The Empirical Argument Against Virtue

Abstract

The virtues are under fire. Several decades’ worth of social psychological findings establish a correlation between human behavior and the situation moral agents inhabit, from which a cadre of moral philosophers concludes that most moral agents lack the virtues. Mark Alfano and Christian Miller introduce novel versions of this argument, but they are subject to a fatal dilemma. Alfano and Miller wrongly assume that their requirements for virtue apply universally to moral agents, who vary radically in their psychological, physiological, and personal situations; I call this the ‘content problem.’ More troubling, however, the content problem leads to what I call the ‘structural problem:’ Alfano and Miller each structure their argument against the virtues as a modus tollens argument and, owing to the breadth of the content problem, each must constrain their argument with a ceteris paribus clause. But the ceteris paribus clause precludes each argument’s validity. More important, however, the resulting conception of virtue implicitly endorsed by Alfano and Miller holds that virtues are idealized models; but since idealized models do not even purport accurately to describe (much of) the world, neither novel version of EAV gains any empirical traction against the virtues. The upshot is an old story whose moral has yet, within the empirical study of the virtues, adequately to be internalized: it is imperative that the empirical observation of character traits proceed via longitudinal studies.

Appendix: Miller’s Methodological Rigor

Some of Miller’s cited studies employ students as subjects and self-reports. My concern about these studies’ rigor might be best characterized as a mere quibble, as their inclusion in the inductive evidence base supporting Line 2 of EAV-M does not significantly contribute to this premise’s overall strength. Nonetheless, the presence of studies employing students as subjects and self-reports could be vitiating in a distinct argumentative framework, so they clearly merit scrutiny.

Miller explicitly appeals to studies that employ undergraduate students as subjects. For example, one study divided 99 students into two groups, assigning them to complete a variety of tasks, some of them related to helping. (Liljenquist et al. 2010: 381–383) 22 % of the first group expressed interest in donating to Habitat for Humanity (HfH), while only 6 % of the second group expressed a similar interest. And the first group’s level of interest in volunteering for HfH (on a 1–7 scale) was 4.21, while the second group’s interest level was 3.29. The only discernible explanation for the two groups’ differing levels of willingness to donate or volunteer was that the room in which the first group completed their tasks had been cleaned with Windex, while the second group’s room had not. The experimental results are allegedly generalizable to all moral agents, but there are many reasons for thinking that (traditional) undergraduate students’ mental attitudes and behaviors are not representative of the adult population at large. A broad base of anecdotal evidence suggests that teens and young adults are ineffective at assessing the potential consequences of their actions and regulating their emotions. Further, “functional brain imaging studies…suggest that the responses of teens to emotionally loaded images and situations are heightened relative to younger children and adults. The brain changes underlying these patterns involve brain centers and signaling molecules that are part of the reward system with which the brain motivates behavior. These age-related changes shape how much different parts of the brain are activated in response to experience, and in terms of behavior, the urgency and intensity of emotional reactions.”²² While it might not be readily apparent how teens’ and young adults’ diminished cognitive and affective abilities might specifically affect their willingness to donate or volunteer, this is an empirical question that still awaits thorough study. EAV-M explicitly appeals to dozens of experiments that do not use students as subjects; however, to the extent that his argument does rely upon such studies, the inductive base for his Line 2 is correspondingly weakened.

The second problem of methodological rigor attaching to Miller’s empirical studies concerns self-reports. In the above study on volunteering for and donating to HfH, students were asked to report their level of interest in volunteering and to indicate whether they wanted to donate. Liljenquist et al. (2010: 381–383) explicitly state that “because our charity measures captured participants’ intentions, rather than their actual actions, future research should also measure charitable behavior directly”. Someone’s forming an intention to donate or volunteer (assuming the subjects did, indeed, form such an intention) neither entails nor probabilifies that she will actually donate or volunteer, as the study’s own authors imply. And Miller fails to mention this relevant point, even though he explicitly notes several problems with self-report studies. (Miller 2013: 231) Thus, while Miller’s argument explicitly relies on a swath of studies that do involve actual behavior, the inductive base in support of Line 2 is weakened to the extent that it relies upon self-report studies.

To reiterate an important point, Miller’s appeal to studies that use students as subjects and self-reports likely does not undermine the overall force of EAV-M, Line 2, as he also appeals to a wide scope of studies that employ neither students nor self-reports. However, owing to the burgeoning practice of philosophers’ citing psychologists’ works to establish a range of philosophical points, both issues clearly merit careful attention.²³