Groups and Second-Person Competence

1. Second-Personal Reasons and Second-Person Competence

Second-personal reasons constitute a distinct class of reasons to act. Following Stephen Darwall (2009, 5f), consider three possible reasons for removing your foot from mine:

Reason 1. You wouldn’t want other people to step on your own toes, and you want to treat people the way you want to be treated.

Reason 2. I demand that you remove your foot, and you want to fulfill that demand.

Reason 3. Removing your foot would reduce the total amount of suffering in the world, and you want to reduce the total amount of suffering in the world.

Any of these gives you a reason to remove your foot from mine, but there are important differences among them. Reason 1 is first-personal and depends mainly on your relation to yourself. You can thus arrive at Reason 1 through self-reflection. Reason 3 is third-personal and depends mainly on your relation to the world. Even if your desire to reduce the total amount of suffering in the world can be brought about through self-reflection, you can only arrive at Reason 3 by taking into account some impersonal facts about the world, i.e., that the current position of your foot causes a token of pain. Reason 2 is second-personal and depends mainly on your relation to me. It is in principle unavailable to creatures that do not acknowledge, even if only tacitly, that other creatures have a claim to a special kind of treatment from them.

There are several accounts of the nature of second-personal reasons and obligations (for helpful overviews and discussion, see Schaab 2023; S. C. May 2015; Lewis 2024). For present purposes, I’ll rely on an interest-based account of second-personal reasons. According to this view, such reasons are grounded in how our actions affect others’ interests. As R. Jay Wallace writes

Interpersonal morality (…) might be thought of as a set of requirements that reflect the fundamental insight that we share a world with other individuals whose interests are in some sense neither more nor less important than our own. (2019, 3)

The Danish philosopher K.E. Løgstrup makes a similar point:

An individual never has something to do with another human being without holding something of that person’s life in their hands. It can be a very small matter, a passing mood, a dampening or quickening of spirit, a disgust one deepens or takes away. But it may also be of tremendous significance, so that it is simply up to the individual whether the other person’s life flourishes or not. (…) [W]e are each other’s world, and each other’s fate. (1956/2020b, 15f)

In short, the fact that other people depend on me and my actions provides me with second-personal reasons to give them a special kind of treatment.

I take it that this treatment has two features. First, second-personal obligations are directed or bipolar in the sense that they have two distinct poles – obligor and obligee – that are connected in a normative nexus (e.g. Thompson 2004; S. C. May 2015; Wallace 2019). Rather than being obligations period, second-personal obligations are obligations that particular people owe to other particular people. Second, we are obligated not just to perform certain actions but also to perform these actions with a specific attitude. As Wallace writes: “It matters to the question of whether A has wronged B with what attitudes A acted. (…) [T]o flout a requirement is to act with a distinctive attitude toward it, one of knowing and even open disregard” (2019). Although the actions that are prescribed by a second-personal obligation sometimes coincides with the actions prescribed by moral first- or third-personal reasoning (à la Reason 1 and Reason 3), we still hold each other accountable for the underlying reasoning. Suppose I demand that you remove your foot and that you then do so, but you later tell me that you initially did not know what to do until you remembered some abstract moral principle that you learned in your night class in moral philosophy. Intuitively, I am slighted by the fact that you needed to search for a reason to remove your foot when I had already given you one. As Løgstrup puts this point, first- and third-personal reasons to act only appear in moral life if we have already “become indifferent to some of the requirements and challenges that in a very immediate way arise from the relationships we have to other people” (1971/2020a, 27).

Second-person competence requires different agential capacities than our susceptibility to first- or third-person reasons. For example, we might follow Kant in saying that susceptibility to legitimate first-person reasons requires the capacity of rational agents to make judgments and decisions in accordance with universal laws, but this capacity is clearly not sufficient for second-person competence. Instead, I suggest that we understand second-person competence in the following way:

A has second-person competence if and only if

1. A is able to sympathize with others;

2. A is able to act upon their sympathy;

3. A is able to assess the moral worth of their actions and inactions by way of self-reactive attitudes.

Let me unpack this. First, second-person competence requires that we understand and are immediately motivated by the plight of others. Darwall calls this empathy, understood as a “simulation or imaginative projection into the other’s point of view” (2009, 45), and Løgstrup often calls it “care” (e.g. 1956/2020b, 18). Both terms are contested, however, and for present purposes it is important not to overcommit our account of second-person competence to unnecessarily demanding agential capacities, because this leads to problems downstream where our theory of group accountability risks becoming unnecessarily restrictive.⁴

In the following, I will use the term ‘sympathy’ to refer to the relevant interpersonal relations.⁵ I do this because sympathy has many of the same connotations as ‘empathy’ and ‘care’, but is less entangled in on-going technical debates. This minimizes the risk of overcommitment. An added benefit is that ‘sympathy’ has a rich history for being classified as an important interpersonal moral emotion in, e.g., Smith and Hume.

So, by definition,

A sympathizes with B if and only if

1. A has an approximate understanding of what it is like for B to be in B’s situation;

2. A has emotional and conative attitudes towards B that express a disposition aimed at promoting B’s well-being.

To take a trivial example: A understands that B is in pain (thereby satisfying condition (1)) and responds emotionally with distress at B’s pain and conatively with an inclination to help B (thereby satisfying condition (2)). A thereby sympathizes with B.

Turning briefly to the second condition of second-person competence, although sympathy has a conative component (often in the form of an urge to help), it is nonetheless possible to sympathize with someone without being able to act upon that sympathy. Suppose, for instance, that a friend and I have been kidnapped by terrorists. The terrorists tie us up, and I am forced to watch as they torture my friend. I sympathize with my friend in the sense that I feel distress at his suffering and have a strong urge to help. But as I am tied to a chair, there is nothing I can do. In this case, I am clearly able to understand second-personal claims by way of sympathy, but I am unable to act upon them. Following the ‘ought implies can’-principle, I am hardly fit to be held accountable for not helping my friend. Therefore, second-person competence requires not just sympathy but also the ability act upon that sympathy.

Finally, second-person competence requires the ability to assess the moral worth of one’s actions and inactions by way of self-reactive attitudes. That is to say, second-personally competent agents must be seen as loci of accountability not just by others but also by themselves. As Darwall (2009, 79) puts it, “we can intelligibly address demands through reactive attitudes only to those we assume able to take the very same attitudes toward themselves.” Levinas makes a similar point when he claims that “the welcoming of the other is ipso facto the consciousness of my own injustice” as manifested in “shame” (1961/2012, 86).⁶ Formulated differently, second-person competence requires an awareness that the second-personal reasons imposed upon us have normative force.

Some might object that this third condition is too demanding and argue that even if some kind of moral assessment is indeed necessary, we can assess the moral worth of our actions by other means than reactive attitudes. There are two reasons to assume that second-person competence requires self-reactive attitudes. First, as noted, this assumption is widespread in the literature on second-personal obligations. Second, and more importantly, it seems plausible that there is a tight connection between our capacity to sympathize with others and our capacity to have self-reactive attitudes. Sympathy requires an emotional concern for the other person, and it seems plausible that self-reactive attitudes such as guilt and remorse are rational projections of such emotional concern about the second person. I’ll draw upon this connection between sympathy and self-reactive attitudes in Section 4, but my argument does not rely on there being a strict conceptual connection between the capacity for sympathizing with others and for having self-reactive attitudes.

2. Joint Capacities

I have suggested that a group must have the capacities for sympathy, acting on sympathy, and self-reactive attitudes to be fit to be held accountable. In Sections 3–5, I discuss each of these capacities in more detail. First, however, I’ll focus on the more general point concerning what it is for groups to have capacities.

Like other types of dispositions, it is notoriously difficult to give a precise account of what a capacity is. A plausible necessary but not sufficient condition is the following:

The Manifestation Requirement: X has the capacity to Φ only if X would Φ in a suitably broad range of possible worlds satisfying the manifestation condition C.

This is similar to how we think about dispositions. When we say that a flask is fragile, we mean roughly that the flask would break when struck where the state of being struck is the manifestation condition.

In contrast to dispositions more generally, however, capacities involve agency in the sense that exercised capacities must continually adapt to feedback from the world. This leads to the following requirement:

The Adaptivity Requirement: X exercises their capacity to Φ only when X Φ-s in a way that suitably tracks and responds to relevant changes in the world.

Whereas the flask’s shattering is just a passive effect of its being struck, someone exercises a capacity only if they continually respond to the environment throughout the duration of that exercise (cf. McGeer 2018). For example, to say that someone has the capacity for driving implies both that they are likely to drive under certain manifestation conditions (e.g., they have a car at their disposal and have to run an errand) and that they drive in a way that is suitably adaptive to changes in the world (e.g., they don’t hit people and objects with the car, stop at red lights, etc.).

Now, we often talk about the capacities of groups, e.g., a dyad’s capacity for moving a table or a company’s capacity for filing a law suit. But groups are composite entities consisting of various member components standing in specific relations to each other, and groups can only exercise their capacities by way of the lower-order capacities of their members. For example, a country exercises its capacity for waging war by having its soldiers exercise their lower-order capacities for firing their guns, etc.

How do we distinguish cases where several people exercise their capacities as individuals from those cases where several people exercise their capacities as a group? The details are bound to differ from capacity to capacity in the same way that the manifestation and adaptivity conditions for driving are very different from those of playing the cello, but it seems plausible that a group only has a capacity if the relation between the members is essential to how they satisfy the manifestation and adaptivity requirements.

Let’s start with the manifestation requirement. Suppose that a group G consists of the members M₁, M₂ … M_n who has the capacities for Φ₁-ing, Φ₂-ing … and Φ_n-ing respectively, and that these capacities might but need not be interpreted as lower-order capacities realizing G’s higher-order capacity for Ψ-ing. It is obviously too weak to simply say that G has the capacity to Ψ only if M₁, M₂ … M_n would Φ₁, Φ₂ … and Φ_n in a suitably broad range of possible worlds satisfying C. To see why this is so let Φ₁-ing, Φ₂-ing … and Φ_n-ing refer to isolated instances of gun₁, gun₂, and gun? being fired, and let Ψ-ing refer to the group’s attacking. The fact that several gun owners fire their guns does not necessarily mean that the gun owners constitute a group exercising its capacity for attacking. Intuitively, a group exercises its capacity for Ψ-ing only if group membership makes an explanatory difference to how the component members exercise their lower-order capacities. In such cases, it is warranted to say that the members exercise their lower-order capacities because they are part of the group. More formally, I suggest the following:

The Group Manifestation Requirement: G has the capacity to Ψ only if M₁, M₂ … M_n would exercise their lower-order capacities for Φ₁-ing, Φ₂-ing … and Φ_n-ing in a suitably broad range of possible worlds satisfying C, where M₁, M₂ … M_n Φ₁-s, Φ₂-s … and Φ_n-s because they are part of G.

This might seem puzzling at first since explanations typically track causal relations, and it is unclear how the higher-order capacity of the group can be causally related to the lower-order capacities of the members. But the fact that the lower-order capacities are causally efficacious in producing some outcome (e.g., shooting people) doesn’t necessarily imply that the group-level capacity is explanatorily and causally irrelevant. Jackson and Pettit (1990) have thus suggested that groups can be explanatorily and causally significant in the sense that they can program for an outcome that is then implemented by one or several group members. The group programs for the outcome when the group constitutes a set of conditions that ensures that the outcome is likely to come about even if these conditions themselves are not causally efficacious. As List and Pettit formulate this point:

The group may control for the performance of a certain action by some members, maybe these, maybe those. It does this by maintaining procedures for the formation and enactment of its attitudes, arranging things so that some individuals are identified as the agents to perform a required task and others are identified as possible back-ups. (2011, 163)

However, since they focus on groups with well-functioning decision-making procedures, List and Pettit only accounts for what we might call explicit programming. A group explicitly programs for an outcome when the decision-making procedure of that group ensures that the outcome is likely to be brought about by some of the members of that group. This contrasts with implicit programming, where an informal group dynamic ensures the outcome (Rachar 2024; Shockley 2007). Examples of such informal group dynamics include tacit social norms, group narratives, emotional contagion and other forms of interbodily resonance. Such informal group dynamics are central to much social scientific reasoning, e.g., Durkheim’s influential account of collective effervescence (1912/1964; cf. R. Collins 2004).

Program explanations typically focus on how actions events can be attributed to groups rather than (or in addition to) the causally efficacious members of that group. My suggestion is that the same line of reasoning applies to joint capacities more broadly. Thus, to say that the group members exercise their lower-order capacities “because they are part of G” is similar to saying that the members exercise their lower-order capacities in those particular ways because the organization or interpersonal dynamic of the group programs them to do so.

There is also another important way in which the higher-order capacity of the group can explain the lower-order capacities of the group members. If we take into account the internal perspective of the group members, they will often themselves explain why they exercised their capacities in some particular way by referring to their group membership. Indeed, the informal group dynamic animating LSGs will often program for some kind of plural self-awareness so that the members are tacitly aware that they are part of a group and that they, by exercising their lower-order capacity, contribute to the group’s realizing a higher-order capacity.⁷

This leaves us with the second issue: How can groups satisfy the adaptivity requirement? Like individuals, groups must track and adapt to relevant changes in their environment in order to exercise their capacities. But since groups are composite entities that realize their capacities by way of the lower-order capacities of their members, groups must not only be adaptive externally in terms of how they relate to the environment but also internally in terms of how each member tracks and responds to how the other group members exercise their capacities. To exercise its capacity for attacking, the army must track and respond to the external world (e.g., the terrain and the movements of the enemy forces), but this also requires that the people within the group track and respond to each other (e.g., the soldiers must track and respond to the orders of their superiors, who again must track and respond to the reports of the scouts, etc).⁸

Taking stock, I suggest that in addition to the group manifestation requirement, groups must also satisfy the following requirement to have a capacity:

The Group Adaptivity Requirement: G exercises its capacity to Ψ only when M₁, M₂ … M_n Φ₁-s, Φ₂-s… and Φ_n-s in a way that suitably tracks and responds to (a) how the other group members exercise their lower-order capacities and (b) relevant changes in the world.

In the following, I use the term joint capacities to refer to capacities that satisfy both the group manifestation requirement and the group adaptivity requirement. With this account of joint capacities in place, we can turn to the specific suite of capacities required for second-person competence.⁹

3. Sympathy in Loosely Structured Groups

When do groups sympathize? As an illustration, consider the following case:

Good Samaritans. During an adventure race, a loosely structured group of people witness a child fall into a lake. Each member is only slightly bigger than the child, and it is therefore impossible for any member to rescue the child alone. The members look terrified at each other before they, with an exchange of glances, initiate a group action that ultimately brings the child to safety.

It appears plausible that the good Samaritans act on a second-personal reason. Yet, it is unclear whether they acted on that second-personal reason jointly or as individuals. Initially, these interpretations may seem similar because the higher- and lower-order capacities in Good Samaritans appear functionally equivalent. I’ll argue, however, that the member’s lower-order capacities for sympathizing with the child realize a higher-order capacity of the group exactly because the example satisfies the manifestation and adaptivity requirements outlined above. Let’s go through them one by one.

The group manifestation requirement suggests, first, that the group has the capacity to sympathize only if the members would exercise their respective capacities for sympathizing in a suitable broad range of possible worlds satisfying some manifestation condition. This might be construed as: it is salient that some person is in danger and the group members have the possibility of helping that person if they cooperate. We can easily imagine that this extends across a suitably broad range of possible worlds, e.g., if the person drowning wasn’t a child but a grown-up or if the person wasn’t drowning but stuck under a tree that had fallen over, etc.

The group members must also exercise their lower-order capacities for sympathizing with the child because they are part of the group. It is fairly easy to see how members of corporations and organizations can exercise capacities because they are part of a group. The rational decision making procedure of the group can simply require that the group members exercise their capacities in a specific way and thereby program for a specific pattern of behavior or a specific outcome. LSGs lack the top-down control characteristic of corporations but they can still satisfy this condition through implicit programming. For instance, interbodily resonance or some other informal group dynamic among the good Samaritans might regulate and constrain how they would exercise their lower-order capacities for sympathizing. If that is the case, we might say that the group members sympathized because they were part of the group where the implicit program explanation adds important causal and explanatory information.

In addition to this external explanation of why the group members exercised their capacities for sympathizing in the way that they did, it also appears plausible that the Samaritans themselves might explain why and how they sympathized with reference to the group. Although the spontaneous actions that are typically demanded of LSGs rarely allow that the group members explicitly form a joint or collective commitment to pursue some goal, the members might still understand their own responses to the situation from a group or we-perspective. This would be the case if each of the members (i) has a particular attitude (e.g., a sense of urgency to rescue the child), (ii) believes that this attitude is wide-spread in the group, and (iii) believes or is disposed to believe that there is mutual awareness within the group that this attitude is wide-spread (cf. Tuomela 2007, 66). In this case, each of the members would be justified in thinking to themselves that “we feel a sense of urgency to rescue the child.”¹⁰ They might then take this joint attitude as their practical point of departure in figuring out how they should act. Indeed, in many cases, this kind of we-sympathizing is likely to be “socially grounded” in the sense that these attitudes are part of what the group implicitly programs, i.e., that the members have the attitude in question partly because of the informal group dynamic (cf. Tuomela 2007, 66).

Taking stock, the good Samaritans would satisfy the group manifestation requirement if they sympathize under the relevant manifestation conditions and do so because of their group membership. This occurs when a group dynamic programs them to sympathize in that way and they understand their own responses to the situation from a group or we-perspective.

The group must also satisfy the adaptivity requirement, according to which the group members must track and respond to both each other and changes in the world in the right way. In terms of how the Samaritans respond to the world, we would, on the one hand, not say that the Samaritans sympathized together if they had very different impressions of what the situation called for. For example, we can imagine that some Samaritans took the child to be drowning, while other Samaritans took the child’s falling into the lake to be part of a role-playing scenario. On the other hand, it seems too strong to require that the Samaritans have the exact same type of attitude. We can thus imagine that some group members panic while others remain cool-headed. The group members can thus sympathize jointly even if their emotional responses diverge. A plausible suggestion is that the members must share the same concern, where concern refers roughly to the “desires and aversions as well as attachments, interests, and cares that ground our emotional appraises and responses” (Salmela 2022, 59; cf. Roberts 2003; Helm 2001). The concern, I suggest, can be shared in the sense that the members are programmed to have that concern in such a way that this is out in the open for the group members. When exercising their capacities for sympathy on the basis of the same concern, the group as a whole adapts to the circumstances in a way that is sufficiently similar to how we imagine an individual would have acted in similar circumstances. As we might say with reference to Gilbert, the group would thereby sympathize “as a single body.”

Sympathizing as a single body does not only require that the members respond in a largely synchronous way to their environment. This external adaptivity is closely related to what we might call internal adaptivity. Internal adaptivity refers to how the members track and respond to each other, and given the fact that divergence in their responses to the situation are likely to arise, internal adaptivity plays an important role in ensuring that the difference between their responses does not grow too large. Imagine, for instance, that all members have a clear sight of the child that has fallen into the lake, but whereas one watch in horror, another laugh heartily. In normal cases, there would arise a kind of affective pressure between them to adapt their responses to each other. The horror that one member sees in the face of the other causes them to reconsider their initial understanding of the situation as a role-playing scenario. This internal adaptivity is often facilitated by the group members’ seeing themselves as group members. The pressure for some to reconsider their understanding of the scenario might thus arise from their (implicit) sense that their way of sympathizing with the child diverges from the way of sympathizing with the child that is widespread within the group. Each member can thus find themselves out of sync with the group perspective in a way that causes them to change their attitude (cf. Schmid 2014).

Summarizing, I have argued that LSGs can satisfy the manifestation and adaptivity requirements, and that this would give them a joint capacity for sympathy. It follows that we can attribute a concern about someone’s well-being to an LSG even if this is realized by way of the lower-order capacities of the group members.

4. Acting on Sympathy in Loosely Structured Groups

Before turning to the thorny issue of reactive attitudes in loosely structured groups, a brief comment about the second condition of second-person competence is in order. As noted above, LSGs must not only be able to sympathize jointly but also to act upon that sympathy. Fortunately, it is uncontroversial in the literature on joint action that LSGs can act jointly in this sense. This means that once we grant that LSGs can have joint sympathy, there is in principle no reason why LSGs should not be able to act jointly upon that sympathy. Depending on your preferred account of joint action, you might say that when the Samaritans in Good Samaritans exchange glances this establishes a joint commitment (Gilbert 2013) or plural self-awareness (Schmid 2014) or shows that the Samaritans act in the we-mode (Tuomela 2013) or pursue a shared goal, have meshing subplans, and are mutually responsive to each other (Bratman 2014).

But it is not enough for second-person competence that the group is capable of acting jointly. As I specified above, second-person competence also requires that an agent is capable of acting for the right kind of reason. Phrased in quasi-Kantian terms, the LSG should be able to act in light of rather than in accordance with second-personal reasons. Fortunately, my account can accommodate this. Consider:

Opportunistic Samaritans. This case parallels Good Samaritans, but differs in two respects. First, a judge follows the group around in order to assess its team spirit and morality and determine which of the groups participating in the adventure race should get a prize. Second, when the child falls into the lake, the members initially look terrified at each other, but then, after a quick exchange of glances, a smug grin spreads across their faces as they realize that this is the perfect opportunity to impress the judge. The group then initiates a group action that ultimately brings the child to safety.

By stipulation, this group has the sympathic capacities of understanding how its action might impact the drowning child, and it could have acted upon that understanding. In this sense, the group had a second-personal reason to act. Yet, the group disregarded that reason and acted instead out of its own self-interest. By acting out of vanity rather than concern for the child, the group acts for the wrong reason. Of course, it is to be preferred that the group saves the child out of vanity rather than leaving the child to drown, but this does not change the fact that the group’s reason for acting is of the wrong kind and that the group thereby slights the child. If this is so, LSGs could be held accountable for acting in light of rather than just in accordance with second-personal reasons.

5. Self-Reactive Attitudes in Loosely Structured Groups

Having established that LSGs can sympathize with others and act in light of that sympathy, we can finally turn to the third condition of second-person competence, namely, the capacity to assess the moral worth of one’s actions and inactions by way of self-reactive attitudes.

There has been some recent discussion of the conditions under which it is appropriate to attribute reactive attitudes to groups. Gunnar Björnsson and Kendy Hess (2017, 274) have argued that groups can have states that are “sufficiently similar” to the reactive attitudes by way of their group-level commitments to certain values and beliefs, while Stephanie Collins (2022; 2023) and Frank Hindriks (2018) have argued that group-level reactive attitudes can sometimes be realized by way of the feelings of the group members. These authors have, however, focused on groups with decision-making procedures, operating under the assumption that reactive attitudes can only be attributed to a group if those reactive attitudes are subject to the top-down control enabled by such procedures. I now propose that LSGs can have (self-)reactive attitude that are realized by their members’ lower-order capacities for having reactive attitudes if the manifestation and adaptivity requirements are satisfied in the right ways.

Starting with the manifestation requirement, the group dynamics of LSGs can implicitly program for the member’s having specific reactive attitudes and thus play a causal role in determining the actions and feelings of the members. Plausibly, this is why it’s natural for us to say that a LSG feels in some way (e.g., “the crowd was excited”) or to say of a group member that they feel something because they are part of a LSG (e.g., “John could feel the excitement of the crowd”). Similarly, group members themselves may explain the fact that they act or feel in certain ways by referring to their group membership. For example, if we imagine that the group of good Samaritans failed to rescue the drowning child, we can imagine the good Samaritans feeling guilty as part of (and, as we shall see, on behalf of) the group because they, as a group, failed. Certain reactive attitudes are thus a function of how we conceive of ourselves as being members of a group. This suggests that group membership is sometimes part of the manifestation condition of certain reactive attitudes.

Yet, the fact that a group dynamic can program for reactive attitudes does not by itself imply that we should attribute those reactive attitudes to the group. In addition, the reactive attitude in question must be appropriately adaptive. To spell out this adaptivity, we have to look closer at the intentional structure of the reactive attitudes. Following Bennett Helm (2017; 2018; 2020), we can analyze the reactive attitudes in terms of their target, formal object, and focus. The target of an emotion is the entity that the emotion is directed towards and which is somehow evaluated in that emotion. The formal object is the distinct way in which the emotion evaluates its target. Finally, the focus of the emotion is the background object of concern that explains why the subject of the emotion evaluates the target in terms of that formal object. For example, in fearing a spoon of tahini, the tahini is the target of my fear, harmfulness is the formal object, while my son, who is wildly allergic to sesame, is my focus.

Importantly, on Helm’s account, emotions are not isolated affective episodes but parts of larger emotional patterns that connect them to other emotions, mental states, and dispositions to act. This means, he argues, that emotions are projectible and rational. They are projectible because a subject’s commitment to a particular focus extends to both future situations and counterfactual scenarios. They are rational because we can determine what someone ought to feel based on what holds import for them.

This account of the reactive attitudes helps us spell out how exactly the group members must track and respond to each other and relevant changes in the world. In order to be self-reactive, the intentional object of the reactive attitude must be the plurality of agents that, due to their specific group dynamic, constitute the LSG. Importantly, this does not mean that the reactive attitude targets the members as an aggregation of independent individuals. Rather, it targets the members under a specific description, namely, as a plurality of individuals possessing and exercising joint capacities.

The members’ reactive attitude must also align with the emotional pattern characteristic of the LSG. This means that the members’ self-reactive attitude must be focused on the second person who gave the LSG a second-personal reason to act.¹¹ To put this point differently, the members’ self-reactive attitude must arise out of a concern for the second person in the same way that the LSG is normatively expected to act out of concern for the second person as this is established in the group’s act of sympathy with that person. There is, in other words, a connection between the LSG’s sympathic concern for the second person insofar as the self-reactive attitude of the members are rational projections of their joint second-personal reason to act.

At this point, it is important to note that two emotional patterns can have the same target and focus without thereby being one or, more precisely, without being connected by way of rational projections. Conceived as an individual, each member can be rationally expected to feel regret that the second person was not saved in much the same way that a mere spectator would, since they may not have been in a position to make a difference as an individual. But as group members, the members can be rationally expected to feel guilt rather than regret because they, as a group, had the capacity to act on the second-personal reason. Now, guilt and regret have the same target and focus – the LSG and the second person, respectively – but they have different formal objects and are part of distinct emotional patterns. One is appropriate from the perspective of the group; the other is appropriate from the perspective of the individual. It seems that the LSG only has the higher-order capacity for having a reactive attitude if the members’ lower-order reactive attitude (which, recall, is supposed to realize that higher-order capacity) is a rational projection of the emotional pattern established in the LSG’s act of sympathy rather than any personal emotional pattern that happens to share focus and target with that of the LSG.

This also implies that the members must have the appropriate reactive attitudes from the point of view of the group so that the members are aware that the emotion is an appropriate expression of the emotional pattern of the LSG as a group and that they have the emotion on behalf of the group. This is important because it is conceivable that a member feels guilt that the group failed to act on a second-personal reason out of concern for the second person as an individual. In this case, the member would be aware only of himself as feeling guilt not on behalf of the group but only by virtue of some fact about his individual agency. This is psychologically possible, but it would also be irrational since feelings of remorse are only appropriate when they involve self-awareness that one has the capacities that would have enabled one to act in the right way. In our case, the relevant kind of self-awareness would be a plural self-awareness, since the members can only satisfy the second-person claim laid upon them jointly.

In this section, I have argued that LSGs have the capacity for self-reactive attitudes necessary for second-person competence if a group dynamic implicitly programs the members of the LSG to have a reactive attitude. Moreover, I have argued that the reactive attitude of the members must be appropriately adaptive in the sense that (i) the intentional object of the attitude must be the LSG itself, (ii) the attitude must be focused on the second person, (iii) the attitude must a rational projection of the emotional pattern established in the LSG’s act of sympathy, and (iv) the members must have the reactive attitude on behalf of the group.

6. Conclusion

I have argued that groups can have second-person competence over and above their members if and only if the groups have capacities for sympathy, for acting on that sympathy, and for self-reactive attitudes. Moreover, I have shown that loosely structured groups can realize these capacities through their members’ lower-order capacities when these satisfy the manifestation and adaptivity requirements.

Two limitations of my account are worth mentioning. First, I have focused only on heavily idealized cases, where the group members act and feel in similar or compatible ways. Yet, due to the decentralized nature of LSGs, we can easily imagine cases where the group members do not act as a single body. To address this issue, we would need to develop a theory of non-ideal group agency capable of modelling the degree to which groups possess and exercise agential capacities, and we would need to consider how such degrees of agential capacities are or should be reflected in our blaming attitudes and practices. This task is beyond the scope of the current paper.

Second, I have argued that there are cases where we would say that LSGs have the capacities required for second-person competence, but I have not directly discussed what this might entail in terms of the praise- and blameworthiness of the groups and its members. A full account of this is also beyond the scope of this paper, but I’ll nevertheless conclude by briefly discussing some of the positions we might adopt on the accountability of LSGs.

Some suggest that groups can be blamed as collective entities, independently of the individuals that comprise them (see, e.g., Feinberg 1968; French 1979; 1984; Isaacs 2011; Pettit 2007). Others think that praise and blame must be wholly distributed among group members so that each member is only evaluated for their own personal contribution to the acts of the group (see Moen 2024 for a recent discussion). Finally, some have suggested that the praise- and blameworthiness of some groups can be understood as evaluative properties that are possessed by the individual group members (rather than the group itself) but only jointly (see Björnsson 2014; 2020; Blomberg and Petersson 2023; Knudsen 2023; Mellor 2024; Schwenkenbecher 2018). Call these the collective, distributive, and joint views, respectively.

The collectivist view is controversial in itself, but even if we grant that it sometimes makes sense to blame groups independently of the members that compose them, it is not particularly promising for thinking about the accountability of LSGs, since it is usually only applied to more robust groups. An underlying reason for this is that the kind of capacities arguably possessed by LSGs are so closely tied to the capacities of the members that a clear separation between the blameworthiness of the group and that of the members becomes difficult to make. The distributive and joint views are, in contrast, both individualist in the sense that they only attribute evaluative properties to the individual group members, yet, they interpret the address of our blame in different ways. On the distributive view, our blame could be broken down to several instances of second-person singular blame (thereby addressing a you, you, and you), while on the joint view, our blame would resist such reduction by addressing its target in the second-person plural (as yous or y’all).

One theory-laden reason for preferring the joint view is that it aligns easily with the account of joint capacities developed above. If their group membership makes an explanatory difference to how we account for the actions and attitudes of the group members, it is plausible that group membership must also play an important role in our blaming practices. Along these lines, Björnsson (2014, 113) suggests that “a group of people share retrospective responsibility and blame for some event insofar as that event has a straightforward normal explanation in terms of the lack of proper concern of that particular group.” The distributive view would, in contrast, require us to ignore the explanatory and causal information essential to the group’s implicit programming and focus solely on causal efficacy.

To give a full account of the accountability of LSGs, we would have to consider in much more detail which view best explains and predicts our moral intuitions. A critical case worth considering would be this: Imagine that in a Good Samaritans-style case, a member does her part of the joint rescue, yet the joint rescue fails for some reason. On the distributive view, the member would be wholly off the hook, and neither blame nor guilt would be appropriate. End of story. The joint view, in contrast, would insist that although the LSG consists of nothing but individuals, these individuals were given a second-personal reason together that they failed to act upon, and that the individual members therefore can be jointly blameworthy for the failed rescue, regardless of their individual contributions. Importantly, this view is compatible with certain forms of ambiguity. For example, a proponent might argue that it is appropriate for the member to feel proud of their own heroic contribution to the failed joint rescue as an individual, while maintaining that it is nevertheless appropriate for them to feel guilty that we, as a group, failed to satisfy the second person’s claim to a special kind of treatment or regard.

It remains a task for the future to uncover exactly how the second-person plural figures in our responsibility practices. Here I have merely argued that groups, even those lacking unifying decision-making procedures, can be second-person competent.¹²