Cruel to Be Kind: Factors Underlying Altruistic Efforts to Worsen Another Person’s Mood

Abstract

When aiming to improve another person’s long-term well-being, people may choose to induce a negative emotion in that person in the short term. We labeled this form of agent-target interpersonal emotion regulation altruistic affect worsening and hypothesized that it may happen when three conditions are met: (a) The agent experiences empathic concern for the target of the affect-worsening process, (b) the negative emotion to be induced helps the target achieve a goal (e.g., anger for confrontation or fear for avoidance), and (c) there is no benefit for the agent. This hypothesis was tested by manipulating perspective-taking instructions and the goal to be achieved while participants (N = 140) played a computer-based video game. Participants following other-oriented perspective-taking instructions, compared with those following objective perspective-taking instructions, decided to induce more anger in a supposed fellow participant who was working to achieve a confrontation goal and to induce more fear in a supposed fellow participant who was working to achieve an avoidance goal.

Keywords

altruistic affect worsening interpersonal emotion regulation goal perspective taking emotion

Sometimes people cause a loved one to experience a negative emotional state if they think that it will increase that other person’s (long-term) well-being. But why would feeling bad be beneficial? According to an instrumental approach to emotion regulation, people may choose to feel a negative emotion in the short term if doing so maximizes the attainment of a specific long-term goal (Erber & Erber, 2000; Tamir, 2009; Tamir & Ford, 2009). For example, people may choose to feel anger when pursuing confrontation goals (e.g., dealing with someone who cheated) or fear when pursuing avoidance goals (e.g., escaping from a scary situation) because these negative emotions are seen as beneficial for achieving these specific goals (Tamir & Ford, 2009; Tamir, Mitchell, & Gross, 2008).

Research on the regulation of other people’s emotions (i.e., interpersonal emotion regulation; Gross & Thompson, 2007) has for a long time followed a hedonic approach that suggests people may attempt to decrease another person’s negative emotions if those emotions are perceived as harmful for that other person (Zaki & Williams, 2013). However, people may also engage in what we term instrumental affect worsening when regulating others’ emotions: An agent may choose to make a target feel bad if (a) this negative emotion allows the target to achieve a goal and (b) the agent him- or herself can benefit from this interpersonal emotion regulation by obtaining a desirable outcome (Netzer, Van Kleef, & Tamir, 2015). In this case, instrumental interpersonal affect worsening would be purely egoistically motivated. But would people choose to induce negative emotions in others for altruistic reasons? Or, put differently, would an agent make another person feel bad in the short term if this negative affect entailed a benefit solely for the target of the regulation process and not for the agent him- or herself? The aim of this study was to investigate conditions for such altruistic affect worsening.

We suggest that three conditions must be met for altruistic affect worsening to happen. First, the agent’s motivation has to be altruistic—that is, the final aim of the agent’s action must be to increase the target’s well-being rather than to obtain a personal benefit or goal, according to the classic definition of altruistic motivation (Batson, 2011). Second, the agent must aim to instill a negative emotion in the target that is beneficial for the target’s goal pursuit (e.g., making the target feel anger to achieve confrontation goals or fear to achieve avoidance goals; Netzer et al., 2015; Tamir & Ford, 2009). In these situations, affect worsening is seen as a means to an end, not an end itself (Niven, Totterdell, & Holman, 2009; Tamir, 2016). Third, altruistic motivation and altruistic affect worsening will happen when the agent empathizes with the target.

To test these claims, we reconciled two different research traditions. First, we drew on Batson and his colleagues’ (e.g., Batson, 2011) experimental methods to manipulate empathic concern through perspective-taking instructions. People who received other-oriented perspective-taking instructions (e.g., to imagine how another person was feeling in a certain situation) have been shown to be more likely to experience empathic concern and to act altruistically than people who received objective perspective-taking instructions (Batson, Early, & Salvarani, 1997). So far, this line of research has focused on behavior alone as a means to increase others’ benefits or decrease others’ suffering (e.g., taking electric shocks on behalf of another person; Batson, Duncan, Ackerman, Buckley, & Birch, 1981). However, it has not been investigated whether empathic concern leads people to use emotions to benefit others (e.g., to help them achieve a specific goal). Second, we relied on Tamir and her colleagues’ (e.g., Tamir, 2016) procedures to study people’s explicit and implicit emotional preferences for others and perceptions of emotion utility. Although this research has shown that people may choose to make others feel bad if they themselves benefit (Netzer et al., 2015), it has not assessed whether people engage in affect worsening for the sole benefit of another person. Thus, we sought not only to extend previous research findings using reliable experimental designs but also to bridge different traditions to expand current knowledge on interpersonal emotion regulation. This may enhance understanding of the dynamics of social interaction and social cognition by providing further information about adults’ emotion-outcome expectancies and how people balance (emotional) costs and benefits when regulating others’ emotions.

To study altruistic affect worsening, we focused on the process model of emotion regulation (Gross, 2007), which posits that people may change their own and others’ emotions by selecting a strategy that influences a particular stage of the emotion process. People may change emotions by selecting or modifying a situation (e.g., not going to a party), diverting attention (e.g., looking away), changing what they think about the situation (e.g., reappraisal), or altering their physiological response (e.g., suppression). We focused on the strategy of situation selection, which involves selecting or avoiding a stimulus or a situation in order to experience a specific emotion (Gross, 2007). Previous research (e.g., Netzer et al., 2015; Tamir et al., 2008) showed that participants selected different emotion-inducing stimuli to change the emotional experience of themselves or others in order to attain specific goals. We hypothesized that, compared with participants in an objective perspective-taking condition, participants in an other-oriented perspective-taking condition should (a) select more negative emotional stimuli for a target at the risk of lowering their own chances of earning £50 (empathy hypothesis), as long as (b) the negative stimuli are consistent with the target’s goal. That is, angry emotional stimuli should be chosen in a confrontation-goal condition, and fearful emotional stimuli in an avoidance-goal condition (beneficial-goal hypothesis). Furthermore, participants in an other-oriented perspective-taking condition (c) should rate their chosen emotional stimuli as more beneficial than the nonchosen stimuli for the target’s pursuit of his or her goals (altruistic-motivation hypothesis).

Method

Participants

One hundred forty adults (86 women, 54 men; mean age = 30.85 years, SD = 13.68, age range: 18–71 years) were recruited from a paid pool at the authors’ institution and completed the study in exchange for payment (£4, approximately $6). An a priori power analysis showed that 35 participants per condition would provide 80% power to detect an effect size ( f² ) of 0.05.

Design

We employed a 2 × 2 between-subjects design with the two independent factors of perspective taking (other-oriented, objective) and goal (confrontation, avoidance). Participants were randomly allocated to one of the four conditions.

Procedure

The study was presented as an examination of performance on one of two computer-based video games. Participants were tested in groups of 4. Closeness between participants was controlled by making sure that participants did not know each other. Each participant completed the study in a separate cubicle. After signing the consent form, participants rated their current mood using the Positive and Negative Affect Schedule (Watson, Clark, & Tellegen, 1988). Then, as a cover story, participants were told that they would be paired with another anonymous person. If assigned the role of Player A, they had to write down a personal statement so that Player B could get to know them before making choices for them in the game. If assigned the role of Player B, they would receive a personal statement from Player A before making choices for him or her in the game. In fact, all participants were assigned the role of Player B.

Participants were told that prior to reading Player A’s note, they would receive instructions to make sure they all had a similar emotional experience. Participants in the other-oriented perspective-taking condition were asked to imagine how the other player would feel in the described scenario, whereas participants in the objective perspective-taking condition were told to remain detached about the note (see Batson et al., 2007). Given that we aimed to test whether people engaged in affect worsening for altruistic reasons and that an altruistic motivation is more likely to be evoked when a person experiences empathic concern (Batson, 2011), we decided to include only those manipulations that would most likely isolate the experience of empathic concern from the experience of personal distress (i.e., the other-oriented vs. objective perspective-taking instructions; see Batson et al., 1988). Therefore, we decided not to include a no-instruction condition because (a) previous research has shown that when not given any instructions, people tend to take another person’s perspective (see Batson, 2011, for a review) and (b) a no-instruction condition could increase both empathic concern and personal distress, as these emotions usually co-occur (e.g., Barraza & Zak, 2009). Hence, in a no-instruction condition, participants might experience empathic concern and personal distress, which might entail both altruistic and egoistic motivation (Batson, 2011).

Next, participants received a sealed envelope with a purported handwritten communication from Player A that described Player A’s recent breakup and how upset and helpless Player A was feeling about it (taken from Batson et al., 2007). This note was intended to provoke empathic concern in the participants. After reading Player A’s note, participants rated how they felt toward Player A using Batson, Fultz, and Schoenrade’s (1987) empathic-response questionnaire.

In addition, participants were tested in one of two goal-pursuit conditions. We had participants in the two conditions play different games so that we could manipulate the goal to be achieved. In the confrontation-goal condition, the actual participants (as well as their supposed partners) were asked to play the game Soldier of Fortune, a first-person shooter game with a clear confrontation goal (i.e., to kill as many enemies as possible; see Netzer et al., 2015; Tamir et al., 2008). In the avoidance-goal condition, participants were asked to play the game Escape Dead Island, a first-person game with the goal of avoidance (i.e., escaping from a room without being killed by zombies). We chose this game to test preferences toward fear as a means of facilitating an avoidance goal, because previous literature has extensively linked avoidance behavior with fear (Carver, 2001; Frijda, 1986; Öhman, 1993). Participants were informed that according to their own individual performance (i.e., number of individuals killed in Soldier of Fortune and distance traveled in Escape Dead Island), they would receive a number of tickets in a raffle to win £50 in Amazon vouchers. After 5 min, participants were asked to stop playing.

Assessment of dependent variables

Participants were told that they had to make several choices before their partner could start playing. They were reminded that their choices might improve or worsen their partner’s performance. Thus, if they selected stimuli that improved the partner’s performance, they might lower their chances of receiving £50, whereas if the stimuli they selected worsened the partner’s performance, their likelihood of getting the prize would be higher. Before making their choices, participants were presented with different descriptions of the video game and different music clips that targeted specific emotions. As a measure of preferences for emotion-inducing stimuli, we had participants rate the extent to which they wanted their partner to read each description and listen to each music clip before or while playing the game. Then, as a measure of explicit emotion preferences, participants had to rate the extent to which they wanted their partner to feel angry, fearful, or neutral. We always asked participants to indicate their preferences for emotion-inducing stimuli before their explicit emotion preferences to avoid demand characteristics. Next, as a measure of perceived utility of emotions, participants had to rate the extent to which they thought anger, fear, or neutral emotion would be beneficial to success in the game. Finally, participants were fully debriefed.

Materials

Manipulation check

Positive and negative affect were measured with the Positive and Negative Affect Schedule (Watson et al., 1988). This is a 20-item questionnaire that assesses participants’ positive (α = .85) and negative affect (α = .84) using 5-point Likert-type scales from 1 (not at all) to 5 (extremely).

The 12-item version of the empathic-response questionnaire (Batson et al., 1987) was used to assess participants’ levels of empathic concern for and personal distress on behalf of Player A. Responses on this questionnaire are made using 7-point Likert-type scales from 1 (not at all) to 7 (extremely). Empathic concern was calculated as the average score for the items warmth, soft-hearted, tenderness, moved, compassionate, and sympathetic (α = .88). Personal distress was calculated as the average score for the items upset, grief, sorrow, distressed, worried, and anxious (α = .87).

Dependent variables

Preferences for emotion-inducing stimuli for the partner

Participants listened to two anger-inducing music clips (“Refuse/Resist” by Apocalyptica; “Mars, the Bringer of War” from Gustav Holst’s orchestral suite The Planets), two neutral music clips (“Treefingers” by Radiohead; “First Thing” by Four Tet), and two fear-inducing music clips (“The Bon Dam” by Julyan D; “The Hand of Fate – Part 1” by James Newton Howard), all used and validated by Netzer et al. (2015). Participants were also presented with three short game descriptions designed to elicit an angry, fearful, or neutral emotional state (Netzer et al., 2015). The anger-inducing game description described the main character fighting enemies after they had destroyed the character’s village. The fear-inducing game description described the main character surrounded by dangerous enemies who want to kill him or her. The neutral game description described the main character monitoring his or her surroundings. Participants were required to rate how much they wanted their partner to read each description and listen to each music clip, on a scale from 1 (not at all) to 7 (extremely). Given that the correlations between stimuli were high (anger-inducing stimuli: r = .63, p = .001; fear-inducing stimuli: r = .57, p = .001; neutral stimuli: r = .54, p = .001) and that we did not find differences in results when we entered type of stimulus as a within-subjects variable in subsequent analyses (see the Supplemental Material available online), we averaged the responses to the music-clip and game-description stimuli for each emotion, thereby creating an emotion-inducing-stimulus preference score for each emotion (anger, fear, and neutral emotion).

Explicit emotion preferences

Participants rated how much they wanted their partner to feel neutral, angry, or fearful while playing the game, using a scale from 1 (not at all) to 7 (extremely).

Perceived utility of emotions

Participants rated the extent to which feeling angry, neutral, or fearful would be helpful to success in the game, using a scale from 1 (not very helpful at all) to 7 (extremely helpful).

Results

Manipulation check

A multivariate analysis of variance (MANOVA) with perspective taking (other-oriented, objective) and goal (confrontation, avoidance) as independent factors and empathic concern and personal distress as dependent variables revealed a significant effect of perspective taking, F(1, 139) = 13.79, p = .001, η_p² = .09. Pairwise comparisons showed that participants in the other-oriented perspective-taking condition (M = 4.47, SD = 0.15) reported significantly higher empathic concern than did those in the objective perspective-taking condition (M = 3.70, SD = 0.14), p = .001. For personal distress, results of a MANOVA did not show a significant effect of the perspective-taking manipulation, F(1, 139) = 2.91, p = .09, η_p² = .02 (see Table 1 for means).

Table 1.

Means and Standard Deviations for Vicarious Emotions and Perceived Utility of Emotions by Experimental Condition

Variable	Other-oriented perspective		Objective perspective
Variable	Confrontation goal	Avoidance goal	Confrontation goal	Avoidance goal
Empathic concern	4.18_a (1.21)	4.75_a (0.96)	3.62_b (1.42)	3.77_b (1.29)
Personal distress	2.20_a (1.30)	2.33_a (1.02)	1.92_a (1.05)	1.99_a (0.94)
Anger utility	5.69_a (1.39)	2.66_b (1.64)	3.44_b (1.59)	3.74_b (2.11)
Fear utility	3.34_b (1.92)	6.09_a (1.31)	3.31_b (1.95)	3.09_b (1.99)
Neutral utility	3.71_a (1.91)	2.60_b (1.33)	3.42_a (1.86)	3.03_a (1.12)

Note: Standard deviations are shown in parentheses. Different subscripts within a row indicate a statistically significant difference (p < .05).

A MANOVA with positive and negative affect entered as the dependent variables showed no significant effect on positive affect for perspective taking, F(1, 139) = 0.97, p = .33, η_p² = .007; goal, F(1, 139) = 0.18, p = .67, η_p² = .001; or their interaction, F(1, 139) = 1.14, p = .29, η_p² = .008. Likewise, there was no significant effect on negative affect for perspective taking, F(1, 139) = 0.63, p = .43, η_p² = .005; goal, F(1, 139) = 0.48, p = .49, η_p² = .003; or their interaction, F(1, 139) = 0.58, p = .45, η_p² = .004.

Main analyses

Explicit emotion preference

Figure 1a shows explicit emotion preferences by condition. A repeated measures analysis of variance (ANOVA) with perspective taking (other-oriented, objective) and goal (confrontation, avoidance) as between-subjects variables and emotion (anger, fear, and neutral) as the within-subjects variable produced a significant Emotion × Perspective Taking × Goal interaction, F(1, 136) = 54.13, p = .001, η_p² = .28. In the confrontation-goal condition, participants in the other-oriented perspective-taking condition reported a higher preference for anger, F(1, 69) = 125.17, p = .001, η_p² = .65, than did participants in the objective perspective-taking condition. There were no differences between conditions for fear, F(1, 69) = 0.65, p = .42, η_p² = .009, or neutral emotion, F(1, 69) = 0.41, p = .53, η_p² = .006. Pairwise comparisons showed that participants in the objective perspective-taking condition did not differ in their ratings of explicit emotional preferences for the different emotions (ps > .09). However, participants in the other-oriented perspective-taking condition reported a higher preference for anger compared with fear, d = 3.20, SE = 0.30, p = .001, and neutral emotion, d = 2.89, SE = 0.27, p = .001. In the avoidance-goal condition, participants in the other-oriented perspective-taking condition reported a higher preference for fear, F(1, 69) = 76.62, p = .001, η_p² = .53, and a lower preference for neutral emotion, F(1, 69) = 7.59, p = .01, η_p² = .10, than did participants in the objective perspective-taking condition. There were no differences between conditions in preferences for anger, F(1, 69) = 0.51, p = .48, η_p² = .007. Pairwise comparisons showed that participants in the objective perspective-taking condition did not differ in their ratings of explicit emotional preferences for the different emotions (ps > .08). However, participants in the other-oriented perspective-taking condition reported a higher preference for fear compared with anger, d = 3.46, SE = 0.31, p = .001, and neutral emotion, d = 3.23, SE = 0.41, p = .001.

Fig. 1.

Experimental results: (a) mean explicit preference for the three emotions and (b) mean preference for the three types of emotion-inducing stimuli, separately for the two perspective-taking conditions and the two types of goals. Error bars indicate ±1 SE.

Preferences for emotion-inducing stimuli for the partner

Figure 1b shows preferences for each type of emotion-inducing stimulus by experimental condition (results for the game descriptions and music clips are combined; for separate analyses of the music clips and game descriptions, see the Supplemental Material). A repeated measures ANOVA with perspective taking (other-oriented, objective) and goal (confrontation, avoidance) as between-subjects variables and emotion (anger, fear, and neutral) as the within-subjects variable produced a significant Emotion × Perspective Taking × Goal interaction, F(1, 136) = 35.45, p = .001, η_p² = .21. In the confrontation-goal condition, participants in the other-oriented perspective-taking condition chose stimuli that were more anger inducing, F(1, 69) = 43.88, p = .001, η_p² = .39, and less fear inducing, F(1, 69) = 4.42, p = .04, η_p² = .06, or neutral, F(1, 69) = 11.70, p = .001, η_p² = .15, than did participants in the objective perspective-taking condition. In the avoidance-goal condition, participants in the other-oriented perspective-taking condition chose stimuli that were more fear inducing, F(1, 69) = 35.38, p = .001, η_p² = .34, and less neutral, F(1, 69) = 15.34, p = .001, η_p² = .18, than did participants in the objective perspective-taking condition. There were no differences between perspective-taking conditions for anger-inducing stimuli, F(1, 69) = 3.69, p = .06, η_p² = .05 (Fig. 1b). See Figure S1 in the Supplemental Material for results for each type of emotional stimuli.

Perceived utility of emotions

A repeated measures ANOVA with perspective taking (other-oriented, objective) and goal (confrontation, avoidance) as between-subjects factors and emotion (anger, fear, and neutral) as the within-subjects factor revealed a significant Emotion × Perspective Taking × Goal interaction, F(1, 136) = 32.04, p = .001, η_p² = .32. In the confrontation-goal condition, participants in the other-oriented perspective-taking condition perceived anger to be significantly more useful than did participants in the objective perspective-taking condition, F(1, 69) = 39.83, p = .001, η_p² = .37. There were no differences between perspective-taking conditions for fear, F(1, 69) = 0.01, p = .94, η_p² = .001, or neutral emotion, F(1, 69) = 0.44, p = .51, η_p² = .01. Participants expected anger to be more effective than fear, d = 2.34, SE = 0.39, p = .001, or neutral emotion, d = 1.97, SE = 0.41, p = .001 (Table 1). In the avoidance-goal condition, participants in the other-oriented perspective-taking condition perceived fear to be significantly more useful, F(1, 69) = 55.35, p = .001, η_p² = .45, and anger to be significantly less useful, F(1, 69) = 5.78, p = .02, η_p² = .08, than did participants in the objective perspective-taking condition. There were no differences between perspective-taking conditions for neutral emotion, F(1, 69) = 2.11, p = .15, η_p² = .03. Participants expected fear to be significantly more useful than anger, d = 3.43, SE = 0.39, p = .001, or neutral emotion, d = 3.49, SE = 0.37, p = .001 (Table 1).

Testing mediation of explicit emotion preferences and emotion-utility beliefs

Given that participants differed in their perception of emotion utility depending on the perspective-taking condition they were in, one might argue that participants’ preferences for emotion-inducing stimuli may have been driven by different emotion-utility beliefs rather than by explicit emotion preferences. In other words, the results may be explained by theory of mind rather than by altruistic affect worsening, because participants in the other-oriented perspective-taking condition may actually have been better at anticipating what emotion would be more beneficial for the target depending on the goal. To differentiate between these two alternative hypotheses, we conducted two moderated mediation analyses (i.e., one for preference for anger-inducing stimuli and one for preference for fear-inducing stimuli; see Fig. S2 in the Supplemental Material for a depiction of the models) using Mplus 7 software (Muthén & Muthén, 2012). Specifically, we investigated whether perceived emotion utility and explicit emotion preferences were significant mediators of the relationship between perspective taking and preferences for emotion-inducing stimuli across different goals (i.e., avoidance vs. confrontation). To formally test the mediation hypotheses, we used a bias-corrected bootstrap approach (1,000 bootstraps) to create a 95% confidence interval (CI) around the mediated path (ab). This method was used because it has more power to detect mediation effects than does the Sobel test (Hayes & Scharkow, 2013). If the 95% lower and upper CI limits did not include zero, we could conclude that the mediated effect was different from zero.

As depicted in Figure S2 in the Supplemental Material, we ran a path analysis in which perspective taking (0 = objective, 1 = other-oriented), goal (0 = avoidance, 1 = confrontation), and their interaction terms were exogenous independent variables. Explicit anger preference and perceived utility of anger were the mediators, and preference for anger-inducing stimuli was the final outcome. The effect of the interaction terms on both explicit anger preference (b = 3.49, SE = 0.49, p < .001) and perceived utility of anger (b = 3.33, SE = 0.58, p < .001) was statistically significant. Accordingly, we probed the effects of perspective taking on explicit anger preference and perceived utility of anger across the two goal conditions. Simple-slopes analysis indicated that for the avoidance condition, the unstandardized effect of perspective taking was not significant for explicit anger preference (b = −0.29, SE = 0.41, p = .49), but was significant for perceived utility of anger (b = −1.09, SE = 0.47, p = .02). Furthermore, only the effect of explicit anger preference (b = 0.26, SE = 0.07, p < .001), not perceived utility of anger (b = 0.06, SE = 0.06, p = .31), was significantly related to preference for anger-inducing stimuli. Next, the mediation analysis indicated that neither explicit anger preference (ab = −0.07, 95% CI = [−0.37, 0.11]) nor anger-utility belief (ab = −0.06, 95% CI = [−0.29, 0.04]) significantly mediated the effect of perspective taking on preference for anger-inducing stimuli.

For confrontation, simple-slopes analyses indicated that perspective taking significantly predicted explicit anger preference (b = 3.21, SE = 0.28, p < .001) and perceived utility of anger (b = 2.24, SE = 0.35, p < .001). Furthermore, only explicit anger preference (b = 0.26, SE = 0.07, p = .001), not perceived utility of anger (b = 0.06, SE = 0.06, p = .35), predicted preference for anger-inducing stimuli. The mediation analysis showed that explicit anger preference mediated the effect of perspective taking on preference for anger-inducing stimuli (ab = 0.83, 95% CI = [0.36, 1.34]). Following Kline (2011), we conducted a further sensitivity analysis by constraining the mediating paths to test whether the fit of the constrained model was significantly different from that of the unconstrained one. Results showed that although the comparative-fit index was good (i.e., .97), the chi-square of the constrained model showed a significant increase, χ²(2) = 9.03, p = .01, thereby attesting to the implausibility of the constraints.

For preference for fear-inducing stimuli, we entered as independent variables perspective taking, goal, and the interaction of both terms. Explicit fear preference and fear utility belief were entered as mediators, and preference for fear-inducing stimuli was the final outcome. There were statistically significant effects of the interaction terms on explicit fear preference (b = −3.55, SE = 0.55, p < .001) and perceived utility of fear (b = −2.96, SE = 0.62, p < .001). Accordingly, we probed the effects of perspective taking on explicit fear preference and perceived utility of fear across the two goal conditions. Simple-slopes analysis indicated that for the confrontation condition, the unstandardized effect of perspective taking on explicit fear preference (b = −0.32, SE = 0.39, p = .41) and on perceived utility of fear (b = 0.04, SE = 0.45, p = .93) were not significant. In addition, neither explicit fear preference (b = 0.19, SE = 0.10, p = .06) nor perceived utility of fear (b = −0.08, SE = 0.09, p = .41) was significantly related to preference for fear-inducing stimuli. The mediation analysis indicated that neither explicit fear preference (ab = −0.06, 95% CI = [−0.34, 0.07]) nor perceived utility of fear (ab = −0.003, 95% CI = [−0.15, 0.10]) significantly mediated the effect of perspective taking on preference for fear-inducing stimuli.

For avoidance, results showed that perspective taking significantly predicted explicit fear preference (b = 3.23, SE = 0.38, p < .001) and perceived utility of fear (b = 3.00, SE = 0.40, p < .001). As for confrontation, neither explicit fear preference (b = 0.19, SE = 0.10, p = .06) nor perceived utility of fear (b = −0.08, SE = 0.09, p = .41) predicted a preference for fear-inducing stimuli. The mediation analysis showed that explicit fear preference mediated the effect of perspective taking on preference for fear-inducing stimuli (ab = 0.62, 95% CI = [0.004, 1.28]), but not perceived utility of fear (ab = −0.23, 95% CI = [–0.88, 0.31]). Following Kline (2011), we applied an equality constraint to the mediating paths to test whether the fit of the model significantly worsened. Results showed that the constrained model did not have a worse fit, as the chi-square statistic was not significant, χ²(2) = 3.13, p = .21. In the constrained model, the mediated path from perspective taking to preference for fear-inducing stimuli via explicit fear preference no longer significantly mediated the effect (ab = 0.62, 95% CI = [0.004, 1.28]). Likewise, perceived utility of fear was not a significant mediator (ab = 0.15, 95% CI = [−0.10, 0.42]).

Taken together, these findings suggest that explicit anger preference was a significant mediator only when confrontation was the goal. For fear, we did not find a significant effect of our mediators. Therefore, our data did not provide empirical support for the alternative account—that is, the results cannot be explained by the possibility that participants following other-oriented perspective-taking instructions were better able to understand what emotions may be more appropriate for certain contexts than were participants following objective perspective-taking instructions.

Discussion

Previous research has suggested that people may avoid making partners feel bad because such negative emotions are perceived as harmful (hedonic approach; Zaki & Williams, 2013). Alternatively, people may engage in interpersonal affect worsening to obtain a personal benefit (instrumental interpersonal affect worsening; Netzer et al., 2015). Our results show that people may choose to be cruel to be kind. That is, agents may make a target feel bad to achieve a desired goal without reaping any benefits themselves (altruistic affect worsening). Participants in the other-oriented perspective-taking condition selected negative-emotion-inducing stimuli that could benefit the target’s performance in a video game, a finding that supports the empathy hypothesis. Furthermore, participants’ selection of stimuli was not random, because they wanted their partner to feel a specific negative emotion. One could argue that this exemplified affect maintenance rather than affect worsening, given that the ostensible partner was initially described as being upset. However, affect maintenance seems an unlikely explanation. Although anger, fear, and upset are all considered negatively valenced emotions, they differ in their levels of arousal (e.g., Feldman Barrett, 2011).

Finally, participants perceived a stimulus that would induce a particular negative emotion to be more beneficial than other stimuli for success in a particular game (a finding that supports the beneficial-goal hypothesis). Although one might argue that between-groups differences in the perceived utility of emotions indicated that participants in the other-oriented perspective-taking condition were better at anticipating which emotion was more suitable for each goal (confrontation vs. avoidance), this alternative hypothesis was not supported. In two moderated mediation analyses, only explicit emotion preferences were a significant mediator. Thus, participants were indeed cruel to be kind because they wanted the targets to experience a specific negative emotional response depending on the goal.

Our results support extensive research on altruism and empathy showing that people help others even when altruistic behaviors may not entail a personal benefit (Batson et al., 1988). This study extends these findings by revealing, for the first time, that people who experience empathic concern not only employ behavior to decrease others’ suffering but also may altruistically manipulate others’ negative affect if doing so will increase the others’ long-term well-being. Thus, the results support the altruistic-motivation hypothesis. Individuals’ efforts at worsening others’ affect may additionally be motivated by egoistic reasons, such as a desire for social recognition (Cialdini et al., 1987), or demand effects. However, these explanations are unlikely to explain the findings of the current study, given that participants were made aware that their choices were completely anonymous. Moreover, if participants’ choices were driven by demand effects (i.e., if participants selected stimuli similar to those they received), then similar patterns of results should have been obtained for participants in the two perspective-taking conditions. Finally, participants’ behavior could have been driven by a motivation to feel better themselves (Hareli & Hess, 2010). Although participants intentionally reduced their chances of receiving £50 by worsening their partner’s affect, this may have constituted only a low-cost action. Future research should therefore consider actions with a higher cost, such as volunteering time (e.g., Batson, 2011).

The findings of our study pose a challenging question: What are the limits of affect worsening when it is for the sake of another person’s well-being? It may be that an agent will initiate the affect-worsening process for another person’s well-being even if it is not necessary, and even if the agent only mistakenly perceives that the other person needs to feel bad in order to achieve long-term well-being (Hareli & Hess, 2010). Investigating the boundary conditions for altruistically and egoistically motivated interpersonal affect worsening will provide more information about adults’ emotion-outcome expectancies regarding others’ emotions, the cost-benefit calculations, and the factors agents may consider when inducing a negative emotion in another person. Next, we suggest a number of possibilities that can be explored in future research to test the boundary conditions of altruistic affect worsening.

In this study, we intentionally made sure that participants were unknown to each other in order to maximize internal validity. Yet research has shown that people may be more motivated to change the affect of people with whom they have close relationships (Butler, 2011). To maximize external validity, future research could employ diary studies in which people have to note information about the situation, the strategy, and the agent involved in the regulatory process (Parkinson & Simons, 2009). Video-recording dyads discussing real-life concerns or worries may also be helpful (Parkinson, Simons, & Niven, 2016). This line of research and the assessment of whether agents feel mixed emotions when engaging in affect worsening would clarify whether agents differentially experience difficulties when worsening a target’s affect depending on how close they feel to the target.

Other variables may impact agents’ responsiveness to a target’s emotions and goals. One of these variables may be empathic accuracy—the ability to optimally infer another person’s internal states (Ickes, 1997). High levels of empathic concern may lead to high responsiveness to a target only when empathic accuracy is high (Win-czewski, Bowen, & Collins, 2016). Another variable that may impact agents’ responsiveness is their perception of the target’s regulatory skills (Parkinson & Simons, 2012). In fact, the experimental procedure used in the present study depicted the ostensible partner as upset and without any hope of getting over a breakup in order to provoke an empathic emotional reaction in the participants, which may have affected the participants’ willingness to engage in affect worsening. Thus, future research could manipulate the agent’s perception of the target’s regulatory skills.

In this study, participants did not have the option to induce positive emotions in the target. Thus, we were unable to test if participants who experienced higher empathic concern might have wanted to increase the target’s well-being by selecting positive emotion-inducing stimuli, as suggested by the empathy-altruism hypothesis (Batson, 2011). Future research may need to include happiness-inducing emotional stimuli to test this alternative hypothesis. Furthermore, in our study and in the one by Netzer et al. (2015), participants could worsen the target’s affect only by selecting or modifying the situation. However, other research has investigated additional strategies that can be used to change other people’s emotional states (e.g., corumination; Parkinson & Simons, 2012), and future research may benefit by studying these strategies. In sum, the present research opens avenues to an exciting research program that may enhance knowledge of social cognition and interpersonal emotion regulation.

Footnotes

Acknowledgements

We would like to thank the School of Psychology at Plymouth University for providing funding for the study, Anita Todd for editing the manuscript, and Antonio Zuffiano (Liverpool Hope University), Belen Fernandez-Castilla (University of Leuven), and Becky Choma (Ryerson University) for providing help and advice with the moderated mediation statistical analysis.

Action Editor

Jamin Halberstadt served as action editor for this article.

Declaration of Conflicting Interests

The authors declared that they had no conflicts of interest with respect to their authorship or the publication of this article.

Funding

M. Gummerum was partly funded by Economic and Social Research Council Grant ES/K000942/1. Additional funding was provided by the School of Psychology at Plymouth University.

Supplemental Material

Additional supporting information can be found at

References

Barraza

J. A.

Zak

P. J.

(2009). Empathy toward strangers triggers oxytocin release and subsequent generosity. Annals of the New York Academy of Sciences, 1167, 182–189.

Batson

C. D.

(2011). Altruism in humans. New York, NY: Oxford University Press.

Batson

C. D.

Duncan

Ackerman

Buckley

Birch

(1981). Is empathic emotion a source of altruistic motivation? Journal of Personality and Social Psychology, 40, 290–302.

Batson

C. D.

Dyck

J. L.

Brandt

J. R.

Batson

J. G.

Powell

A. L.

McMaster

M. R.

Griffitt

(1988). Five studies testing two new egoistic alternatives to the empathy-altruism hypothesis. Journal of Personality and Social Psychology, 55, 52–77.

Batson

C. D.

Early

Salvarani

(1997). Perspective taking: Imagining how another feels versus imagining how you would feel. Personality and Social Psychology Bulletin, 23, 751–758.

Batson

C. D.

Fultz

Schoenrade

P. A.

(1987). Distress and empathy: Two qualitatively distinct vicarious emotions with different motivational consequences. Journal of Personality, 55, 19–39.

Batson

C. D.

Kennedy

C. L.

Nord

L. A.

Stocks

E. L.

Fleming

D. A.

Marzette

C. M.

. . . Zerger

(2007). Anger at unfairness: Is it moral outrage? European Journal of Social Psychology, 37, 1272–1285.

Butler

E. A.

(2011). Temporal interpersonal emotion systems: The “TIES” that form relationships. Personality and Social Psychology Review, 15, 367–393.

Carver

C. S.

(2001). Affect and the functional bases of behavior: On the dimensional structure of affective experience. Personality and Social Psychology Review, 5, 345–356.

10.

Cialdini

R. B.

Schaller

Houlihan

Arps

Fultz

Beaman

A. L.

(1987). Empathy-based helping: Is it selflessly or selfishly motivated? Journal of Personality and Social Psychology, 52, 749–758.

11.

Erber

M. W.

(2000). The self-regulation of moods: Second thoughts on the importance of happiness in everyday life. Psychological Inquiry, 11, 142–148.

12.

Feldman Barrett

. (2011). Constructing emotion. Psychological Topics, 3, 359–380.

13.

Frijda

N. H.

(1986). The emotions. Cambridge, England: Cambridge University Press.

14.

Gross

J. J.

(2007). Handbook of emotion regulation. New York, NY: Guilford Press.

15.

Gross

J. J.

Thompson

R. A.

(2007). Emotion regulation: Conceptual foundations. In Gross

J. J.

(Ed.), Handbook of emotion regulation (pp. 3–24). New York, NY: Guilford Press.

16.

Hareli

Hess

(2010). What emotional reactions can tell us about the nature of others: An appraisal perspective on person perception. Cognition & Emotion, 24, 128–140.

17.

Hayes

A. F.

Scharkow

(2013). The relative trustworthiness of inferential tests of the indirect effect in statistical mediation analysis: Does method really matter? Psychological Science, 24, 1918–1927.

18.

Ickes

(1997). Empathic accuracy. New York, NY: Guilford Press.

19.

Kline

R. B.

(2011). Principles and practice of structural equation modeling (3rd ed.). New York, NY: Guilford Press.

20.

Muthén

L. K.

Muthén

B. O.

(2012). Mplus user’s guide (7th ed.). Los Angeles, CA: Author.

21.

Netzer

Van Kleef

G. A.

Tamir

(2015). Interpersonal instrumental emotion regulation. Journal of Experimental Social Psychology, 58, 124–135.

22.

Niven

Totterdell

Holman

(2009). A classification of controlled interpersonal affect regulation strategies. Emotion, 9, 498–509.

23.

Öhman

(1993). Fear and anxiety as emotional phenomena. In Lewis

Haviland

J. M.

(Eds.), Handbook of emotions (pp. 511–536). New York, NY: Guilford Press.

24.

Parkinson

Simons

(2009). Affecting others: Social appraisal and emotion contagion in everyday decision making. Personality and Social Psychology Bulletin, 35, 1071–1084.

25.

Parkinson

Simons

(2012). Worry spreads: Interpersonal transfer of problem-related anxiety. Cognition & Emotion, 26, 462–479.

26.

Parkinson

Simons

Niven

(2016). Sharing concerns: Interpersonal worry regulation in romantic couples. Emotion, 16, 449–458.

27.

Tamir

(2009). What do people want to feel and why? Pleasure and utility in emotion regulation. Current Directions in Psychological Science, 18, 101–105.

28.

Tamir

(2016). Why do people regulate their emotions? A taxonomy of motives in emotion regulation. Personality and Social Psychology Review, 19, 199–222.

29.

Tamir

Ford

B. Q.

(2009). Choosing to be afraid: Preferences for fear as a function of goal pursuit. Emotion, 9, 488–497.

30.

Tamir

Mitchell

Gross

J. J.

(2008). Hedonic and instrumental motives in anger regulation. Psychological Science, 19, 324–328.

31.

Watson

Clark

L. A.

Tellegen

(1988). Development and validation of brief measures of positive and negative affect: The PANAS scales. Journal of Personality and Social Psychology, 54, 1063–1070.

32.

Winczewski

L. A.

Bowen

J. D.

Collins

N. L.

(2016). Is empathic accuracy enough to facilitate responsive behavior in dyadic interaction? Distinguishing ability from motivation. Psychological Science. Advance online publication. doi:10.1177/0956797615624491

33.

Zaki

Williams

W. C.

(2013). Interpersonal emotion regulation. Emotion, 13, 803–810.