Portal de Periódicos da CAPES

We‐Mode as Layered Agency

2024; Wiley; Linguagem: Inglês

10.1111/josp.12600

1467-9833

Lukas Schwengerer,

Management and Organizational Studies

We humans are social beings.1 As such, our agency can take on goals and means that are not only determined by an agent individually but by cooperating and coordinating with other people. Agents that are part of groups can act with the group in mind. They can act as group members taking on the group's motivational structure. I can act as a member of the philosophy department aiming for what the department agreed on, even if that might conflict with my personal beliefs. I can shut out these personal beliefs for a moment and reason from the group's standpoint. And I can also switch from my reasoning as a group member back to my individual point of view when the situation requires it. I suggest that this ability to switch between agencies is best described as a form of layered agency—a we-mode layer of agency. The idea of layered agency has been proposed by Nguyen (2019, 2020) in his seminal work on striving play—play we partake in with the aim of experiencing a struggle. I argue that the idea can also be explanatory powerful in the social realm more generally. How we reason in many cooperative activities is best described as taking on a new layer of agency that is jointly established with other people. The ability of agents to submerse themselves in this jointly established layer of agency is the key to understanding actions in a group mindset, the nature of social groups, and the mental states of group agents. Much of our social life is shaped by this ability to take on a we-mode as a temporary layer of agency. I can smoothly transition from practical reasoning based on my own individual goals to reasoning based on goals established jointly with other people, such as the goals of the philosophy department. I can also transition between different we-modes as different temporary layers of agency. At one point, I reason as part of the philosophy department, at another point, I reason as part of a football team. The goals of these groups are different, and some accepted cornerstones that guide my reasoning will also be different. These cornerstones are goals and attitudes that provide fixed points for practical reasoning as a department member or a member of the football team. As a philosophy department, we might have jointly accepted that Katherine is a good job candidate and, in my reasoning as part of the department, this acceptance is held fixed to guide practical reasoning. But as part of the football team, nothing requires me to hold that acceptance fixed to guide my reasoning. The football team is not interested in good job candidates for a philosophy department. There has not been any agreement on job applicants at all for the football team. My practical reasoning has different cornerstones depending on what layer of agency I take on board: the philosophy department layer or the football team layer. I can switch between these different kinds of agencies with their individual cornerstones depending on what the situation calls for. I suggest that the we-mode layer of agency is the basic building block of social groups and collective mental states. The we-mode layer itself is only dependent on joint commitments and my abilities of submersion and layering of agency. It does not depend on the prior existence of groups or collective mental states. Moreover, the we-mode layer does not entail any form of plural subject as is accepted in other accounts using the notion of joint commitment, such as Gilbert's (1989, 1990, 2009) proposal. A social group is then narrowly defined in terms of agents sharing a we-mode agency with the same cornerstones by joint commitment. A collective mental state is defined in terms of a sufficient ratio of the same we-mode attitudes in relevant group members (where the relevancy is determined by the cornerstones of the shared we-mode agency). I end these introductory remarks with a note of caution: not every detail nor potential problem for the proposed account will be discussed in detail. My aim is to sketch a big picture view of how to think about the we-mode in a new way. It is an exploration of using the idea of layered agency in the social realm and more work will be needed to fully flesh out the proposal. Agents can adopt a we-mode as individuals. Unlike other proposals of a we-mode (e.g., Tuomela (2005, 2013), Schmitz (2017, 2018, 2023), Schmid (2017)), I understand the we-mode as a temporary layer of agency that allows agents to take on disposable goals and means. These goals and means are based on agential cornerstones jointly committed to by multiple agents under conditions of common knowledge. This is the picture I want to spell out in more detail. When I play a game of Super Mario Kart, I adopt a temporary agency that wishes to win the race. I make that temporary agency dominant. I immerse myself within it, but within certain limitations. […] But so long as the background monitoring processes of our full agency haven't broken through with such a cancellation, we let the temporary agency regulate our decisions and dominate our consciousness (Nguyen 2019, 443). For striving play, I look for a particular experience of struggle. It is not the winning itself that is my ultimate goal in playing Super Mario Kart, but the struggle experienced while trying to win. This is the core of striving play in contrast to achievement play, which would be playing for the sake of winning. What makes striving play so interesting is that I cannot aim for the struggle throughout my practical reasoning while I am playing the game. If struggling was my guiding aim while playing, I would intentionally play worse. Nguyen illustrates this with a story: a 10-year-old enjoyed his experience of beating his father in Monopoly so much that the boy ended up offering free cash to his father to prolong the game. The kid missed something essential about playing a game (Nguyen 2019, 437). A player with the kid's reasoning can never wholeheartedly play the game, trying to win. In a sense, it becomes dubious whether a mature adult using this kind of reasoning would be playing at all. The constitutive goal of the game would not be adhered to. To wholeheartedly take part in striving play, the goal set by the game has to be taken on in a way that allows it to temporarily dominate the player's agency. The background goal of experiencing struggle has to be shut out from practical reasoning for a while. I have to fully submerse myself into the goals and means provided by the game. This is a remarkable ability. Not only can I take on temporary goals and means, but I can do so in a way that changes my practical reasoning and my actions fundamentally. In doing so, I create a new layer of agency.3 This is a motivational two-step process. I have the overall goal of experiencing striving play and in order to do that, I need to take on the temporary goals and means in a way that keeps the overall goal itself temporarily out of my practical reasoning. This ability does show itself in other areas as well. Nguyen (2020, 55) gives the example of relaxing and clearing one's head from work stress. I cannot clear my head and relax by willing myself to do that. Instead, I need to submerse myself in some other task that gets my mind away from work. Perhaps going for a walk, cooking, or reading. And I need to do this in a way that temporarily shuts out the overall aim from my current reasoning and action. I have to wholeheartedly focus on the activity. Just like in the case of games, I manage to take on a temporary layer of agency with its goals and means and I let this agency become dominant. I suggest that adopting a we-mode is another instance of this kind of layered agency. What makes the we-mode a particularly interesting instance of layered agency is that the goals and acceptances that constitute the cornerstones for my reasoning in the newly adopted layer of agency are not determined by myself alone (as in the relaxation case), nor are they determined solely from the outside (as typically in the games case). The cornerstones are determined jointly with other people. They are established via joint commitments under common knowledge. I temporarily seal myself off from my ordinary interests and ends in order to make the new joint commitments dominant in both my reasoning and motivational structure. I do that in order to achieve some overall, individual aim which I can best (and sometimes only) achieve if I take these new, jointly established goals and means on board. I need to fully submerse myself in that new layer of agency with its unique structure of practical reasoning in order to follow my individual goal successfully. Just like in the case of playing a game, it is a motivational two-step process. I start with a usual, individual I-mode goal that motivates the creation of a new layer of agency. In the we-mode case, it is an I-mode goal that I cannot (or not easily) fulfill by myself. This I-mode goal leads me then to joint commitments together with other people that establish a particular we-mode. I take on this we-mode as a new layer of agency and make it dominant. I reason and act in the new layer of agency based primarily on the goals and means jointly committed to. My common I-mode goals and reasons still exist in the background, but they are temporarily shut out and only show up again when the newly adopted layer of agency becomes questionable—either in its functionality or with some unpalatable consequences, or perhaps simply because something more important needs to be taken care of. When I take on the we-mode layer of agency, my action is temporarily guided only by the new goals and means jointly committed to, plus my individual representations of the world (as long as they are not incompatible with the joint commitments). Take Gilbert's (1990) example of two people going for a walk together. I cannot achieve the goal of going for a walk together with someone else just by myself. Moreover, I need the other person to cooperate in a way that is more than just both of us wanting to go for a walk at the same time at the same place. It requires particular commitments to each other that set up tacit rules and expectations. For instance, going for a walk together establishes normative demands that no person just randomly runs off without notice or explanation. In order to go for a walk together, we jointly commit to do so. Going beyond Gilbert, I take this to be well described by each of us adopting a temporary layer of agency of us walking together as a unit. And in contrast to Gilbert's proposal, my model does not entail any sort of plural subject that is formed in virtue of those joint commitments. By jointly committing to go for a walk together we each seal off our individual agencies and set some cornerstones for each of our new we-mode agencies, but these new we-mode agencies are still located within each of us individually. There are individual subjects who act and reason in a we-mode agency based on the set cornerstones, but there is no plural subject in anything like Gilbert's sense. The normative demands of the social practice of going for a walk together are such cornerstones, which each of us individually now takes on board as guides for practical reasoning. For instance, each of us accepts not to leave without voicing the plan to leave and the reasons for doing so. And each of us accepts not to move too fast or too slow compared to the other person. These are examples of cornerstones that now shape the individual we-mode agency for each of us. My actions are determined by my practical reasoning based on these cornerstones and my beliefs about the world (that are compatible with the cornerstones jointly committed to). This idea of one's reasoning and one's actions being partially determined by cornerstones and partially by personal beliefs also fits well with an observation made by Rovane (2002): Joint endeavors do not require that all involved parties achieve overall rational unity, but rather rational unity to a sufficient degree. With "rational unity," Rovane refers to a rational equilibrium of all-things-considered judgments by an agent. An agent has overall rational unity when their all-things-considered judgments have the least contradictions and conflicts possible. It is a maximally rational, maximally coherent set of such judgments. This includes judgments about the world and evaluations of actions available to the agent. Part of rational unity is to choose those actions that fit best with the agent's psychological attitudes and to judge those actions as appropriate for the agent. Rovane takes overall rational unity to be a normative demand on individual agents (Rovane 2002, 215). Agents ought to show overall rational unity—they ought to resolve contradictions and conflicts within their judgments. For group agency and the we-mode, rational unity comes into play as a notion that also applies to people acting and reasoning together. In order to reason and act together, the involved parties need to show some rational unity. Group members cannot hold wildly contradicting beliefs and desires if they want to act and reason together. However, Rovane observes that rational unity to a degree can be sufficient for joint deliberation and action—group members do not need to be perfectly aligned in all judgments. Let me show this with her example of two undergraduate students working together to answer a philosophical question (Rovane 2002, 218–219). In order to successfully do that, they have to build on what each one of them says, work out consequences, and so on. This can only be done if they behave as if they have only one rational point of view with respect to their task. They need to work with the same set of base assumptions and the same goal guiding their joint deliberation. However, this does not require that both students need to share all goals, desires, and beliefs, nor that the students cannot have any contradicting beliefs or desires. They only need to share things that pertain to their joint endeavor. For instance, both need the same goal of answering the philosophical question and they need to agree on a methodology (among other things). But it does not matter if one wants to do so because they are genuinely curious and the other one has a completely different motivation (Rovane 2002, 219). Similarly, two people walking together might want to do so for different reasons, but they need to align to a sufficient degree with respect to the desire to walk together and in (at least some of) their beliefs what walking together entails. Only then they can form a rational point of view together. This is exactly what happens when they commit to particular cornerstones for the we-mode. Those cornerstones capture the jointly accepted point of view and establish a sufficient degree of rational unity between the involved parties. There are many beliefs and desires that each individual party has besides those cornerstones, but for a we-mode, a degree of rational unity is necessary and that rational unity is established insofar as all parties take on the cornerstones as fixed points for reasoning and acting. Those fixed points then determine which other beliefs held by individuals can also be used to aid judgments in the we-mode. Different joint endeavors require different degrees of unity between the involved parties and hence more or less cornerstones that shape the we-mode. Hence, one might say—as Rovane (2002, 221) does—that the joint endeavor imposes normative requirements on the people involved. Some complex tasks require all parties to take on many cornerstones, others only require a few cornerstones to be done successfully. Some tasks and cornerstones are rather general, others specific. We-mode agency is shaped by the task at hand that prompts people to join forces. The proposal of a we-mode as a form of layered agency with cornerstones can shed light on how easily such we-modes can be established, how they relate to I-mode agency and how they fit into the broader picture of our ability to shift between agencies more generally. Layered agency shares some characteristics with Tuomela's (1992, 2013) proposal of a we-mode with positional beliefs (or positional views)—the beliefs (views) that a position-holder has qua a position-holder. My layer of agency proposal and Tuomela's idea of a we-mode approach share that an individual can reason and act from different points of view: their own personal point of view, and a point of view jointly established with other people. In contrast to Tuomela, I am reluctant to put this jointly established point of view in terms of group membership. I want to use the jointly established layer of agency to explain social groups, rather than explaining a we-mode with reference to groups. Tuomela understands the we-mode as including the agents intentionally functioning together as a group for the same authoritative group reason, satisfying the criterion of collective commitment and a collectivity condition (Tuomela 2013, 23–24). This collectivity condition requires that the group members stand or fall together in we-mode activities and attitudes. For Tuomela, a group member cannot satisfy a we-mode goal or attitude if that goal or attitude is not satisfied for all other group members as well. He takes that to be a constitutive principle of the we-mode (Tuomela 2013, 29). My layer of agency model for we-modes does not accept a general collectivity condition. An agent can have we-mode goals and attitudes that can be satisfied only for the agent, as long as they are not cornerstones for the layer of agency. The primary reason for this is that individual agents operating in a we-mode layer of agency can form new goals and attitudes based on the jointly established cornerstones. These newly formed goals and attitudes are still in the we-mode, as they operate fully within the temporary layer of agency. But they need not be shared between everyone who jointly committed to the cornerstones and took on a we-mode layer of agency. In such a case, one Agent A might have a we-mode goal that is not present in another Agent B, and therefore A can satisfy that goal without any such goal satisfaction taking place for B. I might jointly commit with my football teammates on overall aims and strategies for the football game. When on the pitch, I act and reason within the we-mode layer of agency established by such joint commitments. Nevertheless, I can form intentions and goals individually in that we-mode while playing football. I can form a we-mode goal to pass the ball and I can satisfy that goal by acting accordingly. But there might not be a corresponding we-mode goal present in my teammates. They do not have the goal of me passing the ball—neither as an I-mode goal, nor as a we-mode goal. My teammates' we-mode layer of agency has the same cornerstones as mine, but lacks the relevant non-cornerstone goal. There is nothing to be satisfied for them. However, it is still a goal of mine in the we-mode, because it was formed in the we-mode layer of agency. That is a difference between Tuomela's understanding of a we-mode and my layer of agency proposal. Tuomela takes we-mode goals to only be those that are established jointly, either directly by joint commitment or indirectly as a result of joint commitments on positions within a group. In contrast, I take we-mode goals to be any goals present in a layer of agency with cornerstones that were established in a joint commitment.4 Even though I reject the strong collectivity condition, I accept a weaker version. A weakened version holds that a we-mode goal can only be satisfied if the goal is also satisfied for all other group members who have the same we-mode goal. The weaker version respects that we-mode goals can differ between individual members and come into conflict. Forming goals and intentions from the group's point of view is no guarantee that no such conflicts show up. They regularly do. Many intradepartmental conflicts stem from parties forming incompatible goals while nevertheless having the group's best interest in mind. These conflicts constitute practical challenges that prompt some form of conflict management, bargaining, and joint deliberation. A strong collectivity condition as a necessary condition for we-mode attitudes cannot capture these nonideal intragroup conflicts. A weaker version recognizes these conflicts, but nevertheless captures that we-mode goals that have exactly the same conditions of satisfaction will all be fulfilled at the same time. Two we-mode goals with the same conditions of satisfaction cannot differ in their satisfaction. If two agents have the same we-mode goal, it is impossible for only one of them to achieve their we-mode goal. The model of layered agency has clear advantages over Tuomela's model of a we-mode with positional beliefs. The layered agency model is independently applicable to behavior that has nothing to do with groups or social positions. Remember Nguyen's (2020) example of clearing one's head by taking a walk and trying to relax, and the central role of layered agency for the ability for striving play. It is a distinct advantage of the layered agency proposal that the same capacity can explain Nguyen's examples and also the we-mode. Positional beliefs (or views) on the other hand cannot do that. Even with the most charitable interpretation, they do not fit the example of clearing one's head. Moreover, layered agency provides a good explanatory story for how, when, and why one switches between different agencies. I can submerse myself in a temporary layer of agency while in the background, my I-mode agency is still present—sensitive to situational factors that demand me to abandon the temporary layer of agency if I am confronted with something of higher priority. The temporary layer of agency is taken on in order to fulfill a goal that could not (or not easily) be achieved otherwise. The whole temporary layer of agency is justified in the broader agency of the agent. When that is no longer the case, or when the costs of staying within the temporary agency become too high, I abandon it. Moreover, I can regularly check whether that is the case. There will be instances at which I ask myself whether I still want to act with the jointly committed goals and means in mind. This is also something that has parallels outside of the we-mode or any form of cooperation. When I play a game, a situation might come up in which I step back and ask myself whether I want to keep playing—whether it is still fun. Or when I am trying to clear my head by taking a walk, I might stop to ask whether the walk is actually working and helps me clear my head. In all these situations I step back from the layer of agency I took on temporarily, ask whether reasoning and acting within that layer is still promising for my overarching goals, and then decide whether to abandon that layer of agency or submerse myself back into it.5 I monitor—either consciously or unconsciously—whether I should stay in a particular layer of agency. This is exactly what happens when I act under joint commitments with other people as well. There will always be a normative demand to keep acting according to the joint commitment, but that demand can be overridden in favor of leaving the we-mode layer of agency. Again, layered agency captures more than positional beliefs (or views) can. The we-mode is just one particular form of layered agency—one based on joint commitments. I suggest that the we-mode is a form of layered agency with cornerstones determined by joint commitments. However, one may wonder whether it is necessary to rely on the joint commitment idea to get my proposal of the ground. One might think that something less demanding, perhaps akin to Bratman's (1993) proposal for shared intentions can be sufficient. His account is stated as follows: Shared intentions are here understood as interlocking intentions, meshing subplans, and common knowledge. In this framework, shared intentions are explained on an individualistic basis. There is nothing fundamentally collective in the same way that joint commitments are. Can I use a similar, purely individualist basis for my proposal? Perhaps interlocking mental states akin to Bratman's account determine the cornerstones for the we-mode layer of agency. If so, then the proposal functions just the same, except it loses the strong collective nature established with the reference to joint commitments. One might argue that such an account becomes more parsimonious and ought to be preferred. It might even be a better fit with the primarily individualist psychological explanation used to describe an agent's action in the we-mode. Unfortunately, such an account will not do the trick. Joint commitments come with personal cornerstones in the we-mode layer of agency for all parties. But at the same time, joint commitments are more than that. They perform double duty as personal cornerstones in we-reasoning and as a normative constraint on the parties that jointly commit to ϕ. By jointly committing to ϕ, the parties take on a we-mode layer of agency and are committed to reason and act within that layer of agency. If someone does not reason or act in a way that is within that layer and fits with its cornerstones, then that person fails to fulfill the joint commitment and is therefore blameworthy. For instance, when we jointly commit to going for a walk together but I end up running away for personal reasons incompatible with the cornerstones, then I fail to fulfill my commitment because I act from I-mode reasoning rather than we-mode reasoning.6 This explains the normativity that comes from joint commitments observed by Gilbert (1989, 1990, 2017). A similar account that establishes the we-mode layer in virtue of a Bratman-like individualistic ground has difficulties capturing this normative component. This is at the core of the disagreement between Gilbert and Bratman. I find Gilbert's (2008, 2009) arguments convincing, showing that personal intention as proposed by Bratman will not be enough to generate the right sort of normativity that is involved in joint action. Merely pointing to a demand for stability of intentions, as done in Bratman (1993), is not enough. The distinct normativity present in groups and joint action is best explained by a joint commitment. However, it is worth emphasizing here again that endorsing joint commitments and giving them an important role in the model does not entail any sort of plural subject as is usually associated with Gilbert's concept of joint commitments (1989, 1990, 2017). [E]ach of two or more people openly expresses his readiness jointly to commit them all in a certain way, and their having made these expressions is common knowledge between the parties. By this I mean, roughly, that the expressions are entirely out in the open between them, and each knows this. (Gilbert 2017, 131) Agents take on a we-mode layer of agency individually based on joint commitments. They do not form a single body in the strong sense suggested by Gilbert. However, they are acting "as one" insofar as they all now have the same cornerstones for practical reasoning in the we-mode. They all act roughly with the same layer of agency. They are only roughly the same layer of agency because the cornerstones, that have been jointly committed to, leave gaps. Every individual can still have their own representations of the world (and their own subgoals) as long as these representations (and subgoals) do not conflict with any cornerstone jointly committed to. Two people might jointly commit to going for a walk together, but they can still differ in their knowledge of the outside temperature and their beliefs on whether to wear a jacket. The joint commitment is quiet on these issues and the parties can make individual judgments in the we-mode that differ with regard to what is best to wear for both of them walking together. On the other hand, for issues that are covered or practically entailed by the joint commitments such a difference is not possible. If they jointly committed to walking on Princes Street, none of the parties can practically reason with the belief (or acceptance) that it would be best to walk on Rose Street instead. The paradigmatic story for we-mode attitudes goes like this: several agents have a primary goal that each of them cannot reach by themselves. They decide to work together and jointly commit to that primary goal. In order to reach the primary goal, they further jointly commit to a set of acceptances and secondary goals that they agree to as cornerstones for practical reasoning. Each involved party thereby takes on a we-mode layer of agency with these cornerstones as fixed points in their individual reasoning.9 The agents can jointly commit to changes to the cornerstones if needed and their individual we-mode reasoning and we-mode attitudes have to adapt based on changes to cornerstones. They might start with a joint commitment to accept p, but find out later that accepting p as a cornerstone is problematic and hinders them from aiming for the goal they jointly committed to. In this case, they can jointly commit to amend the prior commitment and now accept not-p. All the previous attitudes that were based on the cornerstone acceptance that p now have to change relative to the adjustment in the cornerstone. Moreover, the we-mode reasoning and we-mode attitudes also have to adapt based on changes to the individual's representation of the world. Suppose I am playing football with my team. We have as a football team a bunch of joint commitments on playing football in general and on the strategy in a particular game. My current we-mode cornerstones about our strategy plus my belief that there is no opposing player in front of me might lead my practical reasoning to judge that I should move forward.