Asymmetric Causal Attributions to Environmental Influences for Prosocial Versus Antisocial Behavior

Abstract

Several recent studies have explored how people may favor different explanations for others’ behavior depending on the moral or evaluative valence of the behavior in question. This research tested whether people would be less willing to believe that a person’s environment played a role in causing her to exhibit antisocial (as compared to prosocial) behavior. In three experiments, participants read a description of a person engaging in either antisocial or prosocial behavior. Participants were less willing to endorse environmental causes of antisocial (versus prosocial) behavior when the environmental influence in question was witnessing others behaving similarly, either during childhood (Experiment 1) or recently (Experiment 2), or being directly encouraged by others to engage in the behavior described (Experiment 3). These results could be relevant to understanding why people resist attributing wrongdoing to causes outside of individual control in some cases.

Attempts to understand the reasons for people’s actions—so-called “behavior explanations”—have been called “a fundamental tool of social cognition” (Malle, 2004, p. 1). When people make inferences about the determinants of others’ behavior, the psychological processes involved extend far beyond a simple procedure of dispassionately weighing the evidence for a range of possible causes. In particular, evaluative judgments of a behavior—that is, whether it is perceived as good or bad, helpful or harmful, moral or immoral—seem to influence people’s reasoning regarding what brought the behavior about. For example, people are more likely to say that immoral behaviors, as opposed to other kinds of behavior, are a product of the actor’s free will (Clark et al., 2014). Compared to behaviors that produce positive outcomes, people are also more likely to say that behaviors that produce negative outcomes were performed intentionally or were a product of the actor’s intentions or decisions (Knobe, 2003, 2010; Leslie, Knobe, & Cohen, 2006). For negatively valenced behaviors, people tend to attribute their own behavior to situational causes, but others’ behavior to causes internal to the person, whereas the opposite pattern emerges when people explain positively valenced behaviors (Malle, 2006). Additionally, people are less likely to say that an individual’s genes were involved in causing her behavior if the behavior was antisocial than if it was prosocial (Lebowitz, Tabb, & Appelbaum, 2019, 2021). In the present work, we build on this finding of an asymmetry in genetic attributions for behavior by investigating whether non-genetic (i.e., environmental) attributions for prosocial and antisocial behavior show a parallel asymmetry.

To the extent that the asymmetry in genetic attributions is due to what has been termed “blame-validation processing,” a form of motivated reasoning characterized by a “proclivity to favor blame versus nonblame explanations for harmful events” (Alicke, 2000, p. 568; see also Lebowitz et al., 2019), we might expect a similar reluctance to endorse environmental attributions for antisocial behavior. That is, as with genetic causes, people may be relatively reluctant to attribute antisocial behavior to a wrongdoer’s environment out of concern that such an explanatory account would unjustly deflect blame or responsibility for the misdeeds away from the perpetrator. People might also assume that immoral (e.g., antisocial) behavior reflects an underlying disposition toward immorality (Reeder & Spores, 1983) and may therefore assume that situational or environmental influences are less relevant as causes of antisocial behavior.

Alternatively, people’s demonstrated relative resistance to making genetic attributions for antisocial behaviors may indicate a propensity to attribute them to environmental causes instead, insofar as genetic and environmental causes, or “nature and nurture,” are often seen as competing forces. If this is the case, one would not expect people to show the same kind of prosocial-antisocial asymmetry in environmental attributions that they appear to show for genetic attributions.

In the present research, we investigated whether the asymmetry previously observed in genetic attributions for prosocial and antisocial behavior would emerge when the causal explanations being tested were attributions to environmental influences. In three experiments, we presented participants with vignettes describing an individual engaging in either prosocial or antisocial behavior and examined whether there was an asymmetry in how willing participants were to attribute the two types of behavior to events and environments that the individual experienced. We also experimentally manipulated whether or not the participants were provided with an environmental explanation for the behavior they read about, to test whether this would moderate the pattern of results. All data and experimental procedures are available at https://osf.io/5xnbv/?view_only=14df98b82e044a7b99e6641c0e944ed7.

Experiment 1

Methods

Experiment 1 prompted participants to read the same descriptions of prosocial and antisocial behaviors that have been used used in previous research examining asymmetrical genetic attributions (Lebowitz et al., 2019). Here, however, we measured how strongly each type of behavior was attributed to the effects of an environmental influence that is often assumed to powerfully affect behavior: the actor’s childhood environment.

Participants and recruitment.

Participants in Experiment 1 were 606 U.S. adults (52% male, 47% female, 1% other or unknown gender) recruited through Amazon.com’s Mechanical Turk (MTurk), a platform that allows users to find online tasks to complete in exchange for monetary compensation (Buhrmester, Kwang, & Gosling, 2011). This sample size was estimated to provide more than 95% power to detect a small-to-medium effect (f = .15) of the difference between prosocial and antisocial behavior on environmental attribution ratings. Participants were offered $1 in exchange for completing the experiment.

Stimuli and procedures.

The stimuli used in Experiment 1 were identical to stimuli in prior work (Lebowitz et al., 2019, Studies 3–4). Participants were randomly assigned to read one of six vignettes (see Table 1) describing a woman named Jane in a particular scenario. Each vignette had a stem that was the same for all participants assigned to that vignette, as well as two possible endings: one that described Jane exhibiting antisocial behavior and one that described her exhibiting prosocial behavior; each participant was randomly assigned to one of these two behavior types. All participants then viewed a statement indicating that the behavior described was typical of Jane. In the antisocial condition, it read, “This is consistent with how Jane usually behaves: when she realizes that someone is in a difficult situation, she takes advantage of them for her own benefit, even if it means adding to the other person’s misfortune.” In the prosocial condition, it read instead, “This is consistent with how Jane usually behaves: when she realizes that someone needs help, she will provide assistance, even if it means inconveniencing herself.” This statement was followed by an additional paragraph that began, “Scientists have found that when people repeatedly witness certain kinds of behavior being performed by others when they are children, this can lead them to behave the same way themselves when they become adults.” Participants were randomly assigned to one of two endings for this paragraph. For those randomly assigned to receive an environmental explanation, the paragraph continued: “In fact, when Jane was growing up, she was surrounded by people who behaved the way that Jane did in the situation that you just read about. In other words, Jane’s childhood experiences lead her to behave the way she does in situations like these.” For those randomized to receive no environmental explanation, it read: “However, when Jane was growing up, nobody around her behaved the way that Jane did in the situation that you just read about. In other words, there is no evidence that Jane’s childhood experiences lead her to behave the way she does in situations like these.”

Table 1.

Vignettes used in Experiment 1.

Vignette Stem	Antisocial Ending	Prosocial Ending
Jane’s window looks out onto a busy sidewalk that has a large pothole where people often trip. When she sees elderly pedestrians approaching who look like they may trip and fall,	she gets out her phone and records the accident, then posts the embarrassing video to her YouTube channel to attract more subscribers and advertisers.	she stops whatever she is doing and opens the window to call out and warn them.
Jane works at a large company. When new employees start at the company,	she gives them misleading advice so that she will look better by comparison in the boss’s eyes.	she goes out of her way to check if they have any questions and helps them settle in.
Jane owns a company. When she notices a decline in the work performance of any of her employees,	she fires them to free up some money, which she uses to increase her own take-home pay.	she goes out of her way to check whether they are having any difficulties that she might be able to help with.
Jane works for an airline. When she finds bags that passengers have accidentally left on the plane after a flight,	she goes through the bags to see if there are any valuables that she can sell or keep for herself.	she devotes time outside of her normal duties and hours to help track down the passengers and reunite them with their belongings.
Jane is a taxi driver. When she sees tourists who are lost,	she purposely takes the least direct route to their destinations so that she can increase the fare and make more money for herself.	she picks them up and suggests the route to their destinations that she knows is most direct, to save them money on their fares, even though she will make less money for the trip.
Jane lives in a large city. When she sees homeless people who appear to be unconscious or asleep outdoors,	she steals the cups of money they have collected panhandling.	she goes out of her way to make sure that they are OK and gives them a bit of money.

Next, participants provided an environmental attribution rating. They were asked, “How much of a role do you think Jane’s childhood experiences play in Jane’s patterns of behavior that you just read about?” and provided their responses on a scale from “1 (No role or a very minor role)” to “7 (A very major role).” This response scale was the same one used in prior research measuring genetic attributions for prosocial and antisocial behavior (Lebowitz et al., 2019, 2021). Participants were then asked basic demographic questions, were debriefed as to the fictitious nature of the vignettes used in the experiment, and were provided with instructions for how to receive compensation for their participation.

Results and Discussion

Environmental attribution ratings were analyzed using a 2 (behavior: antisocial vs. prosocial) × 2 (environmental explanation: present vs. absent) × 6 (vignette) univariate ANOVA. This revealed, most importantly, that antisocial behavior evoked weaker environmental attributions (M=4.31, SD=2.11) than did prosocial behavior (M=4.70, SD=2.02), F (1, 582) = 10.80, p = .001, η_p² = .02 (see Figure 1). Unsurprisingly, environmental attributions were also stronger among participants for whom the environmental explanation was present than among those for whom it was absent, F (1, 582) = 486.05, p < .001, η_p² = .46. There were no significant two- or three-way interactions (all Fs ≤ 1.40, all ps > .22), and no significant main effect of vignette, F(5, 582) = 1.17, p = .322.

Figure 1.

Mean environmental attribution ratings, by condition, in Experiment 1. Means are collapsed across other independent variables because the additional experimental manipulations did not significantly moderate the antisocial-prosocial asymmetry in endorsement of environmental attributions. Error bars represent ±1 SE of the mean.

These results indicate that across the six scenarios, participants were less inclined to attribute antisocial behavior, as compared to prosocial behavior, to the perpetrator’s childhood environment. When a statement explicitly stating that the behavior described was influenced by the actor’s childhood experiences was present, this yielded stronger environmental attributions than when such a statement was not provided, but this was not a significant moderator of the asymmetry in environmental attributions between antisocial and prosocial behavior.

Experiment 2

The findings of Experiment 1, suggesting that people are less willing to view antisocial behavior as stemming from the actor’s childhood environments, are similar to the results observed when participants in previous research were asked to rate their genetic attributions for antisocial and prosocial behavior. Indeed, this may be unsurprising, as the impact of childhood experiences may be perceived, like genetics, as affecting a person’s fundamental makeup in a way that is unlikely to vary substantially over time. Thus, in Experiment 2, we examined whether a similar asymmetry in environmental attributions would be observed when the environmental influences in question occurred recently (i.e., during adulthood) and thus seemed less likely to be viewed through an essentialist lens similar to that often applied to reasoning about genetic attributions.

Methods

Participants and recruitment.

Participants in Experiment 2 were 605 U.S. adults (43.8 male, 54.9% female, 1.3% other or unknown gender) recruited via Prolific, an online platform that allows individuals to sign up to participate in online research studies in exchange for payment (Palan & Schitter, 2018). This sample size was estimated to provide more than 95% power to detect a small-to-medium effect (f = .15) of the difference between prosocial and antisocial behavior on environmental attribution ratings. Participants were again offered $1 in exchange for completing the procedures.

Stimuli and procedures.

The stimuli and measures were identical to those used in Experiment 1, except for the wording of the environmental attribution item and the final paragraph displayed under the vignette, both of which in Experiment 2 pertained to the behavior of people around Jane in recent months rather than during her childhood. In particular, the final paragraph displayed under the vignette was reworded to begin, “Scientists have found that when people repeatedly witness others engaging in certain types of behavior, this can lead them to behave the same way.” Participants were randomly assigned to one of two endings for this paragraph. For those who received an environmental explanation, the paragraph continued: “In fact, over the past few months, Jane has been surrounded by people who behave the way that Jane did in the situation that you just read about. In other words, Jane’s recent experiences lead her to behave the way she does in situations like these.” For those randomized to receive no environmental explanation, it instead said: “However, over the past few months, nobody around Jane has behaved the way that Jane did in the situation that you just read about. In other words, there is no evidence that Jane’s recent experiences lead her to behave the way she does in situations like these.”

The prompt for the environmental attribution item was the same as in Experiment 1 except that the word “childhood” was replaced with “recent”: “How much of a role do you think Jane’s recent experiences play in Jane’s patterns of behavior that you just read about?”

As in Experiment 1, after the environmental attribution rating participants were asked basic demographic questions before being debriefed about the fictitious nature of the vignettes and receiving instructions for how to claim their compensation.

Results and Discussion

As in Experiment 1, environmental attribution ratings were analyzed using a 2 (behavior: antisocial vs. prosocial) × 2 (environmental explanation: present vs. absent) × 6 (vignette) univariate ANOVA. This revealed, most importantly, that antisocial behavior again evoked weaker environmental attributions (M=3.82, SD=1.91) than did prosocial behavior (M=4.36, SD=1.92), F (1, 581) = 16.67, p < .001, η_p² = .03 (see Figure 2). Unsurprisingly, environmental attributions were once again also stronger among participants for whom the environmental explanation was present than among those for whom it was absent, F (1, 582) = 252.55, p < .001, η_p² = .30. There were no significant two- or three-way interactions (all Fs ≤ .99, all ps > .42), and no significant main effect of vignette, F(5, 581) = 1.51, p = .186).

Figure 2.

Mean environmental attribution ratings, by condition, in Experiment 2. Means are collapsed across other independent variables because the additional experimental manipulations did not significantly moderate the antisocial-prosocial asymmetry in endorsement of environmental attributions. Error bars represent ±1 SE of the mean.

The results of Experiment 2 suggest that participants were less inclined to attribute antisocial behavior, as compared to prosocial behavior, to environmental influences, even when the environmental influences in question were recent, i.e., witnessing others exhibiting similar behavior during the past few months. Once again, the presence or absence of an explicit statement indicating that the behavior in question was brought about by such recent environmental exposures was not a significant moderator of the asymmetry in environmental attribution ratings between antisocial and prosocial behavior.

Experiment 3

The results of Experiments 1 and 2 suggest that endorsements of a causal role for childhood experiences and recent experiences may have something in common that leads both to be resisted when making causal attributions for antisocial behavior as compared to prosocial behavior. Experiment 3 tested whether this asymmetry would extend to another form of environmental influence on behavior. In particular, we examined whether a similar pattern of results would be observed when the environmental influence whose presence (vs. absence) was experimentally varied consisted of direct encouragement by others to behave in a particular way, rather than mere passive observation of others’ behavior.

Methods

Participants and recruitment.

Experiment 3 was completed by 600 U.S. adults (39.5% male, 59.0% female, 1.5% other or unknown gender) recruited via Prolific. This sample size was estimated to provide more than 95% power to detect a small-to-medium effect (f = .15) of the difference between prosocial and antisocial behavior on environmental attribution ratings. Participants were again offered $1 in exchange for completing the procedures.

Stimuli and procedures.

The stimuli and measures were identical to those used in Experiments 1 and 2, except for the wording of the environmental attribution item and the final paragraph displayed under the vignette. In Experiment 3, the final paragraph referred to others encouraging Jane to behave in prosocial or antisocial ways. It began, “Scientists have found that people’s behavior can be influenced by those around them, such as their peers.” Participants were randomly assigned to one of two endings for this paragraph. For those randomly assigned to receive a “peer-pressure” explanation, the paragraph continued: “In fact, over the past few months, Jane has been surrounded by people, including friends and coworkers, who behave the way that Jane did in the situation that you just read about and who have encouraged her to behave similarly. In other words, the recent actions of those around Jane lead her to behave the way she does in situations like these.” For those randomized to receive no peer-pressure explanation, it instead said: “However, over the past few months, nobody around Jane has behaved the way that Jane did in the situation that you just read about, and nobody around Jane has encouraged her to behave this way. In other words, there is no evidence that the recent actions of those around Jane lead her to behave the way she does in situations like these.”

In Experiment 3, the prompt for the environmental attribution measure was worded, “How much of a role do you think Jane’s recent experiences (that is, the influence of the people around her) play in Jane’s patterns of behavior that you just read about?—i.e., the prompt was the same as in Experiment 2 except for the words in parentheses, which were included to make clear which particular experiences were being asking about.

As in Experiments 1 and 2, after the environmental attribution rating participants were asked basic demographic questions before being debriefed about the fictitious nature of the vignettes and receiving instructions for how to claim their compensation.

Results and Discussion

As in Experiments 1 and 2, environmental attribution ratings were analyzed using a 2 (behavior: antisocial vs. prosocial) × 2 (environmental explanation: present vs. absent) × 6 (vignette) univariate ANOVA. This revealed that, as in the previous experiments, antisocial behavior evoked significantly weaker environmental attributions (M=3.65, SD=1.94) than did prosocial behavior (M=4.12, SD=2.03), F (1, 576) = 15.76, p < .001, η_p² = .03 (see Figure 3). Unsurprisingly and also consistent with experiments 1 and 2, environmental attributions were also stronger among participants who received the peer-pressure explanation than among those who did not, F (1, 576) = 463.17, p < .001, η_p² = .45. There were no significant main effects of vignette, and the effect of behavior type (prosocial vs. antisocial) was not significantly moderated by vignette or by the presence (vs. absence) of the environmental explanation (Fs ≤ 1.46, ps > .20).^‡ The three-way interaction was also not significant, F (5, 576) = .43, p > .82.

Figure 3.

Mean environmental attribution ratings, by condition, in Experiment 3. Means are collapsed across other independent variables because the additional experimental manipulations did not significantly moderate the antisocial-prosocial asymmetry in endorsement of environmental attributions. Error bars represent ±1 SE of the mean.

The results of Experiment 3 replicated the findings from Experiments 1 and 2 that participants were less willing to endorse the idea of behavior being influenced by a person’s environment (i.e., the behaviors of others) when the behavior in question was antisocial in nature, rather than prosocial. Moreover, Experiment 3 extended this finding to cases in which the type of environmental influence being considered involved explicit encouragement from others to behave in a particular way. The presence or absence of an explicit statement indicating that the behavior in question was brought about by peer influence significantly affected environmental attribution ratings across antisocial and prosocial behavior but did not interact significantly with type of behavior.

General Discussion

In light of recent research indicating that people less readily attribute antisocial (vs. prosocial) behavior to genetic causes (Lebowitz et al., 2019, 2021), the present research assessed whether the same asymmetry would be observed for nongenetic, environmental causal attributions—in particular, the influence of others’ behavior (either through passive observation or direct peer pressure). Results indicated that, indeed, participants less readily endorsed childhood experiences of observing others’ behavior and recent adult experiences of either observing others’ behavior or being actively encouraged to behave in a certain way as causes of antisocial than of prosocial behavior. At first glance, this may seem counterintuitive, as one might expect that if people are relatively reluctant to attribute antisocial behaviors to genetic causes, there should be a corresponding increase in willingness to attribute them to environmental causes. However, genetic and environmental attributions may not be “zero-sum”: endorsement of explanatory accounts that would seem to be in competition with one another can actually be independent (Stroessner & Green, 1990). Indeed, people do not always view genetic and environmental explanations as competing accounts for human behavior or characteristics, and attributions to individual choice—the viability of which participants may have been trying to preserve to the extent that they resisted environmental attributions in the antisocial condition of the present work—may be more robustly and consistently negatively associated with genetic attributions than are environmental attributions (Jayaratne et al., 2009).

The present results may be relevant to questions about the mediators of asymmetries in causal attributions for prosocial versus antisocial behavior that have previously been observed. In particular, recent work has suggested that one reason people may endorse genetic attributions more strongly for prosocial (versus antisocial) behavior may be that they perceive prosocial behavior as more “natural,” and genetic causation is fundamentally linked with naturalness (Lebowitz et al., 2021). However, the present experiments found a similar asymmetry for causal attributions to nongenetic causes that likely would not be expected to be associated with concepts of naturalness. This could support the notion that asymmetrical perceptions of naturalness may not be the sole explanation for discrepancies in endorsement of causal attributions for prosocial and antisocial behavior. Other factors, such as a motivation to blame wrongdoers for their actions (Alicke, 2000; Clark et al., 2014) and therefore resist attributing antisocial behavior to causes outside of individual control (Lebowitz et al., 2019), may play a role. While people may be motivated to hold offenders accountable for their misdeeds (e.g., given the evolutionary costliness of being seen as easily exploitable), there may be relatively less downside to ascribing too little personal responsibility to others for their helpful behavior (Clark, 2022). This could help to account for why people—to the extent that they see environmental or other external causal attributions as deflecting responsibility—might be more inclined to reject such explanations in the case of antisocial (as compared to prosocial) behavior. It is important to note that the present research did not measure motivation to blame or any other potential mediating variables, so our discussion of the possible psychological mechanisms giving rise to the observed prosocial-antisocial asymmetry is speculative and awaits testing in future research.

It is also possible that the asymmetry in environmental attributions observed in the present research may be a product of a different psychological process than the one(s) that produce asymmetrical genetic attributions for prosocial versus antisocial behavior. For example, people may be generally inclined to perceive harmful (e.g., antisocial) behaviors as definitively indicative of an immoral intrapersonal disposition on the part of the actor, which may make them unlikely to view environmental factors as relevant causes of such behavior. Previous research has suggested that people will perceive an individual who behaves immorally as dispositionally immoral, regardless of other circumstances (Reeder & Spores, 1983), whereas exhibiting moral behavior is seen as less diagnostic of intrapersonal dispositions, given that “both moral and immoral persons are thought likely to attempt moral behavior when it is to their advantage” (p. 737). That is, people may believe that a morally good person will only ever behave morally, so even when a person’s environment promotes harmful behavior, only a fundamentally immoral person will actually engage in it. Thus, a person described as exhibiting antisocial behavior would be perceived as dispositionally immoral, which might seem to render the environmental influences less causally relevant, leading to the lower environmental attribution ratings observed in the antisocial conditions of the present experiments. By contrast, “beneficial actions tend to be much more ambiguous than harmful actions when it comes to deciding on the actor’s true intention or his ultimate objectives” (Jones & Davis, 1965, p. 259). Participants in the prosocial conditions of our experiments may have perceived the prosocial behavior as possibly caused either by Jane’s prosocial dispositions or by circumstantial factors, leaving room for such actions to be attributed more strongly to environmental causes in some cases.

A tendency to view dispositions, and not situational factors, as primary causes of others’ immoral or harmful behavior has long been discussed in the attribution literature (e.g., Reeder, 1993; Reeder & Spores, 1983). However, when examining what factors shape people’s propensities to make situational attributions, this research has generally focused on the inclination (or lack thereof) to attribute behavior to the circumstances immediately surrounding a behavior. The present work goes beyond such a focus by investigating people’s willingness to attribute a person’s behavior to the individual’s broader social environment (i.e., the behavior of ones’ peers over a period of months or even behaviors observed during childhood). We find that even these kinds of durable environmental influences are generally seen as less relevant in shaping antisocial behavior than in shaping prosocial behavior.

The present research is not without limitations. For one, the environmental explanations tested in Experiments 1–3 included only a narrow set of influences (i.e., witnessing others engaging in prosocial or antisocial behavior, or being encouraged by others to do so). Although the results are sufficient to demonstrate that greater resistance to endorsing causal explanations for antisocial, vs. prosocial, behavior is not unique to genetic attributions, it is possible that other environmental influences (e.g., socioeconomic status, being a victim of violence or coercion, etc.) could yield different patterns of results. Future research could provide additional evidence.

The present experiments add to a growing body of literature indicating that when people seek to explain others’ behavior, they do not simply weigh the evidence for various causal attributions in a dispassionate manner. Rather, seemingly extraneous factors such as the evaluative valence of the behavior appear to play a role in which causal accounts they prefer. In the case of genetic attributions, this may involve the influence of intuitions about the naturalness of the behavior in question, with behaviors that are evaluated positively perceived as more “natural,” and therefore more likely to be genetically influenced, than behaviors that are evaluated negatively. However, the present work reveals asymmetries between prosocial and antisocial behavior in the attributions people make when considering environmental influences that are less likely to be perceived as closely linked to naturalness. This suggests that perceptions of naturalness may not be the only intuition that must be taken into account in understanding the drivers of asymmetrical reasoning about the causes of behavior.