In his new book, John Doris takes on an intimidating interdisciplinary challenge: How does one retain a philosophically intelligible account of moral responsibility that is consistent with current themes in psychology and cognitive science? The contradictions are stark between the two. As Doris notes, typical philosophical accounts of the moral reasoner assert that people are fully functioning agents making moral choices after careful and conscious deliberation. In a phrase, they are reflective moral agents (e.g., Annas Reference Annas1993; Brink Reference Brink1992; Tiberius Reference Tiberius2002) in full control of their moral choices, basing those choices on visible and diligent ethical reasoning, and fully self-aware of the basis for their moral decisions.
Against this logic, contemporary research in psychology, cognitive science, and neuroscience shatters the portrait of the self-necessary for such a “reflectivist” account of moral reasoning. Instead, extant research suggests that the self is anything but a coherent, reflective, and rational being (Stich Reference Stich1990). To be sure, people experience themselves as conscious and deliberative organisms, but the real action leading to their moral choices lies elsewhere (e.g., Haidt Reference Haidt2001).
In particular, people's choices are importantly determined by System 1 processes, that is, mental operations that are quick and that often operate under the level of conscious awareness and control. These are distinct from System 2 processes, conscious and effortful cognitive operations that the moral reasoner can guide and govern (Kahneman Reference Kahneman2011a; Sloman Reference Sloman1996). The operation of System 1 suggests that the true causes for people's moral behavior can be distinct from, or incongruent with what the moral reasoner believes them to be. System 1 can “defeat” moral agency so analytically constructed in System 2.
In the first half of the book Doris guides the reader through decades of empirical data that have accomplished this shattering of the agentic self. He takes readers through findings showing that people often produce mistaken accounts of themselves and the causes of their behavior, showing how conscious thought and judgment about the self are often filled with confabulation, illusion, and misunderstanding (Berlyne Reference Berlyne1972; Dunning et al. Reference Dunning, Heath and Suls2004; Johansson et al. Reference Johansson, Hall, Sikström and Olsson2005; Nisbett & Wilson Reference Nisbett and Wilson1977). The argument Doris makes is rigorous and comprehensive, and contains, I believe, one of the most coherent and instructive narratives about what twentieth-century psychological science has to say about self and agency – in both senses of self that William James talked about: the “I” as the doer and the “me” as the object of reflection (James Reference James1890/1950).
That said, the second half of the book is where the more difficult segment of Doris's task begins: After all this shattering of the agentic and self-aware agent, can one rehabilitate the idea of moral agency and responsibility among the ruins that are left? Here, Doris is less successful simply because he is less complete. He offers directions to take rather than destinations to inhabit.
Those directions, however, are intriguing. The first is the proposal that moral decisions are first and foremost designed to be expressions of a person's values. The second is to reject moral agency as a completely individual exercise and instead emphasize “collaborativism” among people. People do not conduct their moral reasoning in isolation. Instead, they discuss and debate it with other people.
Let's evaluate each of these proposals in turn. The first is that moral choices are primarily expressions of a person's values. This is an intriguing idea, but Doris leaves it underdeveloped. First, why would System 1 processes lead to moral behavior favoring expression of values over all other possible alternatives? There are many to consider.
Dan Katz, in his classic article on attitudes, noted that attitudes and opinions, like all human responses, serve many different functions (Katz Reference Katz1960). To be sure, people endorse certain attitudes because those attitudes express the values that they hold most dear (e.g., of liberty, individualism, or equality). But there were other functions as well. People might hold certain attitudes because they were the ones that tune the person best to objective truth. Attitudes can also protect and bolster the ego of the person, or assist that person to fit in as a proper member of the groups he or she wishes to affiliate with. These other functions seem just as plausible an engine for System 1 driven moral choices as does value expression. So why privilege value expression?
Second, in emphasizing value expression, Doris potentially makes a thought-provoking implicit proposal well worth following up: People are more concerned about the actions they choose than they are about the outcomes those actions might produce (Cushman Reference Cushman2013; Dunning & Fetchenhauer Reference Dunning and Fetchenhauer2013). That is, people want to express being the right person by choosing the correct behavior rather than by ensuring the right outcome. I have seen such behavior in my own lab. People choose to trust complete strangers with their money even though on average they expect they will never see that money come back to them. They do not seem to be concerned with the outcome of losing the money as much as signifying, through their behavior, the “right” social attitude, namely that they respect the character of the stranger until proven otherwise (Dunning et al. Reference Dunning, Anderson, Schlösser, Ehlebracht and Fetchenhauer2014; Reference Dunning, Fetchenhauer, Schlösser and Elliot2016). As such, one can ask under Doris's scheme whether one can cross out not only reflection as an important component of moral behavior but also consequentialism.
But the more important move is the one toward collaborativism, the proposal that moral agency arises not from individual reasoning but thinking among people. Moral reasoning here becomes a social product, one shaped and guided by action and talk among people as they live their daily lives trying to get along in something better than a nasty, brutish, and short interplay.
Doris proposes that authentic moral agency emerges from this social interaction and discussion, that people achieve accurate and reflective self-awareness of the bases that inform their moral choices. First, he notes correctly that groups tend to produce more insightful and sound solutions to problems than do individuals (Hill Reference Hill1982; Schwartz Reference Schwartz1995). Second, he notes that confabulation, if critiqued and directed by others, may come closer to accurately citing the values that truly drive moral choice. As such, through collaborative social interaction, people come closer to moral reasoning that reflects the true causes of their moral behavior. They become more authentically self-aware.
These are reasonable and testable proposals, but there are other, potentially more radical proposals hiding in plain sight within recent psychological literature. For one, despite the move toward collaborativism, Doris still holds to a rather individualistic and Western vision of agency, one in which people freely choose without too much influence from outside forces. People still act as captains of their own decisions, thus imposing their will on the external world. In psychology, this is known as disjoint agency, the type of agency usually presumed in Western cultures (Savani et al. Reference Savani, Markus and Conner2008; Reference Savani, Markus, Naidu, Kumar and Berlia2010). Other cultures hold to a more conjoint model of agency, in which people strive to harmonize their actions with outside forces and constraints, usually social ones, which they are surrounded by. As such, people still view themselves as agentic, but not just as complete free agents (Savani et al. Reference Savani, Markus and Conner2008; Reference Savani, Markus, Naidu, Kumar and Berlia2010). It would seem to me that if people are collaborative, as Doris speculates, they would be more likely to shift to a conjoint stance in their moral agency than a disjoint one.
But more important, Doris emphasizes the impact of collaborativism on moral reasoning. I would instead speculate that collaborativism would have more of an impact on what people actually do (Henrich et al. Reference Henrich, Ensimger, McElreath, Barr, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Henrich, Lesorogol, Marlowe, Tracer and Ziker2010). The major impact of social collaboration would be on those System 1 defeaters that make people act in ways incongruent with their reflective moral agency. After all, a key insight of recent work on prosocial behavior is not that people go out of their way to be compassionate and generous. Instead, they are kind to others because they are bending to social rules and obligations.
As such, prosocial behavior is often not so much about giving as it is about giving in – to norms and social roles (Cain et al. Reference Cain, Dana and Newman2014). For example, if Salvation Army volunteers expressly ask for holiday donations outside of a supermarket, the number of people donating increases significantly. So, however, does the number of people exiting out some other door to avoid the volunteers (Andreoni et al., Reference Andreoni, Rao and Trachtmanin press). Apparently, people will give, but it may not reflect their moral preferences. Our work on trust behavior also exemplifies this. People do not go out of their way to trust other people. Instead, the key factor is that they feel so anxious and tense about not trusting the other person. Trust is what they should do, not necessarily what they want to do, and so they give in (Dunning et al. Reference Dunning, Anderson, Schlösser, Ehlebracht and Fetchenhauer2014).
Here's another example about how collaboration may center more on agentic defeaters than agentic reasoning: If you ask people whether they will challenge someone making a sexist or racist comment, they typically claim that they would confront the person directly and immediately. However, if you give them an actual encounter with such a person, they act instead by ignoring or deflecting the comment (Kawakami et al. Reference Kawakami, Dunn, Karmali and Dovidio2009; Swim & Hyers Reference Swim and Hyers1999; Woodzicka & LaFrance Reference Woodzicka and LaFrance2001). At the moment of moral truth, a lifetime of collaborative social training and interaction appears to have taught people to be polite and respectful of others rather than to stand up for their personal values (Brown & Levinson Reference Brown and Levinson1987; Goffman Reference Goffman1958; Reference Goffman1967). Politeness trumps (i.e., defeats) moral values. Of course, there are socially proscribed occasions when argument and contention are expected (cable television, anyone?), but those instances are few and well-defined.
What this may ultimately mean is that moral agency may not reside in individuals but rather in social groups that make up, teach, and enforce the social rules that people live by. Or instead, moral agency may arise in the cues of social situations that signal one value over its alternatives (Lindenberg Reference Lindenberg, Steg, van denBerg and de Groot2012). People in economic games, for example, adopt rather different stances toward cooperation versus selfishness simply depending on the name of the game they are incidentally playing – for example, “the community game” versus “the Wall Street game” (Liberman et al. Reference Liberman, Samuels and Ross2004). The simple choice architecture of the game itself can suggest a moral value. In simple dictator games people tend to share the money they have with other people. However, adding the option of taking the other player's money stops that sharing. Instead, the modal choice is for players to take the other player's money. Unprompted generosity turns to unbridled greed (List Reference List2007).
In sum, Talking to Our Selves is a pleasing and thought-provoking tour of the philosophical and psychological issues surrounding agency, self-awareness, and moral choice. It is a delight for those wishing a well-curated tour of past scholarship on these issues, particularly psychological research. It also sets a fine table for a potentially roiling but useful discussion about these issues yet to come. I hope it provokes those discussions. It would be collaboration well worth having.
In his new book, John Doris takes on an intimidating interdisciplinary challenge: How does one retain a philosophically intelligible account of moral responsibility that is consistent with current themes in psychology and cognitive science? The contradictions are stark between the two. As Doris notes, typical philosophical accounts of the moral reasoner assert that people are fully functioning agents making moral choices after careful and conscious deliberation. In a phrase, they are reflective moral agents (e.g., Annas Reference Annas1993; Brink Reference Brink1992; Tiberius Reference Tiberius2002) in full control of their moral choices, basing those choices on visible and diligent ethical reasoning, and fully self-aware of the basis for their moral decisions.
Against this logic, contemporary research in psychology, cognitive science, and neuroscience shatters the portrait of the self-necessary for such a “reflectivist” account of moral reasoning. Instead, extant research suggests that the self is anything but a coherent, reflective, and rational being (Stich Reference Stich1990). To be sure, people experience themselves as conscious and deliberative organisms, but the real action leading to their moral choices lies elsewhere (e.g., Haidt Reference Haidt2001).
In particular, people's choices are importantly determined by System 1 processes, that is, mental operations that are quick and that often operate under the level of conscious awareness and control. These are distinct from System 2 processes, conscious and effortful cognitive operations that the moral reasoner can guide and govern (Kahneman Reference Kahneman2011a; Sloman Reference Sloman1996). The operation of System 1 suggests that the true causes for people's moral behavior can be distinct from, or incongruent with what the moral reasoner believes them to be. System 1 can “defeat” moral agency so analytically constructed in System 2.
In the first half of the book Doris guides the reader through decades of empirical data that have accomplished this shattering of the agentic self. He takes readers through findings showing that people often produce mistaken accounts of themselves and the causes of their behavior, showing how conscious thought and judgment about the self are often filled with confabulation, illusion, and misunderstanding (Berlyne Reference Berlyne1972; Dunning et al. Reference Dunning, Heath and Suls2004; Johansson et al. Reference Johansson, Hall, Sikström and Olsson2005; Nisbett & Wilson Reference Nisbett and Wilson1977). The argument Doris makes is rigorous and comprehensive, and contains, I believe, one of the most coherent and instructive narratives about what twentieth-century psychological science has to say about self and agency – in both senses of self that William James talked about: the “I” as the doer and the “me” as the object of reflection (James Reference James1890/1950).
That said, the second half of the book is where the more difficult segment of Doris's task begins: After all this shattering of the agentic and self-aware agent, can one rehabilitate the idea of moral agency and responsibility among the ruins that are left? Here, Doris is less successful simply because he is less complete. He offers directions to take rather than destinations to inhabit.
Those directions, however, are intriguing. The first is the proposal that moral decisions are first and foremost designed to be expressions of a person's values. The second is to reject moral agency as a completely individual exercise and instead emphasize “collaborativism” among people. People do not conduct their moral reasoning in isolation. Instead, they discuss and debate it with other people.
Let's evaluate each of these proposals in turn. The first is that moral choices are primarily expressions of a person's values. This is an intriguing idea, but Doris leaves it underdeveloped. First, why would System 1 processes lead to moral behavior favoring expression of values over all other possible alternatives? There are many to consider.
Dan Katz, in his classic article on attitudes, noted that attitudes and opinions, like all human responses, serve many different functions (Katz Reference Katz1960). To be sure, people endorse certain attitudes because those attitudes express the values that they hold most dear (e.g., of liberty, individualism, or equality). But there were other functions as well. People might hold certain attitudes because they were the ones that tune the person best to objective truth. Attitudes can also protect and bolster the ego of the person, or assist that person to fit in as a proper member of the groups he or she wishes to affiliate with. These other functions seem just as plausible an engine for System 1 driven moral choices as does value expression. So why privilege value expression?
Second, in emphasizing value expression, Doris potentially makes a thought-provoking implicit proposal well worth following up: People are more concerned about the actions they choose than they are about the outcomes those actions might produce (Cushman Reference Cushman2013; Dunning & Fetchenhauer Reference Dunning and Fetchenhauer2013). That is, people want to express being the right person by choosing the correct behavior rather than by ensuring the right outcome. I have seen such behavior in my own lab. People choose to trust complete strangers with their money even though on average they expect they will never see that money come back to them. They do not seem to be concerned with the outcome of losing the money as much as signifying, through their behavior, the “right” social attitude, namely that they respect the character of the stranger until proven otherwise (Dunning et al. Reference Dunning, Anderson, Schlösser, Ehlebracht and Fetchenhauer2014; Reference Dunning, Fetchenhauer, Schlösser and Elliot2016). As such, one can ask under Doris's scheme whether one can cross out not only reflection as an important component of moral behavior but also consequentialism.
But the more important move is the one toward collaborativism, the proposal that moral agency arises not from individual reasoning but thinking among people. Moral reasoning here becomes a social product, one shaped and guided by action and talk among people as they live their daily lives trying to get along in something better than a nasty, brutish, and short interplay.
Doris proposes that authentic moral agency emerges from this social interaction and discussion, that people achieve accurate and reflective self-awareness of the bases that inform their moral choices. First, he notes correctly that groups tend to produce more insightful and sound solutions to problems than do individuals (Hill Reference Hill1982; Schwartz Reference Schwartz1995). Second, he notes that confabulation, if critiqued and directed by others, may come closer to accurately citing the values that truly drive moral choice. As such, through collaborative social interaction, people come closer to moral reasoning that reflects the true causes of their moral behavior. They become more authentically self-aware.
These are reasonable and testable proposals, but there are other, potentially more radical proposals hiding in plain sight within recent psychological literature. For one, despite the move toward collaborativism, Doris still holds to a rather individualistic and Western vision of agency, one in which people freely choose without too much influence from outside forces. People still act as captains of their own decisions, thus imposing their will on the external world. In psychology, this is known as disjoint agency, the type of agency usually presumed in Western cultures (Savani et al. Reference Savani, Markus and Conner2008; Reference Savani, Markus, Naidu, Kumar and Berlia2010). Other cultures hold to a more conjoint model of agency, in which people strive to harmonize their actions with outside forces and constraints, usually social ones, which they are surrounded by. As such, people still view themselves as agentic, but not just as complete free agents (Savani et al. Reference Savani, Markus and Conner2008; Reference Savani, Markus, Naidu, Kumar and Berlia2010). It would seem to me that if people are collaborative, as Doris speculates, they would be more likely to shift to a conjoint stance in their moral agency than a disjoint one.
But more important, Doris emphasizes the impact of collaborativism on moral reasoning. I would instead speculate that collaborativism would have more of an impact on what people actually do (Henrich et al. Reference Henrich, Ensimger, McElreath, Barr, Barrett, Bolyanatz, Cardenas, Gurven, Gwako, Henrich, Lesorogol, Marlowe, Tracer and Ziker2010). The major impact of social collaboration would be on those System 1 defeaters that make people act in ways incongruent with their reflective moral agency. After all, a key insight of recent work on prosocial behavior is not that people go out of their way to be compassionate and generous. Instead, they are kind to others because they are bending to social rules and obligations.
As such, prosocial behavior is often not so much about giving as it is about giving in – to norms and social roles (Cain et al. Reference Cain, Dana and Newman2014). For example, if Salvation Army volunteers expressly ask for holiday donations outside of a supermarket, the number of people donating increases significantly. So, however, does the number of people exiting out some other door to avoid the volunteers (Andreoni et al., Reference Andreoni, Rao and Trachtmanin press). Apparently, people will give, but it may not reflect their moral preferences. Our work on trust behavior also exemplifies this. People do not go out of their way to trust other people. Instead, the key factor is that they feel so anxious and tense about not trusting the other person. Trust is what they should do, not necessarily what they want to do, and so they give in (Dunning et al. Reference Dunning, Anderson, Schlösser, Ehlebracht and Fetchenhauer2014).
Here's another example about how collaboration may center more on agentic defeaters than agentic reasoning: If you ask people whether they will challenge someone making a sexist or racist comment, they typically claim that they would confront the person directly and immediately. However, if you give them an actual encounter with such a person, they act instead by ignoring or deflecting the comment (Kawakami et al. Reference Kawakami, Dunn, Karmali and Dovidio2009; Swim & Hyers Reference Swim and Hyers1999; Woodzicka & LaFrance Reference Woodzicka and LaFrance2001). At the moment of moral truth, a lifetime of collaborative social training and interaction appears to have taught people to be polite and respectful of others rather than to stand up for their personal values (Brown & Levinson Reference Brown and Levinson1987; Goffman Reference Goffman1958; Reference Goffman1967). Politeness trumps (i.e., defeats) moral values. Of course, there are socially proscribed occasions when argument and contention are expected (cable television, anyone?), but those instances are few and well-defined.
What this may ultimately mean is that moral agency may not reside in individuals but rather in social groups that make up, teach, and enforce the social rules that people live by. Or instead, moral agency may arise in the cues of social situations that signal one value over its alternatives (Lindenberg Reference Lindenberg, Steg, van denBerg and de Groot2012). People in economic games, for example, adopt rather different stances toward cooperation versus selfishness simply depending on the name of the game they are incidentally playing – for example, “the community game” versus “the Wall Street game” (Liberman et al. Reference Liberman, Samuels and Ross2004). The simple choice architecture of the game itself can suggest a moral value. In simple dictator games people tend to share the money they have with other people. However, adding the option of taking the other player's money stops that sharing. Instead, the modal choice is for players to take the other player's money. Unprompted generosity turns to unbridled greed (List Reference List2007).
In sum, Talking to Our Selves is a pleasing and thought-provoking tour of the philosophical and psychological issues surrounding agency, self-awareness, and moral choice. It is a delight for those wishing a well-curated tour of past scholarship on these issues, particularly psychological research. It also sets a fine table for a potentially roiling but useful discussion about these issues yet to come. I hope it provokes those discussions. It would be collaboration well worth having.