Analyzing Moral Deliberation During Gameplay: Moral Foundations Theory as an Analytic Resource

How Have Multiplayer Games Been Used to Study Moral Reasoning?

In search of a practical and ethical setting to study how individuals respond to moral dilemmas, social scientists have used games and simulations (e.g., Aguiar, Brañas-Garza, Espinosa, & Miller, 2007; Ambec & Sprumont, 2002; Axelrod, 1980; Beal, Ghintran, Remila, & Solal, 2013; Bogost, 2008; Costikyan, 2010; Dray, Perez, Le Page, D’Aquino, & White, 2007; Eckel & Grossman, 1996; Hennig-Schmidt, Li, & Yang, 2004; Henrich, Heine, & Norenzayan, 2010; Nowak, Page, & Sigmund, 2000; Ostrom, Walker, & Gardner, 1992; Sicart, 2009; Stahl, 1972; Thaler, 1988; Zagal, 2011). Inasmuch as individuals respond to moral dilemmas in these settings in dynamic and visible ways, social scientists can adequately capture their phenomenon of interest: moral deliberation in action. Yet, as we will argue, several of the games traditionally used in these studies fall short of producing dynamic and visible responses to moral dilemmas. This does not mean that players fail to engage in moral deliberation; they sometimes do. Rather, we mean to suggest (1) that the restrictions that researchers place on participants limits the ability of researchers to detect moral reasoning and (2) that it is possible to design games that are better at eliciting a more intense degree and wider range of moral reasoning. We draw upon moral foundations theory (MFT), not commonly used within game theoretic research, but now well known in moral psychology, and a game that mirrors the experiences of participants in climate change negotiations, to illustrate these points.

In this article, we situate our game and study in comparison with games and studies from game theory for several reasons: First, game theoretic games such as the Prisoner’s Dilemma are well known; second, game theoretic games such as Ultimatum and Dictator have a rich tradition of being used to study moral deliberation; third, game theoretic games pit multiple players against/with each other; and fourth, game theoretic games are still being used and innovated upon by moral researchers today.

In the Prisoner’s Dilemma game, two individuals awaiting trial are simultaneously presented with a bargain: they can testify against the other or remain silent. The payoff is dependent on what both players simultaneously and without communication choose to do. If they betray each other, they suffer together; if they remain silent, they also suffer together but to a lesser extent. And, if only one betrays the other, they serve no time, while the other serves the highest possible punishment.

The Ultimatum and Dictator games, by contrast, present players with an asymmetrical power structure. One player is given a set value of goods (e.g., US$10) and must decide how much of the goods to share with a second player and how much to keep for themselves. In Ultimatum, the second player can refuse the offer (e.g., if they think it is too low) in which case neither player receives any goods. Dictator is the same game as Ultimatum with one striking difference: the second player has no veto power. As in the Prisoner’s Dilemma, players are not allowed to communicate with each other.

These games fit the geopolitical events ongoing at their time of creation; some of the earliest applications of game theoretic games by John von Neumann and colleagues at the RAND Corporation in the 1950s were metaphors to Cold War dynamics. Games like these suggest a solution to nuclear threat: Both sides need enough nuclear weapons to have mutually assured destruction, effectively changing the situation from one in which only one side had nuclear weapons (analogous to Dictator) into one where both have the nuclear option (more analogous to Ultimatum and Prisoner’s Dilemma).

Like a Cold War, these games have implicit elements of competition—whichever player you are, you may find yourself wanting to have the least amount of punishment or the most goods by the end of the game. They also have moral elements. In the Prisoner’s Dilemma, the morally significant question is “will you betray the other player” and a strategic question is “do you think the other player will betray you?” In Ultimatum, the morally significant question for the first player (if there is one) is how much to give, and a strategic question is how greedy they can be before the other player refuses an offer out of spite? The morally significant question for the second player in Ultimatum is when to refuse an offer, and if the game is played in iterative (repeated) rounds, a strategic question is if to refuse any offer, thus if, and when, to hold the first player at ransom? If Ultimatum is played over iterative rounds, another morally significant question can emerge for the first player: How much to give after a refused offer? In the Dictator game, the morally significant question for the first player is how much to give knowing that the second player does not have the ability to refuse the offer?

To examine the moral dimension of these games, researchers often draw upon game theoretic models from behavioral economics which hypothesize that individuals will always act to maximize their winnings. Yet, researchers find anomalies in empirical studies of gameplay; players do not always act to maximize their winnings, which has sparked research into the motives of players and the role that morals play in decision-making (Camerer & Thaler, 1995; Dalbert & Umlauft, 2009). For example, the second player in Ultimatum will sometimes reject low payoff offers, even though doing so means that player scores no goods in the end. In this case, researchers speculate that a moral concern for fairness might be in play (Güth, Schmittberger, & Schwarze, 1982). Sometimes, the first player in Dictator gives some, half, even all US$10 away. In post-interviews, Aguiar, Brañas-Garza, Espinosa, and Miller (2007) revealed reasons for this including equity (50/50 is proportionally fair), selfishness (I want the money), hardship (I need the money), and charity (I have plenty, others need it more, etc.).

Henning-Schmidt, Li, and Yang’s (2004) empirical research on Ultimatum revealed that a wide range of moral thinking, and at times inconsistent thinking, can take place during a player’s decision-making process. The researchers positioned Ultimatum participants into teams of two participants each—one team had to decide together how much to offer and the other team deliberated about whether to accept or refuse the other group’s offer. This participant structure allowed for rich talk within the teams, from which the researchers used discourse analysis to identify a variety of player-generated justifications for their choices, since the team deliberation can be used as an analog for the inner conversation of a single person. Their study of Chinese participants, uncovered justifications based upon emotional upset, Chinese cultural conventions around reciprocity, avoiding unlucky numbers in Chinese culture, and strategic planning. As we will see, many of these considerations can be made sense of by MFT.

Addressing Challenges in Game-Based Moral Studies

Many game theoretic researchers gravitate toward conceptions of morality accepted by behavioral economists. They may also utilize games such as Ultimatum and Dictator because they are parsimonious and lead to easily quantifiable results. In fact, Ultimatum and Dictator have been used to quantify cross-cultural comparisons of fairness in well-cited studies that assume players will be motivated to attain more rewards (e.g., cigarettes, money, or other tokens) at the end of the study (e.g., Henrich et al., 2005, 2010). Yet, we believe more can be done (1) in the design of games, (2) in the methodological design of game-based moral research studies, and (3) in the adoption of moral theory that can support these studies.

On the first point, the design of games for moral study can be innovated by including game elements that align with current metaphors; while contemporary international negotiations do have some of the same elements as Cold War negotiations including non-zero-sum features, they additionally integrate dynamics such as the tragedy of the commons, vast inequality among participants, deliberative discussion among participants, asymmetric goals, difficulty in determining fairness among participants, more than two key sides, stochastic elements (e.g., uncertainty around climate events), and delayed impact of decisions on a complex system. Games with some of these traits have been developed and studied by game theorists, yet games with these traits are rare and not used in studies of moral thinking (Fennewald, 2015). The questions implicit in traditional examples are quite limited in dynamic scope and seem to be most relevant to deliberation involving considerations of reciprocity and trust (just as the moral issues in the Cold War were focused on reciprocity and trust). Those designs do not allow players to demonstrate a wide variety of moral considerations during gameplay. Unexplored questions include: What would negotiations between players look like if there were multiple decisions to be made? What if the game were a system with multiple moving parts that required attention to both the parts and to how they balance together? These elements are likely to encourage players to deliberate about how they will share and manage resources that are commonly available to all. This brings us to our next point, which regards research methodology.

On the second point of research methodology, we point out that players’ moral justifications and deliberation is rarely rendered visible in game-based moral research studies. Most studies only capture player’s actions and miss their verbal communication altogether. Two factors contribute to this issue. First, as just mentioned, the games themselves might not be rich enough to give players something to have a moral discussion about. Second, researchers might not allow participants to talk. This may be done for any number of reasons, possibly including because verbal data are messy or because they are not currently employing theories that emphasize the process of oral moral deliberation. Methodologically, some studies of moral justification have used think aloud procedures or conversation between participants, as in Henning-Schmidt et al.’s (2004) study where two participants form a team to control a single player. Yet, even in this case, the participants were still not able to communicate with the participants who control the other players. As such, players have little opportunity to verbally justify their decisions and try to persuade other players—in other words, negotiation is not possible; this is like a Cold War with no Red Phone. We expect that these games could be rich with moral discourse if players were allowed to negotiate and deliberate with one another. Even in the case that players do not talk during the game, postgame interviews or surveys of moral reasoning can support inquiry into moral reasoning.

While we suggest that methodology can be improved by accounting for dialogue and justification because these methods can reveal more information, we also caution against relying solely on quantified methods because these can be misleading until participants are consulted on how they interpret the game/research experiment as an artifact. We aim to avoid the methodological procedure of several highly cited studies including Henrich et al. (2005) and Henrich, Heine, and Norenzayan (2010), which rely solely on quantified results from games to compare the aggregate fairness of groups such as societies and genders while not appropriately acknowledging or accounting for how players might be interpreting these games. We acknowledge that cultures and individuals vary in how they interpret and use games and that games can act as play spaces in which players can exhibit traits that do not reflect their true selves (Huizinga & Hull, 1949). An individual can vary in fairness from one gameplay to the next based upon a variety of reasons. It is therefore not enough to simply administer games like Ultimatum or Dictator across populations (such as gender, age, race, religion), and draw claims based upon game results about what society or person is more “fair” or “moral” than the next; we must understand what the game means to the players and calibrate accordingly. Until we do, the results say little about the players outside the game (Fennewald, 2015).

On the third point of moral theory, we point out that most game theoretic studies assume people are selfish and rational agents and focus on elements such as reciprocity and trust. Yet, many behavioral studies have found that actual players do not act as predicted (e.g., Camerer & Thaler, 1995). In the next section, we explore MFT as a guide for thinking about additional moral dimensions.

In summary, we recommend creating games that enable rich moral decision-making and studies that allow participants to voice their decision-making. A benefit of researchers in utilizing simulations that are rich is that individuals’ responses may show more sides of their moral reasoning. Furthermore, if researchers allow opportunities for individuals to justify their responses or use their responses to persuade others, then their thinking may be easier to study. As well, if the simulations employed represent contemporary contexts, the findings might be more relevant to current events. In response to these realizations, we developed Troubled Lands, a bargaining game that contains non-zero-sum, tragedy of the commons, and open-communication mechanics. The result is a rich setting for dynamic and visible moral deliberation. But, before we discuss Troubled Lands, we need to move beyond the pitfall of thinking about morals in narrow terms of reciprocity and trust. Thus, we first turn to discussion of MFT, a lens that makes a wide range of moral motivations visible.

What Is MFT?

MFT was developed through a process of asking participants from many cultures questions about morality regarding numerous moral taboos and values: care for vulnerable individuals, distribution of rewards, respect of elders, sexual chastity, respect for the environment, personal autonomy, group loyalty, and more (Haidt, 2012). MFT focuses on a plurality of values that lie at the foundations of individuals’ moral reasoning. The six moral foundations that Haidt (2012) finds to be the clearest cut are care, fairness (proportionality), liberty, authority (respect), loyalty (in-group), and sanctity (purity). Table 1 presents Haidt’s identified six moral foundations.

OPEN IN VIEWER

Table 1.

Haidt’s Six Moral Foundations Upon Which Decisions Are Justified.

Foundation	Description
Care	A propensity to empathize with people who suffer, take care of those who are weaker, and who need assistance. Care is believed to have been evolutionarily important, as more caring individuals are more likely to protect young. Today, it is often expressed as a form of empathy for the weak, old, and sick, in addition to those who are young.
Fairness or proportionality	Involves justice—that all get their just earnings and are rewarded appropriately. Fairness/proportionality is expressed when one strives for people to get what they deserve and when one feels negatively when others do not get what they deserve.
Liberty	A value for one’s right to autonomy and self-determination. Liberty is the foundation most commonly used by libertarians who contend that it is not morally required to do or not do anything unless it causes an active role in reducing another’s liberty. Liberty arises when individuals believe they and others are responsible for being self-sufficient.
Authority or respect	A value for traditions and those with status. These figures can include elders such as parents and teachers or leaders. Social conservatives tend to have stronger authority foundations than social liberals.
Loyalty or in-group	A value for the group to which one belongs or an idea for which one believes. It can be expressed in the form of nationalism and group pride, including when people have team or school spirit. Some of the strongest expressions of loyalty include when individuals risk their lives for their family, nation, tribe, or a cause.
Sanctity or purity	A value for avoiding disgusting things and protecting what is pure. Social conservatives and social liberals often contrast in how they express sanctity, with conservatives typically focusing on sexual purity through abstinence and liberals focusing on purity from nutritious foods and/or purity of the environment.

Note. Adapted from Fennewald (2015).

There are several qualifiers to keep in mind when examining Haidt’s depiction of morality. First, different populations and worldviews will emphasize different moral foundations as guiding their decision-making and justifications for what is right and wrong. Second, Haidt and his fellow researchers agree that they have not yet discovered all the foundations. Their “list is not the final list,” as they put it (Graham et. al., 2013, p. 36). Indeed, they report that “we are constantly arguing among ourselves over changes to existing foundations and considerations of new candidate foundations” (p. 36).

The research group uses several formalized criteria for vetting potential moral foundations. The criteria touch upon third-party normative judgments, automatic affective evaluations, cultural profusion, as well as biology and evolution. The group openly invites critics to pushback on the various categories, so that they can continually test and vet more potential foundations. The research group calls this the method of coevolution: “in our vision of method theory coevolution, critics are especially needed on the theory side…offering competing conceptualizations of the moral domain.” The method theory of coevolution has challenged the research group to consider alternative categories such as honesty, industry, and modesty.

Third, these categories are derived from a one-on-one interview protocol. Scant attention has been paid to the morals that are brought into play during group interactions. Yet, negotiations are often rich with moral discourse as individuals attempt to persuade others to act in certain ways and to justify their own actions to those who must face the consequences of their actions. We have wondered what would examining the discourse of negotiations with consequences tell us about potential moral categories? Reliance on an interview protocol may be one reason why MFT has not yet identified many interpersonal and relational factors (Graham et al., 2013) and why it has little to say about real-world instances of moral persuasion and disagreement.

This is precisely why games and simulations that position players to negotiate across difference in situations where they feel the impact of their own and others’ actions can be a powerful methodological tool to make visible additional moral categories. In other words, games and simulations can act as vehicles for the method theory coevolution of MFT.

Likewise, MFT can make sense of the plurality of values that players draw on during gameplay. By contrast, the conceptions of morality held in many behavioral economics/game theory studies are not designed to be sensitive to a plurality of values. In other words, MFT is a vehicle for the method theory coevolution of the field of game studies.

What Is Troubled Lands?

Troubled Lands is a 30-45 minute game designed to emulate many of the power dynamics present in a climate change debate. It is playable by ages 10 and up and is played in groups of three players. Each group has the same three roles: a farmer, a rancher, and a lumberjack. Each group shares a common area of 12 plots of land. Each plot can be occupied by desired resources such as fields, pastures, and forests, or by fallow (empty) or eroded (unusable). Each of the three roles has unique abilities for planting, harvesting, and scoring points from the plots of land. For example, the farmer scores the most points from harvesting fields but is best at creating pastures, whereas the rancher is best at creating fields and scores the most from harvesting pastures. The lumberjack role is unable to score points from anything but harvesting forests, which are a critical resource; without enough forests, the land is susceptible to erosion. A random element die roll after each round determines how many forests are needed to avoid erosion each round. Erosion permanently alters the commonly shared area of land.

Players compete in a tournament style such that each farmer aims to have more points than farmers who are in other groups and each rancher aims to have more points than ranchers in other groups, and so on. Groups play simultaneously and are only allowed to interact within their own group (and do not have access to other groups). The result is a situation that in a group none, one, two, or all three roles might win. Importantly, players within a group are not competing against one another (non-zero-sum interactions), but they are managing a shared area of land together (a common resource pool in which a tragedy of the commons is possible), and they are allowed to freely communicate with one another (open communication). These are three of the qualities that we believe makes Troubled Lands an ideal research setting for rendering moral deliberation dynamic and visible.

Furthermore, Troubled Lands was specifically designed so that there is no easy way to balance the game such that all players end with the same number of points and so that players cannot simply perform the same pattern of actions to maintain a stable condition (in game theoretic terms, there is not a steady-state solution). This stands in significant contrast to most bargaining games such as The Prisoner’s Dilemma or Ultimatum in which, over iterative rounds of play, players can devise and enact a perfectly balanced solution. It also makes moral deliberation more lifelike and dynamic (a solution that works one round will have to be revised in the next round due to changes in the land and in player’s trust relations) and more visible (players are continually deliberating throughout each round of the game rather than deliberating once and executing a plan the remaining rounds).

Of course, moral dilemmas surround gameplay for most, if not all, games in general: How to balance competition and fun? What to do if cheating occurs? As well, moral questions can be asked of games in general: Is this game violent? Does it play forward patriarchal norms? Note how these moral concerns are generally extrinsic to the game, hence, why they appear in most if not all games, including sports. But, in our research, we have found that few games generate moral dilemmas where players feel genuinely, and continuously, torn between multiple competing values, due to design elements intrinsic to the game (Phelps et al., 2016). In fact, the design challenge for Troubled Lands was to create a game in which players truly are torn between working toward selfish goals (to outcompete their counterpart roles in other groups) and working toward selfless goals (being considerate and helpful to others, and showing care and respect for the land, even if doing so does not directly impact their own chances of winning) in a way that needs to be continuously reevaluated.

This differentiates Troubled Lands from Ultimatum and Dictator where steady-state conditions are not only easier to identify (in the strategic sense), yet also easier to maintain (because moral reasoning for any given player can be gray). After many prototypes, we finally developed a game in which players had independent win conditions (a rancher wins regardless of whether the lumberjack or farmer in their same group wins) and orthogonal goals (lumberjacks compete against other lumberjacks, etc.) and where there was not a steady-state solution. The result is a game that is neither purely competitive nor purely collaborative and one in which the relative success of any two players at the same table can never be fully assessed (just as in real life, there are many factors to success and it is always up for debate which of two individuals or nations is better off). These conditions leave players with a legitimate and ongoing situation in which they can genuinely debate what is fair in terms of economic justice, just as politicians have for millennia. For those readers who want to learn more about Troubled Lands or see it in action, a video explanation and game materials are freely available at www.troubledlands.com

Methodological Twist: Troubled Lands and MFT in Conversation

Troubled Lands is designed to be a location for rich, proof-of-concept inquiry into in-game moral justifications, as it intentionally positions players to engage in continual discussion and negotiation. Indeed, often the difference between winning and losing is how well players negotiated with one another during gameplay, but not merely in deception. Winners are those who coordinate as often as they are those who deceive. For this reason, Troubled Lands generates quite a bit of deliberation and surfaces several justifications that players give for their plans and actions. The specific design moves made in creating Troubled Lands, as documented above, answers our first research question: How to design games that engender dynamic and visible moral deliberation. Our second research question is which moral foundations—care, fairness, liberty, and so on—do players invoke during their gameplay discussions. Our third research question is what additional moral justifications do players invoke that are not readily captured by Haidt’s six categories. Due to the conceptual nature of this piece, we are only focusing on proof-of-concept. More detailed analyses of player’s moral justifications are reported in Fennewald’s (2015) empirical presentation of this same data.

We can situate the experience of playing Troubled Lands in terms of how it compares with Haidt’s interview method. Haidt’s interviews begin by asking participants a question such as “Would you eat your pet?” or “Would you have intercourse with your sibling?” When interviewees say no, he pushes them to give a justification. In response to their justification, Haidt offers a counterexample: what if you were starving or what if your sibling was not biologically related. If interviewees say no, Haidt pushes for another justification and again offers more counterexamples. This method is very clever and it often leads to a phenomenon that Haidt calls moral dumbfounding—in which interviewees feel that something is grossly wrong but cannot articulate why. This is usually a sign that an interviewee’s deep moral foundation has been violated.

Whereas Haidt’s method asks participants to imagine hypothetical situations in an interview, our participants are placed in a simulation of resource scarcity. Instead of replying to an interviewer, as in Haidt’s technique, in our Troubled Lands simulation, participants are responsible to each other and are also a source of interrogation and dispute. Also unlike Haidt’s technique, players will anticipate and observe actual outcomes from their actions played out in the simulation, including potential direct impacts on themselves and their peers.

From the first turn of Troubled Lands, players have several conflicting considerations. For example, the farmer may notice that they can score more points per turn than any other player and do so relatively easily by harvesting just one wheat. Yet, other players may point out the farmer’s privileged position—the farmer is able to score more points, create more pastures, and plant more forests than any other role—and try to persuade the farmer to use their positionality to support the shared common land rather than exploiting it (an appeal to care and fairness/proportionality). Furthermore, the rancher and the farmer often notice that they are in a relationship of comparative advantage. The rancher has the strongest ability to plant fields, which the farmer benefits most from harvesting, and the farmer has the strongest ability to plant pastures, which the rancher benefits most from, leading them to consider whether to make appeals to reciprocity. This mutual reciprocity seems to approximate an ideal steady state, but often the lumberjack will realize that this mutual reciprocity is not fair at all—not to the lumberjack’s role nor to the commonly shared plots of land which start to degrade because the farmer and rancher are not focused on protecting it from erosion, a burden that usually falls on the farmer to attend to since the farmer is best at planting forests.

The lumberjack is dependent on harvesting forests for points, is not able to plant any fields or pastures, but can destroy fields and pastures. This puts the lumberjack in a unique position with numerous moral choices to make: do they acknowledge their dependency on others and follow their lead helping them to win (similar to deferring to authority or demonstrating group loyalty), do they modestly pursue their own collection of resources even though they’ll never gain a high score, do they threaten to destroy the land if others do not plant trees and help them to win, or do they threaten to destroy the land to sanction the other players for their greedy actions. All of these scenarios play out in the various sessions of Troubled Lands.

The game usually ebbs and weaves with these dynamics and questions leading to a variety of moral claims: Throughout the experience, conversations include justifications of care for others (I want you to win/I see you as weak and in need of my assistance), fairness/proportionality (in at least two forms: let’s split the earnings, let’s take turns), liberty (you cannot tell me what to do), and loyalty to the group (talking about the effort to plant forests, avoid erosion from lack of forests, and reach common negotiations as teamwork). As the game ends, players have a final moral choice, clear-cut the forests for points, or leave them for aesthetic reasons (in-game sanctity/purity). In debriefs, some players describe the forests as beautiful, and harvesting forests as painful, whereas players from other groups are genuinely surprised with responses to the effect of “really, I just cut them down, I wanted the points so I could win” (note we suggest winning/accomplishment as a moral/motivational driving force, see below). Other players seem less concerned with the sacred aspect of the trees, but more concerned with the balance of resources and their personal responsibility in maintaining balance (we also suggest that personal responsibility could be a unique moral/motivational foundation, see below).

Moral Foundations in Action

In this section, we explore transcripts from eight games of Troubled Lands for examples of moral claims. Methodologically, our approach to studying data from Troubled Lands is guided by our desire to bring moral theory research and game-based moral research into dialogue with one another, seeing how moral theories can encourage the design and use of games with more moral dimensions and seeing how morally rich games might elicit categories that moral theories such as do not contain. Therefore, we aim to identify examples of moral categories much as Aguiar et al. (2007) and by Henning-Schmidt, Li, and Yang (2004) do in their studies of Dictator and Ultimatum. This research can be conducted with other games, but we chose to use Troubled Lands, because the game more closely approximates climate change debates than do classic game theoretic games and because Troubled Lands is designed to open opportunities for a wide range of moral dialogue; we wanted to start searching for moral claims with a simulation that was both relevant to current events and a likely location to find diversity.

Our data set is described in full in a dissertation which collected data on eight groups of players who played Troubled Lands (Fennewald, 2015). Four of the groups were undergraduates who played the game as part of a political science class on sustainability, two groups were from a graduate seminar in mathematics education, and two groups were adults who met at a private residence. In our exploration of the eight transcripts, we first explored for examples from in-game conversations and postgame focus groups for any statements, interplayer dialogues, or actions that could be deemed as justifications for actions, be they overtly moral or not. Then, we considered whether each example of a claim or moral dialogue could align with moral foundations in MFT or an emergent category that did not fit within these well-established moral frameworks. We found examples that both authors agreed matched most of the foundations in MFT as well as a few that did not fit easily with MFT.

Table 2 crosswalks Haidt’s (2012) six moral foundation categories with gameplay descriptions and actual examples of transcript from Troubled Lands game sessions. These examples illustrate that MFT can be used to make sense of the many moral dimensions that players drew upon in making claims (to justify their actions or persuade others). Note that we found clear dialogue between players for five of the six categories. The category of authority was not directly referenced during in-game talk, but it appeared to us to be enacted whenever players deferred to the lead of others or followed the advice of more experienced players. It also appears when players make bids to be an authority figure, including when players try to map out the entire course of gameplay and develop a plan for how each party should act on their turn.

OPEN IN VIEWER

Table 2.

Moral Foundations in Gameplay.

Category	Gameplay Description	Examples Across Troubled Lands Play Sessions
Care	Care is upheld when players make a genuine sacrifice (lower their own chances of success) to help another player do well. Care is violated when players do not support other players in need. Care is invoked in discussions about whether or not to help others.	Lumberjack: “I’ll lose so that you guys can win” Farmer: “I can plant trees, yes. Um, just trying to help out” Lumberjack: “I think you [at Farmer] should take a wheat” Rancher: “yeah, I think so”
Fairness or proportionality	Fairness is upheld when players are rewarded or punished in proportion to their contribution to an effort. Fairness is violated when players are rewarded or punished disproportionately to their contributed effort. Fairness is invoked in discussions about whether to reward or punish others, and in discussions about reciprocity.	Lumberjack: who hasn’t, who hasn’t played a card, to restore land. Farmer: “We need two people to volunteer and not lie about restoring the land card down for this next round.” Lumberjack: “well, since I lost one of my actions, I don’t want to do it.” Farmer: “Alright, that’s fair, that’s fair. So do you want to restore land?” Farmer: “I’ll do some pasture, you do some wheat?”
Liberty	Liberty is upheld when players are allowed to act and speak freely without coercive force or manipulation from others. Liberty is violated when players lose control of their actions (their turn is skipped or their actions are manipulated by others) or their voice is silenced. Liberty is invoked in discussions about whether or not force or manipulation is being used to affect someone’s actions.	Rancher: I just want to wait until there’s a little more destroyed land Farmer: Alright, if you wait until there’s more destroyed land then there’s all this destroyed land. Alright. You can do whatever you want, I can’t force you.
Authority or respect	Authority is upheld when players follow the rules, proposals, and/or advice of those they deem to have authority such as more advanced or engaged players. Authority is violated when players go against rules, proposals, and/or advice of those they deem to have authority such as more advanced or engaged players. Authority is invoked in discussions about whose rules, proposals, and/or advice should be followed or not.	Authority was evidenced when players made bids to lead the rest of the group, and when others in the group followed another’s lead (including their advice, plans, and moral considerations).
Loyalty or in-group	Loyalty is upheld when players work to the benefit of their allegiances or group identity. Loyalty is violated when players work against their allegiances or group identity. Loyalty is invoked in discussions about how well specific allegiances or the group overall is performing. This can be expressed through celebratory remarks when things are going well, negative remarks when things are going poorly, or through collaborative planning for how to help the group overall.	Rancher: I don’t know how we did individually, but I think as a group we did really well. Farmer: I mean we just completely worked together the whole time. Rancher: Yeah! We did it! Farmer: well, we’re fucked. Farmer: I planted two trees for us. Farmer: we can get rid of the empty land and we could have more trees, if you guys want. Rancher: we should do that and then take care of the destroyed land, because, what would that give us. I mean we have 3 right now, of 4, 5, we’d have 7.
Sanctity or purity	Sanctity is upheld when players defend that which is deemed reverent or sacred within the game. Sanctity is violated when players attack that which is deemed reverent or sacred within the game. Sanctity is invoked in discussions about what is worthy of respect and value.	Farmer: okay. now we’ve got bunches of trees [laughs] Rancher: ok. Observer: that’s beautiful Farmer: it was like this whole game we were trying to keep things green and then these last two round we were. I felt like we were doing something really bad because we don’t want to hurt the earth.

Motivations: Beyond the MFT Categories

Just as our second research question asks how MFT can inform game studies and game theoretic investigations, our third research question flips the question and inquiries into ways games can be used as an exploration grounds for discovering moral categories. Can games act as vehicles for method theory of coevolution of moral categories? We argue in the affirmative and provide evidence for our argument by exploring alternative moral values that were drawn on during in-game player deliberations.

Discussion

In our investigation, we explored three questions: First, how can we design games and research studies that generate dynamic and visible in-game moral deliberation among players. Second, what can moral theory from the field of moral psychology, and specifically MFT, teach the fields of game theory, behavioral economics, game studies? Third, what can an investigation of morals in gameplay of morally rich games and games about morally complex events like climate change, such as Troubled Lands, teach the field of moral psychology? To answer these questions, we contrasted the mechanics of bargaining games from game theory to those of Troubled Lands, and we selected examples of player dialogue and action from Troubled Lands that supported as well as pushed back on the plurality of values identified in MFT.

An inherent limitation of the methods used in MFT research is that the theories arise from the questions asked. MFT researchers ask questions about onetime events that involve the choice of a single individual (Is it ever right to steal medicine? Is it ever right to eat a pet?). These onetime event scenarios do not position participants in situations where they are acting over time and in relation to others. As such, participants are not placed in situations where achievement is on the line, where the weight of their consequences is felt by themselves and others, and where they must navigate trust and conflict in developing relationships and decide whether to persevere or give up in the face of setbacks. If moral theories arise from the methods used to develop them, we can begin to see why games and simulations are such a fruitful opportunity to further develop moral theories.

We are not arguing against interview protocols as a tool for probing into individuals’ moral positions. We are arguing for the value of games and simulations to extend the inquiry into moral foundations even further. Our main finding here is methodological: When games are designed to make players’ moral deliberations dynamic and visible, we find that games can act as a petri dish for generating moral claims (reasons, justifications, persuasions, etc.).

Regarding moral theories, we find evidence supporting many of the categories in MFT and evidence to go beyond these categories. We see strong examples of care, fairness, liberty, and sanctity. We find weak examples of authority and loyalty (in group) but attribute our weak findings to the limitations of Troubled Lands rather than to limitations of the theory or the games-based research medium. Indeed, other research (Holahan, 2009) on a fishing boat game found that players with smaller boats were more likely to defer decisions to those with larger boats, suggesting that power as authority can be drawn on in gameplay. We also find that players’ actions are not purely motivated by moral reasons—power relations and strategic calculations also impact how players act in a game. Again, this is not a limitation to games-based research, rather it demonstrates just how readily games and simulations can model real-world negotiation situations that are often mired in messy issues of power and strategy.

We do not claim that how players act in such games reflects their morality outside of the game. Troubled Lands players have told us as much, making statement during gameplay or afterward such as “I’m not like this in real life,” “Don’t worry, I recycle,” and “I don’t like to compete in games, but in real life I’m very competitive.” Furthermore, we have observed how the same groups of players’ in-game cooperative levels can dramatically shift in response to simple rule changes (Fennewald, 2015). This suggests to us that, in line with a sociocultural perspective on identity (Holland, Lachicotte, Skinner, & Cain, 1998; Wortham, 2006), players’ identities are dynamic and coconstructed achievements across different situations.

The examples we find using Troubled Lands stand as counterexamples to research that focuses primarily on player behavior rather than on player reasoning. Many behavioral economists may already know this from their own unpublished work. For example, the abovementioned fishing boat example in which weak players defer power to powerful players is not mentioned in the article itself; one author told us that the observation that players with smaller boats will defer to players with larger boats remains unpublished because in the field of behavioral economics, which often uses games, any mention of normative analysis is frowned upon (even when it comes from the most prestigious research groups). We view this as an area in which MFT stands to aid the field. In our work, we find that players do not make decisions as purely rational agents; they are not purely Homo economicus but rather also Homo ludens and Homo moralists. Bolstering this idea, in other studies (Fennewald, 2015; Fennewald & Kievit-Kylar, 2013; Phelps et al., 2016), we explore Troubled Lands play in competitive, collaborative, and cooperative forms and find that the same players adjust their play styles accordingly while changing the number of moral claims they engage in depending on the rules.

Players often adapt to the rules and situations of the game; what players do in a game has as much, if not more, to do with the game or context than with the players as evidenced by greater variance within players across changing contextual conditions than between players within similar contextual conditions. Thus, moral researchers should be cautious when, or even avoid, using games to make claims about individuals or societies. It is safer to make moral claims about how games and situational contexts influence humans.

Conclusion

This methodological theory-focused piece has worked through a way to move the fields of game studies and moral psychology forward to create richer games and simulations that can act as petri dishes to observe moral deliberation in action. The high-level answers to our research questions are summarized below:

How can games and game research studies be designed to support the elicitation of moral claims?

Several design mechanics were effective for generating robust moral deliberation: independent and orthogonal goals, a common pool resource/tragedy of the commons mechanic, and an open-communication mechanic, as well as the difficult to balance game state.

We will end on a note of optimism—that the intersection of the fields of game studies and ethics will continue a fruitful dialogue that coevolves our theories of morality.