1 Introduction

The statement that we are currently failing to address some of humanity’s greatest challenges seems uncontroversial: our apparent failures include not limiting global warming to a maximum of 2 °C, not slowing down anti-microbial resistance and run-away environmental degradation, and exposing vulnerable people to preventable diseases when failing to produce herd immunity against infectious diseases. We regularly lament our failures to collectively produce highly desirable public goods: but how meaningful are such self-incriminations? When today’s teenagers accuse their parents’ generation of jeopardising their futures by leaving them a ruined planet are they justified in claiming that we—collectively—are failing them?

There is, naturally, much vagueness in ascriptions of lamentable collective failings. Individual inactions and omissions (more on the distinction later) have received a fair amount of attention in the literature (Bernstein, 2015; Clarke, 2010, 2012a, 2014; Sartorio, 2009; Talja, 1985; Willemsen, 2016; M. J. Zimmerman, 1981; D. Zimmerman, 1994). But hardly anything has been written on collective inactions and omissions (Copp, 1991; May, 1990, 1992; Petersson, 2008). The aim of this paper is to outline a meaningful notion of collective inaction: when does it make sense to say that we, collectively, failed at something?Footnote 1

The first thing to note is that many of the things we collectively fail to do are not actions any of us consciously choose not to perform. At times, we may not know enough about the issue at stake or be unaware of the potential consequences of our actions and simply not give the matter any serious thought. Other times, however, we might deliberately choose not to make our contribution to the realization of important collective goals.Footnote 2

In the first section of this paper, I distinguish between different ways of not doing something: non-action, inaction, and omission. Clearly, the list of things that we could be doing—individually and collectively—is infinite. But there is no point in including in our inquiry all the morally relevant (and valuable) things we could potentially do. When we think of noteworthy inactions, we do not usually use mere hypotheticals, but actions that are solidly possible and are relevantly connected to the agent in question.Footnote 3

In the second section, I develop a notion of collective inaction. In his 1990 paper, Larry May argued that “[f]or inaction to be collective there must be some sense in which the group as such failed to act” (p. 273). I will show that much more detail is needed here and will eventually come to define ‘collective inaction’ as a group not performing a collective action x (or not producing a collective outcome x)Footnote 4 that was collectively feasible at time t where each agent in that group had a reason to contribute to performing (or producing) x or others had reasonable expectation that they would perform (or produce) x. As is clear from the above definition, I will use ‘fail to act’ as the verbal phrase for ‘inaction’.Footnote 5

The third section illustrates my results using two examples of large-scale collective non-action: closing the global emissions gap and localized failures to produce herd immunity in the context of the 2017 measles outbreak in Europe. This will also serve to elucidate one of my main findings: collective inaction can be the by-product of individual omissions, that is, deliberate non-actions as well as the combined result of individual inaction (contra May, 1990).

At this point one might ask who is meant by ‘we’ and what kind of ‘collectives’ I am focusing on. This question goes to the very heart of my discussion, so rather than clarifying my usage of the term here, I assure the reader that this will be addressed in the course of the discussion. For now I can say that the focus is on so-called unstructured groups, that is, pluralities of agents that are not (yet) organised with regard to the collective task at hand: bystanders, passers-by, the affluent, etc.Footnote 6 That is, I do not discuss inaction of so-called group (or structured) agents, which are the focus, for instance, of Christian List’s and Philip Pettit’s 2011 book Group Agency.Footnote 7 The kind of pluralities of agents I am focusing on are not unified agents. I will dedicate some time to discussing the question of how one could possibly claim that unstructured collectives can have reasons to perform actions and what it means that they have the ability to act. The answers to these questions are not obvious, because the groups I am focusing on are not agents proper.

Before I continue, let me briefly pause and consider how this discussion is of wider philosophical interest. (i) First, it is adding a missing piece to action theory. If, as the literature on individual omissions and inaction suggests, it makes sense to separate the infinitely large set of non-actions from relevant inaction—more on this distinction in a moment—then it is important to do so for collective inaction also. An examination of the latter concept is currently missing in the literature. As we shall see, collective inaction is, in fact, a more interesting and fruitful concept than individual inaction. At face value, at least, there are many more (and potentially more important) opportunities to fail collectively than there are to fail individually, where we have no intention not to perform the action in question. (ii) Examining the notion of collective inaction can help us identify ascriptions of collective failings that are meaningful and distinguish them from those that are not. (iii) Relatedly, and perhaps most importantly, the notion of collective inaction may serve as the basis for ascribing responsibility and blame for collective failings.

2 What is inaction?

Starting with individual inaction, it may appear that it is easily defined: inaction is simply something one does not do. But a closer look reveals a complication, as John Kleinig illustrates:

At the present moment, for example, I am not walking a tightrope in Moscow or ringbarking a tree. However, though truly said of me, to assert that I am not doing these things is somewhat misleading, since it suggests that my not doing them is something I have failed, omitted or neglected to do. It is the latter which form part of the class of active nondoings. Failures, omissions and neglects can be said to be things that are done. They are acts, attributable to particular agents, and not mere act-descriptions. (1976: 393).

The task of drawing a distinction between attributable and non-attributable ‘nondoings’ is made somewhat more difficult by the fact that not all authors use the terminology in the same way.

Basically, there are three different concepts that need distinguishing:

  1. 1.

    The class of actions an agent does (or did) not perform as opposed to actions she does (or did) perform: I will call the former non-actions. In this I follow Jari Talja, whom I discuss in more detail below (1985).

    Example: John did not walk a tightrope in Moscow last night.

Non-actions can be either intended or unintended:

  1. 2.

    The class of actions an agent decided not to perform, as in actions she deliberately does not perform. I will call these intended non-actions omissions.Footnote 8

    Example: John decided against walking a tightrope in Moscow last night.

  2. 3.

    The class of actions an agent does not to perform without having any intention not to perform them: I will call these unintended non-actions inactions.

    Example: It has not occurred to John to walk a tightrope in Moscow.

My nomenclature departs from the use of these terms in some of the literature. This is inevitable since these terms are not being used consistently. For the purpose of this paper, I will adopt Larry May’s terminology, who in Sharing Responsibility (1992) distinguishes between collective ‘inaction’, which is non-intentional and collective ‘omission’, which is intentional. On his account, a collective omission exists where the group chooses collectively not to act (1992:107). In contrast, collective inaction is the failure of a ‘collection of people’ (roughly, what I am calling an ‘unstructured collective’) to act as a group when they could have acted as a group (ibid.), that is, in a coordinated manner.Footnote 9 I will turn to the specifics of collective inaction later. Inaction here means unintended non-action as opposed to action that is intentionally omitted.Footnote 10 As mentioned before, when I speak of someone ‘failing to act’ I am referring to ‘inaction’ that is attributable to that person. However, my notion of collective inaction will be broader than May’s, as I shall explain below.

Fig. 1
figure 1

Types of non-action

Full size image

At first glance, it might seem that the number of any agent’s inactions is potentially infinite (see May, 1992, pp. 88 and 94). After all, with everything we do, every action we perform, there are endless alternative actions that we could have performed instead without having actively chosen not to perform them. Yet, it seems ludicrous to include every non-action in the group of (morally) relevant inactions.

Philosophers have in fact used different criteria to separate ‘inactions’ from this vastly larger group of ‘non-actions’. Pascale Willemsen argues that inactionsFootnote 11 are those kinds of non-actions (or non-events) that agents can be expected to perform (Willemsen, 2016).Footnote 12 She identifies two types of expectations concerning inactions. The first is morally neutral: anything an agent can statistically be expected to do, e.g. because they regularly do it, like brushing their teeth every day. Further, there are normatively loaded inactions: anything an agent can be morally expected to do such as keeping their promise. Adding to Willemsen, I suggest including a third category that is not covered in her two subtypes. It is also normative and often implicit rather than explicit: the adherence to cultural norms like saying ‘hello’ to someone upon meeting them, or moving aside to let someone pass.

Talja (1985, p. 237) gives the following definition of (unintended) inaction (for which he uses the term ‘omission’, as explained above):

“An agent S omits to perform a iff.

(i) it is not the case that S performs a

(ii) S has the ability of performing a

(iii) S has an opportunity to perform a

(iv) S has a reason to perform a

Let me put the issues of ability and opportunity aside for now and come back to them later. Note that Talja also includes a normative concept into his definition, namely that of ‘reason’. He writes that

[t]he purpose of introducing the concept ‘reason’ is to differentiate between omission and nonaction. Nonaction comprises an infinite number of actions the agent is not doing at the moment. I am not starting a nuclear war (for which action I have no opportunity), I am not performing a 250 kg dead lift (for which I have no ability in my present physical condition), I am not throwing the typewriter out of the window (for which I have no reason). (p. 238).

According to Talja, if I do not have a reason to do something my not doing so is no inaction (omission on his terminology) but ‘merely’ non-action. Both Willemsen and Talja, then, can be said to apply some kind of normative standard to individuate inactions from the infinite set of non-actions. Talja’s and Willemsen’s normative criteria overlap, but not completely, since, firstly, we can have reasons to act in the absence of others’ expectations and, secondly, others’ statistical expectations do not necessarily indicate that we have a reason to act. While such statistical expectations can be thought of as normative in a broad sense (as in something being the ‘norm’ if it usually happens with regularity or frequency) they are not related to an agent’s reasons in the same way that moral and etiquette-related expectations are. Nonetheless, expectations regarding other people’s actions are, ultimately, (based on) beliefs concerning other people’s reasons. In other words, we expect you to act a certain way only if we believe that you have reason to act in that way. To sum up, an agent S’s non-action qualifies as inaction only if S had a reason to perform the action in question, or where others had a justifiedFootnote 13 belief that S had such a reason.

But this is still not satisfying: what kind of reason are we talking about? Talja does not specify whether he refers to pro tanto reasons, as in ‘considerations in favour of performing the action in question’ or all-out reasons, as in ‘overriding reasons to perform the action in question’. The first interpretation, would risk proliferating the cases of inaction to the point of meaninglessness. Almost any action that someone is in a position to perform has something counting in its favour and as such for almost any such action there is some reason to perform it. Even the action of throwing one’s typewriter (or laptop, nowadays) out of the window, to use Talja’s example. The second interpretation, on the other hand, would set the boundaries of inaction much more narrowly. If inaction occurs only where we have an overriding reason to act in the specified way (and we are capable and have an opportunity to do so) then one’s inaction or failure to act becomes shorthand for failing to meet one’s moral or other obligations, or the demands of rationality or expediency. This, in turn, is implausibly restrictive, since it would reduce inaction to what should have been done (where the ‘should’ need not be a moral one). I contend that ‘reason’ in the definition of inaction is best understood as ‘sufficient reason’,Footnote 14 a reason that is strong enough to warrant action and is defeasible:

There is sufficient reason for A to φ so long as there is not more reason for A to not-φ than to φ. There is sufficient reason for A to φ so long as it is not the case that A should not φ even if it is not the case that A should uniquely φ. There is sufficient reason for A to φ even if there is something other than φ-ing that A has just as much reason to do as φ-ing. (Rowland 2019: 14).

Let me briefly sum up: Non-actions are all the actions that do not get performed by anyone (in the broadest sense of ‘action’, including outcomes that are not being produced or attitudes that do not get adopted). Inaction is a type of unintended non-action that is attributable to a specific agent,Footnote 15 whereby that agent is capable of performing the action in question and has an opportunity to do so, but does not perform it, even though that agent has sufficient reason to perform that action or other agents have a reasonable (moral or other) expectation that they perform that action. In my discussion from hereon, I will set aside inaction where there was a statistical expectationFootnote 16, simply because those are not the cases I am interested in.

3 What is collective inaction?

It is now becoming clearer how a notion of collective inaction could be conceptually challenging: How should we understand the claim that unstructured collectives can have reasons to perform actions or expectations placed on them? What does it mean that they have the ability to act or an opportunity to do so? Since the groups we are focusing on are unstructured groups and not agents proper, the answers to these questions are not straightforward.

Let me narrow in on collective inaction by using a simple, small-scale example. For reasons that I will explain in a moment I will focus on actions (or outcomes) that necessarily require two or more agents in order to be successfully performed (or produced), such as singing a duet. Let us say that last Christmas two sisters—Mona and Hannah – did not sing a duet at their family Christmas dinner. Did they collectively fail to perform that duet? Since singing a duet is something that by definition requires two agents, individually, none of the two could be said to have failed to sing the duet.

What would make their not singing the duet a case of collective inaction? If individual inaction is understood as the non-performance of an action (or non-production of an outcome) by an agent where the agent has a reason to perform that action (or to produce that outcome) then a question concerning the nature of reasons for collective action cases arises.Footnote 17 Should these be understood as individual reasons or as collective reasons? A reason to act is always someone’s reason. So whose reasons are ‘collective’ reasons? I will abstain from positing (or rejecting) a notion of collective reasons as reasons of the group.Footnote 18 For my purposes it is sufficient (and completely uncontroversial) to assume that—individually—we regularly have reasons to perform joint activities or produce collective goods together with others.

We can think of these as group-based reasonsFootnote 19 in that they refer to actions (or outcomes) that can only be performed (produced) by several agents together. Even if such reasons are held by individual agents, they contain an essential collective element. The distinction between individual and group- (or pattern-) based reasons is described as follows by Christopher Woodard:

[A] pattern of action by the group is capable of providing a reason to perform a constituent part of the pattern, so long as the group could perform this pattern of action—where that means that each member is willing to play her part. (2003: 225)

In other words, performing a duet—a pattern of action by a group—can provide a reason for each of the potential singers to perform their part. These individual reasons to perform one’s part are group-based in the sense that they are reasons to perform an action that is part of a collective (or group) pattern.Footnote 20 On my view, these group-based reasons for one and the same collective action or outcome may well differ in their exact content. That is, one person’s reason for singing their part may differ from the other person’s reason. But they are nonetheless reasons to take part in the whole.

Collective inaction, then, obtains only where the agents who fail to act (together) each have group-based sufficient reason for that collective action (or for producing the collective outcome) or where others have legitimate expectations on those agents to act collectively. Otherwise, their failure to act will be mere non-action. Though Woodard’s group-based reasons seem to be primarily conceived as moral or justificatory reasons I see no problem with speaking of group-based reasons of expediency or other non-moral reasons.Footnote 21 In contrast, where agents have the collective ability to act but no relevant group-based reasons, their non-action would not be collective inaction.

At this point, we need to specify what it means to be collectively able to perform some action x. ‘Inaction’, Talja argued, is non-action that an agent has both the ability and opportunity to perform:

Here ‘opportunity’ refers to the agents’ physical (or human) possibilities, i.e., the agent has an opportunity for doing an action if nothing prevents his doing it, and if it is not impossible for him to do it. (1985: 237)

And “an agent S has the ability to perform an action, if in case he has an opportunity to perform it, he can (normally) if he so chooses.” (p. 238) According to Talja’s terminology, “‘ability’ is taken to refer to the ‘internal’ state of an agent, whereas ‘opportunity’ is the external situation in which the agent finds himself. ‘Ability’ must be relativized with respect to time.” (p. 238). I will henceforth speak of feasibility as combining both of these aspects—opportunity and ability.Footnote 22

Let me now turn to the question of what collective feasibility is—how should we understand it? If at a certain moment in time t, Hannah knows how to play guitar but Mona does not, and hence playing guitar is an action in Hannah’s option set but not in Mona’s, we may say that it is feasible for Hannah. But is it in any meaningful way feasible for them? It seems that an action that is feasible for Hannah in the above sense is also feasible for Hannah and Mona. This may seem odd but it need not concern me here since I focus exclusively on cases where a particular action or outcome must be performed or produced by two or more agents. Collective inaction then is non-action where it is feasible for two or more agents to collectively produce an outcome or to collectively perform an action (and they each have group-based reasons to do so). Collective feasibility minimally requires that the individual contributory actions are compossible.

Note that—so far—my way of framing collective inaction includes cases where agents produce an outcome together with someone who is unknown to them, someone of whose very existence they may be ignorant.Footnote 23 It may be collectively feasible for you—the reader—and a Tuvaluan whose acquaintance you have never made to produce some outcome that can only be collectively produced. This is, ultimately, how I will understand collective feasibility. Those who are uncomfortable with this idea probably prefer what I call a strict reading of collective feasibility.

On the strict interpretation, collective feasibility means the feasibility of collective action in the narrow sense. By collective (or joint) action in the narrow sense I mean action where the individual group members form plural intentions.Footnote 24 The strongest form of plural intentions are ‘interlocking’ intentions, which feature prominently in Michael Bratman’s account of shared agency (2014), but also appear under a different description in Pettit and Schweikard (2006). On Bratman’s account, two agents’ intentions regarding a shared activity j are ‘interlocking’ if each agent intends that they j “by way of” the other’s intention that they j. This is the case when “the joint activity both is in accord with and is in part a result of that intention of the other” (2014, pp. 49–50). Such an account of plural intentions and joint agency is fairly demanding.

It is this type of collective action that Björn Petersson seems to have in mind when he argues that we can only be collectively responsible for failing to act where we could have performed a collective action (2008). Petersson thinks that joint action is not possible in large-scale scenarios where no direct communication between group members is possible precisely because plural intentions of the kind he and Bratman have in mind require communication between members of the collectivity. On a strict understanding of collective feasibility then, collective non-action only qualifies as collective inaction where it was feasible for the members of the collective to perform a collective action in the narrow sense just described. In this narrow sense, collective action features intentions that are interlocking and this, in turn, requires second-or-higher-order knowledge amongst the members. On the strict reading, then, collective inaction occurs only where agents are in a position to have some form of interdependent, second-or-higher-order knowledge.Footnote 25

However, weaker notions of plural intentions are available. Minimalists (or reductionists) like Ludwig (2016) argue that plural intentions are individual intentions with a special plural content (e.g. a collective action). In contrast to Bratman, Ludwig does not think that individuals necessarily hold plural intentions “by way of” another agent’s plural intentions. That is, while plural intentions can be interlocking in that sense they need not be. Minimalists like Ludwig see little problem in claiming that agents can form plural intentions and perform collective actions even where people are unlikely to (be able to) communicate with one another.Footnote 26

On the wide interpretation of collective feasibility, a collective outcome or action is collectively feasible if there is a significant probability of a combination of feasible individual actions generating that collective action or joint outcome. These include the above-described collective actions in the maximalist and the minimalist sense (and everything in between), but also aggregate actions (and outcomes) such as the (aggregate) reduction of greenhouse gases or of our global meat consumption generated by individual behavioural change.Footnote 27 On this view of collective feasibility, the question of plural intentions is irrelevant. For my purposes, it makes sense to work with the wide interpretation of collective feasibility so as to not be unduly restrictive.Footnote 28

Let me return to my example. Suppose that it just did not occur to Mona to sing a duet with Hannah, while Hannah was keen and ready to sing with Mona (but did not make this known to her sister). Imagine that Hannah was singing her part of the duet while Mona simply did nothing. Was it not Mona who was failing to act and to whom the inaction should be attributed (rather than attributing it to them both)? After all, Mona is the one who undermined the joint performance. Clearly, they did not sing a duet, but it seems odd to suggest that they collectively failed to do so. After all, Hannah is part of the collective and she did something—she sang her part, that is, she made her contribution to the joint endeavour. And yet, this is precisely how I will understand collective inactions in this paper: the sisters together—or jointly—failed to sing a duet and they did so without any intention to not sing a duet (though their individual intentions differ: Hannah intended that they sing a duetFootnote 29 but Mona had no intention either way concerning the duet), all the while there was sufficient reason to sing a duet,Footnote 30 and singing a duet is a collective good that necessarily requires two agents in order to be produced. Taken together, these features make their non-performance of a duet a case of collective inaction. The example shows that collective inaction is perhaps not an intuitive category at the first glance. However, at the second glance, the fact that they failed at something when one of them did not contribute should make more sense: Whereas Hannah did not fail to sing her part of the duet, together they failed to sing the duet. There was collective inaction as far as the duet was concerned.

While I defined collective inaction as collective non-action without any intention not to act, it need not be the case that the individual contributory non-actions are inactions (that is, unintended). In other words, collective inaction is not necessarily the product of (or combination of) individual contributory inaction, even though it can be. Collective inaction—as I understand it here—can be the byproduct of individual omissions, that is, deliberate non-actions.Footnote 31 For instance, take the case where people intentionally refrain from vaccinating their children because they are worried about the health impacts of vaccines. Let me further assume that they are unaware of herd immunity—the eradication of a pathogen from a population and the ensuing protection of those who are not vaccinated—and are therefore not intending to undermine that public good. Unintentionally, by refusing to vaccinate they collectively compromise (or fail to generate) herd immunity. Their collective failure to generate (or maintain) herd immunity is not intended and counts, therefore, as collective inaction even when the individual contributory actions are omissions, i.e., intentional non-action.

In sum, it makes sense to think of collective inaction as the kind of inaction that concerns the production of collective goods or the performance of joint activities. Collective inaction proper is to be distinguished from cumulative individual inaction, which is merely the aggregation of individual inaction. For instance, where you and your neighbour each fail to trim your hedges in autumn the sense in which you both failed to trim your hedges is that of cumulative individual inaction. The phrase: “you and your neighbour failed to trim your hedges” is grammatically equivalent to statements concerning collective inaction proper such as “you and your sister failed to sing a duet”. However, in the first case we are merely expressing that two people each failed to perform the same act (type) while in the second case we assert that two people failed to perform a token act together.

Here is a potential challenge to my account: One might also think that the following failures to act jointly qualify as collective inaction: Think of all actions (or outcomes) that can be performed (or produced) by two or more agents together as joint capacity cases. These can be contrasted with actions that can only be performed by maximally one agent—individual necessity cases.Footnote 32 Amongst joint capacity cases, there are countless actions that could be performed either by an individual or a group. They are the things that we can but need not do together. For instance, lifting a not-too-heavy table: we could each do it on our own, but it is easier to do together. Now imagine that there are two of us in a room with a table and neither acts: we do not lift the table together or by ourselves. It is unclear at this point whether our not-lifting-the-table should be considered individual or collective inaction, or possibly both. It would seem odd if much of our individual inaction were simultaneously considered collective inaction purely for the fact that many things that can be done individually could in principle also be done collectively. For simplicity’s sake I focus only on actions and outcomes that require two or more agents to be realized, that is, on joint necessity cases. The two sisters not performing a duet is collective inaction because it is impossible to individually fail at this kind of thing. Singing a duet is—by definition—a type of action that can only be performed by two agents.

4 Two examples and a challenge

Can we meaningfully claim that we are failing to save the planet? Let me narrow this down: are we failing to close the emissions gap? That is, is not closing the emissions gap collective inaction attributable to a concrete set of agents—‘us’? The emissions gap is the gap between the greenhouse gas emission reductions required for limiting global warming to a maximum of 2 °C and the reductions that countries across the globe have currently committed to (UNEP, 2020). Given current GHG reduction trajectories based on UNFCCC member states’ reduction commitments we will most likely not curb global warming at or below 2 °C. This is my first example of large-scale non-action.

It is safe to assume that this is no collective omission since there is no deliberate (collective) choice to not close the gap. But who is meant by ‘we’? Is there an identifiable group (or set) of agents currently failing to close the emissions gap? Scientists and policy experts identify a range of agents across the globe who can jointly close the emissions gap, these include governments (through regulations and incentives), agents in the private sector (through setting industry standards and self-regulation), but also individuals—through behavioural change and lifestyle choices (Dietz et al., 2009; Ostrom, 2010; UNEP, 2020; Wynes & Nicholas, 2017).

Above, I suggested that an action or outcome is collectively feasible if there is a significant probability of a combination of feasible individual actions generating that action or outcome. I specified that an agent’s inaction ¬x at time t is the non-performance of action x where x is feasible for the agent at time t, in the sense of the agent having both opportunity and ability to x at t. It is difficult to see how even for the loose set of agents containing all the abovementioned actors—governments, corporations, ordinary citizens—the closing of the emissions gap is not feasible. Based on expert analyses such as the annual UNEP Emissions Gap reports one can only conclude that both opportunity and ability exist.

Further, the individual (and group) actors that form part of this loose set of agents have at the very least pro tanto reasons to contribute to (or partake in) a pattern of action that would correspond to the closing of the emissions gap. However, one might doubt that we can have reasons that are sufficiently strong to warrant action. For instance, it could be suggested that in some collective action scenarios we have no reason to partake either because our individual mitigation actions make no difference to the better (the ‘no-effect-view’) or because our individual failure to mitigate makes no difference for the worse (the ‘no-harm-view’) (see Schwenkenbecher, 2014). In the case of reducing GHG, adopting individual behavioural changes, which may be costly for us, is not going to make a difference to whether or not the emissions gap will be closed. Further, an individuals’ decision not to change their behaviour and reduce their carbon footprint is not going to prevent the emissions gap from being closed, either. Clearly, this will not apply to everybody, since those in positions of power can well make a difference individually. But it applies to the vast majority of individual citizens and consumers who by changing their habits may in aggregation substantially reduce GHG.

These sceptical views have been variously defended by a small number scholars, most prominently Sinnott-Armstrong (2005), while a significantly larger number of scholars have tried to find solutions to the ‘inefficacy problem’, as Nefsky (2019) calls it (e.g. Kagan 2011; Nefsky, 2011, 2017; Parfit, 1984; Pinkert, 2015; Woodard, 2003). I do not have the space to discuss the issue in detail, though I clearly sympathise with the latter, non-sceptical view and have argued for such a position in more detail elsewhere (Schwenkenbecher, 2014, 2021). As such I will only be able to draw a conditional conclusion regarding our first example:Footnote 33

To the extent that we can have sufficient reasons to partake in collective endeavours even where our individual actions make no difference to whether or not the desired collective outcome is produced, it appears that not bridging the emissions gap (so far) is truly our collective failure where ‘us’ means the very large and diverse set of agents who have opportunity and ability to perform actions that would together result in a significant reduction of GHG. Saying this does not prevent us from ascribing a similar failure to subsets of agents: the set of governments who form part of the UNFCCC; the set of large corporations that have been falling short of taking responsibility or—worse—have been actively undermining efforts to curb global warming. Today’s teenagers, then, can plausibly maintain that their parents’ and grandparents’ generations are failing to adequately address climate change. This does not (yet) settle the question of responsibility or blame. However, that some inaction can be thus ascribed to an agent (or set of agents) seems to be a necessary condition for apportioning responsibility and blame. I will briefly comment on collective blame below and again in the final section.

My second example was already mentioned: it is the failure to realise herd immunity against an infectious disease in a specific community as a result of falling vaccination rates. Herd immunity exists where a group is protected from an infectious disease because there are not enough potential carriers of the disease. As a result, those who for some reason cannot be vaccinated (e.g. they are too young to be vaccinated or are immunocompromised) are protected from those diseases. Local outbreaks of diseases for which effective vaccines exist, such as measles, for instance, can and do occur when too many individuals in a community are not vaccinated. This is often a result of individual agents’ decisions rather than lack of public health infrastructure. For example, in 2017, contractions of measles in Europe increased by 400%. In order to realise herd immunity a large-enough proportion of the population must be vaccinated. The threshold for herd immunity differs for each disease; for measles a 95% vaccine coverage rate is required (Funk et al., 2019). Herd immunity is only jointly achievable.

This presents us with a tricky case: the 2017 measles outbreak across Europe was the result of declining vaccination rates. According to Zsuzsanna Jakab, WHO’s Regional Director for Europe at the time, the outbreak was threatening the previous progress made towards elimination (World Health Organisation, 2017). Was the compromising of herd immunity (and the setback regarding elimination of the measles pathogen) a case of (or the result of) collective inaction?

Presumably, those who decided against the vaccine did not aim to undermine herd immunity, that is, their intention was not focused on this collective outcome. Typically, those opposed to vaccines cite other reasons for their refusal to vaccinate: worries concerning detrimental health effects, concerns regarding the freedom of persons and a scepticism towards government interventions in ‘personal’ issues such as bodily health and integrity. The loss of herd immunity—to the extent that people are familiar with the concept—is usually accepted as a form of collateral damage rather than being an intended outcome. If that is so can we safely conclude that this was no collective omission, even if at the individual level the refusal to vaccinate would classify as an omission?

Debates surrounding the Doctrine of Double Effect regularly concern what counts as ‘intended’ versus ‘taken into account as a side effect’. In discussing the ethics of war and violence and the principle of noncombatant immunity, Tony Coady warns of the dangers of ‘double-think’ (2008: 138): declaring non-desired effects of one’s action to be unintended, thereby narrowing the meaning of the term ‘intention’ to include only the desired effects of one’s action. Are those who refuse vaccines engaged in double-think? I think the case against them is weakened by the fact that no individual defection makes a discernible difference to the collapse of the public good. It is fairly plausible to intend to be unvaccinated but to also not intend to undermine herd immunity—just as plausible as saying that one did not intend to create a desire path when crossing the lawn (even if fully aware that if many people cross the lawn in the same spot a path will form). If this is so, then it is plausible that the loss of herd immunity in the 2017 measles outbreak was not intended and therefore not the result of collective omission.

But does it make sense to say that the above-described outbreak that resulted from compromised herd immunity was the consequence of collective inaction? This seems peculiar because the vast majority of people did get vaccinated. They contributed their share to the public good. Would it not be more appropriate to say that those who did not vaccinate failed to do their share rather than saying that all members of affected populationsFootnote 34 were collectively failing to produce herd immunity? I believe that this concern mainly reflects how easy it is to confound what it means to fail to do something (that is, to identify inaction) and what it means to be responsible and to be blameworthy for such failure or inaction (that is, to evaluate inaction). Even if people were collectively failing in producing herd immunity this does not mean that everyone was (equally) to blame for that (or held equal amounts of individual responsibility).

Let me now turn to a challenge: Given that I adopt a wide notion of feasibility, does my analysis invite random ascriptions of collective failures after all? For instance, are the author of this paper, King Charles III, and Fijian rugby player Akuila Uate currently failing to jointly perform a(n online) dance routine? I should not think so. There are (pro tanto) reasons that generally speak in favour of dancing. But even if the agents in the above set are capable of performing a dance ‘together’ and might be said to have an opportunity to do so (given the freedoms of online participation in joint action), we are not currently ‘failing’ to do so. First and foremost, the aforementioned agents do not share a reason to perform a dance routine together. In other words, none of us have reasons to perform a dance routine with the members of this specific group that go beyond general pro tanto reasons to dance. Neither do there appear to be any reasonable expectations by others that we do so.

This suggests that in order for sets of unstructured agents to have the right kind of (sufficient) reasons to collectively act or produce outcomes, there must be something that singles out the specific joint action from the vast array of feasible joint actions. Naturally, there is no cut-and-dry formula for how pro-tanto reasons turn into sufficient (or even all-out) reasons for an action, so it is along these lines that ascriptions of collective inaction will likely be most contested.

5 Summing up

I argued that we can meaningfully speak of collective inaction whenever it is the case that a group of agents does not perform a collective action x (or does not produce a collective outcome x) that was collectively feasible at time t where each agent in that group had a reason to contribute to performing (or producing) x or others had reasonable expectations that they would perform (or produce) x.

As we have seen, our account produces a couple of mildly surprising, but hopefully acceptable insights:

  1. (1)

    Collective inaction can take place where only one person fails to do their share. I suggest that whether or not it is appropriate to focus on the joint aspect of some collective failures over the individual inaction or omission causing it depends on the context; and

  2. (2)

    We may not always be able to determine whether a particular failure is really inaction: We may lack the knowledge of whether or not other agents had sufficient reason for contributing, for instance.

Other, perhaps less surprising insights include the following:

  1. (3)

    Collective inaction can result from individual inaction or individual omission. We—collectively—may fail at producing morally beneficial goods regardless of whether or not we—individually—intend to do so. Unintended social outcomes can still be relevantly linked to us in a way that allows for ascribing responsibility (backward-looking and forward-looking) and potentially blame. This holds true even for very loose set of agents. Again, take the issue of collective failures to produce herd immunity. While those who do not get vaccinated against an infectious disease do not necessarily intend to undermine herd immunity, they nonetheless are party to the collective failure. In fact, raising awareness for the collective dimension of our individual choices frequently lies at the core of attempts to overcome collective action problems. The formulation and propagation of a shared goal, an emphasis on how ‘we are all in this together’, and clear messaging on how each of us can individually contribute are elements of successful mobilization strategies. Singling out specific joint actions and individual contributory patterns as a way of overcoming collective inaction is at the core of public information campaigns, including health campaigns.

And, finally,

  1. (4)

    For very large and dispersed sets of agents there is often no meaningful way in which they can be said to have collectively failed where these agents lack collective ability and opportunity to produce the outcome in question. Thus separating collective inaction from mere non-action prevents pointless proliferation of collective failings and potentially counterproductive ascriptions of blame and responsibility.

I did not discuss the question of blameworthiness for collective inaction—this issue surely merits a paper in its own right. So let me finish off with only the briefest of comments on the direction such a paper could take and the questions it would need to address. Footnote 35 Suffice it to say that ascriptions of collective inaction narrow down the field of non-action sufficiently to allow for meaningful investigations into a collective’s blameworthiness. In order to ascribe blame for collective inaction one would need to establish whether or not the set of agents in question ought to have acted all-things-considered, that is, whether or not the reasons for acting were sufficiently strong to yield an all-out obligation to achieve the collective outcome. Naturally, holding sets of agents collectively blameworthy comes with its own challenges.