Does Could Lead to Good? On the Road to Moral Insight

Abstract

Dilemmas featuring competing moral imperatives are prevalent in organizations and are difficult to resolve. Whereas prior research has focused on how individuals adjudicate among these moral imperatives, we study the factors that influence when individuals find solutions that fall outside of the salient options presented. In particular, we study moral insight, or the discovery of solutions, other than selecting one of the competing moral imperatives over another, that honor both competing imperatives or resolve the tension among them. Although individuals intuitively consider the question “What should I do?” when contemplating moral dilemmas, we find that prompting people to consider “What could I do?” helps them generate moral insight. Together, these studies point toward the conditions that enable moral insight and important practical implications.

Full text

rAcademy of Management Journal
2018, Vol. 61, No. 3, 857–895.
https://doi.org/10.5465/amj.2014.0839
DOES “COULD”LEAD TO GOOD? ON THE ROAD TO
MORAL INSIGHT
TING ZHANG
FRANCESCA GINO
JOSHUA D. MARGOLIS
Harvard Business School
Dilemmas featuring competing moral imperatives are prevalent in organizations and
are difficult to resolve. Whereas prior research has focused on how individuals adju-
dicate among these moral imperatives, we study the factors that influence when in-
dividuals find solutions that fall outside of the salient options presented. In particular,
we study moral insight, or the discovery of solutions, other than selecting one of the
competing moral imperatives over another, that honor both competing imperatives or
resolve the tension among them. Although individuals intuitively consider the question
“What should I do?”when contemplating moral dilemmas, we find that prompting
people to consider “What could I do?”helps them generate moral insight. Together,
these studies point toward the conditions that enable moral insight and important
practical implications.
Imagine that you learned some confidential in-
formation from a friend about an impending event
that would put your own company and one of your
clients at great risk of considerable loss (Badaracco &
Useem, 1993). This situation—based on the actual
experience of a financial services employee—presents
two possible courses of action: (1) disclose the infor-
mation to your boss, who cantake action to prevent the
loss, but, in so doing, breach confidentiality and loy-
alty to your friend; or (2) uphold your commitment
to confidentiality and friendship by remaining silent,
thereby risking tremendous damage to your employer
and one of its clients. In this moral dilemma, the com-
peting principles of confidentiality to your friend and
duty to your firm and client are both highly valued,
and choosing one value seems to necessitate forgoing
the other, making the problem difficult to solve with no
obvious “right”answer.
Among the myriad ethical challenges that em-
ployees face, a survey of 162 corporate executives
(M
tenure at firm
58.85 years, SD
tenure at firm
55.73; 28%
female) revealed that these types of situations—
moral dilemmas involving tradeoffs between
competing values or principles (Badaracco &
Useem, 1993; Hogan & Paine, 1997; Palmer, 2012;
Toffler, 1986)—are the most difficult and among
the most prevalent type of ethical challenge at
work (see Appendix A for more details).1Forty-five
percent of executives ranked dilemmas that in-
volve tradeoffs between two moral principles
(“right vs. right”dilemmas) as the most difficult type
of ethical challenge they had experienced. Addition-
ally, the majority of situations (70%) raised by exec-
utives involved moral tradeoffs (i.e., vying principles
that compel an individual to act, such as honesty vs.
duty to the organization or fairness vs. loyalty to a co-
worker). In contrast, only 30% of recalled ethical
challenges featured “right-versus-wrong”situations
that involved temptations to violate moral principles in
order to benefit oneself.
Despite both the reported difficulty and prevalence
of right-versus-right moral dilemmas, a small propor-
tion of the papers (18%) published in management and
psychology journals from 2000 to 2015 has studied
The authors are grateful to Max Bazerman, Adam
Galinsky, Michael Slepian, and Eric Anicich for their in-
sightful feedback and comments on earlier drafts of this
paper. The authors also thank associate editor Dave Mayer
and three reviewers for substantive comments that signif-
icantly shaped this manuscript. We are grateful to mem-
bers of SINGO Lab at Columbia University and NERD Lab
and Non-Lab at Harvard University for their valuable
suggestions and to Nicole Ludmir, Judith Ezike, Katherine
Mentzinger, Elena Helgiu, Arleen Chien, Grace Goodby,
and Nikolos Gurney for their research assistance.
1Executives considered the ethical issues that they—
and the employees they manage—have faced, and then
rated different types of ethical dilemmas based on preva-
lence and difficulty.
857
Copyright of the Academy of Management, all rights reserved. Contents may not be copied, emailed, posted to a listserv, or otherwise transmitted without the copyright holder’s express
written permission. Users may print, download, or email articles for individual use only.

these types of ethical challenges (see Table 1 for details
about the literature review we conducted to identify
this estimate). The majority of papers on ethics (82%)
has focused on right-versus-wrong situations, illumi-
nating how and why people are tempted to cheat and
lie (Chugh & Bazerman, 2007; Moore, Detert, Treviño,
Baker, & Mayer, 2012; Shu, Gino, & Bazerman, 2011) as
well as the myriad organizational sources of mis-
conduct (Brief, Buttram, & Dukerich, 2001; Darley,
1996; Palmer, 2012; Vaughan, 1999).
Existing research on right-versus-right dilemmas
has largely investigatedhow we, as individuals, make
tradeoffs across moral imperatives once we recognize
the principles in conflict (Gilligan, 1982; Greene,
TABLE 1
Research on Ethics Published between 2000 and 2015, Categorized by Dilemma Type
Type of dilemma Publications Total papers Percent of papers
Right vs. wrong Aquino, Freeman, Reed, Lim, & Felps, 2009; Barnes, Schaubroek, Huth, &
Ghumman, 2011; Boles, Croson, & Murnighan, 2000; Cohen, Panter,
Turan, Morse, & Kim, 2014; Cohen & Rozin, 2001; Cullen, Parboteeah, &
Hoegl, 2004; Detert, Treviño, & Schweitzer, 2008; Ding, Wellman, Wang,
Fu, & Lee, 2015; Effron, Lucas, & O’Connor, 2015; Effron, Miller, &
Monin, 2012; Exline, Baumeister, Zell, Kraft, & Witvliet, 2008; Frimer,
Schaefer, & Oakes, 2014; Gillath, Sesko, Shaver, & Chun, 2010; Gino &
Ariely, 2012; Gino, Ayal, & Ariely, 2009; Gino & Desai, 2012; Gino &
Galinsky, 2012; Gino & Margolis, 2011; Gino & Mogilner, 2013; Gino &
Pierce, 2009a; Gino & Pierce, 2009b; Gino, Schweitzer, Mead, & Ariely,
2011; Gino, Shu, & Bazerman, 2010; Gino & Wiltermuth, 2014; Gunia,
Wang, Huang, Wang, & Murnighan, 2012; Hershfield et al., 2012; Keck,
2014; Koning, Steinel, van Beest, & van Dijk, 2011; Kouchaki, 2011;
Kouchaki & Desai, 2015; Kouchaki & Smith, 2013; Kouchaki, Smith-
Crowe, Brief, & Sousa, 2013; Kouchaki & Wareham, 2015; Lee, Talwar,
McCarthy, Ross, Evans, & Arruda, 2014; Miller, Visser, & Staub, 2005;
Monin & Miller, 2001; Moore & Tenbrunsel, 2014; Mulder & Aquino,
2013; Mulder, Jordan, & Rink, 2015; Neville, 2012; Pearsall & Ellis, 2011;
Peer, Acquisti, & Shalvi, 2014; Pierce, Kilduff, Galinsky, & Sivanathan,
2013; Pitesa & Thau, 2013b; Pittarello, Leib, Gordon-Hecker, & Shalvi,
2015; Reynolds & Ceranic, 2007; Reynolds, Dang, Yam, & Leavitt, 2014;
Rixom & Mishra, 2014; Roccas, Klar, & Liviatan, 2006; Ruedy, Moore,
Gino, & Schweitzer, 2013; Schweitzer, Ord ´
oñez, & Douma, 2004; Shalvi,
Dana, Handgraaf, & De Dreu, 2011; Shalvi, Eldar, & Bereby-Meyer, 2012;
Sharma, Mazar, Alter, & Ariely, 2014; Shu & Gino, 2012; Song Hing,
Bobocel, Zanna, & McBride, 2007; Spicer & Bailey, 2007; Spicer, Dunfee,
& Bailey, 2004; Teper, Inzlicht, & Page-Gould, 2011; Umphress,
Bingham, & Mitchell, 2010; Vincent, Emich, & Goncalo, 2013; Wang,
Zhong, & Murnighan, 2014; Welsh, Ellis, Christian, & Mai, 2014; Welsh &
Ord´
oñez, 2014a; Welsh & Ord´
oñez, 2014b; Welsh, Ord´
oñez, Snyder, &
Christian, 2015; Wiltermuth, Bennett, & Pierce, 2013; Winterich, Mittal,
& Morales, 2014; Yam, Chen, & Reynolds, 2014; Yam, Reynolds, & Hirsh,
2014; Yap, Wazlawek, Lucas, Cuddy, & Carney, 2013; Zhang, Cornwell,
& Higgins, 2013; Zhong, 2011; Zhong, Bohns, & Gino, 2010
74 82%
Right vs. right Amit & Greene, 2012; Anteby, 2010; Conway & Gawronksi, 2013; C ˆ
ot´
e, Piff,
& Willer, 2013; de Hooge, Nelissen, Breugelmans, & Zeelenberg, 2011;
Feinberg, Willer, Antonenko, & John, 2012; Lammers & Stapel, 2009; Lee
& Gino, 2015; Levine & Schweitzer, 2015; Moore, Clark, & Kane, 2008;
Reynolds, 2006; Smith, Aquino, Koleva, & Graham, 2014; Swann, W. B.,
Buhrmester, M. D., G´
omez, A., Jetten, J., Bastian, B., et al. 2014; Thau,
Derfler-Rozin, Pitessa, Mitchell, & Pillutla, 2015; Pitesa & Thau, 2013a;
Waldmann & Dieterich, 2007
16 18%
Notes: We searched for empirical articles published in Academy of Management Journal,Administrative Science Quarterly,Journal of
Personality and Social Psychology,Psychological Science,Organizational Behavior and Human Decision Processes,Personnel Psychology,
and Journal of Applied Psychology that contained one or more of the following terms: ethics, ethic, ethical, unethical, ethically, moral, morality,
morals, immoral, amoral, dishonest, honest, deception, dishonesty, honesty, dishonestly, honestly, misconduct, wrongdoing. Papers were
included if the primary decisions or behaviors that were studied involved ethics and were categorized based on whether the majority of studies
in the paper featured right-versus-right or right-versus-wrong situations.
858 JuneAcademy of Management Journal

Morelli, Lowenberg, Nystrom, & Cohen, 2008; Greene,
Sommerville, Nystrom, Darley, & Cohen, 2001; K ohlb erg,
1971; Levine & Schweitzer, 2014). In these dilemmas,
individuals are often caught in a conflict that entails
sacrificing a moral principle (e.g., acting ethically and
fairly, being loyal, and avoiding harm) in order to pro-
tect or uphold duty to another individual (e.g., a co-
worker, supervisee, manager) or entity (e.g., team or
organization). To resolve these dilemmas, individuals
may prioritize one imperative over another, resulting
in deontological or utilitarian choices (Greene et al.,
2008), prosocial lies (Levine & Schweitzer, 2014, 2015;
Lupoli, Jampol, & Oveis, 2017), pro-organizational
unethical behaviors (Umphress et al., 2010), taboo
tradeoffs (Fiske & Tetlock, 1997), and necessary evils
(Margolis & Molinsky, 2008; Molinsky & Margolis,
2005).
Whereas prior research on dilemmas has investi-
gated how individuals choose one moral imperative
over another (Gilligan, 1982; Kohlberg, 1971), our
research focuses on when individuals realize they
need not make these tradeoffs.2We examine instances
of “moral insight,”defined as the discovery of solu-
tions, otherthan selecting one of the competing moral
imperatives over another, that honor both competing
imperatives or resolve the tension among them.
To study moral insight, we integrate research on
insight and creativity (Csikszentmihalyi & Sawyer,
1995; Miron-Spektor, Gino, & Argote, 2011; Schilling,
2005; Smith, 1995), conflict resolution (Bazerman,
Curhan, Moore, & Valley, 2000; Harinck & De Dreu,
2008; Malhotra & Bazerman, 2007), and ethics
(Gilligan, 1982; Greene et al., 2001; Kohlberg, 1971;
Rest, Bebeau, Narvez, & Thoma, 1999). In the domain
of interpersonal conflict, research has shown that
creative thinking (Kurtzberg, 1998) and implicit ne-
gotiation beliefs (Kray & Haselhuhn, 2007) enable
individuals to leverage different prioritization of
values between two negotiators in order to trade value
on those issues and find integrative value (i.e., expand
the pie). However, in intrapersonal moral conflicts,
leveraging differences in priorities between individ-
uals, by definition, is longer a viable option. Finding
integrative value is also particularly difficult when
two competing imperatives are often seen as equally
important or when competing moral imperatives are
difficult to rank in terms of their relative priority.
Because of the unique challenges in resolving
moral dilemmas, we classify moral insight as a dis-
tinct type of integrative problem-solving within the
moral domain. For example, in the dilemma posed at
the beginning of this paper, a moral insight solution
that seeks to honor both values might be to raise
a question to one’s boss that would lead her to in-
vestigate the possibility of this impending event
without breaking confidentiality to the friend. In the
famous Heinz dilemma (Kohlberg, 1971), for exam-
ple, rather than choose between stealing a drug your
spouse needs for survival and obeying laws that
protect property, some people suggest that Heinz
might speak to the druggist who owns the medicine
(Gilligan, 1982)—a solution that seeks to resolve the
tension between the competing moral principles
involved in the dilemma.
Thus, moral insight involves a form of creative
thinking applied to ethical challenges, yet, to date,
research has found an inverse relationship between
creativity and ethicality (Gino & Ariely, 2012; Gino &
Wiltermuth, 2014). Research in the domain of self-
dealing—in which ethical principles are pitted
against the wrong of self-dealing—has revealed that
creativity sparks greater mental flexibility in justi-
fying unethical behavior (Gino & Ariely, 2012; Wang,
2011). Our research explores the possibility that
creative cognition operates differently in the domain
of dilemmas, where individuals face competing
moral imperatives rather than an opportunity to gain
personally at the expense of ethics. We study how
encouraging creative thought in the context of moral
dilemmas—by challenging individuals’default ap-
proach toward them—helps individuals find solu-
tions that do not compromise ethics.
To investigate how individuals might reach moral
insight, we turn to their default approach when
contemplating moral dilemmas. When encountering
difficult ethical challenges, people generally ask
themselves the Socratic question “What should I
do?”(Victor & Cullen, 1988). In a pilot study, we
randomly presented individuals with either moral or
amoral dilemmas and found that most people
thought about what they “should”do for moral sit-
uations (61%), whereas a smaller proportion of
individuals (36%) considered “should”when con-
fronting an amoral dilemma (see Appendix B for
more details about this study). And guidance is typ-
ically cast in terms of “should.”For example, we
created a data set with the ethics codes of Fortune 50
companies sampled in 2013 and counted the number
2Underlying the study of these ethical dilemmas is
a long philosophical debate on whether dilemmas actually
exist (Conee, 1982; Marcus, 1980). Whether or not people
are facing actual dilemmas, we are primarily interested in
situations in which individuals perceive the existence of
competing moral imperatives.
2018 859Zhang, Gino, and Margolis

of times the word “should”appeared in each ethics
code. We found that “should”appeared approxi-
mately 30 times on average. Organizations, it seems,
also frame the principles to guide managerial con-
duct in terms of “should.”
Taken together, these findings demonstrate that
should mindsets appear to be the unique default and
recommended approach to contemplating moral di-
lemmas. A “mindset”is commonly defined as cogni-
tions that, once activated, persist over various contexts
(Luchins, 1942; Luchins & Luchins, 1959) and have
the potential to influence later choices on a variety of
tasks (see, e.g., Galinsky, Gruenfeld, & Magee, 2003, for
a discussion of a power mindset). Despite the perva-
siveness of having a should mindset when confronting
moral dilemmas, we challenge this default approach
and suggest that it curbs the generation of moral in-
sight. Whereas a should mindset may encourage in-
dividuals to think analytically in weighing the vying
moral claims of the most apparent courses of action,
people may often benefit from a could mindset that
involves a more expansive exploration of possible
solutions before making a final decision. We propose
that considering what one could do shifts people from
analyzing and weighing what they assume to be fixed
and mutually exclusive alternatives to generating
options that might reconcile underlying imperatives.
When facing ethical dilemmas, shifting individuals
from a more conventional should mindset to a less
conventional could mindset encourages greater ex-
ploration of possibilities, increasing individuals’
ability to discover practical solutions to moral di-
lemmas that move beyond conceding one or more
moral principles to meet another.
The present research makes three main contribu-
tions to the literature. First, it extends how ethical di-
lemmas have traditionally been studied. Prior research
on dilemmas has presented individuals with a choice
between two possible decisions, showcasing the ten-
sion between imperatives—at least one of which is
moral—and revealing the factors that lead individuals
to choose one value or principle over the other (Jones,
1991; Kohlberg, 1971). Rather than investigate which
of two predetermined options is chosen and why, we
examine how novel options that satisfy both impera-
tives might be generated. Additionally, we not only
investigate individuals’preferences of what they
would do in hypothetical dilemmas—as studies on
right-versus-right dilemmas have often done—but also
push the study of moral dilemmas to examine behav-
iors, investigating which actions, out of the multiple
possibilities available, individuals generate to resolve
the moral dilemmas presented.
Second, we integrate research from ethics, negotia-
tion, and creativity to study moral insight, or uncon-
ventional solutions that honor or resolve competing
moral imperatives. Thus far, research linking creativity
and ethics has focused on the domain of self-dealing,
showing that creativity increases the likelihood that
individuals will engage in unethical behavior (Gino &
Ariely, 2012; Wang, 2011). When individuals are
tempted to cheat for personal gain, creativity generates
greater mental flexibility in service of justifying un-
ethical actions. But, creativity may operate differently
in the domain of dilemmas, where individuals face
competing moral imperatives rather than an opportu-
nity to gain personally at the expense of ethics. We
suggest that, for moral dilemmas, creative thinking—
prompted through having a could mindset—helps in-
dividuals engage in divergent thinking, which may
ultimately help them resolve the collision of ethical
imperatives.
Third, we offer a simple-to-implement psychological
intervention that helps individuals address moral di-
lemmas more constructively. Among the array of pos-
sible interventions, ranging from psychological changes
to redesigning organizations (Zhang, Gino, & Bazerman,
2014), our proposed intervention is based on a subtle
psychological shift in an individual’s mindset that
reframes how they can resolve ethical dilemmas. Based
on research in negotiations showing that an in-
dividual’s mindset with respect to interpersonal con-
flict is malleable and impacts their ability to reach
integrative solutions (e.g., Kray & Haselhuhn, 2007), we
seek to understand how shifting individuals’mindsets
in intrapersonal moral conflicts impacts their approach
to these dilemmas—from a contest between two alter-
natives to an open-ended quest for possibilities. In
particular, we study how reframing an individual’s
approach to competing moral imperatives—from ask-
ing “What should Ido?”to “What could Ido?”—
prompts a shift from choosing between these competing
imperatives to generating insight that honors both im-
peratives or resolves tension between them.
THEORY AND HYPOTHESES
To study how individuals face these dilemmas,
organizational and psychological research has inves-
tigated the factors that influence how individuals (a)
recognize the moral issues at stake (i.e., reach moral
awareness; Jennings, Mitchell, & Hannah, 2015;
Reynolds, 2008; Sparks & Hunt, 1998; Tenbrunsel &
Smith-Crowe, 2008), and (b) choose among the com-
peting moral imperatives in conflict. In particular,
research from psychology on moral decision-making
860 JuneAcademy of Management Journal

has shown that there are five primary dimensions on
which individuals recognize that their behaviors have
moral implications: harm, fairness, loyalty, authority,
and purity (Graham, Nosek, Haidt, Iyer, Koleva, &
Ditto, 2011). While individuals vary on the extent to
which they are morally attentive (i.e., chronically
perceptive of the moral aspects of situations;
Reynolds, 2008), environmental factors also affect
moral awareness. For instance, approaching a situa-
tion with a legal or economic lens mitigates the extent
to which individuals recognize that their behaviors
have ethicalimplications(Messick, 1999; Tenbrunsel
& Smith-Crowe, 2008).
In addition to studying individuals’awareness and
judgments of the moral implications in moral dilemmas
(Haidt, 2001; Monin, Pizarro, & Beer, 2007; Uhlmann,
Pizarro, Tannenbaum, & Ditto, 2009), research from
psychology on moral decision-making has examined
how individuals adjudicate among competing imper-
atives (Greene et al., 2004; Kohlberg, 1971). Individuals
broadly approach dilemmas with either a utilitarian
(based on consequences of actions) or deontological
approach (based on adherence to rules, principles, and
duty) (Conway & Gawronski, 2013; Greene et al., 2001;
Reynolds, 2006). For example, in one commonly used
dilemma, referred to as “the trolley problem,”in-
dividuals are asked whether they should actively divert
a runaway trolley onto a path that would kill one person
in order to save five lives (the utilitarian choice), or
whether they should choose inaction, spare one life,
and leave the train on course to kill five people (the
deontological choice) (Foot, 1967; Thomson, 1986).
Examining which options individuals select has pro-
vided insight into moral cognition and the conditions
under which deliberation or intuition are respectively
more likely to guide moral decisions. For example, re-
searchers have learned that emotions and intuition often
influence ethical judgmentsanddecisions(Ditto,
Pizarro, & Tannenbaum, 2009; Greene, Nystrom,
Engell, Darley, & Cohen, 2004; Haidt, 2001; Lee &
Gino, 2015; Monin et al., 2007; Uhlmann et al., 2009),
which had long been assumed in psychology to be
a product of logical and deliberative reasoning. Moral
intuition (or “gut reactions”to moral dilemmas) typi-
cally leads to deontological decisions (i.e., “killing is
wrong, regardless of the consequences”), whereas de-
liberation may lead individuals to override these in-
tuitive responses to make utilitarian decisions (i.e.,
killing one life is permissible in order to save the lives of
many; Greene et al., 2008). Which form of cognition is
best suited to a particular ethical challenge has also been
subject to debate in the literature (Greene, 2013; Haidt,
2001; Monin et al., 2007; Zhong, 2011).
In addition to investigating individuals’moral in-
tuitions and reasoning when facing tradeoffs in di-
lemmas, researchers have investigated how individuals
form moral judgments of others facing these decisions.
In general, individuals perceive choices driven by em-
pathic concern as more valued (Conway & Gawronski,
2013; Everett, Pizarro, & Crockett, 2016; Levine &
Schweitzer, 2015). Levine and Schweitzer (2015) found
cases in which sacrificing certain moral principles
(e.g., being honest) is desirable: when individuals tell
prosocial lies out of empathic concern to protect others
from unnecessary harm. Additionally, individuals are
more valued as social partners and perceived as more
trustworthy when they make deontological choices—
related to empathic concern (Conway & Gawronski,
2013)—over utilitarian options (Everett et al., 2016).
Whereas empirical studies on dilemmas to date
typically involve forced tradeoffs between moral
imperatives, our research complements these ap-
proaches by focusing on the courses of action in-
dividuals develop when they are not forced to decide
between imperatives, but, rather, when they might
formulate solutions. By relaxing constraints on
which courses of action individuals can choose, we
provide individuals with the opportunity to reach
moral insight, moving beyond the two most apparent
courses of action colliding in a dilemma to honor
both competing imperatives or resolve the tension
among them. To reach moral insight, moral aware-
ness (Reynolds, 2006, 2008) is a necessary, but in-
sufficient, condition: individuals must be aware of
the ethical considerations in these dilemmas, but
then they must go on to find creative solutions that
honor those ethical considerations.
Developing Moral Insight
We suggest that, when facing moral dilemmas,
people generate moral insight in the same way they
generate insight when facing creativity problems
(Isen, Daubman, & Nowicki, 1987). Moral insight is
enacted through the generation of novel and practi-
cal solutions that move beyond conventional re-
sponses requiring the sacrifice of one or more moral
principles to meet another. Research on insight
problems has indicated that the cognitive processes
that lead to solutions neither occur to people im-
mediately on presentation of the problem, nor
emerge from analysis and deliberation. Rather, the
insight strikes like a bolt—the proverbial “Eureka!”
moment—changing the solver’s mental representa-
tion of the problem (Schilling, 2005; Seifert, Meyer,
Davidson, Patalano, & Yaniv, 1994; Sternberg, 1988).
2018 861Zhang, Gino, and Margolis

In the context of moral dilemmas, we argue that
generating moral insight involves shifting individuals’
mental representation of dilemmas away from making
a choice between forced tradeoffs. Research on insight
has focused on solutions that require individuals to
relax the assumptions of a given problem. For example,
in the classic Duncker’s (1945) “candle problem,”in-
dividuals are asked to fix a lit candle to a wall using
a box of matches and a box of thumbtacks. The solution
is not to tack a lit candle to the wall directly, as most
individuals initially attempt, but to tack an emptied
thumbtack box to the wall instead, allowing the box to
serve as a stand for the lit candle. Just as insight in
problems like Duncker’s (1945) candle entails relaxing
the assumption that the functions of objects are fixed,
moral insight entails relaxing assumptions that there
are fixed options from which to choose. In contrast to
research on non-moral insight (Duncker, 1945) and
creativity (Mednick, 1962; Rowe,Hirsh,&Anderson,
2007), which typically examines convergence around
a single but non-obvious relationship between objects
and concepts, central to our definition of moral insight
is the possibility that multiple solutions exist beyond
selecting one side of the dilemma or the other. These
solutions can vary both in their feasibility and how
they meet those multiple competing imperatives.
However, development of one or more of the solutions
reflects the individual’s moral insight to move beyond
simply determining which imperative to prioritize at
the expense of another.
Shifting to a Could Mindset
Existing research has shown the importance of
matching cognitive processes to the nature of the
ethical challenge faced. For example, extended
contemplation has been shown to increase the like-
lihood that individuals will make ethical decisions
in both individual and group contexts, particularly
when individuals lack justifications for their un-
ethical behaviors (Caruso & Gino, 2010; Gunia et al.,
2012; Shalvi et al., 2012). However, in other condi-
tions, most notably when another’s welfare is pitted
against an actor’s own interests, quick intuitive re-
sponses increase the likelihood of an ethical choice
(Rand, Greene, & Nowak, 2012; Zhong, 2011). In
sum, interventions that tailor cognition to the nature
of the ethical challenge are essential (Bennis, Medin,
& Bartels, 2010; Moore & Tenbrunsel, 2014).
In the context of moral dilemmas, individuals
generally approach these problems by contemplat-
ing what they should do (see Appendix B). By in-
voking a should mindset, individuals may conjure
their “ought”selves, representing a mental model of
their duties, obligations, and responsibilities (Higgins,
1987; Rogers, 1961). In the context of moral dilemmas,
individuals adopting a should mindset may focus on
the relative importance of different injunctive norms
(i.e., based on what individuals believe others find
permissible [Cialdini, Kallgren, & Reno, 1991; Elqayam,
Thompson, Wilkinson, Evans, & Over, 2015]), under-
lying the tension in these moral dilemmas.
Consequently, adopting a should mindset implies
forcing individuals to narrowly focus on weighing
and choosing among two possible courses of action
(De Dreu, Giacomantonio, Shalvi, & Sligte, 2009;
Kurtzberg, 2009; Trope & Liberman, 2010). For ex-
ample, when contemplating the Heinz dilemma with
a should mindset, individuals may immediately start
weighing the moral costs of stealing against the moral
costs of letting a loved one die. Furthermore, Higgins
and colleagues have found that, when individuals
think about “should”and “oughts,”they typically
adopt more of a prevention-oriented mindset, lead-
ing them to consider the avoidance of negative out-
comes (Higgins, 1997; Higgins, Roney, Crowe, &
Hymes, 1994). Taken together, adopting a should
mindset to resolve dilemmas may lead individuals to
satisfice in settling for the least undesirable solution
that meets one ethical priority while providing jus-
tifications for failing to honor the other value.
In contrast, shifting individuals’contemplation to-
ward what they could do—a mindset that recognizes
other possibilities might exist—may enable in-
dividuals to gain psychological distance from the focal
tension. By seeing the problem from a more distant
perspective, individuals with a could mindset may
realize that the tradeoff between saving a life and not
stealing is not necessarily irreconcilable, increasing
the likelihood that individuals might arrive at insight
solutions (e.g., solicit donations through social media).
Langer and Piper (1987) found that considering what
objects could be, as opposed to what objects were,
helped individuals transcend the problem of “func-
tional fixedness,”or the inability to use objects beyond
the purposes for which they were originally designed
(Isen et al., 1987; Luchins, 1942). When confronted
with the need to erase a pencil mark without using an
eraser, individuals who merely considered what ob-
jects could be were more likely to recognize that
a rubber band could be used in lieu of an eraser,
compared to those who considered what these objects
were. Just as thinking about what objects could be
influenced individuals to perceive the problem from
a distance and overcome the rigidity of considering
only conventional uses of objects, we propose that
862 JuneAcademy of Management Journal

contemplating what one could do in moral dilemmas
helps individuals think beyond the rigidity of making
forced tradeoffs between moral principles and reach
solutions that honor both imperatives featured (see
Figure 1 for conceptual model).
Hypothesis 1. A could mindset relative to a should
mindset increases the propensity to reach moral
insight.
Research on decision-making has shown that, when
individuals shift from a narrow to a broad decision
frame, they consider multiple objectives, alternatives,
and outcomes (Larrick, 2009). By unlocking the recog-
nition that multiple possibilities exist, developing
a could mindset may also unlock individuals’ability to
engage in “divergent thinking,”or the development of
multiple solutions that span boundaries (McCrae, 1987;
Runco, 1991; Silvia et al., 2008). In Langer and Piper’s
(1987) study, those who considered what the objects
could be were able to consider a greater variety of al-
ternative uses for the focal object. Although interlinked,
there is a conceptual difference between having a could
mindset and divergent thinking: whereas a could mind-
set involves the awareness that multiple solutions might
exist, divergent thinking is the ability of individuals
to consider multiple approaches to a problem.
Because divergent thinking involves thinking “with-
out boundaries”or “outside the box”(Thompson, 2008:
226), it leads to the discovery of insightful solutions
(Baer, 1994). By considering multiple possibilities, di-
vergent thinking helps individuals make new connec-
tions and associations (Guilford, 1968, 1982), reducing
individuals’propensity to settle upon obvious answers
and increasing their ability to reach innovative solu-
tions to problems (Williams, 2004; Woodman, Sawyer,
& Griffin, 1993).
In research on interpersonal conflict, thinking cre-
atively beyond conventional options has been found
to generate integrative solutions (De Dreu et al., 2009;
Kurtzberg, 1998), so we have adapted principles from
negotiation to study the intrapersonal conflict that
individuals experience when confronted with ethical
dilemmas. When negotiators perceive their environ-
ment as competitive, they often assume that goals
across negotiating parties are negatively related, im-
plying that they must make distributive tradeoffs in
order to find solutions (Carnevale & Probst, 1998;
Tjosvold, 1998). This competitive mindset often leads
individuals to reach purely distributive solutions that
assume the size of the economic pie as given (Baron,
Bazerman, & Shonk, 2006; Demoulin & Teixeira,
2010; Fisher, Ury, & Patton, 2011; Malhotra &
Bazerman, 2007). In contrast, negotiators who
realize the potential to reconcile competing sides are
more likely to discover integrative solutions that ex-
pand the size of the overall pie and often maximize
outcomes for both negotiating parties (Brandenburger
& Nalebuff, 1996; De Dreu, 2003; Harinck & De Dreu,
2008). Taken together, we propose that applying
a could mindset toward ethical dilemmas prompts
people to develop a wider set of possible options,
enhancing individuals’ability to look beyond forced
tradeoffs and ultimately helping them formulate
moral insight solutions that uphold colliding moral
imperatives. We thus hypothesize:
Hypothesis 2. A could mindset increases divergent
thinking—or the formulation of multiple possibilities—
in moral dilemmas relative to a should mindset.
Hypothesis 3. The effect of a could mindset on the pro-
pensity to reach moral insightis mediated by the extent
to which individuals engage in divergent thinking.
While divergent thinking may often lead to moral in-
sight, we note that divergent thinking and moral insight
are distinct constructs. Not all possibilities generated
from divergent thinking are necessarily moral insights.
For example, ideas generated from greater divergent
thinking might entail creative justifications for selecting
one moral imperative over another, selecting an un-
ethical course of action, or even abstaining from taking
action. More concretely, someone who engages in di-
vergent thinking in the Heinz dilemma could consider
multiple different ways of stealing the drug (e.g., hiring
someonetostealthedrug,orgoingintothedrugstoreto
buy a cheap drug and then swapping out the cheap one
for the expensive drug). Among the many possible so-
lutions, none of those possibilities would be considered
moral insight, which we define as solutions that (a) move
beyond the two salient options presented, and, critically,
(b) resolve or honor multiple competing imperatives.
Overview of the Present Research
We test our predictions regarding the impact of
a could mindset on generating moral insight across
four experiments. Despite the default approach of in-
dividuals to contemplate moral dilemmas with a
should mindset, we find, in two studies, that con-
templating “What could I do?”leads individuals to
engage in more divergent thinking as they explore
possible options, helping them to find more moral
insight solutions across both hypothetical contexts
(Study 1) and an incentive-compatible context (Study
2). We then explore the impact of a could mindset in
two other settings of ethical challenges common for
managers. Study 3 shows that adopting a could
2018 863Zhang, Gino, and Margolis

mindset in interpersonal contexts increases collective
divergent thinking, helping individuals generate moral
insight. Finally, Study 4 examines boundary condi-
tions based on the type of ethical dilemma featured,
investigating how a could mindset fares when there
exists a temptation to cheat for self-gain.
STUDY 1: CONTEMPLATING
MORAL DILEMMAS
In Study 1, participants considered a series of di-
lemmas and indicated either what they “could”or
“should”do in response before finally reporting
what they “would”do. We hypothesized that, rela-
tive to a should mindset, a could mindset would in-
fluence individuals to engage in divergent thinking
(Hypothesis 2), better equipping individuals to gen-
erate solutions that concede neither imperative em-
bedded in the dilemma (Hypotheses 1 and 3).3
Participants
Two hundred and ten individuals (M
age
535.02
years; 59.2% female) recruited through Amazon’s
Mechanical Turk participated in an online study in
exchange for $1.50.4Five participants were not eli-
gible to complete the study because they incorrectly
answered the attention check,5leaving 205 partici-
pants included in the analysis.
Design and Procedure
Participants were randomly assigned to adopt ei-
ther a could mindset or a should mindset while
contemplating four ethical dilemmas (see Dilemmas
1–4 in Appendix D). Those randomly assigned to
adopt a could mindset provided written responses to
the question “What could you do?”for each of the
four dilemmas, whereas those assigned to think in
a should mindset provided written responses to the
question “What should you do?”
All participants then answered the question
“What would you do?”as well as follow-up ques-
tions about their responses to each dilemma. Two
independent coders blind to the study’s hypotheses
rated each of the participants’written responses for
their divergent thinking and whether the proposed
solutions were considered moral insights that satis-
fied the competing values in each of the dilemmas.
Measures
Manipulation check. As a manipulation check,
two independent coders blind to the hypotheses of
the study recorded the number of instances in which
participants used the words “could”and “should”in
FIGURE 1
Conceptual Process Model of the Factors that Generate Moral Insight
Divergent thinking
Could vs. should
mindset
Moral insight
solution
H2 H3
H1
+
+
+
3In a pilot study (see Appendix C), we compared
adopting a could mindset against should and would
mindsets. Individuals in a could mindset perceived moral
imperatives featured across ethical dilemmas as more
compatible relative to those in a should or a would mind-
set. However, we did not observe differences between
should and would mindsets, providing further evidence
that individuals approach moral dilemmas with a should
mindset as their default. That is, differences between
adopting should and could mindsets were not due to
should mindsets reducing perceptions of moral impera-
tives as compatible, but, rather, could mindsets increasing
perceptions of compatibility. Consequently, we did not
include a would condition in Studies 1–4.
4This amount was considered a standard market rate
at the time the study was conducted. Past research has
shown that the Mechanical Turk service provides reliable
data for research purposes (Buhrmester et al. 2011).
5Participants were presented with a series of pictures
and instructed to select the last option.
864 JuneAcademy of Management Journal

their written responses to the question “What [should/
could]Ido?”across the four dilemmas. Because we
achieved high inter-rater reliability, we averaged the
ratings from the two raters for the number of times
participants mentioned “could”(ICC
3
5.93, p,.001)
and “should”(ICC
3
5.94, p,.001).
Divergent thinking. To measure divergent think-
ing, we used two different measures based on prior
research (Guilford, 1967). First, participants self-
reported the number of solutions they considered as
they answered the question “What [should/could] I
do?”Second, two independent coders rated the ex-
tent to which participants’solutions were “outside of
the box”and “spanned different categories of solu-
tions”(1 5not at all,45somewhat,75extremely)
based on Amabile’s (1996) consensual assessment
technique. We aggregated the two rater’s responses
into a single measure since the two raters achieved
high reliability (ICC
3
5.89, p,.001). Below, we
present results of both self-reports and external raters
separately. We also averaged the standardized scores of
both internal and external ratings of divergent thinking
as they were highly correlated (ICC
1
5.75, p,.001).
Moral insight. We triangulated on moral insight in
two ways: using one measure that is more concrete
and another that is more abstract. For the concrete
measure of moral insight, two coders categorized re-
sponses to the question “What would I do?”for each
dilemma as either one of the conventional solutions
(e.g., “steal the money”or “do not steal”in the Heinz
dilemma) or one of the set of solutions that two sep-
arate coders had identified as seeking to address both
imperatives. For example, solutions in response to the
Heinz dilemma such as “bring the story to the local
media”and “start a charitable foundation for my
spouse”were considered to represent moral insight.
Because we obtained high reliability between raters
across the four dilemmas, we averaged these ratings
into a single measure for each dilemma and summed
these scores across all four dilemmas to obtain a single
measure of moral insight (ICC
3
5.86, p,.001).
For our abstract measure of moral insight, two
coders determined whether each solution generated
met the primary moral imperatives featured in each
dilemma. Because we obtained high reliability be-
tween raters across the four dilemmas, we averaged
these ratings into a single measure of moral insight for
each dilemma and summed these scores across all
four dilemmas (ICC
3
5.82, p,.001).
Below, we present these measures separately. Be-
cause our concrete and abstract measures of moral in-
sight were correlated (ICC
3
5.90, p,.001), we also
averaged the two standardized abstract and concrete
measures to obtain a single measure of moral insight that
we used in our mediation analysis.
Results
Table 2 presents the means, standard deviations,
and correlations of the variables we measured. Ap-
pendix F reports all additional variables collected
and analyses conducted for this study and all sub-
sequent studies in this paper.
Manipulation check. We found that those in the
could mindset used the word “could”(M5.90, SD 5
1.02) more times in their responses than those in
a should mindset (M5.24, SD 5.45), t(203) 56.16, p,
.001, d5.86, whereas those in a should mindset wrote
more responses containing the word “should”(M5
.29, SD 5.67) than did those in a could mindset (M5
.12, SD 5.36), t(203) 52.32, p5.02, d5.33, sug-
gesting that our manipulation was indeed effective.
Divergent thinking. We conducted a repeated mea-
sures analysis of variance (ANOVA) with participant
mindset (should vs. could) as the between-subjects
factor, the dilemma as the within-subjects factor, and the
number of solutions participants reported considering
for each dilemma as the dependent variable. Across the
four dilemmas, participants reported having considered
more solutions in the could mindset (M55.11, SD 5
1.68) than in the should mindset (M54.41, SD 51.07),
F(1, 194) 510.70, p,.001, h
p2
5.05. The interaction
between the dilemma type and mindset was also sig-
nificant, F(3, 582) 56.53, p,.001, h
p2
5.03.6
We conducted a similar analysis using the coders’
ratings of how participants’solutions spanned
boundaries across the four dilemmas and found that
having a could mindset increased divergent thinking
(M53.19, SD 51.22) relative to having a should
mindset (M52.42, SD 51.01), F(1, 202) 522.99, p,
.001, h
p2
5.10. We did not find a significant in-
teraction between the dilemma type and mindset,
F(3, 606) 52.08, p5.11, h
p2
5.01.
6Pairwise comparisons reveal that those in a could mind-
set reported more ideas generated in Dilemma 1 (M
could
5
1.69, SD
could
51.04, M
should
51.31, SD
should
5.74), p
1
5.004
and Dilemma 2 (M
could
51.28, SD
could
5.61, M
should
51.06,
SD
should
5.24), p
2
5.001. There did not appear to be a dif-
ference in the self-reported number of solutions generated
for Dilemma 3 (M
could
51.07, SD
could
5.26, M
should
51.08,
SD
should
5.30), p
3
5.83, and Dilemma 4 (M
could
51.09,
SD
could
5.29, M
should
51.05, SD
should
5.22), p
4
5.25.
Participants may have generated fewer ideas for Dilemmas
3 and 4 as they were toward the end of the experiment.
2018 865Zhang, Gino, and Margolis

Moral insight. Using our concrete measure of
moral insight, we found that a could mindset (M5
1.59, SD 51.12) generated more moral insight across
the four ethical dilemmas relative to a should
mindset (M51.17, SD 5.96), F(1, 196) 58.99, p5
.003, h
p2
5.04. We did not find a significant in-
teraction between the dilemma type and mindset,
F(3, 588) 5.89, p5.45, h
p2
5.005.
Using our abstract measure of moral insight, we
found that a could mindset (M5.97, SD 5.89) gen-
erated more moral insight across the four ethical di-
lemmas relative to a should mindset (M5.68, SD 5
.75), F(1, 196) 57.05, p5.009, h
p2
5.04. We did not
find a significant interaction between the dilemma
type and mindset, F(3, 588) 51.02, p5.38, h
p2
5.005.
Mediation analysis. We examined whether di-
vergent thinking—based on our triangulated measure
from internal and external ratings—mediated the effect of
a could mindset on participants’ability to generate moral
insight solutions, based on our triangulated measure of
moral insight (Baron & Kenny, 1986). A could mindset
was positively associated with divergent thinking (b5
.32, t54.80, p5.001; see Table 3). When controlling for
divergent thinking, the effect of adopting a could mindset
was reduced to non-significance (from b5.20, t52.86,
p5.005 to b5.001, t5.005, p..99), and divergent
thinking predicted participants’ability to generate moral
insight solutions (b5.61, t510.46, p,.001). A boot-
strap analysis indicated that the 95% bias-corrected
confidence interval for the size of the indirect effect ex-
cluded zero [.12, .28], suggesting a significant indirect
effect (MacKinnon, Fairchild, & Fritz, 2007). Analyses
with either external or internal ratings of divergent
thinking produced similar results (see Appendix F).
Discussion
When asked what they “would”do, could thinkers
were better able to reach moral insight by generating
solutions that did not simply select one side of a
dilemma at the expense of the other, supporting Hy-
pothesis 1. We also found evidence in support of Hy-
potheses 2 and 3, which predicted that, relative to those
in a should mindset, could thinkers would be more likely
to engage in divergent thinking, which would in part
explain individuals’propensity to reach moral insight.
STUDY 2: “COULD”MINDSET INCREASES
MORAL INSIGHT
Study 1 provided evidence consistent with our pre-
dictions. However, the prior studies relied on hypo-
thetical scenarios rather than real dilemmas faced by
participants. To address this issue, in Study 2, full-time
employees wrote speeches announcing their decision
about an ethical dilemma adapted from a real ethical
challenge that a former executive faced. Employees
were incentivized such that the top 10% of solutions
featured in these speeches would be awarded a mone-
tary prize, as rated by a panel of independent judges.
We predicted that, even when incentivized to reach
a higher-quality solution, individuals adopting a could
mindset would be more likely to reach moral insight
than those adopting a should mindset.
Participants
Three hundred and ten full-time employees(M
age
5
35.81 years; 49% female) recruited from an online
Qualtrics panel completed this study.
Design and Procedure
Whereas we manipulated participants’mindsets in
the previous two studies by directly asking them what
they could or should do after reading each dilemma, in
this study, we sought to instill a could or should mindset
in an unrelated context prior to experiencing a moral
dilemma. Participants were informed that they would
TABLE 2
Means, Standard Deviations, and Correlations for the Variables Measured (Study 1)
Variable Mean SD 123 45
1. Response contains “could”0.56 0.84
2. Response contains “should”0.21 0.55 20.02
3. Divergent thinking: self-reported total
number of solutions considered
4.75 1.44 0.52*** 0.02
4. Divergent thinking: based on ratings
from two independent coders
2.76 1.19 0.41*** 0.04 0.60***
5. Moral insight solutions (concrete) 0.82 0.83 .23*** 0.02 0.36*** 0.63***
6. Moral insight solutions (abstract) 1.37 1.06 .28*** 0.001 0.39*** 0.70*** 0.80**
**p,.01
***p,.001
866 JuneAcademy of Management Journal

complete a series of different tasks and that the first task
would entail evaluating a video for clarity of content.
Participants were randomly assigned to watch one of
two videos that we created to instill either a could or
should mindset (see Appendix E for a transcript of the
video). In these videos, participants learned about one
way to approach solving a difficult ethical dilemma—by
asking oneself either “What could I do?”or “What
should I do?”After watching the video, participants
briefly summarized the main message of the video and
provided comments about the clarity of its content.
In the second part of the study, participants read
about their new role as the president of a non-profit
organization that “advocates for the elimination of
child labor in Southeast Asia and aims to increase
efforts to keep these children in school.”As presi-
dent, they faced a dilemma that featured a tension
between upholding the organization’s mission to
reduce child labor practices and obtaining funding
from a source known to violate child labor laws.
In recent years, you have struggled to get funding as
political governments have diverted resources to en-
vironmental concerns and natural disasters taking
place in the region. ... You need $100,000 to keep
your organization open in the next week or you risk
having to shut down your operations.
Yesterday, TechGen, a multibillion-dollar company that
produces technology products, approached you and
offered to donate $200,000 to your organization. This
donation would not only help you reduce your debt, but
also infuse your organization with cash, allowing you to
start additional projects that would help keep children
in school. Initially, you were excited about this dona-
tion, until you learned about TechGen’s unethical
practices of using child labor in other countries. You
suspect that this company has approached you to make
a donation to reduce any negative opposition to this
company opening factories in the Southeast Asia region.
Based on this information, participants wrote a
speech announcing their decision to the public. The top
10% of solutions—as rated by other participants—
received a $5 bonus. Participants then answered a series
of questions measuring divergent thinking as well as
their mindset as they wrote the speech (a measure we
used as our manipulation check).
A separate group of 258 individuals (M
age
537.69
years; 52% female) from Amazon’s Mechanical Turk
rated five randomly selected speeches such that an
average of 4.19 judges (SD 5.89) evaluated each
speech. For each speech, judges blind to the hy-
potheses of the study rated whether the solutions
were considered moral insights.
In Study 1, it was possible that adopting a could
mindset might have encouraged participants to meet
the primary moral imperatives through unethical
means (e.g., evading taxes could be one way to avoid
stealing from the druggist while also saving the
spouse’s life). To understand whether adopting
a could mindset might encourage justifications of
unethical behavior, judges rated the ethicality and
overall quality of the solutions provided.
Measures
Manipulation check. At the end of the study,
participants rated the extent to which they consid-
ered what they “could”and “should”do while they
were facing the dilemma (1 5not at all to 7 5very
much).
Divergent thinking. After writing their speeches,
participants rated the degree to which they engaged in
divergent thinking based on their responses to the
following three items: “While writing your speech, to
what extent did you try to open the set of possibilities,”
“explore alternatives,”and “explore different possi-
bilities before narrowing on your approach?”(a5.83).
TABLE 3
Mediation Analysis on Moral Insight Solutions Formulated (Study 1)
Variable
Divergent thinking Moral insight solutions Moral insight solutions
X→MX→YX,M→Y
Could mindset 0.32*** 0.20** 0.001
Divergent thinking 0.61***
Adjusted R
2
0.10 0.03 0.37
95% bias-corrected CI [0.12, 0.28]
Notes:CI5standardized confidence interval for the indirect effect. The table reports standardized coefficients for each regression.
*p,.05
**p,.01
***p,.001
2018 867Zhang, Gino, and Margolis

Moral insight. As in Study 1, we triangulated on
moral insight using both concrete and abstract
measures. To measure moral insight more con-
cretely, judges categorized the employees’solu-
tions as either one of the conventional solutions
(i.e., accept the donation or decline the donation) or
a solution that incorporated any one of the follow-
ing elements enabling the organization to remain
viable without forgoing its value to eliminate child
labor: accept the money and work with the donor to
remove child labor from its practices, accept the
money contingent on the donor changing its prac-
tices, and directly asking the audience for funding.
Based on a pilot test,research assistants had identified
these responses as solutions that sought to address
both imperatives.
To measure moral insight more abstractly, judges
explicitly rated (a) whether the solutions fully hon-
ored both of the following imperatives, and (b) the
extent to which the solutions met each of these ob-
jectives: “allowing the organization to continue op-
erating”and “upholding [its] value of eliminating
child labor”(1 5not at all to 7 5a great deal). Across
these items, judges achieved high inter-rater re-
liability (ICC
1
..76, ps,.001).
Because the correlation between the abstract and
concrete measures of moral insight was moderately
high (ICC
1
5.73, p,.001), we triangulated on the
construct of moral insight by averaging the stan-
dardized scores of these measures.
Ethicality. Raters also determined the extent to
which the solution was ethical (1 5not at all to 7 5
extremely).
Decision quality. Raters also evaluated solutions
based on how good the solutions were to determine
the speech writers who would receive a monetary
prize (1 5not at all to 7 5extremely).
Results
Table 4 presents the means and standard de-
viations of the variables we measured.
Manipulation check. As expected, those adopting
the could mindset (M55.69, SD 51.43) considered
what they could do more than those in the should
mindset condition (M55.02, SD 51.64), t(307) 5
3.82, p,.001, d5.44. Similarly, those in the should
mindset condition (M55.75, SD 51.57) considered
what they should do more than those in the could
mindset condition (M55.38, SD 51.49), t(307) 5
2.14, p5.03, d5.24.
A repeated measures ANOVA with mindset as the
between-subjects factor and responses on the two
manipulation check questions as the within-subjects
factors revealed a significant interaction, F(1, 307) 5
25.09, p,.001, h
p2
5.08. Those in the could mindset
thought about “could”more than “should,”t(148) 5
2.37, p5.02, d5.27, whereas those in the should
mindset thought about “could”less than “should,”
t(159) 54.58, p,.001, d5.52.
Divergent thinking. In support of Hypothesis 2,
employees in the could mindset condition (M54.98,
SD 51.51) engaged in more divergent thinking rel-
ative to those in the should mindset condition (M5
4.57, SD 51.41), t(307) 52.52, p5.01, d5.29.
Conventional solutions. Those in a should mind-
set (M5.48, SD 5.41) were more likely to decline the
money than those in the could mindset (M5.32,
SD 5.41), t(307) 53.53, p,.001, d5.40. Employees
were no more likely to accept the money in the could
condition (M5.12, SD 5.25) than in the should
condition (M5.10, SD 5.23), t(307) 5.53, p5.53,
d5.06.
Moral insight. Based on the concrete measure of
moral insight, employees in the could mindset con-
dition were more likely to provide a solution re-
search assistants had identified as meeting both
imperatives (M5.47, SD 5.42) than those in the should
mindset condition (M5.29, SD 5.37), t(307) 53.41,
p,.001, d5.44. We provide specific types of solutions
that were classified as moral insights in Table 4. Based
on the abstract measure of moral insight, judges
were more likely to classify the solutions from
those in the could mindset (M5.40, SD 5.31) as
honoring both imperatives compared to those in
the should mindset condition (M5.32, SD 5.29),
t(307) 52.75, p5.006, d5.31.
More specifically, solutions based on those in
a could mindset better enabled the organization to
satisfy its mission to continue operating (M54.30,
SD 51.69) relative to those in a should mindset
(M53.82, SD 51.63), t(307) 52.55, p5.01, d5.29.
At the same time, could solutions did not do so by
sacrificing the organization’s values: that is, should
mindset solutions (M55.34, SD 51.42) did not
uphold the organization’s values more than could
solutions (M55.16, SD 51.50), t(307) 51.08, p5
.28, d5.12.
Ethicality. Judges did not rate could mindset so-
lutions (M55.11, SD 51.53) as less ethical than
should mindset solutions (M55.25, SD 51.41),
t(307) 5.89, p5.38, d5.25.
Decision quality. Judges also rated could mindset
solutions as higher in quality (M54.25, SD 51.08)
than should mindset solutions (M54.00, SD 51.09),
t(307) 52.09, p5.04, d5.24.
868 JuneAcademy of Management Journal

Mediation analysis. We examined whether di-
vergent thinking mediated the effect of adopting
a could mindset on the triangulated measure of
moral insight (Baron & Kenny, 1986). Adopting
a could mindset was positively associated with di-
vergent thinking (b5.14, t52.48, p5.01) (see
Table 5). When controlling for divergent thinking,
the effect of adopting a could mindset was signifi-
cantly reduced (from b5.20, t53.56, p,.001 to b5
.16, t52.99, p5.003), and divergent thinking pre-
dicted the likelihood of reaching moral insight (b5
.26, t54.71, p,.001). A bootstrap analysis indicated
that the 95% bias-corrected confidence interval for the
size of the indirect effect excluded zero [.01, .07], sug-
gesting a significant indirect effect (MacKinnon et al.,
2007).
Discussion
Consistent with Hypotheses 1 to 3, participants
were more likely to reach moral insight in incentive-
compatible contexts. In this study, participants were
incentivized based on how a panel of independent
judges would rate the quality of the solution proposed.
Consequently, it is possible that participants’solu-
tions did not reflect their true intentions, but, rather,
the solutions that participants believed would gener-
ate the highest reward. To address this issue, in Study
3 (detailed below), we incentivized participants to
adopt a could or should mindset, but did not attach
a financial reward to the solutions proposed. Addi-
tionally, in Study 4, participants made a decision that
would ostensibly impact someone else in the study.
We also note that external raters did not judge
could mindset solutions as less ethical relative to
should mindset solutions. These findings demon-
strate that, when individuals are prompted to think
broadly, the ethicality of their solutions does not
decline. We note that adopting a could mindset (a)
increased moral insight by decreasing the in-
dividuals’propensity to reject unethical money and
(b) did not affect the extent to which individuals
were likely to accept the unethical money. In follow-
up analyses, rejecting the money and moral insight
solutions did not differ in their perceived ethicality,
b52.07, t51.09, p5.28, and both actions were
perceived as more ethical than accepting the money,
bs..65, t57.97, ps,.001. Taken together, these
findings demonstrate that adopting a could mindset
did not encourage solutions that were perceived as
less ethical.
We also note that it is intriguing that could solu-
tions were not perceived as more ethical even though
they were more likely to satisfy both moral impera-
tives and were perceived to be better in quality.
When individuals adopt a should mindset while
facing right-versus-right dilemmas, they are often
choosing to uphold one moral imperative at the ex-
pense of another. Choosing to forgo one moral im-
perative may provide an indication for how much
the decision-maker values the upheld imperative. In
contrast, moral insights may sufficiently relieve or
preempt the ostensible moral tension such that the
ethical nature of the situation itself and the proposed
solution is less pronounced. By removing the ten sion
in the ethical dilemma, moral insight may also
TABLE 4
Solutions Proposed in Speeches Based on Ratings from 258 Independent Coders (Study 2)
Should Could
Conventional Responses
Accept donation 0.10 0.12
Decline donation 0.48 0.32
Total 0.58 0.44
Moral Insight Solutions
Appeal to audience for donations 0.12 0.17
Work with the company to remove child labor from
its practices.
0.13 0.24
Contingency deal (only accept money if donor
removes child labor from its practices)
0.05 0.11
Concrete measure of moral insight 0.29 0.47
Abstract measure of moral insight 0.32 0.40
Divergent thinking 4.57 (1.41) 4.98 (1.51)
Ethicality 5.25 (1.41) 5.11 (1.53)
Decision quality 4.00 (1.09) 4.25 (1.08)
Note: Standard deviations are presented in parentheses.
2018 869Zhang, Gino, and Margolis

remove salience of the ethicality of the decision at
hand.7Consequently, it may be possible for a moral
insight solution that satisfies more moral impera-
tives to be rated as no more ethical than a conven-
tional solution that satisfies fewer moral imperatives
because the moral insight solution reduces or even
removes the salient ethical tension in the original
problem.
STUDY 3: “COULD”CHANGES
THE CONVERSATION
Two studies showed that could mindsets increase
divergent thinking and consequently the likelihood
of reaching moral insight in private contemplation.
Because individuals often discuss ethical challenges
with others, we investigated, in Study 3, the extent to
which a could mindset changes exploration of viable
alternatives in interpersonal discussions. We con-
ducted this study in a laboratory setting and in-
centivized individuals in dyads to adopt either
a could or a should mindset. We expected to find that
dyads adopting a could mindset would generate
a greater number of solutions (Hypothesis 2). Fur-
ther, we hypothesized that these dyadic interactions
would in turn influence individuals’private decision-
making, leading individuals to generate moral insight
(Hypotheses 1 and 3).
Participants
Two hundred and two individuals (M
age
522.97
years; 49.5% female) forming one hundred and one
dyads participated in a lab study at a university in the
northeastern United States in exchange for $20 and
the opportunity to earn an additional $2 based on
their execution of instructions provided.
Design and Procedure
Participants were randomly assigned to adopt
either a could mindset or a should mindset while
discussing an ethical dilemma with a randomly
assigned partner. All participants read an adaptation
of the analyst’s dilemma (Badaracco & Useem, 1993),
which involved deciding whether to tell their boss
strictly confidential information that would help the
company, but would hurt the individual who di-
vulged this information (see Dilemma 4 in Appendix
D). Dyads were given the opportunity to discuss this
dilemma for 15 minutes in a virtual chat room, and
were informed they could end the conversation
earlier if they finished discussing the dilemma in
the allotted time. To ensure participants would
TABLE 5
Mediation Analysis on Propensity to Generate Moral Insight (Study 2)
Variable
Divergent thinking Moral insight solution Moral insight solution
X→MX→YX,M→Y
Could mindset 0.14*** 0.19*** 0.15**
Divergent thinking 0.25***
Adjusted R
2
0.02 0.04 0.09
95% bias-corrected CI [0.01, 0.07]
Notes:CI5standardized confidence interval for the indirect effect. The table reports standardized coefficients for each regression.
*p,.05
**p,.01
***p,.001
7For example, consider an individual, who, after con-
templating the Heinz dilemma, decides to steal the drug in
order to save their spouse’s life. When judging the ethicality of
the decision to break into the drugstore, individuals may infer
that the decider took a deontological approach that de-
termined the duty to uphold one’s right to life was more im-
portant that the duty to uphold the law. Thus, the significance
of upholding the duty to life is measured by what the indi-
vidual is willing to give up to uphold that value. However,
when individuals adopt a could mindset, they relieve or
preempt the ostensible moral tension so that the ethicality of
the solution is less pronounced. Evaluators are not weighing
a battle of principles anymore. By not seeing a sacrifice of
either moral imperative, it may be more difficult for observers
to determine whether the moral insight solution is somehow
ethically better than a solution that selects one imperative
over the other. In the Heinz dilemma, a moral insight solution
such as using social media to raise money for the drug—a
solution that relieves the tension between saving a life and
breaking the law—removes ethical salience in the problem
that would otherwise provide an indication for how much the
decision-maker valued each of the ethical imperatives.
870 JuneAcademy of Management Journal

remember to adopt either a could or should mindset
during their conversations, we incentivized partici-
pants such that those randomly assigned to adopt
a could mindset received $2 if they asked their partner
“What could we do?”during the interaction, whereas
should mindset participants received $2 if they asked
their partner “What should we do?”At the end of
the discussion, all participants then answered the
question, “What would you do?”Participants were
instructed that they did not need to agree with their
partner on their decision of what they “would”do.
Two independent coders blind to the hypotheses
of the study coded for divergent thinking based on
the solutions discussed and moral insight based on
solutions individuals chose (see Table 6, above for
examples).
One hundred independent coders (M
age
531.85;
41% female) then rated each of the unique categories
of solutions provided based on the perceived crea-
tivity and quality of the solution.
Measures
Table 6 presents the means and standard devia-
tions of the variables we measured.
Manipulation check. Two independent coders
blind to the hypotheses of the study recorded
whether participants used the words “could”and
“should”in the conversations. The agreement be-
tween raters was significantly above and beyond
chance agreement (k
could
5.74, p
could
,.001, k
should
5
.50, p
should
,.001).
Divergent thinking. We triangulated on our mea-
sure of divergent thinking based on the total number
of solutions discussed as well as the extent to which
each of these solutions spanned different categories.
Two coders identified the number of instances dur-
ing each conversation when individuals suggested
a new solution (ICC
3
5.76, p,.001; see Table 6).
Additionally, these two coders rated the extent to
which the discussions contained ideas that spanned
different categories of solutions (1 5not at all,35
somewhat,55extremely;ICC
3
5.61, p,.001).
Because there was high inter-rater reliability be-
tween these two measures of divergent thinking
(ICC
3
5.76, p,.001), we averaged the standardized
scores of these measures to obtain a single rating of
divergent thinking.
Duration of discussion. To understand whether
the content of thought or the length of time discus-
sing dilemmas was a stronger predictor of reaching
moral insight, we measured the amount of time that
participants spent conversing with their partners
about the moral dilemma.
Moral insight. To triangulate on our measure of
moral insight, we used similar concrete and abstract
TABLE 6
Solutions Discussed in Conversations with Partner (Study 3)
Categorization of each unique
solution discussed in
conversations
Categorization of individuals’
decisions
Should Could Should Could
Conventional Responses
Tell boss confidential information 0.67 0.78 0.32 0.26
Keep information confidential 0.55 0.54 0.34 0.19
Total Conventional Responses 1.22 (.34) 1.32 (.46) .65 (.45) .44 (.48)
Moral Insight Solution
Ask roommate for permission to tell boss
confidential information
0.12 0.13 0.12 0.23
Convince the roommate that information ought to be
revealed to boss
0.18 0.27 0.12 0.16
Make information anonymous (provide an
anonymous tip, making source of information
anonymous, or both)
0.19 0.40 0.12 0.23
Obtain same information through other channels
first before telling boss
0.03 0.07 0.03 0.09
Notify boss that he should investigate the company
and arrive at same information independently
0.04 0.12 0.07 0.12
Total Moral Insight Solutions 0.57 (0.65) 0.99 (0.84) 0.32 (0.44) 0.53 (0.48)
Note: Means represent the averaged binary coding of two independent coders.
2018 871Zhang, Gino, and Margolis

measures as in Studies 1 and 2. For our concrete
measure of moral insight, two independent raters
coded the solutions as either a conventional solution
(i.e., tell the boss the information or keep the in-
formation confidential) or as one of the many un-
conventional solutions that two separate research
assistants had identified as upholding both the duty
to the organization and loyalty to the roommate (see
Table 6 for examples). Solutions were coded as
moral insight if they contained at least one of these
unconventional solutions. Because the two raters
achieved high agreement (k5.81, p,.001), we av-
eraged the two coder’s ratings to obtain a sin gle rating
of moral insight.
For our abstract measure of moral insight, partici-
pants rated the extent to which they believed their
solutions met multiple imperatives presented in the
dilemma (1 5met only one objective,75met all
objectives).
The correlation between external coders’ratings of
moral insight and individuals’perception of moral
insight was moderately high (ICC
3
5.67, p,.001).
To triangulate on the construct of moral insight, we
averaged the standardized scores between external
judges’perceptions of moral insight and internal
raters’self-assessment of the extent to which their
solutions met both moral imperatives.
Decision quality. One hundred independent
raters blind to the hypotheses also judged these cat-
egories of alternative solutions based on the crea-
tivity (1 5not at all creative to 7 5extremely
creative) and quality of the solution (1 5extremely
bad solution,45neither good nor bad,75extremely
good solution).
Results
Manipulation check. Participants in the should
mindset were more likely to mention “should”(M5
99.0%, SD 5.07) in their discussion with their
partner than those in the could mindset (M529.8%,
SD 5.28), t(99) 516.53, p,.001, d53.32; similarly,
those in the could mindset were more likely to
mention “could”(M594.2%, SD 5.16) in their
discussion than those with a should mindset (M5
7.1%, SD 5.18), t(99) 525.89, p,.001, d55.20.
Duration of discussion. Dyads discussing what
they “could”do spent more time discussing the di-
lemma (M5601.91 seconds, SD 5310.5) than did
dyads discussing what they “should”do (M5482.85
seconds, SD 5246.97), t(99) 52.17, p5