motives for interpersonal emotion regulation. In both studies, we formed situations in which it would be useful to either be happy (collaborative or affiliative situations) or angry (aggressive or confrontational situations) and asked participants to select stimuli for a friend facing these situations. We selected these two types of situations to be sure that we were examining generalizable motives underlying interpersonal emotion regulation, rather than specific motives pertaining to a particular type of emotion or context. In our first study, we put half of our participants in a situation in which they could benefit themselves and half in a situation where they could benefit a coursemate to examine the relative strength of instrumental and prosocial motives. In our second study, we directly compared the two motives by putting all participants in a situation in which they had to choose between benefitting either themselves or a friend.
2 STUDY 1
In Study 1, we sought to examine the relative strength of effects of instrumental and prosocial motives for interpersonal emotion regulation. We paired student participants with members of their course. Those allocated to the interpersonal emotion regulation role were asked to select emotion-inducing newspaper stories for their partner to read before playing a computer game in which a particular emotion (happiness or anger) would be beneficial to performance. Regulators were told that either their own gains (in terms of lottery tickets for a prize) or the partner’s gains were tied to the partner’s performance in the game, meaning that they had a chance to either benefit themselves (instrumental condition) or their partner (prosocial condition) through their regulatory behavior.
2.1 Participants
Two hundred and seventy-eight undergraduate students from the same course took part in an experiment that was ostensibly about how media affects game performance. The experiment took place during a series of laboratory testing sessions, each of which involved the same series of studies, with students receiving credits for taking part. At the start of each session, the participants in that session were asked to pair up with another member of their course and were then randomly assigned to one of two groups: choosers (N = 139, 65 males and 74 females, Mage = 19.80 years, SD = 1.33) and players (N = 139, 62 males and 77 females, Mage = 19.70 years, SD = 1.25).
2.2 Procedure
At the start of each testing session, participants were asked to pair up with another participant. We asked participants to pair up with a friend, if possible, or otherwise someone they knew, so that the interpersonal emotion regulation we studied would be more meaningful to participants. Each member of a pair was randomly allocated an even or odd number. Those who received an even number were told they had the role of “chooser” while those who received an odd number were allocated the role of “player.” All choosers were sat in individual experimental cubicles on one side of a corridor, while all players were sat in individual cubicles on the opposite side of the corridor, in order to make the effects of interpersonal emotion regulation more salient. Both the chooser and player versions of the experiment were administered using an online survey in Qualtrics.
All participants were then told that the study was about the effects of media on game performance. The role of choosers would be to select the media that would be given to players to engage with before playing a game, while the role of players was to play the game. Participants were also told that there would be two lotteries run across all experimental sessions for this study, one for choosers and one for players.
Choosers engaged in their part of the study first, while players were assigned a filler task. Choosers began by rating their current feelings, and then read a description of the game that their partner would be playing in the experiment. All teams of choosers and players were allocated to one of two game conditions. The first was an aggressive game condition, where choosers were told that their partner would be playing “an aggressive road rage game in which the goal is to destroy as many cars on the road as possible.” The second was a collaborative game condition where choosers were told that their partner would be playing “a collaborative game in which the goal is to deliver pizzas safely to customers.” The descriptions were of real games available free to play online (Total Takedown and Rush Rush Pizza, respectively, both from agame.com). A pilot study run with a different sample of participants (N = 70) confirmed that anger was expected to be more useful in the aggressive game (M = 4.48, SD = 1.79) than in the collaborative game (M = 2.16, SD = 1.28, t(69) = 8.46, p < 0.01), while happiness was expected to be more useful in the collaborative game (M = 5.77, SD = 1.02) than in the aggressive game (M = 3.93, SD = 1.77, t(69) = 9.52, p < 0.01). In the present study, allocation to conditions was quasi-random, in that all participants during certain testing sessions (counter-balanced for time of day effects) were given the same game.
After reading the description of the game, choosers were asked to select the media for their partner to engage with before playing the game. They were told to select carefully because how their partner performed in the task would influence the amount of lottery tickets obtained for the session (which would therefore increase the chances of winning a prize). Choosers were then randomly allocated to one of two benefit conditions. The first was the benefit self-condition, where participants were told that the amount of lottery tickets they themselves received would depend on their partner’s score (i.e., a higher score will give choosers more tickets). The second was the benefit other condition, where participants were told that the partner’s game performance would determine the amount of lottery tickets their partner would obtain (i.e., a higher score will give players more tickets).
Choosers were then presented with a series of nine emotion-inducing newspaper headlines (three expected to induce anger, three expected to induce happiness, and three expected to induce neutral emotion). After each headline, they indicated the extent to which they wanted their partner to read the full story associated with that headline prior to playing the game, and were reminded that their preferences would determine the media given to players during the experiment. Finally, choosers indicated their explicit emotional preferences for the partner and rated the perceived utility of different emotions for game performance.
For players, the experiment began with a rating of their current emotional experience. They then read a full newspaper article. Players were randomly assigned to read one of three full articles designed to elicit either anger, happiness, or neutral emotion. Note that choosers’ preferences for the headlines had no actual bearing on the articles the players actually received. Next, players had to rate their emotions again, before they were presented with a description of the game they were going to play and brief instructions. They were then asked to play the first level of the game while trying to attain the highest score possible in a single round. Scores on the game were recorded.
2.3 Materials
2.3.1 Current emotional experiences
In line with prior research (Netzer et al., 2015), both choosers and players indicated the extent to which they currently felt angry and irritated (mean α = 0.63) and happy and cheerful (mean α = 0.81; 1 = not at all, 7 = a lot).
2.3.2 Preferences for emotion-inducing stimuli in others
Choosers were asked to rate the extent to which they wanted to expose their partners to the full articles that corresponded with a series of newspaper headlines (1 = not at all, 7 = a lot). The headlines and corresponding articles were adapted from two UK newspapers and were selected based on the emotional tone of their content. We selected three headlines to elicit anger, three to elicit happiness, and three to elicit neutral emotion.
In a pilot study, we asked a separate group of undergraduate participants (N = 142) to rate the extent to which they expected the articles corresponding to each headline to make them feel angry, fearful, happy, and neutral (1 = not at all, 7 = a lot), in order to check that the articles were expected to elicit the emotions of interest. Paired t-tests confirmed that each of the headlines was expected to elicit the correct emotions.1
For the players version of the experiment, we shortened the original newspaper articles corresponding to the headlines (one for each emotion state) to 250–300 words and one representative picture as provided in the original versions. These articles served as stimuli for players to engage with prior to playing the game. Independent samples t-tests using the players’ data from the present study confirmed that exposure to the news articles had the expected effects on players’ emotions.2
2.3.3 Preference for emotions in others
Choosers indicated the extent to which they wanted their partnering players to feel angry and irritated (α = 0.70) and happy and cheerful (α = 0.91) before playing the game (1 = not at all, 7 = a lot).
2.3.4 Expectations about the usefulness of emotions
Choosers rated the extent to which they expected anger and irritation (α = 0.61) and happiness and cheerfulness (α = 0.86) in their partner to result in better game performance (1 = not at all, 7 = a lot), so that we could test whether perceptions of utility were driving regulatory behavior.
2.4 Results
2.4.1 Manipulation checks
In order to ensure that choosers were paying attention to their instructions and specifically to their manipulations, we included two questions at the end of the chooser survey. The first asked whether their partner was playing the pizza delivery or road rage game (96% answered correctly) and the second asked who would win more lottery tickets depending on the partner’s score, oneself or the partner (68% answered correctly).3
2.4.2 Preferences for others’ emotions
A repeated measures ANOVA on choosers’ preferences for how players would feel before playing the game (Figure 1), with game (collaborative, aggressive) and benefit (other, self) conditions as between subjects factors and emotion (happiness, anger) as the within subjects factor, revealed a significant main effect of emotion (F(1, 135) = 75.04, p < 0.01, 
= 0.36) and an interaction between emotion and game condition (F(1, 135) = 69.64, p < 0.01, 
= 0.34). There were no other significant effects (Fs < 1.50, ps > 0.22), suggesting no differences in preferences for partners’ emotions based on whether oneself or one’s partner is the beneficiary.
Independent samples t-tests confirmed that in both benefit conditions, choosers preferred for their partners to feel more angry before playing the aggressive road rage game (Mbenefit other = 3.71, SDbenefit other = 1.89; Mbenefit self = .03, SDbenefit self = 1.88) compared with the collaborative pizza game (Mbenefit other = 1.60, SDbenefit other = 0.96; Mbenefit self = 1.81, SDbenefit self = 1.06, ts>5.59, ps < 0.01). Choosers also preferred for their partners to feel more happy before playing the pizza game (Mbenefit other = 5.76, SDbenefit other = 0.92; Mbenefit self = 5.91, SDbenefit self =0.95) compared with the road rage game (Mbenefit other = 4.05, SDbenefit other = 1.80; Mbenefit self = 3.88, SDbenefit self = 1.83, ts > 4.75, ps < 0.01). Thus, participants’ preferences for how their partners would feel were comparable whether they were given the chance to benefit themselves or their partners.
2.4.3 Preferences for emotion-eliciting stimuli in others
An equivalent ANOVA on preferences for the newspaper headlines (Figure 2), with the between subjects factors of game (collaborative, aggressive) and benefit (other, self) conditions and the within subjects factor of headline-type (happy, angry, neutral), revealed a significant main effect of headline-type (F(2, 134) = 37.89, p < 0.01, 
= 0.36), a main effect of game condition, (F(1, 135) = 7.19, p < 0.01, 
= 0.05), and an interaction between headline-type and game condition (F(2, 134) = 23.38, p < 0.01, 
= 0.26). There were no other significant effects (Fs < 2.90, ps > 0.09), suggesting that actual interpersonal emotion regulation behavior also did not vary depending on the beneficiary.
Supporting the idea that people may be prosocially motivated when regulating others’ feelings, planned comparisons using independent samples t-tests confirmed that when it would benefit their partner (i.e., in the benefit other condition), choosers expressed a stronger preference for their partners to read the anger-inducing headlines before playing the aggressive road rage game (M = 4.37, SD = 1.40) compared with before the collaborative pizza game (M = 3.31, SD = 1.36, t(68) = 3.15, p < 0.01). They also expressed a stronger preference for their partners to read the happiness-inducing headlines before playing the pizza game (M = 4.87, SD = 1.33) compared with the road rage game (M = 3.55, SD = 1.41, t(68) = 3.95, p < 0.01). Paired samples t-tests further confirmed that when their regulatory actions would benefit their partners and their partners were playing the road rage game, choosers’ preferences for the angry headlines were stronger than their preferences for both happy headlines (t(40) = 2.60, p < 0.05) and neutral headlines (M = 2.85, SD = 0.87, t(40) = 6.55, p < 0.01). Moreover, when their partners were playing the pizza game, their preferences for happy headlines were stronger than their preferences for angry headlines (t(28) = 4.13, p < 0.01) and for neutral headlines (M = 3.71, SD = 1.33, t(28) = 3.43, p < 0.01).
Equivalent analyses in the benefit self-condition revealed findings in line with an instrumental account of interpersonal emotion regulation. Specifically, when it would be of personal benefit to do so, choosers expressed a stronger preference for their partners to read the anger-inducing headlines before playing the aggressive road rage game (M = 4.27, SD = 1.40) compared with before the collaborative pizza game (M = 3.43, SD = 1.40, t(67) = 2.46, p < 0.05). They also expressed a stronger preference for their partners to read the happiness-inducing headlines before playing the pizza game (M = 4.47, SD = 1.26) compared with the road rage game (M = 3.30, SD = 1.69, t(67) = 3.14, p < 0.01). When their regulatory actions would have a personal benefit and their partners were playing the road rage game, choosers’ preferences for the angry headlines were stronger than their preferences for happy headlines (t(39) = 2.45, p < 0.05) and for the neutral headlines (M = 2.43, SD = 1.05, t(39) = 6.47, p < 0.01). Moreover, when their partners were playing the pizza game, their preferences for happy headlines were stronger than preferences for angry headlines (t(28) = 3.38, p < 0.01) and for neutral headlines (M = 3.30, SD = 1.12, t(28) = 4.20, p < 0.01).
2.4.4 Usefulness of emotions in others
To establish whether choosers’ selection of newspaper headlines was influenced by their expectations about how useful anger and happiness would be for playing the games, we first conducted an ANOVA on expected utility of emotions for game performance (Figure 3), with the between subjects factors of game (collaborative, aggressive) and benefit (other, self) conditions and the within subjects factor of emotion-type (happiness, anger). The results revealed a significant main effect of emotion-type (F(1, 135) = 67.00, p < 0.01, 
= 0.33) and an interaction between emotion-type and game condition (F(1, 135) = 83.58, p < 0.01, 
= 0.38). All other effects were not significant (Fs < 1.73, ps > 0.19).
Consistent with their regulatory behaviors, in both benefit conditions choosers expected anger to be more useful for playing the aggressive road rage game (Mbenefit other = 3.90, SDbenefit other = 1.55; Mbenefit self = 4.21, SDbenefit self = 1.64) than for playing the collaborative pizza game (Mbenefit other = 1.86, SDbenefit other = 0.89; Mbenefit self = 2.14, SDbenefit self = 1.16, ts > 5.81, ps < 0.01). Choosers also expected happiness to be more useful for playing the pizza game (Mbenefit other = 5.57, SDbenefit other = 1.38; Mbenefit self = 5.76, SDbenefit self = 0.84) compared with the road rage game (Mbenefit other = 4.09, SDbenefit other = 1.48, Mbenefit self = 3.63; SDbenefit self = 1.68), ts > 4.24, ps < 0.01).4
Correlation analyses revealed that expectations about happiness being useful for game performance were positively related to preferences for happy headlines, both when one’s partner (r = 0.46, p < 0.01) and oneself (r = 0.49, p < 0.01) would benefit from improved game performance. Similarly, expectations about anger being useful for game performance were positively related to preferences for angry headlines, whether one’s partner (r = 0.45, p < 0.01) or oneself (r = 0.33, p < 0.01) stood to gain from improved game performance. Regression analyses confirmed that there were no significant interactions between expected usefulness of emotions and either game or benefit conditions in predicting preferences for emotion-eliciting stimuli. Thus, people prefer to elicit happiness and anger in others in whatever context they believe these emotions will be useful, whether that leads to benefits for the other person or for themselves.
2.5 Discussion
In this study, we provided evidence consistent with both prosocial and instrumental motives for interpersonal emotion regulation. When enhancing a coursemate’s performance would ultimately benefit the coursemate (and not themselves), students selected performance-conducive emotion-inducing stimuli, and their regulatory behavior choices were shown to be related to their expectations about how useful particular emotions would be for the coursemate’s performance. Likewise, when enhancing the coursemate’s performance would ultimately benefit themselves (and not the coursemate), students also selected performance-conducive stimuli because they expected that these stimuli would be useful for the coursemate’s performance. These effects were observed even when enhancing the coursemate’s performance would involve inducing anger, suggesting that our findings were not purely driven by the generalized desire to make others feel pleasant.
The lack of effects observed relating to the beneficiary of interpersonal emotion regulation (oneself or the coursemate) suggests that the strength of prosocial and instrumental motives may be equivalent. However, without a direct comparison, in which people are forced to choose between acting prosocially or instrumentally, it is not possible to draw firm conclusions about which motive is a stronger driver of interpersonal emotion regulation.
3 STUDY 2
Our second study was designed to directly contrast prosocial and instrumental motives for interpersonal emotion regulation, by giving participants a situation in which only themselves or a friend, but not both, could benefit. As well as testing whether the overall tendency is stronger toward prosociality or instrumentality in interpersonal emotion regulation, we additionally sought to explore whether we could predict who would be more or less likely to regulate others’ feelings in prosocial or instrumental ways.
Values are conceptions of desirable ways to live, such as living healthily, respecting tradition, or being ambitious, that contribute to people’s sense of identity (Feather, 1992). Research suggests that values play a strong role in driving people’s motivations and in turn their behavior (e.g., Verplanken & Holland, 2002). As such, values are likely to be an important determinant of people’s interpersonal emotion regulation behavior, with people likely to express regulatory preferences that are in line with their core values. Here, we expected that having core values of care and compassion for others would differentiate between those people who choose to regulate others’ emotions prosocially and those who choose to regulate others’ emotions instrumentally.
3.1 Participants
One hundred and thirty-seven undergraduate students (64 males and 73 females, Mage = 20.34 years, SDage = 1.23) took part in an experiment about how they would want their friends to feel in the context of a hypothetical scenario. The experiment took place in a laboratory setting and students received course credits for taking part.
3.2 Procedure
The experiment was administered using an online survey in Qualtrics and began with asking participants to name a close friend of theirs. They were then presented with a hypothetical scenario in which they were asked to imagine being in direct competition with the named friend, such that only one of the two (i.e., themselves or the friend) would achieve a desired goal at the expense of the other. Participants were randomly assigned to one of two scenario conditions. The first was an “affiliative” scenario (N = 68) in which it was anticipated that participants would believe that it would benefit the friend’s chances, but harm the participant’s chances, if the friend were to feel happy. The second was a “confrontational” scenario (N = 69) in which it was anticipated that participants would believe that it would benefit the friend’s chances, but harm the participant’s chances, if the friend were to feel angry.
After reading the scenario, participants were shown three emotion-inducing newspaper headlines (expected to induce anger, happiness, and neutral emotion, respectively) in turn, and were told that they could show one story from the newspaper to their friend before the friend entered the situation described in the scenario. They then indicated their preferences for each of the headlines. Finally, participants rated the perceived utility of different emotions for their friend in the given scenario and completed a measure of their values.
3.3 Materials
3.3.1 Scenarios
We developed two scenarios in which participants were competing with a friend for a valued resource and had to choose between benefitting themselves or the friend through their interpersonal emotion regulation behavior. Based on the literature about the emotions that are most useful in different types of contexts (e.g., Frijda, 1986), we developed an “affiliative” scenario, in which the friend experiencing happiness was expected to benefit the friend and harm oneself. This scenario explained that the participant and their named friend were competing for a single place in a sorority and that they each individually had to play a social game with members of the sorority to determine which of the two would get the place. The “confrontational” scenario, in which the friend experiencing anger was expected to benefit the friend and harm oneself, explained that the participant and their named friend were competing for a single internship in a company and that they each individually had to take part in an aggressive negotiation with members of the company to determine who would get the internship.
A pilot study with 166 student participants who were randomly assigned to one of the two scenarios confirmed that participants expected happiness to be more beneficial for the friend’s performance in the affiliative (M = 5.96, SD = 1.20) than the confrontational scenario (M = 4.23, SD = 1.83; t(164) = 7.19, p < 0.01) and anger to be more beneficial for the friend’s performance in the confrontational (M = 3.87, SD = 1.98) than the affiliative scenario (M = 2.25, SD = 1.69; t(164) = −5.70, p < 0.01).
3.3.2 Preferences for emotion-inducing stimuli in others
Participants were presented with three emotion-eliciting newspaper headlines. The headlines used were those from Study 1, specifically the three that were developed into full articles for “player” participants. We collected two measures of preferences for these stimuli. The first was a rating of the extent to which participants wanted to expose their named friend to the full articles that corresponded with each of the headlines (1 = not at all, 7 = a lot). The second was a forced choice selection of which of the three news stories to show to their friend.
3.3.3 Expectations about the usefulness of emotions
Participants rated the extent to which they expected anger and irritation (α = 0.61) and happiness and cheerfulness (α = 0.85) in their partner to result in better performance in their given scenario (1 = not at all, 7 = a lot).
3.3.4 Values
Participants completed a four-item measure of values, which was adapted from the prosocial values factor from the Volunteer Functions Inventory (Clary et al., 1998). The original measure comprised items specifically relating to values that drive volunteering behaviors. In our adaptation, we reworded some items slightly to refer to more general values. The final items were: “I feel compassion toward other people”; “I am genuinely concerned about my friends”; “I feel it is important to help others”; “Helping my friends is important to me.” Participants rated their agreement with each statement (1 = strongly disagree, 7 = strongly agree; α = 0.73).
3.4 Results
In order to check that participants did expect happiness to be useful for the friend in the affiliative scenario and anger to be useful in the confrontational scenario, we conducted a repeated measures ANOVA on expectations about the usefulness of emotions, with scenario (affiliative, confrontational) as a between subjects factor and emotion (happiness, anger) as the within subjects factor. The results revealed a significant main effect of emotion (F(1, 135) = 83.52, p < 0.01, 
= 0.38) and an interaction between emotion and scenario (F(1, 135 = 20.49, p < 0.01, 
= 0.13). There was no main effect of scenario (F(1, 135) = 2.70, p = 0.10). Independent samples t-tests confirmed that participants believed that happiness would be more useful in the affiliative (M = 5.43, SD = 1.19) than the confrontational scenario (M = 4.54, SD = 1.82; t(135) = 3.36, p < 0.01). Conversely, they believed that anger would be more useful in the confrontational (M = 3.46, SD = 1.69) than the affiliative scenario (M = 2.20, SD = 1.31; t(135) = −4.83, p < 0.01).
3.4.1 Preferences for emotion-eliciting stimuli in others
We then examined whether there was any overriding tendency within our sample to act prosocially (i.e., for the friend’s benefit) or instrumentally (i.e., for their own benefit) in their interpersonal emotion regulation. A repeated measures ANOVA on participants’ rated preferences for the newspaper stimuli, with scenario (affiliative, confrontational) as a between subjects factor and headline-type (happy, angry, neutral) as the within subjects factor, revealed a significant main effect of headline-type (F(2, 134) = 13.86, p < 0.01, 
= 0.17), but no other effects (Fs < 2.48, ps > 0.09). Independent samples t-tests confirmed that there were no significant differences between participants’ preferences for the happy headline in the affiliative (M = 4.62, SD = 1.63) and confrontational scenarios (M = 4.29, SD = 1.96), t(135) = 1.07, p = 0.29), nor between their preferences for the angry headline across the two scenarios (Maffiliative = 3.26, SDaffiliative = 1.96; Mconfrontational = 3.30, SDconfrontational = 2.02), t(135) = −0.12, p = 0.91). A chi-squared test on participants’ ultimate choice of newspaper story for the friend (Table 1) likewise suggested no significant differences between the two scenarios in terms of the newspaper headlines selected for the friend (χ2(2) = 0.99, p = 0.61). Across both conditions, 55 participants acted prosocially (choosing the happy headline in the affiliative scenario or the angry headline in the confrontational scenario), while 46 acted instrumentally (making the reverse pattern of choices). The remaining 36 participants selected the neutral stimulus in their respective conditions. Together, these findings indicate that there was no overall tendency among the sample toward prosocial or instrumental interpersonal emotion regulation.
| Headline choice | |||
|---|---|---|---|
| Scenario type | Happy | Angry | Neutral |
| Affiliative | 38 | 13 | 17 |
| Confrontational | 33 | 17 | 19 |
3.4.2 Effects of participants’ values on interpersonal emotion regulation
To investigate whether participants’ preferences for emotion-eliciting stimuli in others could be predicted by their values, we first explored the effects of values on participants’ rated preferences for the newspaper stimuli, using a repeated measures ANCOVA, with scenario (affiliative, confrontational) as a between subjects factor, headline-type (happy, angry, neutral) as the within subjects factor, and values as the covariate (Figure 4). The results showed a significant two-way interaction between headline-type and scenario (F(2, 132) = 4.84, p < 0.01, 
= 0.07) and a three-way interaction between values, headline-type, and scenario (F(2, 132) = 4.94, p < 0.01, 
= 0.07). No other effects were significant (Fs < 0.81, ps > 0.42).
Between-subjects ANCOVAs, with scenario as the between subjects factor and values as the covariate, revealed a significant two-way interaction between scenario and values on preferences for happy headlines (F(1, 133) = 6.83, p < 0.01, 
= 0.05, and for angry headlines (F(1, 133) = 6.10, p < 0.05, 
= 0.04), but not neutral headlines (F(1, 133) = 1.65, p = 0.20). Correlation analyses showed that in the affiliative scenario, values of care and concern for others were negatively related to preferences for the angry headline (r = −0.32, p < 0.01), but unrelated to preferences for the happy headline (r = 0.21, p = 0.09). In the confrontational scenario, such values were negatively related to preferences for the happy headline (r = −0.25, p < 0.05), but unrelated to preferences for the angry headline (r = 0.08, p = 0.51). Thus, those participants with higher values of care and concern for others showed significantly lower preferences for inducing goal-inconsistent emotions in their friends (which would harm their friends, but benefit themselves) in both scenarios.
Next, we explored the effects of values on participants’ explicit choices of the newspaper stimuli. To do this, we selected only those participants who chose either the happy or angry headline for their friend in their given scenario (thus omitting the 36 participants who selected the neutral headline), then conducted a binary logistic regression to determine whether the combination of scenario and values could predict which of the two headlines participants would select. The regression, run using Hayes and Matthes’s (2009) macro, revealed a significant main effect of scenario (−2LL = 12.07, p < 0.05), and a significant interaction between scenario and values (−2LL = −2.07, p < 0.01), but no main effect of values (−2LL = 1.06, p = 0.06). Tests of the Johnson–Newman significance regions indicated two Johnson–Newman points. Scenario-type had a positive effect on headline choice at low to moderate levels of values (<4.80), meaning that people with lower to moderate values of care and concern for others were more likely to act instrumentally, choosing the happy rather than the angry headline for their friend in the confrontational scenario or the angry rather than the happy headline in the affiliative scenario. At very high levels of values (>6.45), scenario-type had a negative effect on headline choice, meaning that people with high values of care and concern for others were more likely to act prosocially, choosing the angry rather than the happy headline for their friend in the confrontational scenario or the happy rather than the angry headline in the affiliative scenario.
3.5 Discussion
The findings of our second study provide further evidence that both prosocial and instrumental motives underlie interpersonal emotion regulation. Crucially, when these motives were directly compared, such that participants could benefit a friend or themselves in a hypothetical scenario, no overall tendency emerged within the sample. Thus, consistent with Study 1, our findings suggest that both motives may be equally strong.
A possible alternative explanation for the findings that we inferred as being indicative of prosocial or instrumental motivation for interpersonal emotion regulation is that participants wanted to create a fair playing field. For example, we interpreted selection of an anger-inducing stimulus in the confrontational condition as being prosocially motivated (because it would boost the friend’s performance), but it is possible that participants wished to win the internship fairly, without unduly handicapping the friend’s performance. However, because we gave participants a neutral option in the stimuli they were able to select for a friend, we gave them a clear means by which to avoid influencing the friend’s performance. While a minority of participants selected this option when asked to choose a single newspaper story for the friend to read (suggesting that they were motivated by fairness), the majority chose to either boost or damage the friend’s chances, consistent with our interpretation of their regulatory actions as being prosocially or instrumentally motivated.
The findings of this study also give some indication as to what drives one motive over another, by identifying a key difference between those who elect to benefit others versus themselves. Specifically, we found that individual differences in values relating to care and concern for others may be associated with making more prosocial and less instrumental choices when people regulate their friends’ emotions.
4 GENERAL DISCUSSION
Helping others to achieve their goals may be an important process in forming and maintaining cooperative relationships, which explains why people often act in an altruistic manner, putting others’ concerns ahead of their own (Batson, 1995; Finkel & Rusbult, 2008; Grant, 2007). Given that emotions can drive adaptive behaviors that facilitate goal pursuit (e.g., Frijda, 1986), people may therefore regulate others’ emotions prosocially, in order to induce the emotions that are most useful for those others to achieve their goals. In the present research, we explored this possibility and, across two studies, demonstrated evidence that prosocial motives underlie people’s attempts to regulate the feelings of their friends. In Study 1, there was no obvious benefit for participants themselves to engage in prosocial interpersonal emotion regulation. Although participants might have anticipated feeling good when choosing to improve their coursemates’ emotions (e.g., Niven, Holman, & Totterdell, 2012), participants in this study also expressed preferences (and actions) for their coursemates to feel worse to help them achieve their goals. Moreover, in Study 2 there was actually a direct cost to participants of acting prosocially, in that benefitting their friend would harm their own goal pursuit, yet over half of the participants preferred to give their friends emotion-eliciting stimuli that were consistent with their friend’s (and not their own) goals.
We further tested the alternative possibility that people’s choices of interpersonal regulatory behavior might be driven by more instrumental concerns of self-benefit. It is well established that people regulate their own feelings in order to induce the emotions that will support their goals (e.g., Tamir, 2009). Yet we may need the support of others in order to achieve our goals and we may recruit that support in a proactive and strategic way, through the instrumental regulation of others’ emotions (Netzer et al., 2015). Our two studies also provided support for this motive driving interpersonal emotion regulation. In Study 1, people expressed preferences and regulatory behaviors consistent with inducing emotions that would benefit their own goals, even when those emotions would be experienced as unpleasant to the coursemate who they had been partnered with. In Study 2, just under half of our participants indicated a choice to give their friends emotion-eliciting stimuli that would benefit themselves, even though doing so would be costly to the friend.
A key contribution of the present research was to contrast instrumental and prosocial motives for interpersonal emotion regulation. Our findings suggest that these tendencies may be approximately equivalent in strength. In Study 1, we observed no differences in effects between the tendencies to benefit oneself versus others in interpersonal emotion regulation behavior. In Study 2, where participants had the opportunity to act either instrumentally or prosocially, there were no overall differences in people’s preferences toward or their rates of choosing particular emotion-eliciting stimuli, suggesting that the tendencies may occur equally among the population sampled. Our research also contributes by offering a potential explanation as to why some people may have a greater tendency toward prosociality versus instrumentality, with values of care and concern for others predicting differences in people’s interpersonal emotion regulation behavior.
Given that both prosocial and instrumental tendencies may underlie people’s attempts to regulate others’ emotions, an important question concerns the possibility of conflict between these two motivations. In close relationships, people often want to benefit their partner but at the same time to benefit themselves, and the regulatory behaviors needed to achieve these goals may sometimes be in direct conflict. For example, a person who feels worried might want to engage in interpersonal alerting to make his or her partner to feel worried in order to appreciate and act on the underlying concerns (instrumental motive) yet might also want to engage in interpersonal calming so that the partner can focus on his or her own goal pursuit (prosocial motive) (Parkinson, Simons, & Niven, 2016). In our research, we identified an individual difference factor (i.e., values of care and concern) that may predict people’s regulatory preferences in such cases of motivational conflict. Future studies should consider whether there may be other such individual differences (e.g., Machiavellianism, narcissism, competitiveness, or achievement orientation) that influence which motives prevail in directing people’s interpersonal emotion regulation behavior.
Future research may also wish to explore whether the type of relationship influences motivational tendencies. In our studies, we focused on interpersonal emotion regulation used toward friends. However, it is possible that instrumental tendencies weigh out when regulating the emotions of rivals (Netzer et al., 2015), whereas prosocial tendencies might be more prominent in closer relationships, for example, with romantic partners (Parkinson et al., 2016). Contextual factors might also be important to consider. For example, some contexts might be typically characterized by a more instrumental focus in social interactions, such as commercial settings (Henkel, Bögershausen, Hoegg, Aquino, & Lemmink, 2018). Changes in context might also shape people’s motives for interpersonal emotion regulation; for example, events that elicit empathy might enhance prosociality (Henkel, Bögershausen, Rafaeli, & Lemmink, 2017).
Another important direction for future research concerns the effects of interpersonal emotion regulation. Previous research has indicated that interpersonal emotion regulation may have important relational implications, for instance influencing the extent to which relationship partners trust and like each other (Niven, Holman, & Totterdell, 2012). However, it seems likely that people’s motives might shape the effects of interpersonal emotion regulation on relationships (Niven, 2016). When interpersonal emotion regulation is used with prosocial motives, the regulator may be more attentive to the target’s needs (Bolino, 1999) and the target may appraise the regulator’s goals in a more positive light (Van Kleef, 2009), leading to positive consequences for the relationship. Yet when interpersonal emotion regulation is used with instrumental motives, the regulator may pay scant attention to the target and the target may appraise the regulator more negatively, which could in theory undermine the quality of the regulator–target relationship.
As well as providing insight into why people regulate others’ emotions, a further contribution of the present research is to highlight a possible alternative account of why people regulate their own emotions. Combining the insights here with those from existing research (e.g., López-Pérez et al., 2017; Netzer et al., 2015; Tamir & Ford, 2012) suggests that people might regulate their own and others’ emotions motivated by instrumental concerns and that people might also regulate others’ emotions motivated by prosocial concerns. An additional possibility, therefore, is that people might also regulate their own feelings prosocially. Examples of prosocial emotion self-regulation might include feigning enthusiasm toward an unwanted gift (Gross, 1999), or suppressing anger to avoid hurting someone’s feelings (Martini & Busseri, 2012). The idea that emotion self-regulation might have prosocial motivational roots has not been explored in depth within the emotion regulation literature, but shares parallels with the related coping literature, in which researchers have advocated recognition of both antisocial (instrumental) and prosocial coping tendencies (e.g., Dunahoo, Hobfoll, Monnier, Hulsizer, & Johnson, 1998; Monnier, Hobfoll, Dunahoo, Hulsizer, & Johnson, 1998). Future research could therefore consider this alternative motivation for emotion self-regulation, to establish whether people’s goal-oriented self-regulatory behavior might extend toward benefitting the goals of others as well as themselves.
An important limitation of the present research is that in one of the studies we conducted (Study 2), participants only indicated their preferences for how another person would feel and for particular emotion-eliciting stimuli, rather than actively engaging in interpersonal emotion regulation. In Study 1, participants did believe that they were choosing an emotion-eliciting stimulus for a specific partner who they could see throughout the study; however, this is still one step away from actually regulating that partner’s emotions during a spontaneous interaction. The complexity of motives for interpersonal emotion regulation is likely to be greater than we have been able to simulate in experimental studies (e.g., there may be a complex interplay between motives in everyday situations). Studies of people’s interpersonal emotion regulation in the field will therefore be important to further enhance our understanding of the motives that underlie interpersonal emotion regulation, and would also allow insight into how the motives behind interpersonal emotion regulation influences its social consequences.
REFERENCES
- 1 The results suggested that each of the angry headlines was expected to elicit more anger (Ms ranged between 5.65 and 5.97) than fear (Ms ranged between 3.15 and 3.94, ts(141) > 12.02, ps < 0.01), happiness (Ms ranged between 1.26 and 1.49, ts(141) > 26.61, ps < 0.01), or neutral emotion (Ms ranged between 1.87 and 2.17, ts(141) > 18.69, ps < 0.01). Similarly, each of the happy headlines was expected to elicit more happiness (Ms ranged between 4.98 and 5.47) than anger (Ms ranged between 1.31 and 1.78, ts(141) > 20.35, ps < 0.01), fear (Ms ranged between 1.24 and 1.58, ts(141) > 24.99, ps < 0.01), or neutral emotion (Ms ranged between 3.00 and 3.65, ts(141) > 5.27, ps < 0.01). Finally, each of the neutral headlines was expected to elicit more neutral emotion (Ms ranged between 4.24 and 5.04) than anger (Ms ranged between 1.70 and 2.21, ts(141) > 8.72, ps < 0.01), fear (Ms ranged between 1.64 and 2.45, ts(141) > 7.49, ps < 0.01), or happiness (Ms ranged between 2.16 and 3.38, ts(141) > 6.85, ps < 0.01).
- 2 Those players who were given the angry news article experienced an increase from baseline (Mpre = 2.51, SDpre = 1.13) in anger after reading the article (Mpost = 4.64, SDpost = 1.52, t(42) = −7.80, p < 0.01) and a decrease in their happiness (Mpre = 4.65, SDpre = 1.16, Mpost = 2.92, SDpost = 1.38), t(42) = 5.95, p < 0.01. Those players who were given the happy news article experienced an increase from baseline (Mpre = 4.71, SDpre = 1.10) in happiness after reading the article (Mpost = 5.30, SDpost = 1.04, t(40) = −4.37, p < 0.01), and a decrease in anger (Mpre = 2.43, SDpre = 1.20, Mpost = 1.98, SDpost = 1.26, t(40) = 2.71, p < 0.01). Finally, those players who were given the neutral news article experienced no change in their happiness (Mpre = 4.78, SDpre = 1.08, Mpost = 4.58, SDpost = 1.28, t(44) = 1.15, p = 0.26), but a small increase in anger after reading the article (Mpre = 2.31, SDpre = 1.07, Mpost = 2.70, SDpost = 1.35, t(44) = −2.14, p < 0.05). Follow-up analyses confirmed, however, that players in the neutral news condition were significantly less angry after reading their article compared with those in the angry news condition (t(86) = −6.34, p < 0.01).
- 3 Repeating the analyses with a reduced dataset wherein all those who failed one or more manipulation check were removed did not change the directionality or significance of our results.
- 4 Exploratory analyses suggested that exposure to the news stories did not actually have the intended effects on players’ game performance. We standardized players’ scores around the mean score achieved within the sample for their respective games, in order to eliminate differences arising from variation in difficulty of the games. A 2 (game: aggressive, collaborative) by 3 (news story: angry, happy, neutral) ANOVA on standardized game score revealed a significant interaction between game and news story (F(2, 123) = 3.88, p < 0.05,
= 0.06), with no other significant effects (Fs < 1.89, p > 0.17). However, breaking down the interaction, we found that there was only a significant effect of news article on game score in the pizza game condition (F(2, 63) = 4.25, p < 0.05, 
= 0.12), with post-hoc tests suggesting that those players exposed to the neutral story (M = 294.77, SD = 664.62) outperformed those exposed to the angry story (M = −85.45, SD = 238.86, p < 0.05).
