Edited by: Jack van Honk, Utrecht University, Netherlands
Reviewed by: Barak Morgan, University of Cape Town, South Africa; Peter A. Bos, Utrecht University, Netherlands
*Correspondence: Benoit Bediou, Swiss Centre for Affective Sciences, University of Geneva, 7 rue des Battoirs, CH-1205 Geneva, Switzerland. e-mail:
This article was submitted to Frontiers in Emotion Science, a specialty of Frontiers in Psychology.
This is an open-access article subject to a non-exclusive license between the authors and Frontiers Media SA, which permits use, distribution and reproduction in other forums, provided the original authors and source are credited and other Frontiers conditions are complied with.
Research on self-serving biases in judgments and decision-making suggests that individuals first evaluate the outcomes they get, and then the procedures by which these outcomes were obtained. Evidence also suggests that the appraisal of the former (outcome favorability) can bias the appraisal of the latter (procedural fairness). We investigated the nature of the emotions that are elicited by these appraisals by using a new paradigm in which participants performed a choice task between pairs of competing gambles against a virtual opponent. Conflicts (when the participant selected the same gamble as his virtual opponent) were resolved by a neutral arbitrator who either confirmed the participant’s choice (“
Emotions are central to decision-making processes, both as an input and as an output (Han and Lerner,
Third, the comparison of one’s outcome with a better or worse outcome received by someone else, like in competitive settings, elicits envy or, respectively, gloating and schadenfreude (Heider,
An important determinant of one’s reaction to the outcome of a situation is the appraisal of justice. Perceived injustice is a primary elicitor of anger and guilt (Mikula et al.,
An extensive body of research suggests the existence of self-serving biases in the perception of justice (Walster et al.,
Studies examining the impact of outcomes and procedures on specific emotions also suggest that there may be some interactive effects (Brockner and Wiesenfeld,
The present study examined the impact of random arbitration (the modification of a party’s decision using a random rule) on the emotional reactions to outcomes in a competitive gambling task. We used a simple choice task between pairs of competing gambles, similar to previous research (Coricelli et al.,
Our goal was to directly compare how individuals evaluate and respond to the outcomes of their decisions under three arbitration contexts: (i)
The fact that each arbitrator was not awarding an outcome but a gamble was important in order to dissociate the effects of outcomes from the effects of arbitration and study the impact of the former on the latter. Participants were informed that the arbitrator would do a 50:50 random choice (e.g., flip a coin) to decide whose choice would prevail. Thus, the arbitration procedure itself was neither intended to be perceived as fair or unfair, nor was it intended to be perceived as advantageous or not.
Twenty paid healthy volunteers (14 women, mean age 24.2 ± 5 year olds) recruited among the student population of the University of Geneva participated in this experiment in return for payment. The study was approved by the Geneva Psychology Research Ethical Committee, and written informed consent was obtained from all participants. Individuals with symptoms or a history of psychiatric care, neurological disease, or head injury were not included.
Subjects sat in a dimly lit sound-attenuated cabin, in front of a 17′′ computer screen placed at a viewing distance of 90 cm. Participants performed a computerized two-player gambling task in competition with a virtual opponent (see Figure
We used four different gambles that differed in terms of probabilities of winning (20, 40, 60, and 80%) and magnitude of potential gain (120, 60, 40, and 30 points) but the product of both was constant, i.e., equal expected value. For each pair of gambles, the probabilities of winning added up to 100% to ensure that there was a winner and a loser in each trial. In 32 trials, one of the gambles offered a 20% chance of winning 120 points and the other offered an 80% chance of winning 30 points. In another 32 trials, one of the gambles offered a 40% chance of winning 60 points and the other offered a 60% chance of winning 40 points. Thereafter, these trials are referred to as risky and safe trials respectively, because the variance between potential outcomes is respectively high and low (Figure
To increase participants’ motivation and involvement in the task and thereby their emotional reactions, the points gained during the task were cumulated across trials and paid in cash after the experiment (100 points = 1 CHF). To increase competition, we told our participants that they would receive a bonus, but that this bonus would only be granted if they scored higher than their opponent, whereas they would receive only a base participation fee if they scored below their opponent.
Participants were instructed to keep their non-dominant forearm resting on the table throughout the duration of the experiment. Once a participant had signed his consent and completed the personality questionnaires, physiological sensors were attached. After that, participants were asked to find a comfortable sitting position and reminded to avoid any unnecessary movements and speech during the procedure. Moreover, they were instructed to stay alert, to keep their eyes open, and to breathe normally. Participants were provided with a computer mouse and the instructions for the experimental task. Physiological signals were assessed using TEL 100 Remote Monitoring System of Biopac Systems (Santa Barbara, CA, USA) with separate settings for electromyographic and electrodermal activities. Signals were transferred from the experimental room to the MP100 Acquisition Unit (16 bit A/D conversion) in an adjacent room and stored on computer hard disk (sampling rate 1000 Hz).
To examine whether arbitration decisions affected physiological arousal in our experimental paradigm, we recorded electrodermal activity continuously during the task and analyzed skin conductance responses offline. Electrodermal activity was recorded (high-pass filter: 0.025 Hz) by the constant voltage method (0.5 V). Beckman Ag–AgCl electrodes (8-mm diameter active area) filled with a skin conductance paste (Biopac) were attached to the palm side of the middle phalanges of the second and third fingers of the participants’ non-dominant hand. Specific skin conductance responses (SCRs) to outcomes were measured in microSiemens (μS) and analyzed offline. They were scored as changes in conductance (greater than 0.02 μS) starting in the 1000–4000-ms interval after the onset of the stimulus (Dawson et al.,
To determine whether arbitration decisions affected emotion expression in our experimental paradigm, surface EMG activity of facial muscles was additionally recorded. As an index of facial expression, activity of the Corrugator Supercilii, Zygomaticus Major, and Frontalis muscles was measured on the left side of the face at a sampling rate of 1000 Hz. Prior to application of the electrodes, the designated sites on the skin surface were cleaned with distilled water and cotton pads and abraded using fine emery paper. EMG recordings were obtained with 4-mm miniature Beckman Ag/AgCl electrode pairs filled with conductance gel (Biopac). In a bipolar configuration, electrodes were placed above the left eyebrow for assessment of Corrugator Supercilii muscle activity, on the left cheek in the middle of the mouth-to-ear tip line for Zygomaticus Major activity and above the eye brows for Frontalis activity. Sensor placement followed recommendations by Fridlund and Cacioppo (
Analyses were confined to the 32 risky trials in which the variance between possible outcomes was high (Figure
Before testing the effect of arbitration, it was important to show that our new paradigm indeed elicited emotional reactions in our participants. Therefore, we first examined whether the outcomes of gambles chosen by participants elicited the predicted pattern of emotional reactions, in the absence of arbitration. These analyses therefore focused on the no-conflict trials.
Because in social settings, and contrary to private ones, gains loom larger than losses (Bault et al.,
We hypothesized that the same outcome would be appraised differently and hence would lead to differential emotional responses under the various arbitration conditions. We predicted that participants’ ratings of anger and guilt would reflect the interactive effect of outcome favorability and procedural injustice (Weiss et al.,
In addition to self-reported subjective emotions, we also expected an effect of outcomes and arbitration decisions on physiological arousal and facial expressive behavior. In the component process model (CPM) of emotion (Scherer et al.,
We expected effects of outcomes and arbitration on physiological responses and facial expressions, mirroring self-reports (i.e., increased zygomaticus activity and SCR for gains compared to losses (Bault et al.,
Statistical effects of outcomes on emotions are summarized in Table
Self-reports | Physiology | ||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|
Ang. | Dis. | Hap. | Reg. | Env. | Guilt | Zygo.* | Corru. | Front. | S.C.R.* | ||
Valence | High-loss (0–120) vs. High-win (120–0) | ↗ | ↗ | ↘ | ↗ | ↗ | ↗ | ↘ | ↘ | ||
Low-loss (0–30) vs. Low-win (30–0) | ↗ | ↗ | ↘ | ↗ | ↗ | ↗ | |||||
Magnitude | High-loss (0–120) vs. Low-loss (0–30) | ||||||||||
High-win (120–0) vs. Low-win (30–0) | ↗ | ↗ |
*
Analyses of the no-conflict trials showed a significant effect of outcome on all subjective assessments of emotions (
Outcomes had a significant effect on the activity of the zygomaticus (
In sum, results show an effect of outcome valence on self-reported emotions, SCR, and zygomaticus activity. Additional effects of outcome magnitude were observed on self-reports of happiness and SCR.
Figure
Arbitration had a significant effect on self-reported anger for high-loss outcomes (0–120:
Arbitration significantly affected self-reports of guilt for low-loss and high-win outcomes (0–30:
Arbitration significantly affected self-reports of happiness for low-win outcomes (
Arbitration did not affect self-reports of disappointment or regret. Friedman analyses revealed a significant effect of arbitral awards on envy in low-win outcomes (30–0:
For technical reasons, data from two subjects could not be recorded. Hence, the analyses were performed on 18 subjects instead of 20. Friedman analyses showed an effect of arbitral awards on the activity of the frontalis for low-win outcomes (30–0:
In sum, results show an effect of random arbitration on self-reports of anger, happiness, and guilt, depending on the final gambling outcome. Importantly, these effects were observed only when the participant’s initial choice was affected by the arbitrator’s decision.
This study investigated the consequences of random arbitration (using a 50:50 rule, such as flipping a coin) on the emotional reactions to outcomes of gambles in a competitive task.
In no-conflict trials, the outcomes of the gambles elicited the predicted pattern of emotional responses. Participants reported greater happiness on wins compared to losses (valence effect) and they reported increasing levels of happiness with increasing amount of gains (magnitude effect). In contrast, they reported more anger, regret, and disappointment with losses compared to wins (valence effect). Participants also reported more envy after a loss than after a win. These results can be considered as evidence for the validity of our task because they suggest that the emotions induced by our task are consistent with the current literature (Mellers,
Our results question the specificity of the effects that have been attributed to regret and envy in some previous studies (e.g., Coricelli et al.,
Consistent with an impact of arbitration on appraisals of injustice (Weiss and Cropanzano,
Our study replicates and extends the existing literature on distributive and procedural justice, and their interaction (Brockner and Wiesenfeld,
The present results are consistent with appraisal theories of emotions, which postulate that emotions arise from the evaluation of a situation with regards to an individual’s concerns (Frijda,
Appraisal theories also discuss the consequences of emotions for individual and social behavior (Davidson et al.,
To conclude, the present study provided two major findings. First, we showed that the effect of outcome favorability on participants’ emotions depends on both the valence (winning or losing the trial) and the magnitude (low or high amount) of the outcome. Winning increases positive emotions and decreases negative emotions compared to losing, and vice versa, but there was no significant difference between disappointment, regret, and envy. Further research should address more directly whether counterfactual thinking and social comparison affect these emotions differentially, as some previous studies suggest. Second, we showed that biases in the appraisal of outcome favorability can influence the appraisal of procedural fairness in an egocentric way. Although the outcomes were determined by gambles and were thus independent from the arbitration process, we found that arbitration using a neutral, objective, and impartial rule (a computer programmed to give equal numbers of pro-self and pro-competitor decisions to all participants) affected self-reports of anger, and to a lesser extent guilt, when the outcome of the received gamble was unfavorable and favorable, respectively. Importantly, these effects were observed for pro-competitor awards only, and despite the explanation that the arbitrator would decide randomly in each case, using a 50:50 coin flip, who would keep the selected gamble. Moreover, the gambles were also programmed to ensure an equal number of wins and losses after each arbitration decision. These results are consistent with the literature on self-serving biases in the perception of justice, and may be explained by the impact of outcome favorability on procedural justice. Critical factors are (a) whether one is personally affected by the process (e.g., self-serving biases in attribution, which may be activated by the felt dispossession), and (b) whether the received outcome is favorable or not (e.g., self-serving biases in appraisals of outcome favorability). Future research should clarify the role of expectations and attributions (e.g., causality, intentionality, responsibility) in appraisals of distributive and procedural justice, and their impact on emotions and self-serving biases in third-party conflict resolution. More significantly, compared to no-conflict trials and to pro-self awards with similar outcomes, pro-competitor decisions by a neutral arbitrator resulted in increased subjective reports of anger for unfavorable outcomes, and increased happiness and guilt for favorable outcomes. Despite the use of a neutral computer, which was programmed to ensure outcome equality, but was presented as being an arbitrator flipping a coin in each case with even odds, individuals emotional reactions may reflect their tendency to perceive a loss as being due to a bias in the arbitrator’s decision, and to attribute their unsuccessful outcomes to an unfair arbitrator, rather than to an unfortunate gamble. This suggests that it may be impossible for arbitrators to ever be perceived as being neutral, objective, and independent, all the more so when they are not constrained to a random outcome generation process (e.g., flipping a coin), but when having to apply a legal syllogism whereby the facts and the law determine an outcome. Further research on this is important, as well as assessing emotional perceptions of bias in other forms of dispute resolution compared to arbitration (e.g., mediation or conciliation).
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
This study was supported by the National Competence Centre for Research (NCCR) in the Affective Sciences, hosted by the University of Geneva. We would like to thank Leonie Koban and Patrik Vuilleumier for discussions concerning the task and the different types of conflict.
1In case of loss, the participant won nothing, whereas his or her opponent won the amount of his gamble. Although the outcome is a no-gain, we call it a loss trial considering both the competitive context and the social comparison of one’s own outcome with the opponent’s outcome.
2We chose a −500 to 10,000-ms time window to include the whole contraction until the return to baseline. Analyses with a −500 to 3,000-ms showed similar results.
3We performed additional repeated-measures ANOVAs on our data after transformation to fill missing values with linear trend interpolation and with mean substitution in order to test for the interactive effects of outcomes and arbitration. These analyses confirmed our results and are kept available to the reviewers if required.