R1.1. Skepticism and surprising effects

The skepticism I develop is empirically motivated, and is substantially dependent on the credibility of the science from which it draws. In both the book (pp. 44–49) and the précis, I discuss the recent replicability controversy in psychology, and I won't reprise my remarks here. But Lambert & Dennett mention a selection effect in the production of science that should be addressed. Psychologists don't get paid, to borrow their example, for demonstrating that people prefer $1,000 to a pin prick: “generating small, surprising effects may be rewarded in psychology,” they say, “but it is not clear whether or how the common lore of everyday psychology that psychologists never bother to investigate is undermined” by this enterprise. Reward structures in academic psychology reflect, to some degree, the Surprising Effect Bias (p. 47), and like Lambert & Dennett, I'm confident that where psychologists set out to demonstrate Obvious Effects, they'll have decent luck in finding people behaving expectedly, rationally, and, perhaps, agentially.

Does this mean the surprising effects I recount are the unlikely exceptions that prove the agential rule? Maybe not. Unexpected effects appear regularly throughout psychology (Roediger & Butler Reference Roediger, Butler and Kapur2011), which makes me think that they are comparatively easy to get, and are not merely an artifact of the Surprising Effect Bias. I'm therefore guessing we should expect unexpected phenomena in life, where stimuli are likely to be vastly more various and potent than in controlled laboratory experiments.

While it is repeatedly observed that effect sizes in psychology tend to be of a size conventionally regarded as small (Cohen Reference Cohen1988, pp. 77–81), it should also be noted that the manipulations themselves are often rather slight, with nothing like the potency of real-world stimuli. The pictures used for inducing disgust in the lab, for example (e.g., Cameron et al. Reference Cameron, Payne and Doris2013), are the palest simulacra of something truly disgusting, like a dumpster festering with rot and squiggling with maggots on a sweltering August day. So, too, the “prison” in the Stanford Prison Study is hardly more than a parody of an actual correctional facility.

Given the limited force of many laboratory manipulations, isn't the small size of many unexpected effects exactly what we should expect? The remarkable thing, one might say, is not that the surprising effects are small, but that they can be gotten at all. By the same token, shouldn't we expect larger impacts from the rough and tumble of real-world influences? All this disinclines me to conclude that bypassing defeaters are largely an artifact of the Surprising Effects Bias.

R1.2. Skepticism, effect size, and defeaters

Fowers, Anderson, & Lang (Fowers et al.) contend I have overstated the skeptical challenge. I suspect this conviction can be traced to their understanding of the philosophical enterprise: “philosophers incline to search for absolute truth, which fuels the skeptic's challenge.” Certainly, some philosophers, among them many historical Greats, incline to absolutism, but this is emphatically not my orientation; as I said in the book, I'm making a “wager” in conditions of scientific uncertainty, not searching for absolute truth (pp. 12–14, 48–49; cf. Machery & Doris, Reference Dorisforthcoming). Moreover, my epistemic orientation is explicitly fallibilist (p. 65), meaning I think knowledge has its basis in defeasible justification, not “absolute” justification. The skeptical challenge is not ginned-up by philosophical absolutism.

Fowers et al. also misunderstand how the skeptical argument works. They say the experimental effects are “generally mild and inconsistent,” and we are in at least partial agreement here; I acknowledge (pp. 61–64) that the relevant effects will often be small (while observing that small effects can be practically and/or theoretically important). But Fowers et al. also say “there is a small probability that a given individual act will involve ‘bypassing’ the actor's reason,” and conclude that this “small” probability (whatever it is) means that the skeptical worry does not go through;Footnote ³ rather, the most we can conclude is that “humans are imperfect moral reasoners.”

The skeptical argument does not require that the probability of defeaters be greater-than-small, but rather that the skeptical hypothesis be a live one, that requires ruling out (p. 66). If Fowers et al. could show that the probability of a defeater obtaining was trivially small, like the odds of winning a multistate lottery (lottery paradoxes noted!) maybe the skeptical argument would be a merely academic exercise. But that isn't what they show, nor does it seem to be entailed by their guesstimate that the probability of defeaters is small.

When Fowers et al. turn to my suggestion that defeaters may aggregate to undermine agency (Doris Reference Doris2015b, p. 64), they say “it is just as likely that situational factors cancel one another out as cumulate” in a way that undermines agency. I'm not sure what supports their estimate. But suppose they are right, and it is exactly as likely that an aggregate canceling of defeaters obtains as it is that an aggregate defeater obtains. On this supposition, it is equally likely that agency is, or is not, undermined, and attributions of agency are the epistemic equivalent of flipping a coin; here, Fowers et al.'s scenario just is a skeptical scenario.

May presses the issue of effect size with an elegantly posed dilemma: “influences on many choices tend to be either substantial or arbitrary but not commonly both.” For something to threaten agency, on May's picture, it would have to be simultaneously rationally arbitrary and a substantial influence on behavior, but most of the effects I mention, he thinks, are either insubstantial or non-arbitrary. Without the other, neither condition unsettles agency, so there aren't enough troubling effects to make trouble.

My argument focused on arbitrary influences (pp. 54, 64); roughly, influences that make unlikely justificatory material, like Ballot Order Effects. But, May argues, many of the experimental effects, like certain bystander effects, “aren't necessarily” considerations “we'd reject as non-reasons.” True enough; being in a hurry, for example, might reasonably, in some contexts, count as a reason. But that a consideration can serve as a reason in some contexts does not mean it can't count as a defeater in others. If being in a hurry leaves you unresponsive where you otherwise would have helped, given your values, it's plausible that something has gone wrong for your agency, even if being in a hurry might have served you as a justification in another context, or for a person with different values. So, rationally non-arbitrary considerations sometimes count as defeaters.

It's not clear how to gloss “substantial,” but makes a practical difference is a plausible reading. Again, the effects I'm considering often are often small, but even small effects of the sort typical in much psychology can make a practically meaningful difference, as they do in medicine (pp. 61–64). I suspect that most complex psychological effects are the function of many individually small influences, an observation I dub the lotta-little principle (Doris, Reference Dorisin preparation). The truth of this principle does not give us reason to think that none of these effects can make a practical difference. No straw by itself is the straw that breaks the camel's back. But each straw, given all of the other straws, has an effect.

Once more, I'm not sure how one could substantiate May's assertion that the combination of arbitrary and substantial doesn't “commonly” occur. He's surely right that “many of our choices are influenced by good reasons,” but this is compatible with many choices being undermined by defeaters: that there are many short people is compatible with there also being many tall ones. Furthermore, it's not obvious to me how much confidence “not commonly” should yield, even if that estimate be granted. Everywhere I've lived in the United States, you see red-shouldered hawks a lot less than you see red-tailed hawks; the former, tragically, may be in decline, while the latter are still ubiquitous. That is, red-tails are “common;” red-shoulders, not so much. Still, red-shoulders are common enough to reduce confidence in identification when one sees a buteo at sufficient distance in places where both birds range. Even an uncommon occurrence can have epistemic import.

Because I'm reticent about speculating frequencies, May wonders if “we didn't need all of the empirical evidence,” and might just as well have generated skepticism by “imagination alone,” as is done in the many philosophical intuition pumps featuring such unlikely scenarios as Epistemically Malicious Demons or Envatted Brains. Well, I use the empirical evidence as a kind of “possibility pump,” elevating defeaters from a mere possibility that might be reasonably discounted, to a live possibility that requires ruling out. Interestingly, some skeptical intuition pumps famous in epistemology may likewise draw their force from empirical observations. For example, early on in his First Meditation (Descartes, Reference Descartes1641/2008, paragraphs 3–5), Descartes mentions several possibilities that, while perhaps not commonplace, happen often enough to give pause: perception is inaccurate or distorted, people hallucinate, people are unsure whether they're dreaming. I speculate that if such empirical possibilities were unknown, and there were no evidence of inaccurate perception, the imaginings induced by the skeptical intuition pumps would lose much of their power to provoke. Matters are the same, I think, for the contemplated agency skepticism; if the curious disruptions of agency I discuss were empirically unheard of, the skepticism would have a weaker bite.

Levy is likewise skeptical about the skeptical problem, which he dismisses as “largely illusory.”Footnote ⁴ He argues that questions of agency are decided not by the reasons we do endorse, or would endorse were we aware of them, but rather by those “we should endorse.” What we should endorse, according to Levy, is often fixed by what was “adaptive in our ancestral environment,” because “a large proportion of these processes continue to track reasons.” Therefore, he thinks, “very many, perhaps the overwhelming majority, of the processes that Doris identifies as defeaters are better seen as realizers of our agency than as defeaters of it.”

I do not deny that there are many cases where automatic processes support rather than subvert agency (p. 51); some goals are better accomplished by automaticity. But many does not get us to “overwhelming majority,” and yet again, it's quite unclear how the frequency of defeaters in the great corpus of human behavior can be confidently estimated.

There's more trouble for Levy's suggestion than that. Adaptation, infamously, is not overly troubled by the demands of morality and rationality: do selfish genes make reasonable persons, who generously donate to worthwhile charities and prudently save for a secure retirement? Take the Cinderella Effect – stepchildren are more likely to be mistreated than are biological children, and this circumstance may well be an adaptive product of natural selection (Daly & Wilson Reference Daly and Wilson1996; Reference Daly, Wilson, Crawford and Krebs2007). But surely the existence of this “adaptive” tendency does not generate a moral prescription to mistreat stepchildren. Such examples are among the many reasons that attempts to find an evolutionary basis for ethics are often met with well-considered suspicion (e.g., Machery & Mallon Reference Machery, Mallon and Doris2010).

Levy's suggestion is problematic not only with respect to morality, but also with respect to agency. I formulated my account in terms of “subjective” reasons, rather than “objective” reasons, and maintained that the question for agency concerns whether the agent herself would treat the causes of her behavior as justifying reasons, not whether these reasons are reasons she ought have from some perspective other than her own (pp. 43–44, 70, 135). If a parent is committed to treating all of their children equally, but is biased against their stepchildren as a result of the Cinderella Effect, there are questions about the extent to which they're exercising agency. Their behavior runs counter to their values, and is not regulated by their reasons. The fact that there is some perspective which counts them as acting on good reasons, however abhorrent those considerations are to them, does not make it the case that they are acting agentially. Of course, one might seek to build an account of agency around a theory of objective reasons, the apparent agential significance of the actor's subjective perspective notwithstanding. But even if one takes on such demanding work, they should, by dint of familiar examples like the Cinderella Effect, hesitate to source reasons in adaptations.

Hirstein & Sifferd also propose estimated frequencies as a bulwark against skepticism: inaccurate self-awareness, they allow, is “fairly common,” but inaccuracies that go uncorrected because the subject cannot correct them are less common. They insist that only the latter, less common cases, make trouble for responsibility: “in cases where we have the capacity to correct for our mistaken perceptions, using our brain's prefrontal executive processes, it would seem we are responsible for them.” Because they think the responsibility-threatening cases where people lack this capacity to correct are pretty rare, Hirstein & Sifferd think the skeptical challenge is less serious than I suggested.

I won't say more about anti-skeptical frequency guesstimates, but I will say something about Hirstein & Sifferd's appeal to capacities, because I argued in the book that this expedient cannot disarm skepticism (pp. 37–40). My argument is pretty simple: in many cases, folks occupy excusing conditions, and are therefore not responsible, when there is little doubt they have the responsibility-relevant capacities. Certainly I have the capacity to act another way when manipulation or coercion undermines my exercise of agency, it's just that circumstances prevent me from exercising that capacity. Indeed, here lies a way of understanding the distinction between excusing and exempting conditions (Doris Reference Doris2002, pp. 129–30). Excuses are at issue when the actor is assumed to have the capacities requisite for agency, but are somehow prevented from exercising them, while actors in exempting conditions are supposed to lack the relevant capacities altogether, and are globally non-responsible. Thus, the presence of responsibility-relevant capacities is not sufficient for the attribution of responsibility, because people in excusing conditions, who are paradigmatically (though locally) non-responsible may be assumed to have them.

A promising approach, which I should have considered in the book, links responsibility to fair opportunity (Brink & Nelkin Reference Brink and Nelkin2013). On this picture, when I'm in excusing conditions I have the capacity to exercise agency, but factors like coercion and manipulation deny me fair opportunity to do so. Here, if defeaters are to count as underminers of responsibility, it must be the case that their subjects lacked fair opportunity to resist them. And here, self-ignorance can matter, because it is not clear that people have fair opportunity to resist influences of which they are unaware. One might disagree about this – if your vote is really a vote of conviction, maybe you can fairly be expected to resist ballot order effects, whether or not you know about them. But working this out will involve more than simply observing that the subjects of defeaters have the capacity to resist their influence.

Unlike other anti-skeptical commentators, Ward & Machery's resistance is not based on guesstimates of defeaters' rarity. Indeed, they allow the possibility that “the diversity and ubiquity of defeaters is even more staggering than Doris himself acknowledges,” and recognize, as I do (p. 72), that the causal history of any behavior is likely to be shot through with influences those so influenced would be unlikely to treat as justifying reasons. What Ward & Machery doubt is that these influences threaten agency: there are huge numbers of what I term defeaters, but according to them, most of these “so-called defeaters” aren't agency-undermining. As they observe, “even behavior that expresses individuals' values is likely to be influenced by rationally arbitrary causes that are not mediated by” their values; if these causes have to be counted as defeaters, there would seem to be an objectionable scarcity of agency.

Ward & Machery propose this fix: the “valuational account of agency should . . . only require that, in order for an agent to be morally responsible for an action, the action must express at least one of the agent's values. It may also reflect numerous arbitrary causal influences (mediated or unmediated) that the agent would not recognize as justificatory; but that is not enough to disqualify it as agentive.”

If meeting this requirement is supposed to be sufficient for agency, difficulty ensues.Footnote ⁵ Say I value both gustatory pleasure and health, but value gustatory pleasure far less. Say next that someone slips me a science fiction-y medication that briefly amplifies, in some rationally arbitrary fashion, the value I place on gustatory pleasure, and I end up elbow deep in a tray of Winslow's salt caramel donuts. On Ward & Machery's amendment, I'm responsible, because my snarfing expresses at least one of my values. But manipulation of this kind is a classic responsibility negating excuse. (More on manipulation below.) The point generalizes: even manifestly non-agential behavior may express one (or more) of an actor's values. While Ward & Machery's amendment stems the unwelcome proliferation of defeaters, it may succeed too well, countenancing the unwelcome proliferation of agency.

My own attempt to curb the population of defeaters is to say defeaters obtain when the actor would be “unwilling to cite in defense of her behavior the factors figuring in the most perspicuous psychological explanation of her behavior” (précis, p. 1). What's meant to do the needed work is the admittedly vague “most perspicuous psychological explanation.” The most perspicuous explanations won't include irrelevant or extraneous information: many rationally arbitrary causal factors that do not undermine agency, like my parents' first meeting, are in this way excluded. Also doing some work is the requirement that the explanations be “psychological”; when we ask about motives and reasons, we're not asking about distal natural causes like the big bang, so these candidate defeaters are excluded. Is my way of culling defeaters ad hoc or good sense?

Both. Assessing explanatory relevance proceeds on a case-by-case basis, and there's unlikely to be anything tidy and general to say about whether something implicated in a behavior destabilizes agency, or is merely an agentially irrelevant feature of that behavior's causal history. Perhaps this amounts to an on-the-fly “sniff test,” but I'm guessing such appeals are inevitably part of assessing explanations, in as much as explanation has pragmatic goals like producing understanding.Footnote ⁶ There may be some biting of bullets here, where I have to recognize defeaters where I'd rather not. That is, I may get a bit less agency than I wish. But without further elaboration, Ward & Machery's proposal seems to leave us with agency where we shouldn't have it, so I'll stick to my approach, messy as it is.

R1.3. Skepticism and practical utility

Speaking of explanations, Lambert & Dennett demand one of me: if rationally arbitrary influences on behavior are as robust a phenomenon as I suppose, “how do we manage to hold it all together?” People do pretty well at things like interpretation, prediction, and coordination – and more generally, just getting by – but this practical success would be miraculous, the argument goes, if agency were regularly undermined by unexpected “goofy” influences.

Here, it is useful to distinguish rigid and fluid contexts. Rigid contexts are highly constrained (whether implicitly or explicitly), and any exercises of agency in these domains are likely to be less easily destabilized. Fluid contexts are less constrained, and present more latitude for defeaters. Under ordinary conditions, driving is a fairly rigid context: not often do goofy influences have you speeding the wrong direction on the interstate. But there are many (explicit) protections in place, like routing and signage, to prevent such mishaps. Similarly, I doubt there's some goofy manipulation of the sort found in the psychology literature that could surreptitiously induce me to lecture naked, in face of the countervailing (often implicit) social expectations. Once more, this looks like a rigid context, and there's doesn't seem to be much question about defeaters.

Now think of the complex, emotionally freighted contexts I'm calling fluid, such as romantic relationships, political preferences, and career choices. In these cases, which are often of great practical interest, there are serious questions about defeaters (pp. 75–76). Regarding Lambert & Dennett's question, I'm inclined to think that much of our practical success in endeavors like prediction and coordination can be attributed to the behavioral regularity enforced by rigid contexts, while the troubling failures of agency, I'd wager, more often occur in fluid contexts, where there are weaker forces for keeping people on the rails. If so, we have the paradoxical result that defeaters are less likely, and the exercise of agency more likely, where behavior is more constrained. Given that my theory of agency “positively celebrates constraint” (p. 12), this air of paradox is not untoward. It may also be that the rigid/fluid distinction provides some traction on the epistemological problem: we may more confidently attribute agency in certain rigid contexts.

Hirstein & Sifferd offer another sort of explanatory demand: “Doris's view amounts to saying that the entire upper level that has been designed into our brains, including the executive processes and consciousness itself, is of little use or import.” I'm generally leery of arguments about what it makes sense for Mother Nature to do, but if it were the case that my theory implies She designed-in an expensive system without assigning it meaningful work, I agree I'd have some explaining to do. I didn't, as Hirstein & Sifferd observe, say much about brain science (pp. 88–89), but in my dialogic theory the collaborative exchange of rationalizations makes plenty of toil for the “upper level” (as, presumably, do those occasions where behavior conforms to a traditional reflectivist model). Additionally, the brain does many things that are not in the service of morally responsible agency, and any number of these might employ the upper level. I'm inclined to think, with theorists like Mercier and Sperber (Reference Mercier, Sperber, Evans and Frankish2009) and von Hippel and Trivers (Reference von Hippel and Trivers2011) that nature didn't design the fancy human brain with heavy emphasis on accurate self-awareness, but that doesn't mean there's nothing for it to do.

Hammond leverages developmental psychology, which I left largely untapped, into a suggestive defense of reflection. His central example is a study where Piaget (Reference Piaget1974/1976) had children and adults model crawling. Not everyone hit on the correct “X” pattern; for example, some subjects modeled crawling on an “N” pattern. But when subsequently asked to crawl, some of the mistaken modelers crawled in accordance with their erroneous model. As Hammond has it, “our reflections can shape our actions, even when these reflections are inaccurate representations of the state of the world,” and these “subsequently reorganized actions may create a world that more closely resembles what was in error.”

I take a similar line (pp. 135–37, 143–45): the fact can be the child of the fiction, in something like the way Hammond supposes. But now, as a self-styled anti-reflectivist, I have a branding issue. For what Hammond is proposing, in effect, is reflectivism minus the accuracy corollary. While I'm apt to caution that much agential behavior is unreflective (p. 69), perhaps my view ought be characterized as a kind of reflectivism, in so far as I recognize the kind of process Hammond describes. On the other hand, jettisoning the accuracy corollary might cause the reflectivist branding problems of her own: the Oracle said Know thyself, not Reflect inaccurately about thyself, and structure future behavior with these inaccuracies. A view like mine (and I think Hammond's), which centrally features inaccurate self-awareness, betrays the spirit of traditional reflectivisms.

Collerton & Perry defend the practical efficacy of reflection by means of an analogy with vision. Vision is prey to many distortions and inaccuracies, but is nevertheless enormously helpful for getting us along; likewise, Collerton & Perry say, for reflection, which needn't be perfect to be useful. I'm inclined to agree (pp. 129–33), but I'd caution against supposing that our success in getting along is evidence for our successful exercise of agency. After all, many species that presumably lack morally responsible agency, like the pathetically robotical Sphex wasp made philosophically famous by Dennett (Reference Dennett1984, pp. 10–11), manage just fine (when not made examples of by meddling naturalists),Footnote ⁷ and it's not clear that we wouldn't be able to do the same if we lacked morally responsible agency – as indeed some philosophers (e.g., Pereboom Reference Pereboom2014) have argued.

The analogy with vision actually presses this point. Noting the practical value of vision does not necessarily answer skepticism directed at the deliverances of the senses; the skeptic will be quick to grant practical efficacy, while continuing to question the possibility of knowledge. Indeed, comparatively simple animals successfully navigate their world, but epistemologists may hesitate to ascribe full-blooded knowledge to such organisms (e.g., Sosa Reference Sosa and Sosa1991, p. 240). Likewise, the agency skeptic isn't doubting that human beings get by, she's doubting that they get by while exercising agency. And this doubt, she thinks, doesn't get allayed by insisting on our pragmatic success.

R1.4. Skepticism and individual differences

Several commentators suggest I should pay more attention to individual differences. While I've sometimes criticized character and personality theory, I've always acknowledged the importance of individual differences (e.g., Doris Reference Doris2002, pp. 25–26), and I participate in empirical work investigating them (Bollich et al. Reference Bollich, Doris, Vazire, Raison, Jackson and Mehl2016; Cameron et al. Reference Cameron, Payne and Doris2013;Footnote ⁸ Mooijman et al., Reference Mooijman, Meindl, Oyserman, Dehghani, Monteresso, Doris and Grahamforthcoming). Although individual differences were not a primary focus of Talking to Our Selves, I assumed throughout that there are individual differences with respect to agency (e.g., pp. 34–35, 39–41, 48, 156, 162).

It is an interesting, and underexplored, question what exactly those differences might be; one major obstacle is operationalizing amorphous and contested notions of agency for empirical work. Lambert & Dennett suggest there are likely individual differences in susceptibility to the goofy influences that may function as defeaters; if so, there should by individual differences in the extent to which people exercise agency. For example, there's considerable research on individual differences in suggestibility (e. g., Frost et al. Reference Frost, Nussbaum, Loconto, Syke, Warren and Muise2013; Marotta et al. Reference Marotta, Tinazzi, Cavedini, Zampini and Fiorio2016). Intuitively, the less suggestible might be less susceptible to goofy influences, so maybe the less suggestible are, all else equal, better able to exercise agency. This, it seems to me, has the makings of a worthwhile research program integrating moral psychology and personality psychology.

Niemi & Graham argue that individual differences affect not only the exercise of agency, but also the attribution of agency. Work by Graham and his colleagues (Graham et al. Reference Graham, Haidt and Nosek2009; Graham et al. Reference Graham, Nosek, Haidt, Iyer, Koleva and Ditto2011) on moral foundations suggests that different people may adopt very different moral perspectives; for example, conservatives may favor binding values like loyalty and purity, which support group solidarity, while liberals may favor individualizing values like harm prevention and fairness, which focus on the protection of individual persons. Niemi and Young (Reference Niemi and Young2016) have shown that the differing values may be implicated in differing perspectives on responsibility; for example, those higher in binding values may attribute more responsibility to the victim of sexual assault. This result might be thought to generate another sort of skeptical challenge: if variation in responsibility attribution is attributable to foundational differences in values, perhaps we cannot expect agreement on responsibility attributions between people embodying these differing evaluative perspectives. And if there is such “fundamental” evaluative disagreement (Doris & Plakias Reference Doris, Plakias and Sinnott-Armstrong2007), perhaps we cannot converge on objective assessments of responsibility.Footnote ⁹

While consideration of individual differences may be thought to exacerbate the skeptical difficulty, Taylor thinks it may instead suggest a solution: “the development of expertise is a widely applicable self-regulative strategy” so “people should develop expertise in certain domains if they want to consistently express their values.” It's a truism of the human performance literature that where there's expertise, there are individual differences, at least in complex domains (Gobet & Campitelli Reference Gobet and Campitelli2007, p. 159; Howe et al. Reference Howe, Davidson and Sloboda1998, pp. 399–400), so if exercising agency involves a kind of expertise, some folks should be better agents than others. This may sound a bit undemocratic, but in as much as the development of expertise is sensitive to practice (e. g., Ericsson Reference Ericsson2014), perhaps the less agential may become more so, if they put in the work.

Taylor's suggestion is intriguing, not least because expert performance is, oftentimes, not so much undermined by automatic processing as supported by it (Christensen et al. Reference Christensen, Sutton and McIlwain2016), and the performance of experts often seems appropriately agential. One of Taylor's examples is a study where nurses were less subject to the “group effects” that mute helping behavior when other bystanders are present (Cramer et al. Reference Cramer, McMaster, Bartell and Dragna1988). Arguably, the nurses' training made them less susceptible to defeaters, and more agential, in the performance of their profession. Expertise, we might say, involves a kind of educated automaticity: if you want to exercise agency in an area, and squelch lurking defeaters, develop expertise.

Unfortunately, the development of expertise is not so well understood as we might wish, regarding such questions as the relative contributions of talent and practice (for an overview, see Doris, Reference Dorisin preparation). Additionally, expertise is domain limited (Chi Reference Chi, Ericsson, Charness, Feltovich and Hoffman2006): given the large amount of practice required, and the unequal distribution of domain-relevant talents, people are seldom truly expert in multiple complex domains. Now, Taylor's intriguing proposal has generated another intriguing proposal. Perhaps, if people are only expert in a few (or fewer) domains, people only exercise agency in a few narrowly circumscribed areas of their life: maybe I exercise agency as an academic, but not as an athlete. If so, Taylor's solution to the skeptical problem may suggest that for most people, the scope of agency is sharply curtailed.

Fowers et al. also urge attention to individual differences as a way of defanging skepticism. As I say, I welcome research on individual differences, but I'd caution against proceeding in terms of virtue, as they propose: “Behavioral research on virtue and character is just getting under way . . ., but the expectation is that virtue and character will directly reduce defeaters' influence as well as moderate defeater effects.” (After critiquing the philosophical method, Fowers et al. here appeal not to empirical evidence, but to the philosophical authority of Aristotle.)

Apparently, Fowers et al. wish to address the skepticism about traditional conceptions of character I developed in an earlier book (Doris Reference Doris2002). Although I have thoughts about how the literature subsequent to that book is developing (Doris, Reference Dorisforthcoming), in the book that is the topic of the present discussion, I explicitly declined to revisit the issue (pp. 14–16). The character skepticism debate in philosophy (like the person-situation debate in psychology that inspired it) is substantially a debate about cross-situational behavioral consistency – the skeptics contend behavior is much less consistent than traditional notions of character would have us predict. But that issue is orthogonal to the issue of agency. To see this, consider some non-human organisms that exhibit limited behavioral variation, like sea anemones. An organism could be perfectly consistent, and not be an agent; in my frame, a person could be both perfectly consistent and invariably subject to defeaters.

Moreover, linking virtue (or moral probity more generally) and agency has unpalatable results (p. 16): if virtue is what facilitates agency, it would appear that the vicious are unlikely to exercise agency, which flies in the face of the apparently viable practice of holding less-than-virtuous people responsible for their bad deeds. Or so I argue (pp. 156–59). You needn't be convinced. But if you link virtue and agency, you must either (1) develop distinct accounts of responsibility for wrongdoing and rightdoing, or (2) contend that non-virtuous people occupy mitigating or exempting conditions and eschew, contrary to ordinary practice, holding them responsible. Because neither of these options is completely appealing, you'd want to offer considerable argument. Fowers et al. do not attempt to do so.

R2.1. Do we need a theory of agency?

Patrzyk decries the “moralistic appeal of agency attribution,” and any attempt to vindicate existing practices of responsibility attribution. Anti-skeptical approaches to moral responsibility, he contends, are “rescue missions,” which can only be explained by theorists' “need to devise a theory allowing them to justify their retributive instincts and take credit for what they do not deserve.” Existing responsibility practices, for him, are “objectively unfair, creating a situation in which blame judgments depend on factors that cannot be controlled by the actor.”

While Patrzyk finds my moralizing distasteful, he himself is moralizing, because he finds non-skeptical thinking on morality “objectively unfair.” I suspect that charges of moralism themselves reliably lapse into moralizing – the more so when deploying markers of certitude like “objectively.” Given what he says, perhaps Patrzyk should be embarrassed by his moralizing, but I readily admit that my own theorizing is morally freighted.

This is inevitable, in as much as notions like agency and responsibility are “thick,” and simultaneously bear normative and evaluative commitments (pp. 14, 195–96). Thus, asserting that people sometimes exercise morally responsible agency, as I do, is an ethical exercise, as is denying that they do so; the former is to say that people are due a certain kind of moral regard, the latter is to say they are not (as indeed Patrzyk's “credit for what they do not deserve” makes obvious). For example, I'm of the altogether familiar conviction that people deserve a different kind of regard for their good deeds than for their good looks, and I'd mark this divide with the language of agency. Patrzyk may either reject this conviction (as philosophers with skeptical leanings about responsibility, such as Smart [Reference Smart1961], are perhaps inclined), or develop a way of grounding it that does not appeal to agency. In doing either, he would be engaging in a recognizably moral enterprise; in debating agency and responsibility, there's no escape from “moralism.”

R2.2. Reactive attitudes

I anchor my approach in the “reactive attitudes,” but as Viciana, Gaitán, & Aguiar (Viciana et al.) observe, my account here is “importantly undertheorized.” Strawson (Reference Strawson1962), who initially proposed understanding responsibility via the reactive attitudes, was himself more suggestive than systematic, and I was perhaps overly content to emulate him in this respect. I supposed that reactive attitudes like anger and admiration are “symptoms” of morally responsible agency (p. 24), while Viciana et al. assert that reactive attitudes “can sometimes be apt even if they are not tracking morally responsible agency.” They suggest, for example, that we ought be outraged at perpetrators of wartime atrocities, even if the conditions of war often destabilize miscreants' agency (as Doris & Murphy [Reference Doris and Murphy2007] also argue).

I take the point: perhaps the reactive attitudes, like the assignment of criminal responsibility, are sometimes governed by something like strict liability, and appropriately target perpetrators of non-agential deeds. Conversely, there may be cases of responsible agency that don't prompt the reactive attitudes; for example, some agential behaviors may be too inconsequential to provoke the kind of emotional responses, like anger, associated with the reactive attitudes (possibly, this departs what I said in the book, on p. 24).

These are important points, but I doubt they vitiate my approach. I propose treating reactive attitudes as a sort of heuristic – a way of loosely delineating the contours of responsibility practices (plural practices, because as Viciana et al. note, not all responsibility systems are the same). Just as it is often the case that no one symptom is necessary or sufficient (and still less, necessary and sufficient) for diagnosing disease, the presence of reactive attitudes is not necessary or sufficient evidence that agency is at issue. Still, if one wants to look for responsibility attributions, and determine whether those attributions are legitimate, one could do far worse than looking for the presence of the reactive attitudes, together with the kind of justifications people offer when they experience and express them. The reactive attitudes are, as I treat them, a beginning – a way into a theory of morally responsible agency, not the theory itself (p. 23).

R2.3. Agency and values

On the theory I propose, exercises of agency are associated with expressions of value: when my declining a donut expresses the value I place on my health, I exercise morally responsible agency. Dunning wonders why I privilege values, among the various psychological states and processes that structure behavior. My thought is that the relevant psychological states must have a certain authority. For example, lots of desires I have are unimportant, or even repellent, to me, and many others may be fleeting and infirm; they're not the sorts of things around which it makes sense to structure my behavior – and still less, my life. My account of values attempts to identify “desiderative complexes” (p. 183) that are fit to play this structuring role: “values are associated with desires that exhibit some degree of strength, duration, ultimacy, and non-fungibility, while playing a determinative-justificatory role in planning” (p. 28). Such desires, we might say, are worth taking seriously: They have normative heft, or rational authority. I agree with Dunning that I might have gone more pluralist here, and not limited my account to values; perhaps it's possible to “upgrade” other psychological states, like attitudes, in the way I attempted for desires, so that these states can also be seen as integral to agency. Nevertheless, I'd want this pluralism to reserve a large role for values, because they are plausibly supposed to bear the needed normative heft.

Maibom worries that on my account, “there may be no causal chain from your values to the expression of them”; if no causal relation is required, even things I didn't do might count as exercises of my agency (like someone I've never met bringing about an outcome I value after my death). For this reason, I distinguished conforming to a value and expressing a value, arguing that expression requires more than mere conformity (p. 70). Part of the something more, I said, is a causal relation: for expression, “that the actor holds the value should be causally implicated in her undertaking a behavior suited to realize the value” (p. 135). But a causal relation will not be enough, because it might be fortuitous or fluky (p. 25): it shouldn't count as an exercise of agency if the value I place on health causes me to faint in horror rather than lighting up when you offer me a smoke as I'm trying to quit. My fix was to propose that the expression relation is manifest in intentional behavior, which I construed as goal-directed behavior (pp. 25–26); thus, “a behavior expresses a value . . . when that behavior is guided by a value-relevant goal.” This helps exclude the fortuitous and fluky from the offices of agency.

In addition to worrying that my theory makes for too much morally responsible agency, Maibom worries that it makes for too little (as I myself do [p. 70]). On my view, she thinks, “people are rarely, if ever, responsible for wrongdoing,” for they are “unlikely to have put desires in the driving seat that are the sorts of values that we see expressed in wrongdoing.” If putting in “the driver's seat” involves conscious decision, I agree that many, perhaps most, wrongdoers do not act as Milton's Satan (“Evil, be thou my Good”). But crucially, on my view people do not need to consciously entertain the values on which they act to exercise agency; indeed, they might be quite unaware of having the value expressed in these exercises (pp. 27–28, 160–61). Moreover, a value does not need to be intrinsically bad to be expressed in wrongdoing, which will often depend on context: valuing sexual gratification is not wicked, but expressing that value while betraying an explicitly monogamous relationship can be. And of course, in many cases there will be more than one value in play, and sometimes wrongdoing stems less from badness of one's values than from the way one integrates or weights them (cf. p. 162). Hopefully, then, my theory makes for neither too little agency nor too much.

Murray argues that the valuational theory cannot properly account for inadvertent “slips,” like neglecting to turn off the stove and starting a fire, because many slips are omissions for which people are appropriately attributed responsibility, despite the fact that these omissions may “not express a relevant subset of the agent's values.” Like Maibom, Murray worries that my approach is overly exculpating: The unlucky cook is properly held responsible, he thinks, despite not valuing house fires.

Many slips, it seems to me, admit of valuational explanations (some Freudian slips, for example). Perhaps the forgetful cook insufficiently values prudential precautions like double checking to see that the stove is off, and the slip can be characterized by the valuational theory as responsible negligence, despite it being the case that a house fire does not reflect what Murray calls the actor's “overall balance of preferences.” In other cases, perhaps there is no relevant value anywhere in the vicinity of the slip. But here, I don't find myself much inclined to attribute responsibility (p. 155), as in the unfathomably tragic cases where ordinarily conscientious parents leave their young child in the car to die of hyperthermia (for more on slips, see Amaya & Doris Reference Amaya, Doris, Clausen and Levy2015).

Maybe you disagree with my lenience here; Murray asserts that “we regularly hold people responsible for their slips” and thinks “we should prefer theories of responsible agency that preserve this part of our practices” – a preference he thinks my theoretical conservativism requires (p. 158). I'm not sure what the contours of everyday practices regarding slips are, but I'm betting they're pretty variable; while I don't want to lean too hard on the law, parents implicated in fatal vehicular hyperthermia accidents are prosecuted around half the time (Collins Reference Collins2006, pp. 807 n. 2, 825). My guess is that people are sometimes held responsible for slips and sometimes not; “I just forgot” can, after all, sometimes be an acceptable excuse. (That it isn't acceptable for hyperthermia accidents is probably attributable more to the horrific outcome than the psychological inaccuracy of “I just forgot.”) So I'm betting the valuational account can manage the intuitive verdict for a tolerable percentage of slips. But it wouldn't much trouble me if it can't. My approach is “revisionary in some regards and conservative in others” (p. 158), allowing that there is “endless room for improvement” in habits of responsibility attribution. If Murray's speculation that people are routinely held responsible for slips is right, I'm inclined to say the practice overreaches and perhaps ought be changed, a revisionary suggestion that is entirely compatible with my approach.

Vargas thinks the valuational account is prey to “manipulation cases”: a manipulated person might be expressing their values (indeed, she might be manipulated into expressing her values), but manipulation is a standard excuse, so my approach holds people responsible who are not. My response is dilemmatic (cf. pp. 31–32): either the manipulation effects impairments in the exercise of capacities requisite for responsibility, or it does not. If it does effect responsibility relevant impairments, the manipulation is responsibility negating (or mitigating), and I can say the intuitive thing. If it does not, I must bite the bullet and hold the manipulation victim responsible. To soften the bite, I'd contend that cases of manipulation without responsibility relevant impairments are empirically unlikely, so the awkwardness will be unusual.

But, Vargas argues, there is a phenomenon making similar difficulty for me that is not only empirically likely, but also in fact commonplace: “adaptive preferences,” where people's motives and desires are shaped by oppression. Sexism, racism, or other patterns of prejudice, for example, might instill self-abnegating desires (Westlund Reference Westlund2003). According to Vargas, such desires can be associated with values, which means my account apparently has the victims responsible when their conduct expresses the values that result from oppression.

This is an exceedingly sensitive issue. While it may seem hard-hearted to hold victims of oppression responsible for value driven behavior attributable to their oppression, it may seem paternalistic not to hold them responsible for acting in accordance with their values – the more so, where one person's culture is another's oppression. For one of many examples intimating the difficulty, consider debates about Sharia and feminism: Is observing certain traditional religious practices compatible with progressive notions of self-determination and autonomy?Footnote ¹⁰

Let's suppose, however, that there are clear cases of adaptive preferences where the subject should not be held responsible for the associated behavior. If these preferences should be thought of as values, the valuational theory apparently gets the wrong result, holding the actors responsible for behavior expressing these values. However, I question whether the desires at issue have the status of values. The prisoner might (in some sense) desire the disgusting institutional food without valuing it, just as the torture victim might (in some sense) desire denouncing his country, without valuing this performance. Remember that on my view (p. 28), values are associated with ultimate desires that figure in justification and planning; the adaptive desires of the prisoner and torture victim are not ultimate, but instrumental, in the service of their survival. Furthermore, it may be that people repudiate their adaptive preferences even while in the oppressive conditions, and would not appeal to them in justification and planning.

So I'm inclined to think that adaptive preferences will often fall short of values. Where they are aptly characterized as values, there's still the question of whether the associated oppression results in responsibility negating or mitigating disabilities; for example, oppression may result in exculpating ignorance. But when that is not the case, and the adaptive preference is properly thought of as a value, my account attributes responsibility for behaviors expressing that value. This may be an unhappy result. It is a difficulty common to the many “currentist” theories of responsibility that, like mine (pp. 30–32), deny that historical considerations figure directly into assessments of responsibility (as opposed to being indirectly accounted for by assessment of the resulting current states). But if I'm right, such embarrassment will be acceptably scarce.

R2.4. Evolutionary theory

Beal & Rochat observe that I make little use of evolutionary theory, which, they think, means I join the story too late: Instead of the negotiation of moral responsibility, I should have been speaking of a “renegotiation,” for human organisms have preferences even in the womb. As I'd put it, biology constrains agency (cf. p. 195): even in a world of massive cultural diversity, not all forms of life are available, and whatever agency is open to us, it will be highly canalized. In this spirit, Beal & Rochat suggest that I ought better attend to “primordial” features of our natural and material worlds, which also serve to structure and sustain the self: for example, in attempting to understand the tragedy befalling displaced peoples, we must recognize that in addition to social and cultural rupture, the loss of land, animals, and other elements of the natural world is itself catastrophic. I'm quite happy to adopt Beal & Rochat's observations, which I think can enrich my account; as I said, I'd be perturbed if my approach were incompatible with the best evolutionary theory, but I'm optimistic this isn't the case (p. 145).

R2.5. Culture

Franks & Voyer and Dunning observe that cultures vary not only in their values, but also in their understandings of agency. In the book I tentatively concluded (pp. 192–96), from cross-cultural research on the locus of control, that the notion of agency I had in mind was perhaps more widespread than the most provocative readings of cultural psychology, like those adverted to by Franks & Voyer and Dunning, would have us suppose.Footnote ¹¹ But I also took the view that I'd not much fret if the scope of my topic was more parochial than I supposed. WEIRD people –those who are Western, educated, industrialized, rich, and democratic (Henrich et al. Reference Henrich, Heine and Norenzayan2010) – aren't the most numerous of the world's peoples, but they are numerous enough to be worth thinking about, and they were, I suppose, the (mostly tacit) focus of my book. Identifying theoretically perspicuous underpinnings for a – revisably – viable cultural practice is honest work, I thought, even if that practice is parochial.

Perhaps this was overly complacent. Dunning suggests, and I take it Franks & Voyer would agree, that I overemphasize a Western, disjoint, conceptions of agency, where agency involves individuals “imposing their will on the external world.” He commends more attention to conjoint conceptions of agency, as found in some non-Western cultures, where “people strive to harmonize their actions with outside forces and constraints, usually social ones.” Maybe the notion of me as an individual having my own, independent, values resonates much less in conjoint cultures than it does in disjoint cultures. Maybe in conjoint cultures the primary notion of agency is group agency, rather than the individual agency that seems to be everywhere celebrated in the West. My interest in the book was individual agency (p. 169), but if conjoint, group, agency is prominent across the cultures of the world, it would behoove me to explore extending my approach to group agency; my guess is that my collaborativism is well suited for this task, though I can't take that up here.

Another possibility deserving further attention is that cross-cultural diversity amplifies the skeptical challenge. Uleman, Granot, & Shimizu argue that the role of cultural and contextual influences on attribution of responsibility means there is unlikely to be an “objectively ‘correct’” or “god's-eye” view of moral responsibility grounding univocal answers to questions of who is responsible. Shimizu et al. (Reference Shimizu, Lee and Uleman2017) propose that most important cultural differences occur through automatic processes; taken together with Uleman and colleagues' work on spontaneous social inference (e.g., Uleman et al. Reference Uleman, Rim, Saribay and Kressel2012), this leads me to expect that many moral judgments, including responsibility attributions, will tend to be culturally determined, automatic, and unreflective. The attribution of responsibility may itself be subject to defeaters, no less than are the behaviors that are (together with the actor) the targets of responsibility attribution.

That is, people's attributions may be influenced by factors they would not, were they aware of them, invoke as justifications for their attribution. And where people are unaware of these influences, they may be unable to resist them. Now the skeptical difficulty affixes at a second place: not to the exercise of morally responsible agency, but to the attribution of morally responsible agency. If an attribution of responsibility is to be warranted, the presence of defeaters must twice be ruled out, once for action, and once for attribution. Then the challenge of developing the sort of epistemically robust collaborations I propose in response to skepticism is even more challenging than I'd imagined.

R2.6 Collaborativism: Pitfalls

Pe-Curto et al. raise what they call “the problem of bad company,” a difficulty also marked by Lambert & Dennett. As Couchman, Birster, & Coutinho (Couchman et al.) put it, not all participants in the “dialog of self” I envisage will be “playing nice”: people's values might be “hijacked” by manipulative social interactions, and even where participants have the best of intentions, agential dialog may be undermined by biases like anchoring effects.

Perhaps sociality is as likely to impair as promote agency: sociality plus totalitarianism, for example, might vitiate agency rather than facilitate it. Pe-Curto et al. sharpen this difficulty with an epistemological challenge: “If we are so entangled in our milieu for cognition, agency, and our unity as selves, we appear badly placed to tell whether we should embrace or resist its influence.”

Pe-Curto et al.'s response is reflectivist-individualist: “Should we be in bad company, we might need reflective, individualist humans sufficiently in touch with their values, and so able to disentangle themselves from social influence. If we may be so bold to suggest it, we might need humans of character.” Relatedly, Couchman et al. assert that “the highest ethical standard in [Doris's] system ought to be the process of increasing metacognition – the ability to self-regulate (to beat defeaters) and to avoid biases and hostile narratives (or meta-defeaters, if you will).”

As I see it, both of these solutions are troubled by the vagaries of individual reflection I documented throughout Talking to Our Selves; Pe-Curto et al. and Couchman et al. locate the solution where I've located the problem. Regarding Pe-Curto et al.'s appeal to character as the facilitator of reflection, I'd want to know more about what aspects of character get the call in this role, and how they are developed, and I'd also appeal to my work on frailty of character and the uncertainty of moral education (Doris Reference Doris2002; Reference Dorisforthcoming; Reference Dorisin preparation). Regarding Couchman et al.'s appeal to metacognition, I don't deny that metacognition has a role in agency – I say the same for reflection generally (pp. 74, 171–77). Nevertheless, their proposal is subject to the very difficulties they themselves raise: why think metacognition less liable to bias and hostile manipulation than cognition simpliciter? (If the solution is to reflectively monitor metacognitions, what makes these meta-metacognitions immune? And so on.) This suggests that Couchman et al. have not ameliorated, but instead relocated, the skeptical difficulty.

But Couchman et al. also suggest the kind of ameliorative strategy I favor, one that trades more on external scaffolding than internal cognition. Knowing the vagaries of memory, an organization might provide minutes of a meeting to attendees. Concerned about the possibility of implicit bias, a search committee might appoint a “diversity officer” to ensure that all files from underrepresented groups receive full and fair consideration. Increasing the accuracy of metacognition can have a place in my approach, but it won't be the “highest ethical standard,” because other processes, including the cultural and institutional, will be at least as important to agency. Of course, what's required is the “right kind” of relationships and institutions, and I didn't do nearly as much as needs be done in specifying what the right kinds are (but see pp. 119–23). To go further, I'm guessing, there will need to be much closer connections between moral psychology and such disciplines as political science than are currently evident in the literature.

None of what I've said excludes altogether a role for reflection; just as there aren't tight entailments between reflectivism and individualism, collaborativism does not entail the rejection of reflectivism (p. 110). Reflection has a role in human life, and a role in agency. The trick is to say something about when and how. Collerton & Perry offer a rich example: the distressing hallucinations associated with some forms of mental illness may be improved by “a joint therapist-client investigation of the reality of experiences.” This much, as they note, is congenial to my collaborativism, but they also suggest the process involves accurate reflection, because “insight” may contribute to amelioration. Then defeaters may sometimes be countered by accurate reflection, and people may sometimes better exercise agency in this way. The clinical treatment of hallucinations is a limited context, but I don't deny that there may be others; that such contexts are prominent enough in human life to satisfy the reflectivist, I am inclined to doubt.

It's also worth noting that reflection may have importance beyond facilitating the exercise of agency. Franks & Voyer contend that my focus on “revealed” agency – agency as manifested in (patterns of) overt behavior – neglects experiences of agency, which “figure significantly in people's own normative explanations and justifications, and connect directly to the sense of self.” This seems right to me: that I think of myself as an agent, rather than, say, a puppet, has a lot to do with how I think and feel about myself, and how others do. An illustration of this I find especially compelling is abnormal experiences of control in mental illness, such as the atrophied perceptions of control associated with schizophrenia (p. 134); the significance (and misfortune) of this condition is not limited to any associated impairments of agential behavior.

R2.7. Collaborativism: Processes

Bonicalzi & Gallotti observe that more needs be said about the “mechanics of collaborativism.” A mechanism they suggest is alignment – the development of shared understandings, such as publicly available and validated moral norms, through social interaction. The internalization of these norms may help answer an important question for collaborativism: why should people be moved to justify themselves to each other, and engage in the practice of rationalization at all? Part of the answer, evidently, is that they hold themselves, to some degree, to a set of shared norms, and so feel motivated to live up to the norms, and explain themselves when they do not. (This account of course requires a story about how the norms get internalized [e.g., Sripada & Stich Reference Sripada, Stich, Carruthers, Laurence and Stich2006].) So if Bonicalzi & Gallotti are right, we've got a crucial piece of the needed mechanics: a story about why people are motivated to collaborate in the first place.

I'm pleased that Niemi & Graham considered my account of the self from the book's last chapter, which seemed to draw less attention than other parts of my argument, but is, I think, crucial to understanding collaborative agency. My concern there was destabilizations of self and agency: I contended (pp. 181–86) that cultural devastation might disrupt the personal continuity required for temporally extended “diachronic” (pp. 163–64) exercises of agency associated with the major life projects, like those involving work and family, that imbue human life with meaning.

Niemi & Graham suggest a process by which this might occur: the depersonalization associated with trauma. Work such as Nizzi and Niemi's (Reference Nizzi and Niemiin preparation) suggests that trauma survivors may experience a sense of self rupture and foreshortened future: their pre-trauma self seems to them destroyed, and it is unclear to them how, or if, their post-trauma self may go forward. This strikingly resonates with the remark attributed to the Crow Chief Plenty Coups, who says of the destruction of Crow traditions, “after this, nothing happened” (pp. 180–81). I didn't notice the connection between cultural devastation and trauma research before Niemi & Graham's urgings, and I take their suggestion that my theory has testable empirical implications regarding the experience of self, patterns of moral judgment, and commission of harm. I also welcome their proposal for future directions: collaborations between clinical psychology and philosophy for understanding and addressing trauma. Philosophical theorizing about agency and the self may help us to understand trauma, while understanding the clinical processes by which trauma may be healed can help us to understand the mechanisms that develop and sustain agency and the self.

Hechler & Kessler indicate another important direction for enriching collaborativism. I understood the negotiations characteristic of agency and its attribution in terms of simple dyads, but as they say, agency attributions are often produced by multiple observers, who “validate their perceptions and beliefs with reference to their fellow group members.” Attribution is not limited to straightforward actor-observer pairings: there's also actor-observer s , actor s -observer, actor s -observer s . Also, because observers will often be active rather than passive observers, many attributions may be better described in terms of actor(s)-actor(s) dynamics (that is, the actor-observer distinction is unstable). Also, as Hechler & Kessler say, attribution may express values (cf. Niemi & Graham), which means that the attribution of agency may also be an exercise of agency. Finally, the attribution of agency may facilitate the exercise of agency; agency may involve collaborations among groups who are exercising their own agency while facilitating and constraining the agency of other groups, who are themselves doing the same. Collaboration, then, is likely to be far more complex than my programmatic depictions intimate.

Oftentimes, these complex social dynamics will not be well described as collaborative; as Hechler & Kessler observe, the parties may stand in varying relationships from cooperative to adversarial. This, however, won't always undermine agency, as Mercier shows in augmenting collaborativism with the interactionist account of reasoning he developed with Sperber (Mercier & Sperber Reference Mercier and Sperber2017). On their view, reason “would have evolved chiefly to serve two related functions, which are both social”: (1) justifying our actions, and evaluating other's justifications, for the purpose of social assessment, and (2) arguing for our own beliefs and evaluating the arguments of others, for the purpose of facilitating communication.

At present, the keyword is “arguing.” I had in mind cooperative collaborations, where participants have substantially overlapping interests, as in my favored example of romantic relationships. But Mercier says, “reasons are most helpful when full collaboration cannot be expected”; the call for justification arises more when parties disagree than when they agree. In so far as my justifications structure my behavior in ways that express my values, it appears to follow that contention, no less than collaboration, may facilitate agency. While we get by with the help of our friends, we may also need the help of our frenemies: Bad Company isn't always bad for agency.

For a fully baked rendering of collaborativism, we require an understanding of “reasoning” more developed than my rather programmatic account (pp. 43–44; 104–106). Mercier counsels against equating reasoning, as I was tempted to do (p. 50), with “System 2” effortful, analytic processing, because “finding and evaluating reasons is, in most cases, quasi-effortless and automatic.” I found conflicts between the dumb automatic and smart analytic to be most trenchant (pp. 69–70); but if Mercier is right, many conflicts between reason and unreason may be conflicts within automatic processing. This might, as he suggests, mean my critique of reflectivism is “too generous,” because the role of reflection might be limited even within the class of cognitive activities appropriately considered reasoning.

That said, we're not without anti-skeptical resources. Zinken & Reddy, with a nice turn of phrase, suggest that in addition to defeaters, there are supporters, “causes of our behavior that we are often unaware of, but that would make good-enough reasons for our actions, were we made aware of them.” They launch this suggestion with intriguing work in linguistics on the selection of interrogatives – Can you pass me a plate?– or imperatives – Pass me a plate – in making a request (e. g., Zinken & Ogiermann Reference Zinken and Ogiermann2013). Evidently, the selection depends on how intrusive the request is: if passing a plate is continuous with what you are doing – maybe you are stacking dishes – my using the imperative is fine, but if it is discontinuous – maybe you are stirring the sauce – an interrogative is required. One lesson for my project is that linguistics, which I did not consider in the book, offers materials for helping us better understand collaborativism: language shapes social interactions, so understanding the details of this shaping might help us better understand how sociality facilitates the exercise of agency.

Additionally, these findings from linguistics can enrich the understanding of defeaters. Presumably, the grammatical form of a request – for everyday requests involving plates, if not for life-changing requests like proposals of marriage – is very often selected unconsciously. But if you were made aware that your selection was based on assessments of continuity, would you regard this as a reason? Presumably, you wouldn't spontaneously offer up such a reason; Zinken's (Reference Zinken2016) subjects justified their selection by appeal to politeness, or said it was arbitrary. If you were offered the scientifically substantiated “continuity explanation” would you take it on board as a “good enough reason”? And if you took it on board as a reason, would that suggest your behavior was agential?

I suspect you didn't know about the explanation, and why it makes your behavior make sense, on previous occasions when you've made a request – it's a cutting-edge bit of scientific discovery and theorizing, that, if you are like me, you're just learning about. If you would accept the scientific account as your reason, when it was explained to you, it apparently makes a promising candidate for agency on my view (pp. 27–28), which requires only counterfactual acceptance (a necessary amendment, because I insist one needn't be conscious of his or her reasons). But it's not obvious such esoteric science connects up with anything you would have recognized as a reason, prior to substantial educational intervention. This suggests that the counterfactual test needs to be amended, with a restriction that the counterfactual recognition not require too substantial a change to your cognitive and motivational structures at the time of action. Once again, I'm pressed to think harder about what cognitive processes deserve to be called “reasoning.”

Finally, my approach bears strong affinities to Alfano's (Reference Alfano2013a) suggestive account of moral character, where, as stated in Alfano's commentary here, “tactically deployed fictions about ourselves can become facts” (e.g., falsely attributing honesty to oneself can help one behave more honestly in the future).Footnote ¹² As with my collaborativism, Alfano thinks this process is substantially social, involving iterated “bid-and-accept patterns” where a person announces what “her values, motives, concerns, or drives are” for evaluation and ratification (or, presumably, rejection) by her associates. The process may also go in the other direction: a person may announce what another person is, and that other may accept or reject the attribution.

From here, Alfano makes the intriguing suggestion that we must recognize “a novel class of dispositions – namely, the dispositions associated with being a good echo” – a person who helps others live up to their announced self. If these dispositions are admirable dispositions, they must also involve a readiness not to echo but oppose – for instance, when one's interlocutor makes pernicious announcements, such as those that are harmful to self and others (here, as elsewhere, the agential and the morally good may come apart). Furthermore, as Alfano knows better than most, whatever theory one wishes to construct for these echo-dispositions, it must account for the fact that dispositions are highly liable to situational disruptions. Nevertheless, Alfano is right to think that as we attempt to more completely uncover the psychological mechanisms supporting collaborativism, we must look for the relevant individual dispositional differences, even if they are not so robustly impactful as we might have hoped.

R3.1. Agency and the law

Mattei contends that if the position I develop “is to be taken seriously . . . the proposed conceptual framework should be able to transcend the purely theoretical realm.” It's not obvious that all theorizing must have practical implications to be “taken seriously” (such a demand seems singularly inapposite for work in metaphysics, for example), but I accept this aspiration for my own work, which I expect might inform, and be informed by, the everyday practice of responsibility attribution (pp. 5, 156–59). From there, Mattei and I part ways, because he considers only one practical domain, that of law, while my project is officially neutral on topics like criminal responsibility (p. 24). My focus throughout was on how to think about everyday interpersonal relationships, and because it seems quite undeniable such relationships are of great practical significance, I suppose that the theorizing I do meets such standards of practical relevance as are appropriate for theorizing in moral psychology and ethics.

I do not know if my theory is, as Mattei says, “irreconcilable with key principles of the American common law tradition.” To re-emphasize, this issue is avowedly not my issue, but I'll say a little about it. Assuming Mattei's charge of irreconcilability is right, we've a couple of interesting possibilities: so much the worse for my theory, or so much the worse for the law. Now the relationship between moral responsibility and criminal responsibility is a delicate one; but while matters are controversial, it seems at least to be commonly held that criminal responsibility should “track” moral responsibility (Duff Reference Duff2009; problematic issues like strict liability noted). If this is right, I can say that the law should be accountable to the best going theory of moral responsibility rather than the other way around, and it's unclear why any tension should be mine to ameliorate.

Suppose one asserted the contrary, as we might take Mattei to do. I'd want to proceed by considering concrete cases, but I'd expect to find many places where we'd want to resist the law dictating our theory of agency. For example, given psychopaths' profound deficits in emotional processing and impulse control (Kiehl Reference Kiehl2014), it is highly plausible that they are not morally responsible for what they do. Yet the law does not excuse or exempt psychopaths (who may account for 25% of the prison population), and it is convincingly argued that this state of affairs is unjust (Morse Reference Morse2008). Whether or not you agree – maybe the extreme dangerousness of psychopaths justifies their incarceration – there seems scant ground for letting the current state of criminal law determine our thinking on the moral responsibility of psychopaths, given what is known about the illness.

Perhaps there are instances where my theory of moral responsibility would apply infelicitously to questions of legal responsibility, as Mattei insists. But as a pluralist (pp. 171–77), this doesn't trouble me; I deny that any single theory of responsibility is equally applicable across all contexts, and I'd not be surprised if notions of responsibility that do good work in many everyday contexts, like interpersonal relationships, are not applicable to the particular context of the law. In any event, there is much to be learned by considering moral responsibility independently of criminal responsibility, as a great many philosophers have done (even if it can be shown that the domains importantly interact).

3.2. In Praise of Busybodies

Sommers fears not that my pragmatic aspirations are too limited, but that they are too extensive. He proposes that I should, instead of adopting a pluralism that accommodates numerous theoretical perspectives, adopt a pluralism that “rejects theorizing about responsibility altogether.” According to Sommers, it's not “the philosopher's business to cast judgment” on everyday judgments of responsibility; to do so is to be a philosophical “busybody,” meddling in practices that successfully deliver what participants in the practices require.Footnote ¹³

Philosophers, at least since Socrates, the patron saint of philosophical busybodies, have staged “philosophical interventions” on theoretically suspect practices and beliefs. So Sommers's position is at once conservative, because it argues that responsibility theorists should not attempt to alter existing practice, and radical, because it advocates overturning what is plausibly thought to be the animating spirit of the philosophical enterprise.

In defense of his view, Sommers offers Lee, from the film Manchester by the Sea, whose drunken carelessness causes his children to perish in a fire. According to Sommers, Lee would not be held responsible on my valuational theory, because the deaths of his children clearly do not express his values. I could dispute this: maybe Lee's behavior ought be understood as stemming from his valuing partying too much, and the responsibilities of parenting too little, and maybe then, valuational theory has him responsible.

But the main question for Sommers is not Lee's responsibility, but who gets to decide. Sommers thinks responsibility attributions are for “the people who are involved in the situation to arrive at,” while I think the theorist has a say. However, I'm not arguing that the role of the theorist is to engage in direct intervention. Sommers is probably right that abstract philosophical theorizing is unlikely to help Lee with his crushing guilt (and/or shame), which we may suppose reflects a self-attribution of responsibility. For that kind of intervention, another kind of busybody, the therapist, is likely better suited (assuming ameliorating Lee's pain would be a good outcome, which Sommers may deny).

Suppose I'm right that many forms of mental illness can be understood as impairments of agency (pp. 34–35). Now allow me to speculate that many sufferers of mental illness are held morally responsible for illness-related behaviors, either because they are not recognized as ill, or because the connection between their illness and impairments of agency are insufficiently appreciated. If this is right, many responsibility attributions of “people who are involved in the situation,” are likely to be quite wrong by the standards of the best going theory. Still, the theorist shouldn't expect to directly change minds, or change them overnight. Rather, any practical influence of theory is likely to be diffuse, indirect, and long-term. Theoretical influence might proceed, for example, through popular forms of media; lots of people flip through Psychology Today at the grocery store checkout, and the magazine apparently boasts a circulation of 275,000 (probably not too shabby, in the internet era). Moreover, the practical influence of theory may be salutary influence; if academic theory has had a role in more humane attitudes toward mental illness, that's a good thing. We ought, I think, be thankful for philosophical busybodies, even if we have good reason to ignore some of their recommendations.Footnote ¹⁴