Individual and Collective Action: Reply to Blomberg

1 Introduction

Olle Blomberg challenges three claims in my book From Individual to Plural Agency (Ludwig 2016). The first is that there are no collective actions in the sense in which there are individual actions. The second is that singular action sentences entail that there is no more than one agent of the event expressed by the action verb in the way required by that verb. The third, connected with the second, is that an individual intention, e.g. to build a boat, is not satisfied if you don’t do it yourself, e.g. if you build the front half and someone else builds the back half.

2 Are There Collective Actions?

Some of the differences between Blomberg and me on collective action appear more verbal than substantive. But I think that Blomberg brings to light an important issue in thinking about primitive actions and their composites that I did not take fully into account.

I did not place much emphasis on the concept of an action in the book. Asking what someone did is not like asking what someone kicked. We answer the latter question by supplying a noun phrase that picks out the thing kicked. We answer the former but supplying an action sentence. My view is that all the substantive logical and conceptual questions about the truth conditions for action sentence are answered when we have an account of their logico-semantic form and an analysis of the agency relation. If pressed to say what were someone’s acts, I suggested that we focus on what he does but not by doing anything else. Following Davidson, I called such actions primitive actions (Davidson 2001). In my analysis of action sentences, I identified this as the value of the variable that appears in the agency relation in the representation of the logic form of the action sentence. Thus, for [1] whose logical form we can roughly represent in [1a],

the agent’s primitive actions (with respect to his boat buildings in the past) are just the values of “f” that make the sentence true for values of “e” (the consequent events) and “t”. The relation expressed by “x is an agent of f at t” I called primitive agency. I am an agent in an extended sense of anything that what I am a primitive agent of brings about in turn. Thus, if in moving my finger, I pull a trigger, fire a gun, and kill someone, I am an agent of his death, though not primitively, since I brought about his death by way of standing in the primitive agency relation to a distinct event, the movement of my finger. The movement of the trigger, the firing of the gun, the death of my victim, are all downstream from the finger movement.

Plural action sentences like [2] are ambiguous between a distributive and collective reading.

The collective reading in contrast says not that each of us built a boat but that we built a boat together.^[3] The distributive reading does not entail the collective, or vice versa. Blomberg disputes the second independence claim, which we return to below. I argued that the collective reading derives from giving the quantifier over the consequent event e wide scope with respect to the quantifier over members of the group and adjusting clause (iii) to restrict the agents of the event to members of the group. This is given in [2c].

This says that all of us and no one else contributed directly to bringing about one and the same boat building. Everything about what the account claims is made plain in [2c].

Why say that there are no collective actions? For individuals, actions are what they stand in the primitive agency relation to. In [2c] only individuals stand in the primitive agency relation to any events. So the group does not stand in the primitive agency relation to anything. There is no group action in the sense in which there are individual actions.

Blomberg disputes this–or does he? He does not disagree with the claim that the group does not stand in the primitive agency relation to any event. What does the disagreement come to? He puts it this way in his conclusion (labels added):

(a) Ludwig construes individual and collective action as much more different in nature and ontological status than they really are. (b) First, he chooses to define “collective primitive action” in way that only makes sense in the light of a mistaken group agent account of collective action; then draws the conclusion that there are no collective primitive actions. (c) However, in the light of his own multiple agent account, there is a perfectly intelligible sense in which there can be collective primitive actions. (d) Furthermore, as a matter of empirical fact, it is arguably plausible that many of us sometimes perform such collective primitive actions. (p. 23)

(b) Gives a somewhat misleading impression of my discussion. I asked whether groups perform primitive actions in the sense in which individuals do. I said no in one sense (does not stand in the agency relation) and yes in another (does something but not by doing anything else), and I claimed the former was the primary sense (Ludwig 2016, p. 173–174). I was not offering an eliminativist account, as Blomberg claims (p. 4, 5, 9). This suggests that I was acknowledging that we thought there was such a thing as collective primitive action and had something definite in mind, and that I was arguing that contrary to popular belief there is no such thing. I don’t think “collective primitive action” has an everyday use or that there is a common sense view about it. So far as (c) and (d) go, I agree with both! There is a perfectly intelligible sense in which there can be collective primitive actions: what groups do but not by doing anything else. And I agree that we sometimes perform collective primitive actions in this sense. I said, “the mereological sum of the primitive contributions of the group’s member may count in an extended, or secondary, sense as the group’s primitive action” (p. 174). It is the characterization of this as “an extended or secondary” sense that is the bone of contention. So the central charge is (a): I fail to acknowledge important similarities between what individuals and groups do that grounds the claim that there are collective primitive actions in “the primary and non-derivative sense” (p. 10, 13, 14, 17).

Blomberg pursues two lines of argument. First, he argues by way of examples that there are collective actions. Second, he argues that my discussion of primitive actions focuses too narrowly on simple actions and that once we take into consideration composite actions, we see that when, e.g. we build a boat together, there is something that is a collective action “in the primary sense.”

2.2 Simple and Composite Actions

The second line of argument is independent of the first and has broader scope.

Blomberg distinguishes, in the individual case, between simple and composite actions. A simple action is an event that an agent brings about primitively (that event may itself be complex, but still in the relevant sense a simple action). But there are complex actions which consist of distinct of primitive actions. There are two cases of complex actions to consider. First, there are primitive actions that compose another primitive action toward which I have a contemporaneous intention-in-action. Second, there are primitive actions that compose a complex action that is not as such the target of a contemporaneous intention-in-action.

The first is illustrated by my clapping my hands. Clapping my hands consists in moving my right arm and my left arm. These are distinct primitive actions. If one arm were paralyzed unbeknownst to me, the other arm would still move at my intention. But the clapping, which consists in the movement of both my arms, is also in a sense a primitive action. I bring it about directly, and intend it de re, though it is composed of two distinct primitive actions (at least). Likewise when typing, the fingers of each hand move independently of the other. The ensemble of movements is a primitive action in the sense in which clapping is, as well as at least some of the movements that compose those. I intend the movements of the left hand de re and of the right de re, but also the ensemble of the two. There are also ensembles of diachronic movements that can count as primitive actions in the same sense, that is, towards which an intention-in-action is directed, for example, tying an Alpine butterfly loop or typing a sentence.

In what sense is the ensemble a primitive action as well as its components? When I clap my hands, I do so by moving my left hand and by independently moving my right hand. Does not this fail the test of being something I do but not by doing anything else? Strictly speaking, yes. So this shows that that test needs refinement. The sense in which the ensemble is clearly a primitive action in these cases is that it can be the target itself of a de re intention-in-action.

The second case is illustrated by projects like building a house that are extended in time. If I build a house, I bring about the coming together of the materials into a house by a sequence of primitive actions over an extended period interrupted by sleep and other activities. As Blomberg notes, there is no one intention-in-action directed at the whole ensemble as opposed to various intentions-in-action directed at parts of it. However, as Blomberg also notes, these ensembles are the sorts of things that I allow to be values of the variable “f” in my account of the logical form of action sentences.

Thus, I have given three criteria for being a primitive action which are not coextensive:

C1.	what we bring about but not by bringing about anything else;
C2.	what are the targets of our intentions-in-action;
C3.	what are the truth-making values of the event argument (values of “f”) in the agency relation.[5]

C3 includes the events that C1 and C2 do, C2 includes the events that C1 does, but C1 does not include all the events C2 or C3 do, and C2 does not include all the events C3 does. I’ll call events meeting C1 simple primitive actions, those meeting C2 but not C1 complex primitive actions, and those meeting C3 but not C1 or C2 composite primitive actions (i.e. composites of primitive actions).^[6] See Figure 1. I’ll use the unmodified term “primitive action” to cover all three, and since I endorse the view that actions are the values of the “f” variable, I will also use the term “action” to cover all three. I will then also speak of simple, complex and composite actions. The need to make and keep track of these distinctions is something I was not clear about in the book and I am grateful to Blomberg’s discussion for bringing this to light.

Fig. 1:

Three Rings of Action.

This idea that a composite of primitive actions may also be an action is central to Blomberg’s second line of argument that there is a notion of a collective action which is not merely a façon de parler. The suggestion is that once we see that in the case of individuals, actions can be composed out of simple and complex actions, we can see that there is no reason to resist the principle that actions may be composed out of the simple and complex primitive actions of more than one individual. Call this the General Action Composition Principle. For generality, suppose primitive actions that compose another action are restricted to events that stand in some relation R (left as a placeholder).

General Action Composition Principle

For any simple primitive actions, α₁, α₂, … α_n that stand in relation R, there is an action which is their sum.

Is the General Action Composition Principle true? Call a sum of simple actions that stand in R an R-sum. Let us suppose that an R-sum of any of an individual’s simple actions is also an action of his. It is true that this has in common with the R-sum of any number of simple actions of two or more agents that it is an R-sum of primitive actions. But it does not follow in general that a concept that applies to a sum of a set of things associated with a single individual applies to a sum of a set of things of the same kind associated with more than one individual. For example, the sum of the cells of my body at any time is a human being, but a of sum of cells from different human bodies at any time is not a human body. The sum of the moments of my life is a human life, but a sum of moments in different human lives is not a human life. The question is whether “primitive action” should gather not just sums of the primitive actions of particular individuals, but primitive actions in general. That they are both R-sums of primitive actions does not settle the question.

In the case of an individual, any R-sum of her simple actions involves events which are minimally unified in their simple components being brought about primitively by one and the same agent, which is why it is suitable to be the value of the variable “f” in our representation of the logical form of singular action sentences. Putting aside recherché cases like the Siamese twins, in the case of an R-sum selecting from simple actions of each member of a group of agents, this is not true. There is no agent with respect to whom the sum of primitive actions performed by its members meets this condition. This is why it seems right to say that there is no agent of what the group does in the sense in which there is an agent of what an individual does and no collective actions in the sense in which there are individual actions.

Blomberg says: “Were Ludwig faithful to his own multiple agents account and to common sense, he ought to say that there are collective (primitive) actions in the primary sense, or at least that such primitive actions are possible” (p. 8). I don’t think that common sense (in the sense of what is generally believed among ordinary people) has a verdict on this matter. “Primitive action” is a bit of philosophical jargon. “Collective action” is used loosely and in a variety of ways, often with a narrower sense that philosophers have in mind, and when used in a general sense is simply a proxy for what groups of people do together. So there is not a common sense view on the use of this expression to get at what philosophers are interested in. My multiple agents account, expressed in [2c], is silent on the use of “collective primitive action” and “collective action”, though it gives a complete account of what is going on. This is one reason I don’t think the question of what collective actions are is itself of much theoretical interest.

In the book I said that if a singular action sentence like [1] is true, then there is some description under which the action is intentional, but the same is not true of a plural collective action sentence such as [2], that is, the group that builds a boat need not have done anything that is intentional under any description. Blomberg argues that there are things that individual agents do which are not intentional under any description. Thirty years ago I moved my left index finger. Yesterday I moved my right index finger. If the sum of those events is something I did, then there is no description under which what I did is intentional. That is right, and it is a good point. It shows that I overstated the case for individual action. Not every event which is an action of mine (allowing their sums to be actions) is intentional under some description. However, the underlying point was that in the case of individual action, the source of any form of agency lies in things individual agents do intentionally. It remains true that if an individual did something, then the individual did one or more things that were intentional under some description or descriptions that explains why it was something he did. It is not true that if a group did something, it did one or more things that are intentional under some description or descriptions. As I wrote in the book, “It is therefore possible for a group to do something without doing anything intentionally” (p. 176). It is not possible for any individual agent to do something without doing anything intentionally.

Where does this leave us? These are things that Blomberg and I agree on.

1. The sums of an individual’s simple primitive actions (or some subset of those sums) are themselves actions.

2. The truth conditions of plural collective action sentences involve only individuals standing in the agency relation to actions.

3. When any group acts, there is a sum of the different member agents’ simple primitive actions that contribute to what the group does.

What we seem to disagree on is whether “action” or “primitive action” should be used for the sums involved in 3 in addition to those in 1, or whether, if used, it is used in “the primary sense.” I submit that this is largely a verbal rather than a substantive disagreement. A disagreement is verbal when you agree about all the facts but disagree about how to use a word in relation to those facts. There might be an empirical issue about ordinary usage, but in this case I think ordinary usage does not settle the issue. It is up to us as theorists to judge how best to use “collective primitive action.” I emphasize the differences between individual and group agency to throw the mechanisms of collection agency into relief. Blomberg wants to emphasize similarities between extended ensembles of individual actions and ensembles of actions of distinct individuals. I don’t find myself moved. But I don’t see a deep disagreement between us.

3 The Sole Agent Requirement in the Analysis of the Logical Form of Action Sentences

My account of the logical form of singular action sentences includes a sole agency requirement. This is clause (iii) of the analysis of [1] in [1a], repeated here.

Blomberg maintains that (iii) is not a semantic requirement on the truth of [1]. When I assert [1], I merely implicate conversationally that I was the only agent directly contributing to the building of the boat. Thus, Blomberg would drop (iii) from the analysis of the logical form of [1], and correspondingly clause (iii) from [2c].

On my account, the distributive reading of plural action sentences does not follow from the collective reading. This is because for a singular action sentence like “I lifted the bench” to be true, the agent must be the only direct agent of the bench going up. But the collective reading requires that there be many direct agents of the bench going up for it to be true. So no one of the members of the group is the sole direct agent of it going up. On Blomberg’s account, however, since the singular action sentence only says that the agent is a direct agent of the bench going up (not the only direct agent), and the plural action sentence on its collective reading says that there is an event of a bench going up of which each of us is a direct agent, it follows that for each of us there is an event of a bench going up of which he is a direct agent. Thus, on Blomberg’s view, the collective reading of plural action sentences entails the distributive reading.

Why does Blomberg think (iii) in [1a] is not a semantic requirement? Blomberg says that [3] is true, presumably on the grounds that the antecedent is semantically sufficient for the consequent, but it would not be if my account were correct. He says that [4] is not a contradiction, but it would be if my account were correct.

Furthermore, the conversation in [5] should sound weird if I am right, but Blomberg says it is fine.

In contrast, I think [3] is clearly semantically false and [4] contradicts its first clause in the second. [4] could at best be read as a correction, taking it back, like saying, “I walked downtown, … but I didn’t walk all the way.” The conversation in [5] suggests you shouldn’t leave your wallet unattended around B. Contrast [3] and [4] with [6] and [7].

On Blomberg’s view, [6] and [7] are contradictions. But in [6] and [7] the antecedent entails the consequent. Perhaps Bomberg will say that “x wrote it himself” is not entailed by “x wrote it”. Surely, though, I wrote From Individual to Plural Agency iff I wrote it myself. If I wrote it myself, then I wrote it. And I if I wrote it, it was not written by somebody else. So if I wrote it, I wrote it myself. Suppose I say to someone: I wrote two books on collective action. He says: Did you write them yourself? I say, insulted: Are you deaf? I just told you I wrote them. Are you implying that I am lying? In any case, consider [8], which should be contradictory if Blomberg is right, but it is not.

Change the example to a student paper rather than a book. A student, Billy, turns in a twenty-page paper you think he could not have written. You call him to your office. The conversation goes as in [9].

Does anyone want to maintain that Billy wrote the paper? Does Billy get to say: “Well, my roommate Bud said that he recently read a paper by Olle Blomberg and learned that if I made a contribution to writing the paper it follows that I literally wrote it. So I’m taking this to the Honor Court.” Now, consider a continuation of the conversation [5] in [5+].

On Blomberg’s view, [5+] should sound weird, and A should sound deeply confused, but it does not sound weird, and A is annoyed rather than confused. A calls out B for claiming to do more than he did. Suppose the conversation continues in [5++].

Now who is confused!

Let’s explore further whose account accommodates the linguistic data best.

Suppose that we are raising money for a charity by a walkathon. There are one hundred of us, and we each pledge to walk a mile for every sponsor we get. We each get ten sponsors. So together we walk 1000 miles for charity. Does [11] follow from [10]?

Obviously not. Indeed, in the situation described [12] is true, not [11]. [12] Is flatly inconsistent with [11]. But [11] follows from [10], on Blomberg’s account. Perhaps Blomberg would say that this is an example of the extreme hubris challenge, like claiming that you built the Great Wall of China when you laid a single brick, and that that’s why we react to the question whether [11] follows from [10] in the way we do. (But, really, did I build the Great Wall of China by laying a single brick?) However, that does not help with the fact that [12] is clearly true and incompatible with [11].

Change the example. I carry the bench from the ground floor to the first landing of the stairs. You carry it from the first landing to the first floor. [13] Is true. Does [14] follow from [13]?

No, obviously not. Indeed, [15] is true. But [14] and [15] are contradictory. For [16] is clearly true.

It is not extreme hubris for me to claim to have carried the bench to the first floor. That’s not the problem. The problem is that it is false that I did that, though it is true that we did it together. Let’s think about just lifting a bench. According to Blomberg, if we lift a bench together, we each lift a bench (I mean lift a bench completely off the ground). Is this true? Suppose that I pick up the left end of a bench, and the right end is still resting on the ground. Have I lifted a bench? No. I’ve lifted the left end of a bench. Suppose you come along and while I am holding the left end of the bench, you pick up the right. Then we have lifted a bench. Does it follow that I have now lifted a bench? Clearly, I only lifted the left end of the bench and did not lift the right end of the bench. But lifting the bench required that the left and right ends of the bench be lifted. So what I did is not sufficient for me to have lifted the bench.

Nigel voted for Brexit. 15,570,000 Pro-Brexit Britons voted for Brexit as well. [17] Is true. Does [18] follow given that Nigel is one of them?

Someone who thinks [18] follows from [17] does not understand how voting works. Blomberg might respond to this by saying it is an instance of what he calls the institutional action challenge, but the pro-Brexit voters do not constitute an institution in the sense in which, say, the Supreme Court does. They don’t know each other. They are not as such organized for collective action. They operate largely independent of each other in voting. They have institutional roles but not as pro-Brexit voters.

In any case, the idea that the group agent account must be right for institutional action sentences cannot be sustained. Presumably the idea is that in the case of institutional action sentences, the noun phrase introduces a group into the agency relation, rather than quantifying over its members. Leaving aside how nine justices on the Supreme Court are supposed to constitute a new agent, we can observe that this is incompatible with the linguistic data. In [19], the first clause is read collectively but the second distributively, but the noun phrase with its meaning is passed to the second clause without any change, since the elision is licit only when the meaning is the same across clauses. If ‘The Supreme Court’ is read distributively in the second clause, it is read distributively in the first clause. My account handles this, Blomberg’s suggestion does not. We can also reverse the order as in [20] where the first clause is distributive but the last is collective.

Blomberg notes that I also use examples of collective action verbs like “gather” and “meet” that don’t involve institutional agency to make the point the distributive reading of a collective action sentence does not follow from the collective reading. He admits this is a problem for his account. “Here there is clearly no additional (group) agent besides the individuals involved” (p. 20). But rather than giving up his view, he suggests that this “arguably shows that the scope of the multiple agents account is more limited than Ludwig takes it to be, not that the implicature account is mistaken” (p. 20). It is not quite clear what he means by this. He does not mean that there are group agents per se for gatherings or meetings. So I think he means it shows that clause (iii) in [2c] is appropriate only for noninstitutional collective action verbs (though even without (iii) the account is a multiple agents account). But in fact the problem is not with the plural action sentences but with the singular action sentences involving collective action verbs. “We gathered at the library” implies “I contributed to a gathering at the library” even if you add a requirement to the logical form of the plural action sentences that all and only those among us are agents of the gathering. Since Blomberg is arguing that “I A-ed” means “I made a contribution to an A-ing”, one would expect the entailment to go through smoothly here. Blomberg must maintain that singular action sentences with non-institutional collective action verbs have a sole agency requirement. This is ad hoc. But it is also surprising. Why would a collective action verb trigger a special semantic rule that does not govern non-collective action verbs that automatically made the singular action sentences containing it semantically false when their default meaning would make it true?

But it does not matter. As we’ve seen, the problems can’t be restricted to collective action verbs. We get the same problem with verbs that do not require multiple agents of the events they express. Change the example again. Does [22] follow from [21]?

Clearly not. But [22] does follow from [21] on Blomberg’s account. Blomberg might say that the verb here is collective. But it is not. Mr. Fantastic could surround the house by himself. A giant amoeba could surround the house. If it were a small enough house, I could surround it. So the verb is not collective. Thus, Blomberg is committed to [22] being true given that [21] is. But for ordinary houses, it is not true that I can surround the house.

Suppose my brother and I ate all the cookies before the party. There were thirty cookies. I ate ten of them and my brother ate twenty. Does [24] follow from [23]?

No. Clearly, [25] is true and [25] is incompatible with [24]. From [24] and the fact that there were thirty cookies [26] follows. (If we started with “We ate exactly thirty cookies” we could insert “exactly” after “ate” in [26]).

But [25] and [26] are jointly inconsistent. Worse, if I ate thirty cookies and there were only thirty cookies, it follows that my brother and I did not eat all the cookies because it follows from my eating thirty cookies and there being only thirty that my brother ate none, and if he ate none it follows that we did not eat the cookies, only I did.

Consider a symphony orchestra performing Beethoven’s 7^th Symphony. To perform the symphony, all the parts have to be played. I play the lead part for the French Horn. Have I performed Beethoven’s 7^th Symphony? It is clear that you cannot perform Beethoven’s 7^th Symphony just by playing the lead part for the French Horn. But on Blomberg’s account, I performed Beethoven’s 7^th Symphony by playing the lead part for the French Horn. Is “performing a symphony” a collective action verb? Maybe an orchestral performance is, but in principle someone with enough arms and mouths could perform a symphony by herself, and if we allow a performance to be pieced together by recording, even one person with enough versatility could play and record all the parts and play them back together.^[7]

Examples can be multiplied. In each case, Blomberg’s account makes the wrong prediction and mine makes the right prediction. The problem is that if I have made only a partial contribution to bringing something about, that is the extent of what I have done. To say that I have done more is false. So to say I have done something when I only did part of it involves us in a manifest contradiction. Thus, to say that if a group does something, each member of the group ipso facto does it, commits (in Blomberg’s phrase) a fallacy of decomposition. The bottom line is that the collective reading of plural action sentences does not entail the distributive reading. My account explains why. Moreover, since it is an important distinction, we should expect our language to encode it, and it does. The right conclusion to draw is that (iii) in [1a] is a semantic requirement on the truth of [1], and (iii) in [2c] is likewise a semantic requirement on the truth of the collective reading of [2]. If Bill, Ted, and Rufus carry thirty boxes upstairs, and Bill carries 5, Ted carries 15, and Rufus carries 10, then Bill and Ted carried exactly 20 boxes upstairs, and they cannot claim to have carried all the boxes upstairs because that requires carrying 30 boxes upstairs and we’ve just said they did not do that. But Bill, Ted and Rufus did carry all the boxes upstairs. This is predicted by my account.

4 I-Intentions and Collective Actions

Blomberg argues that I can intend to do something while being neutral about whether I will do it myself or with others. Thus, I can intend to make dinner. If I make dinner by myself as a result of that intention, then I successfully carry it out. If I make dinner with you, or we make dinner together, then I likewise carry out my original intention. I disagree. This disagreement is not independent, however, of the disagreement about the sole agency requirement on singular action sentences. If Blomberg is right, then the intention to make dinner is simply an intention to be a direct contributor to a dinner making event. It leaves open whether there are one or more direct agents of it. However, I have just shown that singular action sentences do have a sole agency requirement. Thus, if I intend to make dinner, I intend I make dinner, and so that I am the sole agent of a dinner making in the sense required by the verb. If I am right about the sole agency requirement, this settles the matter.

Blomberg adduces some independent arguments for his position. In the two assassins example, say Agent 13 and Agent 47 both intend to kill Dr. Otto Wolfgang Ort-Meyer by poisoning him, but are not aware of the plans of the other. They both poison him and he dies as a result of the double dose of poison. Were their intentions carried out successfully? I say that they misfired because of the sole agency requirement of singular action sentences. Each intended that he be the sole direct agent of Ort-Meyer’s death. Neither was fully successful. Blomberg argues:

… hold fixed an agent’s intention and causal contribution to an intended outcome. We then replace the causal contributions made by other agents who intend to bring about the same outcome with similar non-agential causal contributions. Here, I think it would be odd to say that the agent failed in doing what he intended before the replacement, but that he succeeded after it. If the agent succeeded after the replacement, then he arguably did so before the replacement as well. (p. 22)

Why would it be odd though, if “I poison Ort-Meyer” requires me to be the sole direct agent of the death?^[8] You might protest: but was not the assassins’ goal, namely, the death of Ort-Myer, something which each achieved? Yes, but this is not enough for their intentions to have been successfully carried out, as we know from cases of deviant causal chains. So we cannot infer from the execution of the intention bringing about the event expressed by the action verb that the intention was successfully carried out.

5 Conclusion

(1) Blomberg brings out an important point about the relation of simple primitive actions to composite primitive actions which can be the values of “f” in the account of the logical form of action sentences. This requires us to draw distinctions that were not clearly drawn in the book. This does not alter the basic account, but it calls for a refinement. What is the bearing of this on the question whether there are collective actions “in the primary sense”? There are similarities and differences between individuals and groups when the individuals and groups do things. There is something that the group does but not by doing anything else, namely, mereological sums of the simple primitive actions of their members. There are composites of these as well. However, in the ordinary case, no single agent stands in the primitive agency relation to those mereological sums and composites. Those are the facts. As long as we agree on the description of the phenomena that guides its application, the question of how to use “collective primitive action” is verbal rather than substantive. This is not to say that there are no considerations that can be brought to bear. I still think what I said in the book is right:

On balance, it seems best to dispense with the terminology of primitive actions when it comes to group actions. The best candidate is too distant from what counts in the individual case for it to warrant the literal use of the term, and its use in the extended sense invites misunderstanding. (Ludwig 2016, p. 176)

It would be a consideration in favor of using “collective primitive action”, even if not decisive, if it turned out that the category Blomberg has in mind could be shown to have an important distinctive explanatory role in our understanding of collective action. Even if this were so, it wouldn’t be decisive, since “the sum(s) of their primitive actions” would keep track of it as well as “their collective primitive actions” and have the advantage of being more transparent.

(2) Blomberg suggests that singular action sentences don’t semantically encode a sole agent requirement. If this is right, then the collective reading of plural action sentences entails the distributive reading. But this runs into systematic trouble. The trouble is that we recognize that saying you did exactly this much of something (that falls short of the whole) but no more is not compatible with saying you did it. This is well illustrated by the example our eating exactly thirty cookies together by my eating exactly ten and your eating exactly twenty. If I ate exactly ten, I did not eat exactly thirty. Cases in which one simply can’t (as a matter of fact) do the thing that group one is a part of does bring this out also. We surrounded the house but I did not. The collective reading of plural action sentences does not entail the distributive reading. The sole agency requirement blocks this entailment. It is a semantic requirement on singular action sentences that the subject agent is the only agent of the relevant sort of the consequent event.

(3) This settles the question whether one can have an I-intention that is successfully executed if I do not do what I intend by myself (in the relevant way). It may seem that you can hit what you aim at even if unbeknownst to you someone else is aiming at the same in the same way and it is overdetermined, but it can seem that you hit what you are aiming at in deviant causal chain cases as well, but nonetheless something goes wrong.