A pair of smugglers in a car are trying to evade the authorities. The passenger is on the phone with a mole. She tells the driver to turn left. He does. Why? To answer this question, we may (as usual) appeal to the reasons of the driver: he turned because his accomplice told him to, and he trusts her. He might answer something like this if asked why he turned. But his relationship with the passenger also allows for an answer in terms of her reasons: the driver turned because of the customs roadblock ahead. The driver may gesture toward this reason unknown to him if, in reaction to the why question, he turns to the passenger and asks, “That’s a good question, why did I turn?”
There is no competition between the two possible answers. Each identifies a point of the action. Each provides a self-standing teleological explanation. The driver’s individual reason explains why he acted and why his turning is an action at all. Like all actions, this one necessarily has a point for its agent. But, in addition, this particular action has a separate point, which the agent ignores (and to which he may or may not subscribe). This is not a point for which the agent performed the action, but a point of the action nonetheless. The passenger’s reason did not motivate the driver to turn, but it played a role in the etiology of the action important enough for it to count as a reason of the action.Footnote 1 The two explanations are self-standing because they have roots in different moments in the etiology of the action. In this respect, they stand in the same relation to each other as explanations in terms of nested elements in an individual plan. “To boil water” and “to cook dinner” each provide a self-standing teleological explanation for turning on the stove. But, of course, what singles out the present example is that the driver’s and the passenger’s reasons are not nested elements in an individual plan. The second explanation does not appeal to further reasons of the agent herself. It is a social explanation, rather than an individual one. We have two distinct explanations because the perspectives of two socially related agents bear on the same action. This paper seeks to defend, illustrate, and explain the possibility of giving social explanations of (some) actions. Let us call this phenomenon the social teleology of (some) actions—a modest form of social externalism about actions (independent of social externalism about mental states).Footnote 2
Why and under what conditions does the phenomenon of social teleology occur? There are many things going on in the smugglers’ example and they suggest a variety of possible answers to this question. You might think that the joint action or the shared goal between the two characters is crucial. Or that the hierarchy between them (either epistemic or operational) is crucial. Or that the trust the driver places in the passenger is crucial. Or that his tacit or provisional acceptance of the passenger’s reason is crucial. I argue that none of these features are essential to social teleology. What accounts for the possibility of a social explanation of the turn is the fact that the driver turns in response to a demand from the passenger. When the driver responds, he accepts that the reasons for which the passenger asked him to turn govern his action (which does not mean he accepts the reasons themselves). His action, which he does for his own individual reasons (perhaps out of trust, perhaps because he accepts a subordinate role, perhaps to please his passenger, etc.), also has a different point, not in relation to him, but in relation to the person who demanded it. In other words, the joint action, the shared mental states, the hierarchy, the trust, and the tacit acceptance (such as they are) are all contingent features of this preliminary example. In their absence, we can still give a social explanation, although formulated with more care (see section 2).
Let us call the relation between the agent who demands an action and the agent who acts in response a relation of vicariousness. This relation makes the driver (if only briefly and carefully) the vicarious agent of the passenger, who assumes the role of principal of the action she demanded. To give a more precise definition (further explained in section 1), a vicarious action is (by definition) an action A of a vicarious agent V done in response to a demand issued by a principal agent P who had a reasonable expectation that V would successfully respond.
The thesis of this paper is that vicariousness is sufficient for social teleology:
(Social teleology of vicarious actions—STVA): When a vicarious agent V performs an action A in response to a demand from a principal P who had a reasonable expectation that V would successfully respond, the reasons for P to demand A feature in a separate teleological explanation of A (distinct from the explanation that appeals to the reasons for V to perform A).
In other words, I aim to convince you not only that some actions are admitting of social explanations, but that the rather weak relation of vicariousness is enough to make them so. The social explanation of A is possible because A is not just the individual action of V (under a description that reflects her individual perspective) but also the vicarious action of P (perhaps under another description). I believe that vicariousness is also necessary for social teleology (and some arguments below support the necessity of some of the building blocks), but I cannot present a full case. Notice that the claim is that social explanations fit vicarious actions alongside individual ones, not that they are necessary to fully account for vicarious actions. Again, individual and social explanations are self-standing and do not compete.
Abraham Roth (Reference Roth and Shoemaker2017) discusses a phenomenon related to social teleology (and an example similar to the one above), but he concentrates on the entitlement that an executant may have to the reasons of a more knowledgeable associate.Footnote 3 Where Roth looks for reasons the executant may claim as her own and that describe her intentional action, I am interested in third-personal explanations of her action. Much of the work of this paper is meant to show that an executant’s behavior has a purposeful life of its own, separate from her own intentional agency. Where Roth thinks that entitlement to a social reason accounts for (what I call) social explanations,Footnote 4 I propose the weaker relation of vicariousness because I think social explanations do not always heave as closely to the executant’s perspective and agency as her entitlements to the reasons of others. We can sometimes separate the intentional description(s) of the action (which heave closely to the perspective and perhaps to the entitlements of the executant) from its point (which may reflect the intentions behind the instructions the executant follows).Footnote 5 Roth also holds that entitlements only occur within joint actions and where the executant has communicative access to and confidence in the social reason. None of these conditions bear on vicariousness and therefore on social teleology if I am right. None of this is to dispute the interest of the entitlements Roth studies, which occur in a subspecies of vicarious actions, where other features of the relation between the two agents bridge some of the gap between their perspectives.
The defense of (STVA) in this paper proceeds on theoretical and intuitive grounds. Section 1 unpacks the definition of vicarious actions and gives a structural argument for (STVA) by analogy with individual plans. Section 2 explores a series of examples to add intuitive arguments about the use of adverbs and reason explanations. Section 3 shows two examples of the payoffs. First, vicariousness provides resources to defend a mentalistic understanding of intentions against objections that, as it turns out, draw on social rather than individual explanations. Second, vicariousness is a building block in a minimalistic concept of collective action. Vicariousness itself is not meant as an approach to collective action (and social teleology is not a mark of collective agency). But, with additional elements, vicariousness may help us understand how agents act together, especially when they do not share mental states. Although I lack the space to tell the whole story, I sketch the role of vicariousness and the shape of the missing elements.
1. The concept of vicarious action
The definition of vicariousness above places three conditions on the phenomenon. The first captures the nature of the speech act P addresses to V: a demand. The second captures the success condition of the speech act: V responds to the demand. The third captures the practical confidence P should have: she could reasonably expect V to successfully act in response.Footnote 6 In clarifying all three, I explain my first, structural, argument in favor of (STVA). Vicariousness connects P’s plan and V’s execution in a way analogous to the link between ends and means within the planning and activity of a single agent. Since we take these links to fit the means within the plans of individual agents, we should, by analogy, take them to fit vicarious actions within the plans of their principals.
1.a A demand
For present purposes, demands run the gamut of speech acts that communicate an expectation in order to get the addressee to conform to it. Examples include second-personal prescriptives (“You should A”), polite requests (“Please A”) and orders (“Do A!”). Examples also include some behavioral cues. Lifting one end of the couch may convey the expectation that your friend will lift the other. Sometimes a behavioral cue is crucial to understand what is expected (“Do you mind?”). Vicarious actions thus range from responses to polite requests to actions done on order (and even under duress).Footnote 7
As the examples illustrate, demands belong to different behavioral and grammatical categories. The same grammatical form can also be used to issue a demand in some contexts and not in others. To borrow an example from Rebecca Kukla and Mark Lance, the sentence “Your dog is hungry” could function as a demand if it communicates an expectation that you’ll respond by feeding your dog. But the same sentence could function as a suggestion or statement of fact (Reference Kukla and Lance2009, 31–32). You’ll figure out the difference by the tone of my voice, the nature of our relationship, etc.
The category of demands is therefore a pragmatic one. It is also broad. Demands may belong to many fine-grained pragmatic categories. Kukla and Lance distinguish between prescriptives, alethic imperatives (that tell the addressee what she otherwise ought to do), and constative imperatives (that make it the case that the addressee ought to do what they tell her) (112). All of them are demands for present purposes. All communicate the expectation of an action to be done in response.
Speech acts that are not demands include pleas, suggestions, and invitations. You may use the grammatical form of an invitation to demand, but the act would not belong to the pragmatic category of invitations. If your relation to the invitee is such that your apparent invitation communicates an expectation that she will accept, then you have demanded her presence. A genuine invitation does not convey the expectation of a favorable answer, although it may come with such an expectation in the speaker’s mind, or even with the common knowledge of such an expectation between speaker and addressee.
What unites the broad pragmatic category of demands is that they all convey an expectation and openly convey their role in getting the addressee to conform. They communicate an intention of the speaker about the addressee and they publicly share the functional profile of this intention. We can think of them as public quasi-intentions. Footnote 8 When the accomplice tells the driver to turn left, she makes it clear that she intends him to turn left and her demand is an open way to fulfill both itself and the intention expressed.Footnote 9 We can see the relation between the speaker’s intention and her demand as analogous to relations between intentions within individual plans. When the accomplice plots the route, she concludes to the need to turn left. She forms an intention for the driver to turn. Her demand continues her plan when it expresses this intention. The first part of my structural argument relies on the analogy between, on the one hand, how the accomplice proceeds from her intention about the route to her intention about the turn and, on the other hand, how she proceeds from the latter to her demand. The demand plugs into her plan as any intention about means would.
1.b Acting in response to a demand
For the vicarious agent to conform to the expectation is obviously not enough for her to meet the demand: her conforming should also appropriately relate to the demand. At first glance, we might think that this involves what motivates the addressee to act. However, a condition about motivation runs into well-known counterexamples. I could leave the room because you revealed yourself to be the kind of person who would ask me to leave the room, and yet not at all in response to your demand. A motivational condition also runs into the variety of demands. Some of them purport to create reasons for the addressee to act while others purport to attract her attention to pre-existing reasons. We can expect this to lead to different motivations.
Moreover, the uptake of a reason to act is not what a demand aims at. A demand lays down a mandatory course of action for the addressee (contrary to invitations, pleas, etc.). What it first and foremost seeks is not motivation but intention. The success of a demand requires the addressee to form an intention whose content translates the content of the demand. To take up a demand is (first) to intend what it demands.
The addressee may ponder whether to take up a demand to A. But, when she genuinely does so, what she considers is not whether to A but whether to accept the intention to A demanded of her. She asks herself “Should I respond to this demand to A?” and not “Should I A?” When I consider whether to agree to your demand that I pass the salt, relevant considerations include the relationship between us, the words you used, our epistemic positions, etc., but not the taste of the food, your blood pressure, etc.Footnote 10 If I take these other considerations into account, I make up my own mind about passing the salt. I may take your speech act as evidence of your desire for the salt, but not as a demand to be responded to. The result cannot be an acknowledgment of your demand qua demand. Even if I pass the salt, I have treated your demand as a plea.
So, the first step in meeting the demand is to translate it into an intention. When the driver hears his accomplice tell him to turn, he translates the instruction into an intention to turn. The response thereby continues the demand into the plan of the vicarious agent. The second step of my structural argument relies on the analogy between, on the one hand, how the driver proceeds from the demand to his intention to turn, and, on the other hand, how he then proceeds from the latter to intentions to signal, apply the brakes, etc. The first two steps of the argument show that the demand and response bridge the gap between the two agents’ plans. The demand functions as an intermediate step between their plans analogous to intermediate intentions within an individual plan—for example, analogous to how an intention to boil water mediates between intentions to cook dinner and to turn on the stove.
A complete response to a demand requires obvious further steps: actions. Here the addressee also (usually) needs a reason to act on the intention she took up. But I do not think it matters whether she acts on any purported reason the demand presents her with. To a demand to raise the flag for crown and country, a sailor forcibly pressed into service might respond with an action whose sole motivation is the avoidance of punishment. As long as he acted on an intention he took up from the demand, he responded to it. Note, however, that this last point is essential neither to my characterization of vicarious actions nor to my structural argument.
1.c Practical confidence
The third condition bears on the principal’s expectations: she should know that her demand is not a long shot. The condition bears on her confidence both about the demand and about the action. First, a long-shot demand may elicit a response, but the speaker would not have made the addressee her vicarious agent. She would have been lucky to get a response given what she knew about the addressee, their relationship, and the context. She epistemically should have thought of her speech act as a plea (though she might have had good reasons to issue a demand anyway). Second, the speaker should also know that the addressee is likely to succeed in her attempt to respond. A confident demand to try a long-shot action would not give the speaker practical confidence over the act.
The practical confidence condition plays a role related to J. L. Austin’s felicity conditions on speech acts (Reference Austin1962, 22–23). Knowledge that felicity conditions obtain often plays a crucial role in practical confidence. But infelicitous demands sometimes meet the practical confidence condition.Footnote 11 Many of the illegitimate demands of hostage takers do. Conversely, a felicitous demand could be a long shot. Hostages may demand to be set free (since they have the moral standing to do so), but they would be fools to expect their captors to oblige. The main point here is the first: there are many sources of practical confidence and not all of them make for felicitous demands. Credible threats are reliable ways to gain practical confidence. Less coercively, the speaker might know enough about the addressee’s weak character to know she can expect infelicitous demands to succeed.
The standard for practical confidence is not certainty. The principal must meet the standard of knowledge any agent should meet about her prospective A-ing to be said to intend to A rather than to intend to try to A. To be clear, I do not assume that the two mental states are different. The distinction tracks differences in the relationship between elements in the agent’s plan, not necessarily in the nature of her commitment to A. For example, the distinction tracks a difference in how she should plan around A: if all she may intend is to try to A, she rationally cannot engage in unconditional planning under the assumption that she will A.
The capacity to intend to A (rather than try) is essential for successful A-ing to fit into the agent’s plan. Suppose you intend to buy a yacht and your plan is to play the lottery. You play repeatedly, and, lo and behold, you win. Winning was, however, not part of your plan. When you won, you did not successfully implement a step in a plan to buy a yacht. The question “Why did you win the lottery?” sounds strange precisely because winning was not something you could intend. And you may not (unironically) answer this strange question with your purpose—“To buy a yacht.” The same issue plagues long-shot demands. When the speaker did not know to expect a successful response that she nevertheless receives, the response does not fit into her plan. By contrast, if you have a solid business plan, earning the money may be part of your plan to buy a yacht. And the same is true of a successful response that the speaker knew to expect. Practical confidence thus makes a difference to what gets drawn into a plan. And drawing the vicarious agent into the principal’s plan is a key to the bridge between them.
1.d The structural argument for the social teleology of vicarious actions
Practical confidence completes the analogy between means-end relations within individual plans and vicarious relations across agents. The first two conditions describe an unbroken chain of intentions and intention-like states between the principal’s goal and the vicarious action. Each step in the chain leads to and rationalizes the next—from the principal’s intention about her end to her intentions about means to her demand and to the vicarious agent’s intentions and actions.Footnote 12 When we see the demand as a quasi-intention, the chain is analogous to the one between an intention to achieve some end (e.g., cooking dinner) and an individual action to implement means (e.g., turning the stove on). The third condition ensures that P knows enough to reasonably expect such a chain (just as she would reasonably expect to succeed in her plan to cook dinner). The chain then makes it possible to appeal to the end to explain the means. In the individual case, the chain and practical confidence are grounds for us to say that I turned on the stove to cook dinner, that the action was part of my plan to cook dinner. In the vicarious case, the three conditions allow us to see the vicarious action as part of the principal’s plan. For example, they are grounds for us to say that the driver turned to avoid the roadblock he did not know about.
You might object that there is a shift in perspective along the vicarious chain. I agree and I think the chain allows us to switch between the two perspectives. You might claim that the switch is impossible. It would follow that there is no social teleology for actions, because only the perspective of the proximal agent contributes to the teleological explanation of what she does. To alleviate this concern, we should keep in mind, first, that there is no conflict or competition between individual and social explanations and, second, that the demand mediates a transaction between the two perspectives. When V responds to the demand, she accepts to let P’s perspective govern her action. She accepts that P will assess the success or failure of her action in light of the instructions, but also in light of what P hopes to achieve, and she accepts that P may issue further instructions accordingly. She does not relinquish her own perspective, but she lets another one bear on her action alongside it. Sailors act out of fear of punishment and for whatever reasons hold on the officers’ deck.Footnote 13
The structural argument is a diagnosis of what transpires between the smugglers: the analogy with individual plans accounts for the social teleology of the turn. If this account is correct, it supports the social teleology of all vicarious actions, since the structural analogy applies across the board. The diagnosis would therefore show that social teleology is more widespread than the particular conditions in the preliminary example.
2. Adverbs, explanations, and social teleology
(STVA) does not entail that we always have as clear-cut a social explanation as in the smugglers’ example. We have to modulate explanations to track differences in what V knows as well as in her (potential) attitude toward P’s purpose. Is it a purpose V shares? Could share? Is indifferent to? Is antagonistic to? However, even in the adverse example of a vicarious agent who unwittingly acts against her own interests, we find clear traces of social teleology and, with appropriate care, we can articulate social explanations of her action.
Throughout this section I contrast examples of vicarious actions with examples in which a nonvicarious manipulator gets a victim to act without demanding it.Footnote 14 I look at cases where the victim and the vicarious agent share the purpose they serve, are indifferent to it, and are antagonistic to it. If this net is broad enough, a systematic contrast in the availability of social explanations would support (STVA). Remember that vicariousness is meant here to account for the possibility of social explanations. The following examples do not show that individual explanations of vicarious actions are lacking, but rather that social explanations are available alongside individual ones. I therefore insist not on any need for social explanations in these examples, but on the need to accept them when offered.
A systematic contrast would also show that the first condition on vicariousness (the demand) is necessary. If it is, then the second condition (the response) is not in need of additional support. However, the contrast does not speak to the third condition: skilled nonvicarious manipulators typically enjoy practical confidence.Footnote 15 The structural argument above is therefore my only argument for the necessity of practical confidence.
2.a A shared purpose
The smugglers illustrate the social teleology of a vicarious action where V does not know how she furthers a purpose she shares with P. The driver may believe that he turns to evade the authorities, but he does not know how. His ignorance would appear to prevent any explanation of his action in terms of the shared purpose. There is a gap in his perspective between the end and the means. Vicariousness lets us bridge this gap by switching to the perspective of the knowledgeable passenger.
If this is right, we should not expect a social explanation when an agent is nonvicariously manipulated to further a shared purpose. Paul, Vincent, and Michael wish to disguise the nature of their presence. In the hope of making them look drunk, Paul tells Vincent to shout at him. Vincent is Paul’s vicarious agent. He shouts because Paul asked him to. But we can add a social explanation: Vincent shouts to make them look drunk (even though he did not get the idea). With the same plan in mind, Paul shoves Michael, rightly anticipating an angry reaction. Paul nonvicariously manipulates Michael. Michael shouts because he is angry at Paul. We cannot add a social explanation to the individual one. The nonvicarious manipulation of which Michael is the victim does not let us reach outside his own perspective when explaining his shouting.
Adverbs offer another clue. Vincent and Michael unknowingly further Paul’s plan to make them look drunk. How should we characterize this description of their behavior? We lack a good adverbial marker for Vincent (besides “unknowingly” and the like). Markers of purpose or intention do not fit. Vincent did not intentionally make the group look drunk. But neither do the markers of accident that apply to Michael. Michael incidentally made the group look drunk. But, because (through Paul’s demand) Vincent’s behavior responded to the aim, we cannot characterize his relation to it as incidental. This signals the social teleology of his action, the teleological (rather than incidental) relation between his shouting and Paul’s plan.
2.b Ignorant indifferent agents
We now turn to examples apparently less favorable to (STVA) and important to rule out explanations in terms of shared mental states, joint actions, or trust. Consider first examples where the vicarious agent or the victim ignore (and therefore do not share) a purpose to which they would be indifferent.
Pia wants her coworkers to witness an inappropriate affair conducted late at night at the office. She orders her assistant Victoria to return to the office at 9 p.m. and she hides Mary’s keys while Mary is about to leave for a dinner with clients, knowing that Mary will have to come back around 9 p.m. to look for them. Pia gets Victoria to return and witness the affair vicariously and Mary to return and witness it nonvicariously. Both are ignorant of Pia’s purpose but would be indifferent to it—they take no interest in the affair but are not afraid to know about it.
When Victoria obeys the order, both individual and social explanations are available to account for her return to the office. She comes back to follow orders and to witness the affair. By contrast, Mary returns to find her keys. She does not return to witness the affair. If you do not feel comfortable with the italicized social explanation and the contrast, consider a context in which Victoria and Mary speculate, or explain in hindsight, and consider sentences about the point of their conduct (where they are not grammatical subjects). If her partner asks Victoria why she has to go back, she may state her individual reason (she was ordered to return and has good reasons not to disobey), but she may also reach for the unknown reasons of Pia: she may say that she has no idea what the reason is, or she may speculate about the purpose of her own action: “I think I’m going back because Pia wants me to witness something.” Again, notice that these are possibilities; there is no need for Victoria to reach for the unknown social explanation, but it is possible for her to do so. When asked the why question about her return, Mary has no such possibility. All the answers available to her state her individual reasons alone. Even if she is suspicious, her speculations are limited to the behavior of Pia: “I think she hid my keys because she wants me to witness something at the office.” Victoria can speculate about Pia’s reasons too, but Mary can only speculate about Pia. She cannot speculate about an unknown reason for her return. There is no other point than getting her keys (and whatever she wants the keys for). We see the same contrast in how Victoria and Mary react upon understanding Pia’s purpose. Both may say, “That was the point of making me come back,” but only Victoria may also say, “That was the point of my coming back.”
Now, when Victoria agrees to these explanations of what she did, she does not accept an intentional description of an action of hers. First, Pia’s reason does not explain why Victoria acted (because it does not represent what Victoria saw in the action) but explains what she did (because the deed was done in response to Pia and her reasons). This is why social explanations are more intuitive where Victoria is not the grammatical subject of the sentence. With an emphasis on the pronoun, she might not agree that she returned to witness the affair. Yet, she should not deny that it was the point of her return (whatever point it also had for her). In such a sentence, she takes a third-personal (but still teleological) perspective on her behavior and sees it as the vicarious action of Pia. Second, the social explanation of Victoria’s return in terms of the witnessing does not mean that the latter counts as (an intentional description of) an action of hers; it does not. When she agrees to a social explanation, Victoria takes a teleological perspective on her behavior that is not her own. She accepted to let this perspective govern her behavior when she followed the order, but she did not endorse the (unknown) perspective. Social teleology does not break the links between the intentions of an agent, her reasons, and her agency, but identifies two distinct nexuses of intentions, reasons, and agency that bear on the same behavior.
Adverbs paint the same contrast. Mary incidentally witnesses the affair when she gets her keys. But the adverb does not fit Victoria’s witnessing. As the (unknown) purpose of the agent to whom her action responds, witnessing the affair is not an incidental description of her behavior. The witnessing was the whole point according to a perspective to which she responded when she came back. To say that she incidentally witnesses the affair is to ignore this other perspective (as if Victoria was not carrying out Pia’s orders when she came back).
The example of Victoria establishes almost all I argue for. She neither shares nor accepts Pia’s purpose (not even tacitly or provisionally) and she does not engage in a joint action with Pia. If her action displays some social teleology alongside its individual one, it has to be because of her vicarious relation with Pia.
2.c Antagonists
However, antagonistic perspectives raise a challenge for (STVA). Here P gets V to act against her own interests, and, if V knows the purpose, against her own judgment. Tacit or provisional acceptance are even more clearly out of the question. But the careful use of adverbs is crucial to formulate social explanations that make clear that V does not share or endorse P’s purpose. In all the following examples, various characters are trying to steal a Picasso painting and need the museum guards to turn off the alarm.
Ignorant antagonists
The thief poses as a utility technician and asks Vanessa to let him check on a gas leak in the museum. She deactivates the alarm to let him inspect the pipes. He steals the Picasso. What is her relation to his plan? She deactivated the alarm under false pretense. She mistakenly helped him carry away the Picasso. With these adverbial expressions, we can connect the thief’s purpose to her action without suggesting that it is her own. Vanessa deactivated the alarm mistakenly/under false pretense for the thief to steal the Picasso. The social explanation looks at the action from the perspective of the thief, but the adverbial expressions underline the gap with Vanessa’s perspective. In hindsight, Vanessa realizes that the theft was the point of her deactivating the alarm. When asked the question of what she (now) believes to have been the point—“In hindsight, what do you think was the point of your deactivating the alarm?”—she does not reject the premise that there was a point other than her own, so long as we make clear that it was not hers. “In hindsight” in the question and “mistakenly” or “under false pretense” in the social explanation discharge this condition.
Contrast the thief with a nonvicarious manipulator. The burglar lights a fire in the sculpture garden. Falling into the trap, Mitch deactivates the alarm as he rushes to douse the fire. To make sure he knows as much as Vanessa, assume that he saw the burglar light the fire and believes that the burglar (whom he takes to be homeless) will take advantage of the situation to shelter in the galleries. What is Mitch’s relation to the burglar’s plan? The adverbs applied to Vanessa do not fit. He did not turn off the alarm under false pretense. He did not mistakenly help the burglar steal the Picasso. His letting the burglar into the galleries might be characterized as a mistake, but not his helping with the theft. Rather, when he rushed to douse the fire, Mitch incidentally helped the burglar—an adverb applied to Michael and Mary but not to Vanessa.
Is the contrast between “mistakenly” and “incidentally” significant? Does it map onto the contrast between characters in earlier examples? Does Vanessa clearly fall on the side of earlier vicarious agents Vincent and Victoria (who did not act mistakenly)? Consider how J. L. Austin invited us to draw the distinction between mistakes and accidents:
You have a donkey, so have I, and they graze in the same field. The day comes when I conceive a dislike for mine. I go to shoot it, draw a bead on it, fire: the brute falls in its tracks. I inspect the victim, and find to my horror that it is your donkey. I appear on your doorstep with the remains and say—what? ‘I say, old sport, I’m awfully sorry, &c., I’ve shot your donkey by accident’? Or ‘by mistake’? Then again, I go to shoot my donkey as before, draw a bead on it, fire—but as I do so, the beasts move, and to my horror yours falls. Again the scene on the doorstep—what do I say? ‘By mistake’? Or ‘by accident’?
(Reference Austin1957, 11)To shoot your donkey after drawing a mistaken bead on it is to shoot this donkey (which happens to be yours) on purpose. To shoot your donkey by mistake is to do something on purpose that necessarily amounts to shooting your donkey. On the other hand, when the beasts move and Austin shoots yours by accident, there is nothing he aims to do that necessarily amounts to shooting your donkey. Shooting your wandering donkey is not his purpose under any description. The contrasted stories show that accidents are incidental to our purposes whereas mistakes infect them. This is what the adverb signals about Vanessa: she was wrong to let the thief’s purpose govern her action. With the vicarious relation, the description of her action in terms of this purpose (“to help the thief steal the Picasso”) crosses the line from incidental to mistaken. Nothing similar happens to Mitch. Adverbs mark an important difference between the two guards, though in a more subtle way than in earlier pairs of examples.
The contrast between explanations is clearer. Mitch did not deactivate the alarm for the burglar to steal the Picasso. No adverb helps make this explanation palatable. “Incidentally” does not fit the action (deactivating the alarm). It fits the supposed social purpose (Mitch incidentally helped the burglar) but it cannot fit into an explanatory clause (“to incidentally help the burglar” is not a felicitous explanation because incidentals do not explain). The adverbs applied to Vanessa (“mistakenly” and “under false pretense”) do not fit his action either. Even if we judge Mitch’s deactivation of the alarm to have been mistaken, the mistake lacks any teleological relation to the burglar’s trap. No adverb makes room for a social explanation of what Mitch did. And, indeed, if confronted with the question we put to Vanessa—“In hindsight, what do you think was the point of your deactivating the alarm?”—Mitch would insist that the point was to douse the fire. He would reject the premise that there could have been another point to an action he decided upon on his own.
Knowing antagonists
Ignorance and antagonism made the previous pair of examples the most challenging for (STVA). The case is surprisingly clearer when V is aware of the antagonistic purpose of P.
A third character, the robber, puts a gun to Veronica’s head and she agrees to do whatever he asks. He demands that she deactivate the alarm. She turns it off under duress for the robber to steal the Picasso. Remember that there is no competition with the individual explanation. Veronica certainly deactivates the alarm to avoid getting hurt. What matters is the possibility of a social explanation in addition to the individual one. When she testifies about the events, Veronica should endorse the idea that the point of her coerced action was the theft of the Picasso. Again, “coerced” and “under duress” mark the gap between her aim and the robber’s. They prevent a confusion between the point her behavior had for her (as her action) and the point it had for the robber (as his vicarious action). Her contemporaneous knowledge of the gap between the two perspectives rules out even more clearly any account of the social explanation that would appeal to a joint action, shared mental states, or acceptance of the robber’s purpose.
Contrast Veronica with the victim of another nonvicarious manipulation. This time, Mitch knows what the burglar is after. He nonetheless turns off the alarm because he judges it more important to contain the fire than prevent the theft. We still cannot give a social explanation: Mitch did not deactivate the alarm to help the burglar steal the Picasso. He would not learn anything in hindsight, and he would certainly insist that the only point was to douse the fire. We would characterize his relation to the burglar’s purpose in the same terms as before: he incidentally (but knowingly) helped with the theft.
The fire in the sculpture garden, the way the burglar gets Mitch to act according to his plan, is not a way to get Mitch to act on the plan. The thief’s or the robber’s demands and the burglar’s fire explain what the guards did in very different ways. Lighting the fire makes sense in the burglar’s plan. Mitch’s reaction plays into the plan. But the fire does not get Mitch to execute the plan, because the fire does not convey his role in the plan; it only motivates him to perform it.
2.d Bridges and tensions
The examples above illustrate the different ways in which vicariousness affects reason explanations. Although individual and social explanations are not rival explanations, they sometimes appear to conflict because of tensions between the perspectives of the two agents. Vicariousness provides a bridge between two plans, but not between the ends and reasons of two agents. When they do not share a purpose, they often do not see the point of the action in the same light. However, the appropriate use of adverbs allows for social explanations by clearly dissociating them from vicarious agents. Social teleology highlights a reason and a purpose of the vicarious action but not (necessarily) of the vicarious agent. The upshot is a variety of nuanced ways to express social explanations. What they have in common is the appeal to the perspective of the principal and the denial of an incidental relationship between her end and the action.
3. Vicariousness and the philosophy of action
Why care about vicarious actions and their social teleology? This section sketches two potential payoffs.
3.a Individual intentions and social teleology
Social teleology weakens the relation between individual intentions and purposeful descriptions of actions. As a result, it relieves some of the pressure on an understanding of individual intentions as mental states by adding resources to handle examples that seemingly favor teleological theories of intentions (and actions).
A famous argument from Elizabeth Anscombe illustrates this. Anscombe drew an intuitive distinction between a man who pumps water as part of the normal duties of his job and happens to know that the water is poisoned (call him Norman), and a man who was hired to pump poisoned water and knowingly does so (call him Harry) (Reference Anscombe1963, secs. 25–27). Neither cares about the poison. Each only wants to earn his pay. Yet, Anscombe argues that Harry’s pumping poisoned water (and poisoning those who drink it) is intentional whereas Norman’s is not. She then reasons from the difference in their intentional doings to a difference in their intentions. Since she has described them as mentally alike (at least in their attitude to what they do), she concludes that their intentions are not mental states.
Social teleology supports a different diagnosis. Harry is the knowledgeable but indifferent vicarious agent of the poisoner—a combination not encountered so far. I suggest that the resulting social teleology of his pumping, combined with his knowledge of it, drives the contrast with Norman. The intuition that Harry intentionally pumps poisoned water results from his knowledge that, given his vicarious relation to the poisoner, this is the point of his action. The diagnosis blocks Anscombe’s move from intentional doings to intentions. Vicariousness means that the poisoner’s purpose governs Harry’s pumping (but not Norman’s) independently of what he intends. What explains the contrast is not a difference in intentions, but in relation to the poisoner.Footnote 16
Is the asymmetry in social teleology a better diagnosis than Anscombe’s? In her view, the key is that Harry gets what he is after (his pay) by pumping poisoned water, whereas Norman gets it by pumping water. Pumping poisoned water is a necessary means to Harry’s end but a side effect to Norman’s (identical) end. However, this diagnosis faces a dilemma. Either there is a known difference in the means-end relationships the two agents face or the difference lies entirely in the poisoner’s plan. An example of the first horn would involve the poisoner telling Harry that he will collect his pay once the inhabitants have died. On the second horn, Harry believes that he will collect his pay no matter the outcome. The dilemma for Anscombe is the following. The first horn involves an agreed-upon mental difference between Harry and Norman: different beliefs about the necessary means to their ends. Anscombe’s argument from the mental similarity between them cannot go through. This is a reason to think Anscombe had the second horn in mind. But the second horn appeals to a necessary connection between Harry pumping poisoned water and Harry earning the pay that only exists in the poisoner’s plan. Pumping does not make sense for the poisoner if the water is not poisoned. But the same is not true of Harry (who does not care what he pumps). On the second horn, Anscombe’s explanation therefore appeals to (what I call) the principal’s plan. She taps into the same resources as the social explanation, but without any account for why the structure of the poisoner’s plan is relevant to Harry’s action (and not to Norman’s). Vicariousness offers such an account, but one that makes clear that the explanation does not appeal to Harry’s individual intentions. In other words, the best way to flesh out what happens on the second horn appeals to the social teleology of the vicarious action, but social teleology blocks Anscombe’s conclusion precisely because it is social and not individual.
Whatever you think of this rebuttal of Anscombe’s diagnosis, (STVA) at least suggests an alternative compatible with a mentalistic view of intentions. A new defense of Anscombe’s argument would have to appeal to examples where a diagnosis of vicariousness is out of the question.
3.b Vicarious actions and the coordination of collective actions
Vicariousness helps describe social interactions that are not collective actions (like all the examples in section 2). But, with additional elements, it also helps build a minimalistic concept of collective action. I cannot detail this project here, but I want to motivate and sketch the role of vicariousness in it.
To understand what it means for agents to act together is to understand something about the way in which their individual contributions relate to each other. The behavior of coagents displays interrelations beyond the mutual responsiveness characteristic of any interaction between agents (competitors included). To act together is to coordinate individual contributions. Philosophers have therefore rightly paid a lot of attention to the source of the individual contributions of coagents. However, if we approach this question without vicariousness in our conceptual toolbox, and if we want the source of the contributions to feature in reason explanations (as we usually do when we explain actions), we have to think of the source as something each coagent has access to and subscribes to. In other words, we have to follow a dominant trend in the literature and think in terms of shared mental states.Footnote 17 Let me briefly explain the shortcomings of this approach—the importance of shared mental states notwithstanding.
Agents may act together without sharing an intention, let alone a detailed plan. A proper argument for this judgment would engage with different conceptions of shared intentions. However, we can agree on pretheoretical grounds that agents who have no motivation to promote A and/or ignore that their actions promote A do not share an intention (or plan) to A. But coerced, mercenary and/or ignorant agents can act together with enthusiastic leaders and peers. Conscripted sailors can act together with committed officers, even though they might constantly look for ways to defect and have no idea what their commanders are up to. Such examples show that the philosophy of collective action needs modest concepts to describe the coordination of agents who do not share mental states. There are other minimal but adequate sources of coordination. What matters is their role as sources of coordination.
Conversely, even when agents share mental states, what matters for collective agency is that their shared mental states function as an adequate source of coordination among them. Consider a counterexample. Official political campaigns and friendly political action committees (PACs) in the US are legally barred from coordinating. If workers in each organization follow the law, they act alongside one another but not with one another. Their coordinated efforts (such as they are) are due to isolated efforts to act in compatible and complementary ways. There is no collective action among all of them. However, it is not implausible that they all act on a shared intention to get their candidate elected—their commitments are a matter of public record, each would feel betrayed if the others defected, the PAC would give up if the official campaign did (and perhaps the other way round as well). Acting on a shared intention is therefore not sufficient for collective action where the shared intention is not a source of coordination among its bearers. A shared intention commits its bearers to acting in compatible and complementary ways—by way of meshing subplans in Michael Bratman’s influential view (Reference Bratman and Mantzavinos2009, section 3.3). But the example of the PAC shows that, under some circumstances, the bearers might have to rely on their own isolated individual planning (with an eye to one another) in order to actually implement the mesh they committed to, rather than coordinate and act collectively.
None of the above is to suggest that sharing mental states does not describe an important dimension of variation among collective actions. Shared plans are certainly the more promising route to a narrow concept of shared action—a concept not meant to apply to conscripted sailors. Since the narrow concept is Bratman’s avowed target, I do not propose an alternative to his view of shared actions, but a possible route to a minimal (and, I think, useful) concept of collective action. As such, this sketch might be seen as an alternative to the work of Kirk Ludwig (Reference Ludwig2015; Reference Ludwig2016), who shares many of my reservations about the exclusive focus on shared intentions.Footnote 18 I propose to approach this minimal concept not by thinking about the range of adequate mental states for the participants, but about the range of adequate sources of coordination among them. Whatever we choose to call the resulting concept, I think it delineates an interesting range of collective phenomena, including many that are intermediate between shared action and mere interaction.
To delineate this range, we have to characterize what ensures the mesh between individual contributions—the mesh to which those who share an intention have committed, but a mesh that also sometimes occurs among agents who do not share any intention (for example, officers and conscripted sailors). The source of the coordination of coagents lies in the entanglement between the etiologies of their contributions. A shared plan or the individual plan of a leader are two adequate sources of coordination, two adequate mechanisms to make sure that subplans mesh. To share a plan is to collectively take up the work of coordination. But an individual can take up the work too. What matters (in rough outlines) is that all coagents act on the same (numerically identical) coordinated plan, without necessarily subscribing to it or knowing its details. Vicarious actions are crucial to understand how coagents act on such a plan when it is not theirs and not shared. Vicarious relations allow us to trace the etiology of a contribution beyond its proximal agent and all the way to a source of coordination she may have no commitment to. For example, they allow us to see all the sailors as acting on the officers’ coordinated plan for them. I propose that this lies at the heart of their collective action. Vicariousness would then be a crucial building block for an approach of collective agency through coordination, rather than through the details of individual mental states and commitments. Additional work beyond the scope of this paper is needed to spell out conditions on coordinated plans and their numerical identity.
4. Conclusion
I have argued for a phenomenon with intuitive support, the social teleology of some actions, and for an account of this phenomenon in terms of vicariousness. Actions have a teleological life of their own when their agents respond to the reasonably confident demands of others. Although their explanations in terms of the (individual) reasons of their (vicarious) agents remain essential for them to be actions at all, we may also explain them in terms of the (social) reasons of the principals who demanded them. Although the descriptions that capture how these actions further the plans of their principals may not be intentional, they are not incidental either. A structural analogy between the way in which vicarious actions fit into their principals’ plans and the way in which individual actions fit into individual plans adds a theoretical reason to help us understand why they would have a point distinct from what their agent saw in them. I hope my sketch of two payoffs shows the potential value of this research program.
Acknowledgments
I would like to thank the members of my PhD dissertation committee, Ned Block, David Velleman and Michael Strevens, as well as audiences at New York University, the Institut Jean Nicod, the 2017 SIFA midterm conference in Genoa and the 2018 Cork Social Agency Worksop. Particular thanks to Eliot Litalien, Lilian O’Brien, Luke Roelofs, Alessandro Salice, Knut Skarsaune, Joel Walmsley, and two anonymous reviewers for this journal.
Philippe A. Lusson (PhD, New York University) is lecturer in philosophy at NYU Paris. He works primarily in the philosophy of action and mind with a particular focus on decision-making, omissions, and collective actions.
