“False positive” emotions, responsibility, and moral character
Introduction
“Everyone has told him and he knows there was nothing he could do and it's not his fault, but he can't sleep and he feels guilty about living life if she can't. We were to go to the beach yesterday, but he didn't go because he says if she can't go to the beach why should I get to go.”
D., referring to her husband, who accidentally killed another person
The above quote comes from the website accidentalimpacts.org, an online community that provides support for people who have, accidentally and without any fault, caused severe injury or death to another person. The testimonials on the site chronicle the experience of many individuals who live with feelings of deep guilt over the consequences of their accidental actions. Indeed, those feelings of guilt appear to be so ubiquitous that there is a section of the website dedicated to helping people deal with the moral injury caused by their accidental actions.
At first glance, cases like these seem puzzling. If an action was truly accidental, an individual should neither receive blame nor blame themselves for that action.1 There is a large body of work in the psychology of moral responsibility linking intentional action and the attribution of moral culpability (e.g., Cushman, 2008; Malle, Guglielmo, & Monroe, 2014; Shaver, 1985; Weiner, 1995), and an agent is more likely to be blamed when she intentionally brings about a harmful outcome (e.g., Cushman, 2008; Sloman, Fernbach, & Ewing, 2009). Accordingly, an agent who accidentally harms someone is likely to be judged as less blameworthy than an agent who intentionally harms someone (e.g., Armsby, 1971; Darley, Klosson, & Zanna, 1978; Darley & Shultz, 1990; Shultz, Wright, & Schleifer, 1986). These findings are consistent with normative accounts of moral blame or fault in philosophy and law that hold that an agent should only be blamed or faulted if the harm he caused was “in the sphere of the agent's rational control” (Royzman & Kumar, 2004; cf. Badar & Marchuk, 2013; Fischer & Ravizza, 1998; Perkins, 1939; Vargas, 2013).
There is some evidence that similar attributional processes are at work when agents evaluate their own actions. Making an attribution that one is morally responsible – that one intentionally caused a harmful/immoral outcome – often results in a feeling of guilt, suggesting that the agent is assigning at least partial responsibility for the negative outcome to themselves (e.g., Hoffman, 1982; Mandel & Dhami, 2005; Smith, Webster, Parrott, & Eyre, 2002; Weiner, Graham, & Chandler, 1982). For example, Mandel and Dhami (2005) found that the amount of guilt experienced by prisoners convicted of various crimes was strongly associated with their amount of self-blame. In the absence of moral responsibility, however, theories of blame would predict that people should feel little guilt for committing a purely accidental harm.
However, as we described above, there are a great number of people who cannot seem to avoid feeling guilty even when they do not meet the criteria for moral responsibility. The philosopher Bernard Williams discusses cases like these in his essay Moral Luck (Williams, 1981). He asks his readers to imagine an accident in which a lorry driver, through no fault of his own, runs over and kills a child. Distraught over what has happened, the imagined lorry driver feels a great deal of guilt. As Williams points out, it would seem to an observer that the driver should not feel guilty: “Doubtless, and rightly, people will … try to move the driver from this state of feeling, move him indeed from where he is to something more like the place of a spectator”. At the same time, Williams notes, observers would expect that the driver would need to be encouraged to take something more like a spectator's perspective on it, and “indeed some doubt would be felt about a driver who too blandly or readily moved to that position.” (Williams, 1981 p.28). That is, while surely observers would try to dissuade the lorry from feeling this form of guilt for something that was not his fault, Williams believes that if the driver were persuaded too quickly, it would raise some eyebrows.
These cases of guilt for accidental actions highlight two puzzles (Kamtekar & Nichols, 2019). First, why do agents feel guilty for accidental harms when observers would not blame them to the same degree? Second, why do observers both 1) judge that such agents should receive less blame or feel less guilt and 2) disapprove if they do not at least initially feel some guilt?
In the present paper, we aimed to examine this second puzzle by investigating the inferences that observers make of people who express (or fail to express) these “false positive” feelings (Sperber, 1996); that is, feelings that are not normatively appropriate but are nonetheless characteristically triggered by the situation. Feeling guilt for an accidental harm is a false positive response since you do not meet a necessary condition for guilt – that of being at fault. The distinction between false-positive and true positive emotions seems to apply to many kinds of emotions (see Kamtekar & Nichols, 2019 for discussion). Consider fear: if a person comes upon a rattlesnake on a trail, they will likely feel fear, and this is an appropriate or true positive instance of fear. The rattlesnake really does pose a danger. But people also often feel fear when they come upon a harmless garter snake. This would seem to be a false positive instance of fear, since the garter snake does not pose any danger.
One interesting question about false positive emotions is whether they are predictive of true positive emotions. If a person is not afraid of garter snakes does that mean they are likely to be unafraid of rattlesnakes? Will people rely on a person's false negative emotional responses to predict that person's true positive emotional responses? Our goal was to examine how people might use a specific person's display of a false positive moral emotions2 (such as guilt for an accident, or gratitude toward a person who was simply performing a basic duty) – as a predictor of whether that person would feel “true positive” emotions (such as feeling guilty when they have actually committed an intentional harm). We also aim to examine whether false positive moral emotions predict something good about an agent's moral character and behavior. An important reason to investigate gratitude – true positive as well as false positive – alongside guilt is that gratitude is free of one potential confound one might worry about in the case of true positive versus false positive guilt. This is that true positive guilt requires the commission of a wrong, for which the agent may be faulted, and which would by itself result in a lowered assessment of the agent's character, with or without any information about their feelings of guilt. This is not the case for gratitude, since the subject feeling true or false positive gratitude is different from the agent who is going over and above their duty versus merely doing their duty.
We base our hypotheses on a growing body of literature that emphasizes the role of character in our moral judgments – people appear not just to evaluate the morality of particular actions but also the agents who commit those actions (for reviews, see Helzer & Critcher, 2018; Pizarro & Tannenbaum, 2012; Uhlmann, Pizarro, & Diermeier, 2015). Evaluations of moral character play an important role in how we think of other people: people prioritize moral character traits over other traits when judging the general positivity of a person (Goodwin, Piazza, & Rozin, 2014) and define personal identity largely in moral terms (Strohminger & Nichols, 2014). Furthermore, judgments of a person's morality more strongly predict liking and respect for that person than do judgments of that person's competence and sociability (Hartley et al., 2016).
When evaluating an agent's moral character, people are seeking to uncover the agent's “moral-cognitive machinery” (Helzer & Critcher, 2018) – the set of underlying psychological mechanisms that govern how that agent behaves regarding moral situations. People seek to infer the agent's intentions, motives, desires, meta-desires, beliefs, and other mental states (Ames & Johar, 2009; Critcher, Inbar, & Pizarro, 2013; Fedotova, Fincher, Goodwin, & Rozin, 2011; Gray, Young, & Waytz, 2012; Pizarro, Uhlmann, & Salovey, 2003;). From these psychological inferences, observers can then attempt to predict how that agent will behave in the future. This is consistent with what we know about the mechanisms underlying social prediction more generally, where individuals infer an agent's enduring traits and their temporary mental states from observable behavior, and then use those trait and state inferences to predict the agent's future behavior (Tamir & Thornton, 2018).
One specific method used to infer moral character is to attend to the emotions an agent displays regarding their moral behavior (Brandt & Reyna, 2011). Observers treat affective displays as potential sources of information about the agent's intentions and desires (Higgins, 1998). Whereas displays of positive affect might indicate that the agent is claiming ownership or responsibility of the action (e.g., Tracy & Robins, 2008; Weiner, 1985), negative affect might indicate that the agent is distancing themselves or repudiating the action (e.g., Gold & Weiner, 2000). For example, agents are judged more favorably when they perform prosocial behavior with a positive emotional display (e.g., smiling) or harmful behavior with a negative emotional display (e.g., grimacing) compared to when they perform those behaviors without the same emotional displays (Ames & Johar, 2009). This dynamic appears to play out in criminal courts – a defendant's perceived remorse is one of the most important factors in jurors' decisions of whether to give a death sentence (Haney, Sontag, & Constanzo, 1994).
We hypothesized that even though blame and guilt are not normatively appropriate responses to having accidentally caused harm, an agent who fails to feel guilt for the accident will be considered atypical and judged as lacking in moral character, compared to an agent who does feel guilty for the accident. So, while it may be a normative error to feel guilt when one does not deserve blame, it is the sort of error that may benefit the agent because of what it communicates about their moral character.
In the current research, we investigated the relationship between expressions of false positive moral emotions (guilt and gratitude) and judgments of moral character (Studies 1–5) and the relationship between expression of false positive moral emotions and individual differences in moral traits (Study 6). Our main hypothesis was that observers would judge an agent who feels false positive moral emotions – one who feels guilt or gratitude in response to a situation that does not normatively warrant those emotions – to have a more positive moral character and to be more likely to feel those emotions in true positive cases than an agent who does not feel false positive moral emotions. See Table 1 for a summary of the studies and methods. All materials, data, analysis syntax, and preregistration information can be found on the Open Science Framework at https://osf.io/btwsq/. Per our preregistrations, analyses reported here exclude certain participants, although none of our conclusions are substantively altered if these participants are included (see OSF link).
Section snippets
Study 1
Our first study served as an initial test of our hypothesis, allowing us to examine the judgments that observers make of agents who feel the false positive moral emotions of guilt and gratitude. As our central focus, we wanted to test whether false positive moral emotions would be perceived as reliable predictors of an agent's moral character. We presented participants with two scenarios: one scenario involving an agent who felt guilt (or did not feel guilt) for an accident they caused, but for
Study 2
In Study 2 we aimed to both replicate and extend our findings from Study 1 by making several modifications to the materials and design. Specifically, we modified the guilt scenario to include a new set of conditions where the agent might be seen as having greater responsibility for the accident due to their own recklessness (i.e., having knowledge of the potential harmful consequences of an action and yet performing that action anyway). Varying whether an agent appears to have foreknowledge of
Study 3
In Study 3, we expanded our investigation connecting false positive emotions and judgments of moral character by including assessments of a wider array of emotions, in order to assess whether false positive expressions of nonmoral emotions would also be treated as predictors of a person's moral character. For example, if an agent were to feel fear at a harmless stimulus (i.e., a target that should not trigger fear), would observers infer that the agent has good moral character and would feel
Study 4
In Study 4, we investigated whether our previous findings showing that expressions of guilt influence character evaluations of an agent were a result of the agent having been described as expressing any guilt at all. That is, people may form a positive impression of anyone who feels guilty concerning a harmful outcome. In Studies 1–3, the guilt and no guilt conditions differ both in whether the agent expressed guilt for an accidental harm and also whether the agent expresses any guilt at all.
Study 5
In Study 5, we sought to investigate not just the judgments that people make for agents who express “false positive” guilt (or not) over accidental harms, but to explore whether this information influences behavior toward those agents—particularly in their willingness to trust agents in a social, interactive game (the “trust” game; Berg, Dickhaut, & McCabe, 1995). We predicted that individuals would be more likely to trust an agent who displayed false positive guilt compared to an agent who did
Study 6
In Studies 1–5, we found that participants judge agents who feel false positive moral emotions as having a better moral character. However, it is not clear from these results whether these judgments are accurate. Is there any evidence that participants who report false positive guilt are actually better people? To return to Bernard Williams' example (1981), are we right to doubt the moral character of the lorry driver who is too quick to abandon his guilt over having accidentally killed
General discussion
Collectively, our results support the hypothesis that false positive moral emotions are associated with both judgments of moral character (Studies 1–5) and traits associated with moral character (Study 6). We consistently found that observers use an agent's false positive experience of moral emotions (e.g., guilt, gratitude) to infer their underlying moral character, their social likability, and to predict both their future emotional responses and their future moral behavior. Specifically, we
Conclusion
We have provided evidence that observers use the experience of false positive moral emotions as predictors of an agent's underlying moral character, and as a way to predict an agent's future moral behavior. We have also provided initial evidence that individuals who report that they would experience false positive moral emotions may actually be more likely to possess good moral character. This research may help to understand cases in which observers blame agents very little for their accidents,
Open practices
All preregistration documentation, materials, data, and analysis scripts for these studies are available on the Open Science Framework at https://osf.io/btwsq/
Declaration of Competing Interest
None.
Acknowledgements
This work was presented as a talk at multiple academic conferences, including at SPSP 2020 and at Yale University. We would like to thank our audience members for their helpful comments and feedback during these presentations.
References (95)
- et al.
The mismeasure of morals: Antisocial personality traits predict utilitarian responses to moral dilemmas
Cognition
(2011) - et al.
Trust, reciprocity, and social history
Games and Economic Behavior
(1995) - et al.
Experimental methods: Between-subject and within-subject design
Journal of Economic Behavior & Organization
(2012) Crime and punishment: Distinguishing the roles of causal and intentional analyses in moral judgment
Cognition
(2008)- et al.
Who am I? The role of moral beliefs in children’s and adults’ understanding of identity
Journal of Experimental Social Psychology
(2018) Development of prosocial motivation: Empathy and guilt
- et al.
The folk concept of intentionality
Journal of Experimental Social Psychology
(1997) - et al.
“What I did” versus “what I might have done”: Effect of factual versus counterfactual thinking on blame, guilt, and shame in prisoners
Journal of Experimental Social Psychology
(2005) - et al.
Children are sensitive to norms of giving
Cognition
(2017) - et al.
Environmentally friendly consumer choices: Cultural differences in the self-regulatory function of anticipated pride and guilt
Journal of Environmental Psychology
(2014)