Doris invites scholars to investigate ethical questions “in light of the best ongoing scientific picture” (Doris Reference Doris2015b, p. 12) to make their theories more empirically credible. As psychologists who work at the boundary of psychology and philosophy, we heartily endorse this goal. He exemplifies this through a philosophical discussion of numerous social psychological experiments, concluding that there is room for “skepticism about morally responsible agency” (p. 1) because there are numerous “defeaters” of agency (e.g., ballot order effects). He bases his problem diagnosis – that defeaters undermine morally responsible agency – on evidence indicating that apparently (im)moral actions are caused by minor environmental factors outside the actor's reflective attention. The worry is that individuals' reason and choice are “bypassed” by a cause that cannot provide a rational justification of the act. If Doris's diagnosis is correct, we may have a serious problem. In response, he defends moral agency as the expression of one's values, and champions collaborative reasoning as a key source of agency.
We suggest that Doris's worries about defeaters are overstated. Central to his overstatement is that the experimental effects that he cites are generally mild and inconsistent. Ballot order effects are one of his favorites, but the literature is far less worrisome than he suggests. The effects occur more in low information and low visibility elections (Pasek et al. Reference Pasek, Schneider, Krosnick, Tahk, Ophir and Milligan2014). Clearly, the fewer clear reasons to vote for a specific candidate, the more ballot order will sway voters.
Doris wisely recognizes and discusses the mild, aggregate effects problem: “To be sure, identifying statistically small effects does not allow confident conclusions about particular outcomes for particular individuals” (Doris Reference Doris2015b, p. 63). Yet he still tends to overstatement. He concludes that “In sum . . . evidence of incongruence is readily obtained,” which “make[s] plausible the supposition that incongruence is widespread in everyday life” (p. 61). It is unclear how he translates group differences in experiments with strangers into widespread occurrences in ordinary life. He makes this leap partly by adverting to a confabulated cumulation of small effects because “there could be many goofy influences in any particular instance . . . [and] the aggregate effect may be quite potent” (p. 64). This goes considerably beyond the evidence. There is precious little evidence for multiple situational influences operating simultaneously, and virtually none that multiple situational factors move individuals in the same behavioral direction. It is just as likely that situational factors cancel one another out as cumulate.
Because Doris builds his philosophical argument on social science data, it is important to apply scientific skepticism to philosophical skepticism. He states the skeptical challenge thus: “If there is general difficulty in ruling out defeaters, skepticism about agency ensues” (Doris Reference Doris2015b, p. 65). This is a strong claim; too strong, given the available evidence. Given the aggregate effects and statistical likelihood of the influences, the most that can be said is that defeaters show that not all choices by all people are governed by justifiable moral reasons.
Doris's overstatement of this problem may be partly due to different forms of argument in philosophy and social science. Philosophers incline to search for absolute truth, which fuels the skeptic's challenge. Social scientists do not care much for absolute truth and speak in terms of tendencies and probabilities, and this is the appropriate language for the data Doris cites. In this language, there is a small probability that a given individual act will involve “bypassing” the actor's reason. Furthermore, this mild influence is typically observed in rather trivial circumstances with strangers. The skeptical argument has no purchase here because social scientists assume that humans are imperfect reasoners and that individuals vary in reasoning quality. This is not as surprising or earth-shaking as Doris makes it out to be. Even if one reasonably grants that everyone has some vulnerability to defeaters, all that is being conceded is that humans are imperfect moral reasoners, which should not alarm us. This imperfection is only a blemish against agency if one imposes a perfectionistic requirement that choices are always conscious and well-justified. These are unreasonable assumptions. To create an empirically credible moral psychology, Doris must let go of such psychologically unrealistic starting points.
The skeptical challenge is further defanged when we consider what Doris entirely neglects: individual differences in moral agency and in susceptibility to defeaters. His singular focus on situational influences and his omission of the moral psychology of individual differences distorts his problem diagnosis. His choice is not surprising because this omission is pervasive in social psychology, the empirical foundation of his argument. What would happen if we took individual differences seriously in the experiments Doris cited? No one knows because they are rarely studied in this literature. Although social and personality psychologists agree in principle that behavior is best explained by a combination of situations, traits, and their interactions, actual person-X-environment studies remain the exception rather than the rule.
To clarify the importance of individual differences, suppose that the vulnerability (or resilience) to defeaters itself is treated as an individual difference variable. Thus, some people would be more likely to be influenced by defeaters (i.e., act based on trivial situational influences) and others would be less likely to be so influenced. That is, some individuals are better at maintaining and expressing their value commitments than others (a capacity that is central to Doris's viewpoint). The apparent crisis raised by agency skeptics turns out to be an unsurprising variation in capacity to maintain value commitments. In fact, a primary domain of moral psychology investigates this consistency in value commitment under the rubric of moral identity (e.g., Blasi Reference Blasi, Lapsley and Clark2005), which Doris also curiously fails to mention.
One interesting counterexample of individual difference neglect is a study that Doris co-authored, which he cites as an example of induced disgust leading to more punitive moral judgments (Cameron et al. Reference Cameron, Payne and Doris2013). In fact, there was no main effect for disgust induction in this study. There was an interaction of disgust and emotion differentiation such that participants low in emotion differentiation were affected by disgust, but those high in emotion differentiation were not. We do not know why Doris failed to accurately describe this individual differences result, but we do know that this oversight favors his hypothesis and indicates his disregard of individual differences.
Another illuminating example of this problem is Doris's discussion of Mischel and his colleagues' work on delay of gratification (DG) in children. Interestingly, Doris only cites Mischel's early experimental work, which focused on situational influences on children's DG (e.g., Mischel et al. Reference Mischel, Ebbesen and Zeiss1972). This work fits Doris's thesis. However, he does not cite the later studies of individual differences in DG capacity, which show that DG and rejection sensitivity predict educational attainment, self-worth, interpersonal functioning, and lack of drug use over a twenty year period (e.g., Ayduk et al. Reference Ayduk, Mendoza-Denton, Mischel, Downey, Peake and Rodriguez2000). This work suggests that agentic capacity is, at least in part, an individual difference variable, which unsettles Doris's thesis.
A simple thought experiment on ballot order effects can further illustrate the importance of individual differences in agency. We propose a thought experiment because we do not believe that anyone would think the outcome is sufficiently in doubt to recommend actual data collection. First, we stipulate a small ballot order effect, consistent with Doris's presentation. Second, let us examine a consequential, high visibility contest: the 2016 U.S. presidential election. It must be consequential and visible or Doris's worry about moral agency is moot. The less the election matters, and the less people know, the less a vote can be considered a question of moral agency. Third, we divide the voters into three groups: committed Trump voters, committed Clinton voters, and undecided voters. Fourth, we randomize the presented ballots, with half the ballots listing one candidate first and half listing the other candidate first. The outcomes seem obvious. The small, stipulated main effect for ballot order would almost certainly be strongly qualified by an interaction. The ballot order effect would be far stronger in the undecided group than in the committed groups, illustrating the central role that individual differences (in value commitment) play in agency. Those with clear, strongly held values will be far more defeater resilient than those with ill-defined or weakly held values.
It could be objected that this is a too-easy counterexample, but there is a general case available in virtue theory, which provides an explanation of resilience to defeaters of agency. On Aristotle's (340 bce/Reference Ostwald1999) view, there are multiple overall character types, three of which can illustrate its predictions. Individuals with virtuous characters are highly resilient to defeaters because they have strong moral commitments and have made moral agency habitual. Continent characters know what they ought to do and generally act accordingly, but they are not as firmly committed to acting morally as the virtuous, which makes them somewhat more vulnerable to defeaters. Although incontinent characters know what they ought to do, they are weakly committed to acting morally, making them highly vulnerable to defeaters. Behavioral research on virtue and character is just getting under way (e.g., Lefevor & Fowers Reference Lefevor and Fowers2016; Meindl et al. Reference Meindl, Jayawickreme, Furr and Fleeson2013), but the expectation is that virtue and character will directly reduce defeaters' influence as well as moderate defeater effects. If so, this changes the picture substantially. Some people are very vulnerable to defeaters, some are somewhat vulnerable, and some are relatively invulnerable. Importantly, these individual differences in agentic resiliency do not rely on reflection because they manifest in both automatic and reflective action.
Doris could claim that he has already dispensed with virtue and character in his previous book (Doris Reference Doris2002). He argued, based on social psychology experiments, that small situational factors influenced participants to act more or less morally (primarily whether they helped a stranger). He claimed that these effects mean that moral character cannot be very important. That argument has been thoroughly contested theoretically (e.g., Kristjánsson Reference Kristjánsson2013) and drastically undermined empirically in a meta-analysis showing that the experimental effects he cited as evidence against character are empirically insufficient to rule out character as an explanation for helping behavior (Lefevor et al. Reference Lefevor, Fowers, Ahn, Lang and Cohen2017).
Just as the addition of individual differences modifies the problem diagnosis, it also undercuts the value of Doris's remedy. If differences in the capacity to maintain value commitments is part of the problem, then a clear conception of individual differences must be included in any moral psychology. Therefore, Doris's quick dismissal of virtue theory and utter neglect of moral identity theory are misguided. Both theories can account for the variations in vulnerability to defeaters of morally responsible agency. In addition, moral identity and virtue theories can also explain Doris's definition of agency as behavior that expresses the agent's values: The stronger one's moral identity or character, the more one will have well-defined values, and these values will be expressed with greater frequency and automaticity. This is a deeper and more comprehensive account of moral agency than Doris offers.
We do find Doris's argument for a collaborative understanding of reasoning and agency very congenial. However, his view on collaborative reasoning must also give appropriate weight to morally relevant individual characteristics. Clearly, some people will have greater capacity for collaborative reason than others, again highlighting moral character or identity. For us to engage in the best collaborative reasoning possible, we must cultivate the excellences of collaborative reasoning (e.g., openness and honesty). Moreover, we cannot ignore the aims of human collaboration. We agree with Doris that collaborative reason is invaluable to moral agency, but it can also be a source of moral blindness, as amply demonstrated by the “collaboration” that made the totalitarian slaughters of the twentieth century possible. Although we can only raise these thorny issues here, we do not think it is possible to describe or explain moral agency without considering the character of the reasoners or what makes the ends toward which they reason worthwhile.
You have
Access
Doris invites scholars to investigate ethical questions “in light of the best ongoing scientific picture” (Doris Reference Doris2015b, p. 12) to make their theories more empirically credible. As psychologists who work at the boundary of psychology and philosophy, we heartily endorse this goal. He exemplifies this through a philosophical discussion of numerous social psychological experiments, concluding that there is room for “skepticism about morally responsible agency” (p. 1) because there are numerous “defeaters” of agency (e.g., ballot order effects). He bases his problem diagnosis – that defeaters undermine morally responsible agency – on evidence indicating that apparently (im)moral actions are caused by minor environmental factors outside the actor's reflective attention. The worry is that individuals' reason and choice are “bypassed” by a cause that cannot provide a rational justification of the act. If Doris's diagnosis is correct, we may have a serious problem. In response, he defends moral agency as the expression of one's values, and champions collaborative reasoning as a key source of agency.
We suggest that Doris's worries about defeaters are overstated. Central to his overstatement is that the experimental effects that he cites are generally mild and inconsistent. Ballot order effects are one of his favorites, but the literature is far less worrisome than he suggests. The effects occur more in low information and low visibility elections (Pasek et al. Reference Pasek, Schneider, Krosnick, Tahk, Ophir and Milligan2014). Clearly, the fewer clear reasons to vote for a specific candidate, the more ballot order will sway voters.
Doris wisely recognizes and discusses the mild, aggregate effects problem: “To be sure, identifying statistically small effects does not allow confident conclusions about particular outcomes for particular individuals” (Doris Reference Doris2015b, p. 63). Yet he still tends to overstatement. He concludes that “In sum . . . evidence of incongruence is readily obtained,” which “make[s] plausible the supposition that incongruence is widespread in everyday life” (p. 61). It is unclear how he translates group differences in experiments with strangers into widespread occurrences in ordinary life. He makes this leap partly by adverting to a confabulated cumulation of small effects because “there could be many goofy influences in any particular instance . . . [and] the aggregate effect may be quite potent” (p. 64). This goes considerably beyond the evidence. There is precious little evidence for multiple situational influences operating simultaneously, and virtually none that multiple situational factors move individuals in the same behavioral direction. It is just as likely that situational factors cancel one another out as cumulate.
Because Doris builds his philosophical argument on social science data, it is important to apply scientific skepticism to philosophical skepticism. He states the skeptical challenge thus: “If there is general difficulty in ruling out defeaters, skepticism about agency ensues” (Doris Reference Doris2015b, p. 65). This is a strong claim; too strong, given the available evidence. Given the aggregate effects and statistical likelihood of the influences, the most that can be said is that defeaters show that not all choices by all people are governed by justifiable moral reasons.
Doris's overstatement of this problem may be partly due to different forms of argument in philosophy and social science. Philosophers incline to search for absolute truth, which fuels the skeptic's challenge. Social scientists do not care much for absolute truth and speak in terms of tendencies and probabilities, and this is the appropriate language for the data Doris cites. In this language, there is a small probability that a given individual act will involve “bypassing” the actor's reason. Furthermore, this mild influence is typically observed in rather trivial circumstances with strangers. The skeptical argument has no purchase here because social scientists assume that humans are imperfect reasoners and that individuals vary in reasoning quality. This is not as surprising or earth-shaking as Doris makes it out to be. Even if one reasonably grants that everyone has some vulnerability to defeaters, all that is being conceded is that humans are imperfect moral reasoners, which should not alarm us. This imperfection is only a blemish against agency if one imposes a perfectionistic requirement that choices are always conscious and well-justified. These are unreasonable assumptions. To create an empirically credible moral psychology, Doris must let go of such psychologically unrealistic starting points.
The skeptical challenge is further defanged when we consider what Doris entirely neglects: individual differences in moral agency and in susceptibility to defeaters. His singular focus on situational influences and his omission of the moral psychology of individual differences distorts his problem diagnosis. His choice is not surprising because this omission is pervasive in social psychology, the empirical foundation of his argument. What would happen if we took individual differences seriously in the experiments Doris cited? No one knows because they are rarely studied in this literature. Although social and personality psychologists agree in principle that behavior is best explained by a combination of situations, traits, and their interactions, actual person-X-environment studies remain the exception rather than the rule.
To clarify the importance of individual differences, suppose that the vulnerability (or resilience) to defeaters itself is treated as an individual difference variable. Thus, some people would be more likely to be influenced by defeaters (i.e., act based on trivial situational influences) and others would be less likely to be so influenced. That is, some individuals are better at maintaining and expressing their value commitments than others (a capacity that is central to Doris's viewpoint). The apparent crisis raised by agency skeptics turns out to be an unsurprising variation in capacity to maintain value commitments. In fact, a primary domain of moral psychology investigates this consistency in value commitment under the rubric of moral identity (e.g., Blasi Reference Blasi, Lapsley and Clark2005), which Doris also curiously fails to mention.
One interesting counterexample of individual difference neglect is a study that Doris co-authored, which he cites as an example of induced disgust leading to more punitive moral judgments (Cameron et al. Reference Cameron, Payne and Doris2013). In fact, there was no main effect for disgust induction in this study. There was an interaction of disgust and emotion differentiation such that participants low in emotion differentiation were affected by disgust, but those high in emotion differentiation were not. We do not know why Doris failed to accurately describe this individual differences result, but we do know that this oversight favors his hypothesis and indicates his disregard of individual differences.
Another illuminating example of this problem is Doris's discussion of Mischel and his colleagues' work on delay of gratification (DG) in children. Interestingly, Doris only cites Mischel's early experimental work, which focused on situational influences on children's DG (e.g., Mischel et al. Reference Mischel, Ebbesen and Zeiss1972). This work fits Doris's thesis. However, he does not cite the later studies of individual differences in DG capacity, which show that DG and rejection sensitivity predict educational attainment, self-worth, interpersonal functioning, and lack of drug use over a twenty year period (e.g., Ayduk et al. Reference Ayduk, Mendoza-Denton, Mischel, Downey, Peake and Rodriguez2000). This work suggests that agentic capacity is, at least in part, an individual difference variable, which unsettles Doris's thesis.
A simple thought experiment on ballot order effects can further illustrate the importance of individual differences in agency. We propose a thought experiment because we do not believe that anyone would think the outcome is sufficiently in doubt to recommend actual data collection. First, we stipulate a small ballot order effect, consistent with Doris's presentation. Second, let us examine a consequential, high visibility contest: the 2016 U.S. presidential election. It must be consequential and visible or Doris's worry about moral agency is moot. The less the election matters, and the less people know, the less a vote can be considered a question of moral agency. Third, we divide the voters into three groups: committed Trump voters, committed Clinton voters, and undecided voters. Fourth, we randomize the presented ballots, with half the ballots listing one candidate first and half listing the other candidate first. The outcomes seem obvious. The small, stipulated main effect for ballot order would almost certainly be strongly qualified by an interaction. The ballot order effect would be far stronger in the undecided group than in the committed groups, illustrating the central role that individual differences (in value commitment) play in agency. Those with clear, strongly held values will be far more defeater resilient than those with ill-defined or weakly held values.
It could be objected that this is a too-easy counterexample, but there is a general case available in virtue theory, which provides an explanation of resilience to defeaters of agency. On Aristotle's (340 bce/Reference Ostwald1999) view, there are multiple overall character types, three of which can illustrate its predictions. Individuals with virtuous characters are highly resilient to defeaters because they have strong moral commitments and have made moral agency habitual. Continent characters know what they ought to do and generally act accordingly, but they are not as firmly committed to acting morally as the virtuous, which makes them somewhat more vulnerable to defeaters. Although incontinent characters know what they ought to do, they are weakly committed to acting morally, making them highly vulnerable to defeaters. Behavioral research on virtue and character is just getting under way (e.g., Lefevor & Fowers Reference Lefevor and Fowers2016; Meindl et al. Reference Meindl, Jayawickreme, Furr and Fleeson2013), but the expectation is that virtue and character will directly reduce defeaters' influence as well as moderate defeater effects. If so, this changes the picture substantially. Some people are very vulnerable to defeaters, some are somewhat vulnerable, and some are relatively invulnerable. Importantly, these individual differences in agentic resiliency do not rely on reflection because they manifest in both automatic and reflective action.
Doris could claim that he has already dispensed with virtue and character in his previous book (Doris Reference Doris2002). He argued, based on social psychology experiments, that small situational factors influenced participants to act more or less morally (primarily whether they helped a stranger). He claimed that these effects mean that moral character cannot be very important. That argument has been thoroughly contested theoretically (e.g., Kristjánsson Reference Kristjánsson2013) and drastically undermined empirically in a meta-analysis showing that the experimental effects he cited as evidence against character are empirically insufficient to rule out character as an explanation for helping behavior (Lefevor et al. Reference Lefevor, Fowers, Ahn, Lang and Cohen2017).
Just as the addition of individual differences modifies the problem diagnosis, it also undercuts the value of Doris's remedy. If differences in the capacity to maintain value commitments is part of the problem, then a clear conception of individual differences must be included in any moral psychology. Therefore, Doris's quick dismissal of virtue theory and utter neglect of moral identity theory are misguided. Both theories can account for the variations in vulnerability to defeaters of morally responsible agency. In addition, moral identity and virtue theories can also explain Doris's definition of agency as behavior that expresses the agent's values: The stronger one's moral identity or character, the more one will have well-defined values, and these values will be expressed with greater frequency and automaticity. This is a deeper and more comprehensive account of moral agency than Doris offers.
We do find Doris's argument for a collaborative understanding of reasoning and agency very congenial. However, his view on collaborative reasoning must also give appropriate weight to morally relevant individual characteristics. Clearly, some people will have greater capacity for collaborative reason than others, again highlighting moral character or identity. For us to engage in the best collaborative reasoning possible, we must cultivate the excellences of collaborative reasoning (e.g., openness and honesty). Moreover, we cannot ignore the aims of human collaboration. We agree with Doris that collaborative reason is invaluable to moral agency, but it can also be a source of moral blindness, as amply demonstrated by the “collaboration” that made the totalitarian slaughters of the twentieth century possible. Although we can only raise these thorny issues here, we do not think it is possible to describe or explain moral agency without considering the character of the reasoners or what makes the ends toward which they reason worthwhile.