States’ culpability through time

Abstract

Some contemporary states are morally culpable for historically distant wrongs. But which states for which wrongs? The answer is not obvious, due to secessions, unions, and the formation of new states in the time since the wrongs occurred. This paper develops a framework for answering the question. The argument begins by outlining a picture of states’ agency on which states’ culpability is distinct from the culpability of states’ members. It then outlines, and rejects, a plausible-seeming answer to our question: that culpability transmits from a past state’s action to a present state just if the two states share a numerical identity, for example as determined by international law. I advocate a different answer: culpability transmits from a past action to a present state to the extent that the present state ‘descends from’ the aspects of the past state that underpinned the past action. One potential upshot is that some present-day settler-colonies (such as Australia) are culpable for the centuries-ago invasion of their lands by European powers—even though these states did not perform these invasions and indeed did not exist at the time.

1 Introduction

Consider contemporary injustices against Indigenous peoples in settler-colonies, such as the United States, Canada, and Australia. These injustices have a historical dimension: they can be traced to actions of invasion, dispossession, and genocide that occurred hundreds of years ago. The moral status of the ongoing injustices depends, in part, on their historical dimensions: it matters, morally, that the contemporary injustices are perpetrated against people whose ancestors were here first and whose sovereignty was violated. But who bears culpability for those historical wrongs? Is culpability borne merely by the individuals involved, who are now long-dead? Are those long-ago actions attributable to the ideology of the time, which persists in various ways but which seems to lack the agency necessary for culpability? Or are there presently-existing agents that bear culpability?

When answering the last question, it’s natural to turn to states. These are entities that persist through time and seemingly have the agency necessary for blameworthiness. So, we can blame, for example, Britain, France, Spain, and Portugal for centuries-ago invasions, dispossessions, and genocides that have ongoing manifestations.¹ Yet there is a problem with blaming states: the states that perpetrated the initial wrongs in settler-colonies are often not the states that now govern the relevant territories. Lands that were invaded by Britain (amongst others) are now governed by the United States, Canada, Australia, and New Zealand, for example. It is against these latter states that some Indigenous groups express anger, resentment, and indignation—the paradigmatic blaming attitudes.² Are these states morally culpable for the initial actions of invasion, dispossession, and genocide, in the sense of owing apology, redress, and reconciliation for these actions—actions that were not performed by these states and that occurred before these states even existed?

I aim to make progress on this question, by addressing an underlying philosophical question: under what conditions does culpability transfer from a past state’s action to a present state? Section 1 presents a framework on which states are agents whose culpability is irreducible to that of their members. Section 2 asks: what relationship must a contemporary state bear to a past state’s action, for the contemporary state to inherit culpability for that action? I argue against a seemingly-promising answer: that culpability transmits just if the two states are numerically identical, for example as determined by international law. The relation of ‘numerical identity’ holds between an earlier state and a later state just when these (time slices of) states are (time slices of) one and the same entity.

Instead, Sect. 3 argues that culpability transmits from a past state’s action to a present state to the extent that the present state descends from the agential resources of the past state that underpinned the past action. The relationship of ‘descent’ is one in which there is an unbroken cross-temporal agential stream, where the agential resources that underpinned the culpable action morphed into the (central and endorsed) agential resources of the present state, without requiring a relation of numerical identity. Importantly, this descent relation can hold across secessions (fissions) and unions (fusions), and indeed across the founding of (what we would ordinarily treat as) entirely ‘new’ states. I close by briefly addressing an example: Australia’s culpability for Britain’s invasion of the Australian continent.

Before beginning, an initial concern. Given the extensive contemporary manifestations of historical injustices, it might seem unnecessary to concern ourselves with moral culpability for the historically distant actions that precipitated the historical processes. Why not simply turn to those present-day states and individuals who are beneficiaries of historical wrongdoings, who perpetrate contemporary wrongdoings, or who preserve the unjust structures that were created by those historically distant actions? Because culpability for historically distant actions vindicates particular claims in the present: claims to apology, redress, and reconciliation for the historical action.³ On typical approaches to moral wrongdoing, only culpable parties can have obligations of apology, redress, and reconciliation. After all, these processes are about restoring relations between the culpable party and the wronged party. While beneficiaries and preservers of injustice do have obligations, these cannot be obligations to apologise, redress, or reconcile for the initial wrongdoing itself. It is therefore worth establishing this culpability specifically.

2 States as culpable agents

States are not merely collections of citizens, nor merely collections of leaders and officials. This is important: it means the culpability of the state does not imply the culpability of the state’s members, whether citizens or officials.⁴ If Australia is culpable, this does not imply the Prime Minister is culpable—still less that the any given citizen-on-the-street is culpable. If Australia is culpable, then apology, redress, and reconciliation are owed from Australia itself. The relevant actions will be performed by members in their membership capacity—for example, the Prime Minister might give an official apology. This does not necessarily entail—though it does not preclude—that these members are personally culpable.⁵ By asking which states are culpable for historical wrongdoings, we therefore take up a different subject-matter than those philosophers who have asked after the culpability of present-day citizens.⁶

We can conceptualise states as distinct agents by understanding them as realizations of structures.⁷ A state’s ‘structure’ is a recipe: the structure dictates how the state’s components must relate to each other, if the state is to exist (be realised). When the recipe is followed—when the components relate as it dictates—then the structure is ‘realized.’ For states (as for other organizations), the relations between the components might be represented in an organizational chart or constitution. Such a chart or constitution might be formally written down, but need not be. For states, the ‘components’ include humans (e.g., legislators, civil servants, judges, voters), perhaps alongside inanimate objects (e.g., parliamentary chambers, courtrooms, ballot papers) that play roles required by the ‘recipe.’ Just as we have no table when the legs and top are separated and strewn across the room, so we have no state if the relevant humans do not relate in the right way, where ‘the right way’ is specified by the state’s structure, charter, or constitution.

Thus, states are not merely collections of people: they are collections of people structured according to the recipe. Moreover, the state persists through changes in who realizes the structure: when the Prime Minister is replaced, the state persists through this change—much as a table persists when one of its legs is replaced. Indeed, states persist through wholesale changes in their component parts—as when all members die and are replaced through the course of a century. Such replaceability highlights the fact that states are not merely collections of people; they outlive any such collection.⁸

In being realized structures, states are analogous to tables or cars. But states are not inert or mechanical. States are agents. To be an agent is to be capable of action, which requires attitudes such as intentions, beliefs, and preferences. For the state to be an agent distinct from the collection of humans that constitute it, the state must therefore be capable of having attitudes with different content from the attitudes of the humans that constitute it. It must be possible, for example, for the state to want a top tax rate of 45% even though no member wants this. Most often, such possibility arises through compromise: if some Cabinet members want a 50% top tax rate, while others want 40%, then 45% might emerge as the state’s view even though no member wants exactly this. Without the possibility of such departure (of the state’s attitudes from members’ attitudes), we could identify the state-as-agent with the collection of members-as-agents. Furthermore, for the state to be an agent, it must be capable of aiming to approximate norms of rationality.⁹

For a state to have distinct attitudes and aim to approximate norms of rationality, it must have some procedure for arriving at those attitudes and following those norms. Procedures will be important for the arguments to come about culpability through time. Procedures can include formal aggregation rules,¹⁰ alongside bargaining, debate, compromise, command, consultation, conversation, and so on—as seen within legislative chambers, Cabinet meetings, court judgments, regulatory bodies, and polity-wide deliberative processes. A state’s procedures need not be formal or written down, just as its structure (recipe) need not be written in a formal constitution. But some procedures must exist, if the state is to arrive at its own attitudes that approximate rationality.

Such procedures also enable a state to alter its attitudes over time. I shall assume that such alterations are rational only if they are made in accordance with the state’s procedure (including the state’s meta-procedure for changing the procedure, meta-meta-procedure for changing the meta-procedure, and so on—which may be culturally implicit rather than institutionally formalised¹¹). Given my permissive characterisation of states’ procedures, this assumption amounts to the idea that entirely random or exogenously-manipulated changes in attitudes undermine the state’s rationality and, thereby, its diachronic agency.¹²

Alongside structures and procedures, there is a third element to states’ agency: values. While the structures and procedures are the ‘formal’ aspects of a state’s agency, the values are the ‘substantive’ aspect. Values give content, guidance, and purpose to the structures and procedures. I use ‘value’ synonymously with ‘project’ or ‘purpose.’ This substantive aspect is emphasised by Carol Rovane, who argues that each group agent is unified by a project, which might be discontinuous with the projects of members. A group’s project can be “any project that requires a substantial coordinated effort on the part of many persons.”¹³ Peter French’s theory of corporate agency similarly centres purposes: French states that all for-profit corporations have a purpose akin to “make automobiles, airplanes, breakfast cereals, dog food, etc.”¹⁴ For French, “an act is corporate not only because of its form [that it follows structures and procedures], but because of the policy it instantiates, displays, or manifests.”¹⁵ Thus, values—broadly construed to include projects, purposes, and policies—arise in prominent theories of group agency, alongside structures and procedures.

For many states, their guiding value, project, or purpose is (something like) ‘presiding over this territory’ or even ‘presiding over this territory democratically and liberally,’ if its structure (constitution) encodes a commitment to liberal democracy. But states’ values often include more than this—such as a national or religious culture,¹⁶ if such a culture has substantial sway over how the state’s structures and procedures are enacted. A state’s values can include a commitment to an imagined history, an aspired future, or a present ethos. A state’s values can of course change over time (as can the structures and procedures, via meta-procedures). Analogously, an individual agent’s guiding values can change over time: states are no fickler than individuals in this regard. An agent’s values orientate the agent to the world, despite values’ dynamic nature. Values, structures, and procedures give different states distinctive characters.

Understood in this way, states are candidates for culpability ‘in their own right,’ that is, culpability that neither entails, nor is entailed by, culpable members. For a state to be culpable, it’s sufficient that its structures, procedures, and values render it capable of processing moral reasons, where the state has failed to adequately process some moral reason without excuse.¹⁷ When this happens, states are apt targets of negative reactive attitudes, including resentment, indignation, and anger. It is also apt for them to hold self-directed negative reactive attitudes, such as remorse and guilt.¹⁸ I take the aptness of negative reactive attitudes—and the aptness of acts like apology and contrition—to be the marker of moral culpability.¹⁹ Reactive attitudes can be ‘apt’ even if no one has the standing to hold the attitudes (for example, because anyone’s doing so would be hypocritical) and even if no one should all-things-considered hold the attitudes (for example, because doing so would produce bad consequences). It is consistent with my argument that culpability is cancelled by adequate displays of apology, redress, and reconciliation; this is irrelevant since, in the cases I discuss, such actions have not occurred sufficiently to cancel culpability.

States’ irreducible agency and culpability is important for historical injustice. Certainly, many individual agents were culpable for historical injustices. When those individuals were alive, we might have queried the need for irreducible state culpability. But those individuals are now dead. To vindicate blame towards presently existing agents, our best bet is to attempt to vindicate blame towards states as irreducible agents. With this picture of states’ irreducible agency and culpability in-hand, let’s examine their culpability through time.

3 The identity view

Which contemporary states are culpable for which historical wrongs? An intuitive first pass is The Identity View:

(1) A contemporary agent’s culpability for some past act presupposes that the contemporary agent shares a numerical identity with (one of) the agent(s) who was culpable for that past act when it was performed. In a slogan: ‘diachronic culpability presupposes numerical identity.’

(2) States’ numerical identities are constructed through international law.

Therefore,

(3) A contemporary state can be culpable for some past act only if that contemporary state shares a numerical identity with a state who was culpable for that past act when it was performed, where numerical identity is determined by international law.

The identity view allows that one state can be culpable for what another state does. For example, the identity view allows that present-day Britain could be culpable for Australia’s 20th century treatment of Indigenous Australians, if and because Britain (say, in the 19th century) caused Australia to enact that treatment. In that case, 19th century Britain was the—or, at least, one of the—agents who was culpable for Australia’s 20th century treatment of Indigenous Australians, when that treatment was first performed. And if Britain was culpable for that treatment when it was first performed, then Britain can remain culpable for that treatment in the 21st century. In other words: the identity view allows that both Britain and Australia can be culpable for Australia’s actions, and that both can retain that culpability through time. However, the identity view holds that culpability can be transmitted through time—can be ‘diachronic’—only via the relation of numerical identity. On the identity view, I now can be culpable for what you did in the past only if I now am numerically identical to someone (i.e., me) who was culpable for your behaviour when you initially performed it—for example because I caused that behaviour, or was complicit in it, or controlled for it, or so on.²⁰ Once those initial culpability assignments have been made, those assignments can be passed on through time only via the relation of numerical identity. Or so the identity view says.

Several philosophers imply (1) as applied to organizations.²¹ Janna Thompson writes: “States, corporations and churches, through their decision-making and executive processes, deliberate, decide, and act. … Since these collectives persist through time and, in many cases, through the generations, so, it seems, do their responsibilities.”²² Likewise, (1) is presumed in Stephen Winter’s defence of legislatures’ apologies for historical wrongdoings,²³ and in Angela Smith’s defence of universities’ apologies for their involvement in chattel slavery.²⁴ Other philosophers working on collective responsibility are more cryptic. For example, Paul Sheehy writes:

Perhaps it is legitimate to hold a group responsible today for the deeds of the past if it has continued to endorse and sustain the attitudes and practices salient in the production of those past actions.²⁵

Likewise, Mihailis Diamantis proposes:

The essential trait of a criminal corporation is whatever trait caused it to commit the crime. … If a criminal corporation retains its criminal essence to the present day, our intuitions should be that the corporation retains its identity.²⁶

On Diamantis’s view, culpability is retained if ‘identity,’ meaning a criminal trait, is retained. Charitably interpreted, Sheehy’s use of ‘continued’ and Diamantis’s use of ‘retains’ should be taken to imply numerical identity between the earlier and later organization. Otherwise, their views would imply that any organization who endorsed (and still endorses) certain attitudes, or who had (and still has) certain traits, would be responsible for deeds that were caused by those attitudes and traits, but with which the organization has no causal or agential connection.²⁷ I take the language of ‘continued’ and ‘retains’ to imply a commitment to (1).

Others are sceptical about states’ culpability through time, precisely because they presume (1). For example, David Miller writes: “demonstrating the identity of states over time is generally much more problematic than the state-centred approach [to assigning culpability for historical injustice] assumes. … Of how many … European countries, for instance, could one say that they are governed by the same state that governed them in 1750, in the light of the radical disruptions that have occurred meanwhile, including territorial expansion and contraction as well as regime change?”²⁸ On this basis, Miller attributes culpability for historical injustices to nations (understood as cultural groups), rather than states. This further demonstrates that (1) has good standing among philosophers working on collective responsibility.

I will question both (1) and (2). My argument against neither is dispositive, which is why I target both: my combined arguments motivate the search for an alternative picture. My arguments against (1) will function regardless of whether states’ identity is grounded in international law or in something else.

As a first move, notice that (1) is dubious when applied to humans. As David Shoemaker explains, this is true on the two most popular accounts of numerical identity for humans.²⁹ Suppose one holds the ‘somatic’ view, on which two humans (one past, one present) are numerically identical just in case they have the same body. Then, (1) rules out the transmission of culpability in sci-fi cases where a person’s whole consciousness is uploaded to a machine while the human is dying. The present person and the past person are not numerically identical, on the somatic view, since they don’t have the same body. Yet intuitively, responsibility transmits in this case. The intuition undermines (1). Alternatively, suppose one holds the ‘psychological’ view, on which two humans (one past, one present) are numerically identical just in case they are psychologically continuous. Then, (1) rules out the transmission of culpability when someone has a sudden and significant psychological shift. Sometimes, this result is counterintuitive—for example, when the shift affects significant parts of a person’s personality that are irrelevant to the wrongdoing. Again, this counterintuitive result speaks against the slogan. Yet the somatic and psychological views are the two most popular accounts of numerical identity for humans. Thus, even for humans, it’s unclear whether (1) holds.³⁰

There’s a deeper problem with (1), as applied to states specifically: secessions, unions, federations, revolutions, coups, and constitutional revisions are not the stuff of science fiction. State fission and fusion is an everyday political reality. By contrast, fissions and fusions of human agents are fantastical—as well as being a problem for any theory that endorses the slogan.³¹ Yet it is our intuitions about humans that incline us towards (1). Because of this difference between states and humans, it’s doubtful that our everyday intuitive treatments of the connection between numerical identity and culpability for humans—which motivate (1)—will be transferrable to states. Relatedly, Marya Schechtman notes that fission and fusion simply doesn’t happen to humans; therefore, we can’t be sure that our everyday concepts of identity and culpability (which motivate (1)) would remain if these processes suddenly became possible.³² That is: our intuitions about fissioned and fused humans are arguably not reliable in supporting (1) as applied to all agent-types. Even ‘sudden and significant psychological shifts’ in humans are arguably not the target of reliable philosophical intuitions, because the details of such shifts can vary wildly, and look different in states than in humans. The slogan, then, should not cheerfully be imported from humans to states, even if the slogan was found to hold true for humans as we actually find them.³³

What about (2): are states’ numerical identities constructed through international law? States’ agency is undoubtedly socially constructed: in defining the sense in which states have agency, we must refer to social factors, such as laws, institutions, and practices.³⁴ But it’s questionable whether states’ agency is constructed through international law alone, at least when ‘agency’ refers to a feature with moral, social, and political (not merely legal) import. If a state’s agency can render it morally culpable and open it to social and political negative reactive attitudes, we should not simply defer to the law on the nature of states’ agency.

Furthermore, international law is vague on states’ identity. Brownlie’s Principles of Public International Law is the canon on states’ international legal treatment. This text asserts that states persist except under a “permanent displacement of sovereign power.”³⁵ However, first, “displacement” and “sovereign power” are open to numerous interpretations that have filled books in international relations.³⁶ Second, Principles notes that “it is not always easy to distinguish a dependent state with its own personality from a subordinate entity with no independence, a joint agency of states from an organization, or a private or public corporation under some degree of state control from the state itself.”³⁷ If international law is the basis of states’ numerical identity, then the latter is simply not determinate enough to guide attributions of diachronic culpability.

Even if international law were determinate, it’s still doubtful that it should guide our assessment of states’ agency and culpability. Compare philosophical treatments of humans’ agency and culpability. Far from deferring to the law, contemporary philosophy views itself as capable of guiding law on humans’ agency and culpability. The law, philosophers assume, can get humans’ agency and culpability wrong. Laws might require philosophical guidance or correction. Why should this be so for legal treatment of humans, but not legal treatment of states? One might reply that states’ agency is socially constructed, while humans’ is not. Even granting that humans’ agency is not socially constructed (itself doubtful), it’s still true that states’ agency is not constructed by law alone. It is also constructed by moral, social, and political practices. These latter constructions are relevant to moral-political blameworthiness, so they should inform our conception of states’ agency and culpability. International law cannot change our moral conclusions by definitional fiat.

Could we save the identity view with a different account of states’ numerical identity? We might try appealing to states’ distinctive territories. Territory is foregrounded in two influential definitions of states. First, the Montevideo Convention on the Rights and Duties of States (1933) describes states as having “(a) a permanent population; (b) a defined territory; (c) government; and (d) capacity to enter into relations with other states.” Second, Max Weber’s definition characterises a state as “a human community which (successfully) lays claim to the monopoly of legitimate physical violence within a certain territory.”³⁸ Governing (or monopolising legitimate violence within) different territories is what distinguishes contemporaneous states from one another. And governing (or monopolising legitimate violence within) a particular territory is plausibly a value of each state, which is also reflected in their structures (e.g. military that protects specific borders) and procedures (e.g. parliamentary procedures that represent specific regions). So, this proposal gels with Sect. 1’s account of states’ agency.

However, if we tie culpability to numerical identity, and tie numerical identity to territory, then this problematically allows states to remove their culpability by invading new territories. When one country successfully invades another, the territory of the former has changed, but this should not affect its culpability for its past actions. Worse, the territory proposal forecloses the possibility of Indigenous liberation from the legacy of colonial rule. If an Indigenous group successfully lays sovereign claim to its territory that had previously been colonised, suddenly that Indigenous group would inherit the culpability of the coloniser. Thus, the Montevideo and Weber definitions might be useful for differentiating states that co-exist. But they are no good for determining when one state inherits culpability from another state across time.

One could reach for other aspects of these definitions: the Montevideo convention mentions “government” and Weber mentions “human community.” Somewhat similarly, Thompson writes that there are two criteria for determining whether there is “the continuity that depends on each generation passing on responsibilities to the next”.³⁹ The first criterion “focuses on legal and institutional changes. The continuity is likely to be broken with respect to particular relationships and responsibilities when, as a result of political changes, the laws, institutions or practices that have in the past governed these relationships and responsibilities cease to exist or are radically altered.”⁴⁰ Thompson gives the examples of secession and revolution as cases where the continuity is broken.⁴¹

Thus, perhaps the continuity of “government,” “human community,” or “institutions or practices” are the basis for states’ numerical identity through time. But this just pushes the question back: under what conditions does a “government,” “human community,” “institution,” or “practice” maintain its identity through time, in a way that coherently underpins states’ culpability for past injustices? I will provide something of an answer in the next section. However, that answer will not align with common-sense judgments about the conditions under which states retain their numerical identity through time. We should therefore take the question of states’ diachronic culpability to be distinct from the question of states’ numerical identity.

4 The descent view

4.1 Proposal

If we abandon identity, what’s the alternative? My proposal is that culpability transmits from a past state to a present state to the extent that the present state descends from relevant aspects of the past state. Specifically, the relationship of ‘descent’ holds when the present state’s existence and (what I will call) its ‘central endorsed agential resources’ are part of a cross-temporal agential stream that includes those agential resources of the past state that underpinned the culpable action (or other culpable exercise of agential resources). The descent generated by such agential streams can be stronger or weaker, grounding degrees of culpability that track the degree of descent.

Let me explain some key terms in this proposal. ‘Agential resources’ include a state’s structures, procedures, and values: the things that establish its distinct agency, as described in Sect. 1. A state’s ‘agential resources’ also include its decisions, evaluative attitudes, and character traits. These are manifestations of a state’s agency. A state’s decisions, evaluative attitudes, and character traits are sometimes produced by, sometimes produce, and sometimes are manifest in, the state’s structures, procedures, and values. Thus, these six features are intertwined. I use ‘agential resources’ to cover all six features and their interrelations. Although I will focus on culpability for actions, I assume that culpability could concern any exercise of agential resources, broadly construed, of the past state.

The idea of a state’s ‘decisions’ is relatively straightforward: decisions occur when intentions are produced by the state’s procedures, as characterised in Sect. 1. A state’s evaluative attitudes are the attitudes it holds, even if it never decided to hold them, and indeed even if it explicitly disavows them. A state’s evaluative attitudes are manifest in the state’s effects on the world, and in what the state notices and attends to.⁴² Evaluative attitudes include, for example, implicit (i.e., disowned yet existent) racism or a lack of care for society’s most vulnerable. A state can hold these evaluative attitudes even when leaders or public servants—when acting within and because of their role—would profess that the state does not hold these attitudes. Finally, a state’s character encompasses its culture or ethos, which might be virtuous or vicious, and which is somewhat more diffuse and all-encompassing than a given decision or identifiable evaluative attitude. Like for humans, a character assessment of a state involves the attribution of virtues and vices, including vices such as greed, arrogance, or imperialism.⁴³

The intuitive idea behind ‘descent’ is that the present state and its (central endorsed) agential resources fall within a cross-temporal agential stream, where that stream includes the agential resources that underpinned the historical culpable action (or other historical culpable exercise of agential resources—I’ll take this as implicit from now on). Importantly, it needn’t be that the present state’s existence and agential resources were caused by the culpable act itself. Rather, the present state’s existence and (central endorsed) agential resources must be included in an agential stream, where that agential stream also includes the agential resources that underpinned the historical wrongdoing. After all, some historical wrongs might be epiphenomenal; they might have no impact on any later state’s existence or agential resources.⁴⁴ Yet a present state can still be culpable for those wrongs, if the agential resources that underpinned those wrongs are included in an agential stream that also includes the later state.

Perhaps most often, this descent relation will hold when the present state shares a numerical identity (on whatever account one prefers) with the state that committed the historical wrong. More interestingly, the descent relation can hold when the earlier and later state are numerically distinct. I don’t rely on any demarcation between numerical ‘sameness’ and ‘distinctness’—or even presume these are well-defined or defensible notions—for reasons given in Sect. 2.

On the proposal, to determine whether a present state “descends from” the agential resources that underpinned a past state’s wrongful action, we must assess the states’ respective agential resources in context, to determine whether those resources belong in a stream together. The requirement here is not merely that the earlier state’s agential resources must somehow ‘affect’ the later state’s agential resources. After all, many present-day states are affected in myriad ways by the agential resources of almost all earlier states. When one state decides to raise import tariffs, this affects many other states in the future. These other states do not thereby acquire culpability for the first state’s raising import tariffs. And the requirement is not that the past agential resources ‘persist’: any appeal to persistence would introduce problems analogous to those outlined in Sect. 2. Instead, the requirement is two-fold.

First, the agential resources that underpinned the wrongdoing must give rise to the nature, style, or type of the later state’s central and endorsed agential resources. It’s not enough that the earlier state affects peripheral decisions or attitudes of the later state: that would lead to an over-inheritance of culpability, where too many later states would inherit culpability. Instead, the nature, style, or type of the agential resources that are central to the later state must be descended from the earlier agential resources. Moreover, the present state must endorse the central agential resources that were thus descended. This endorsement requirement rules out cases where the later state has the agential resources only because of coercion, threats, and so on that it could not reasonably have resisted. That said, endorsement can be tacit and unreflective: a state can endorse an agential resource simply by not changing that resource when the opportunity arises. Assessing endorsement requires assessing difficult counterfactuals, such as whether the state would change the resource if it could. Thus, unreflective states cannot evade culpability via the endorsement condition.⁴⁵

There are degrees here, which enables the descent relation to come in degrees. There are degrees concerning how many of the later state’s agential resources were determined by the earlier state’s resources, how central the later agential resources are to the later state’s agency, and how strongly the later state endorses those resources (which can be measured by, roughly, under what circumstances the state would voluntarily change the resources—the fewer such circumstances there are, the stronger the endorsement). The stronger the relevant contemporary resources fare on these dimensions, the more culpable the later state is.

Second, the ‘descent’ in question must involve an agential stream or flow that runs through time. This stream is not a stream of mere causation or influence. Instead, the earlier state’s agential resources (that underpinned the wrongdoing) must morph (transform, evolve, grow) or branch (off-shoot, spin-off, grow-out) into the later state’s agential resources—perhaps through a slow process taking many years, with many intervening agential resources, or perhaps in a rapid process of change. In this way, the earlier state is not merely the passive conduit through which some third party causally determines features of the later state; rather, the earlier state is the originating source of the determination (perhaps alongside the later state). Such morphing can occur despite the earlier state going through a process of fission or fusion. The descent can flow through such phenomena as exiled governments, civil wars in which only one side arises out of the earlier state, and so on—the agential stream need not be ‘a state’ throughout (consider that the German agential stream manifested in a nation, not a state, in 1945-49).

How is this morphing or branching relation to be characterised? Notice that states’ agential resources are united into an integrated whole: states’ structures, procedures, values, actions, attitudes, and character traits are mutually supporting in a way that is (somewhat, though never perfectly) rationally integrated. This rational integration is part of what makes states agents. When we assess a later agential resource—asking whether that resource is a “morphing of” an earlier resource—we should ask two things. First, we should ask whether the later resource was arrived at from the earlier one via a path that initially accorded with the integrated whole that was present at the earlier time, including that integrated whole’s informal values, cultures, and traditions, and including its meta-resources for changing its agential resources. Second, we should ask whether each subsequent step in the process likewise accorded with the integrated whole was in-place when those subsequent steps occurred, bearing in mind that the meta-procedures permit changes in the integrated whole (including changes in the number of integrated wholes that exist, for example by allowing secession).

Thus, if an earlier state E had policy A and a later state L has policy B, we should ask: was B arrived at from (the resources that underpinned) A via a path that started off constrained by E’s agential resources, and that subsequently departed from those resources only if such departures were permitted by the agential resources that were in-place at the time the departures occurred? Answering this question requires taking no stand on whether E and L are ‘numerically identical’, or whether policies A and B are numerically identical.

The idea here is not that L itself has to be rationally unified with E, in the sense of abiding by all the procedures and structures that were in-place in E: that would produce a paucity of culpability, because few contemporary states are so rationally unified with their historical forebears. Instead, the requirement is that L’s departures from E’s resources were permitted by the resources that were in-place when the departures occurred. Departures (such as, to foreshadow, Australia’s ‘departure’ from Britain, or Britain’s ‘departure’ from a policy of colonialism to decolonisation) should arise endogenously rather than exogenously, in order for descent to hold. Endogenous changes are changes that abide by the agential resources in-place at the time of change. Endogenous changes enable the earlier state to leave its agential ‘mark’ on the state that results from the change, including the mark of culpability. Exogenous changes are, by definition, attributable to a different agential stream.⁴⁶ I shall defend this idea in Sect. 3.2.

This notion of ‘descent’ may bring to mind the work of Derek Parfit.⁴⁷ For Parfit, psychological “continuity” consists in overlapping chains of strong psychological “connectedness.” Psychological “connectednesss” holds between two person-stages when, for example, one person-stage remembers experiences of the other, or one realizes an intention of the other (therefore, the two share the same token intention), or the two have the same token beliefs, desires, and so on.⁴⁸ “Strong” psychological connectedness holds when two persons (one past, one present) have “enough” connected mental states.⁴⁹

Parfit’s notion of an ‘overlapping chain’ implies periods of strong psychological connectedness between different time-slices, combined with shifts towards a successive series of new psychological conditions that are increasingly disconnected from the original one. In this way, two distant chain pieces might have weak connectedness, but nonetheless be continuous. For example, my five-year-old self was strongly connected to my six-year-old self, who was strongly connected to my seven-year-old self, who was strongly connected to my eight-year-old self, and so on through to me in middle-age. My middle-aged self is, therefore, ‘continuous’ with my five-year-old self, through the intervening chain. But my middle-age self is barely connected to my five-year-old self: we share few memories, intentions, beliefs, or desires.

Parfitian continuity differs sharply from my relation of descent. Parfit does not ask whether earlier agential resources descend from earlier ones, in the specific way I characterised that relation above. Instead, Parfit would look at two time-slices of states and ask whether enough of their mental states are token-identical.⁵⁰ If so, the two time-slices are connected. If not, Parfit would call them ‘continuous’ only if they were part of a chain of time-slices in which each time-slice was connected to its neighbours. That is: Parfitian continuity rests on mental states’ token-identicality, so his account of branching agency also depends on this. To evoke the token-identicality of states’ agential resources would introduce issues of numerical identity that I seek to avoid for the reasons given in Sect. 2.

Moreover, Parfit does not specify how ‘continuity’ or ‘connectedness’ relate to culpability. Parfit uses these relations to link a later person to an earlier person—not to relate a later person to an earlier action (or other exercise of agential resources). I am concerned with the latter relation, which likely does not track the relation Parfit was interested in: on my proposed view, a later state might be culpable for only some (not all) actions of an earlier state. The later state is culpable only for those actions from whose underpinnings it descended. As I explain below, this is intuitively plausible. Parfitian continuity and connectedness cannot capture this, because they link a later person to an earlier person tout court. If a past state, A, ‘fissions’ into two new states, B and C, and a wrongful action of A is underpinned by agential resources that evolve into B’s central endorsed agential resources (but that have no effect on C), then B (but not C) would inherit culpability for A’s wrongful action. However, on Parfit’s account, both B and C are ‘continuous’ with A.

With the proposal on the table, I now provide arguments.

4.2 Arguments

A first argument concerns cases in which there is no doubt that a present state is culpable for the actions of a past state. Take contemporary Australia’s culpability for the Stolen Generations. This was a policy enacted in Australian territory in the 20th century, in which Indigenous children were forcibly taken from their families by state officials and sent to missions, state-run boarding homes, and white adoptive families. Australia is still reckoning with this policy, which many argue requires sustained state action in the present, having led to intergenerational trauma and loss.⁵¹ There is no doubt about contemporary Australia’s culpability—indeed, Australia has officially apologised for those actions. How can this be philosophically vindicated, without recourse to the identity view?

The natural (if vague) answer is that there is some agential lineage between past-Australia and present-Australia. We can tighten up this natural answer: at each moment in the intervening decades, the agential resources of Australia were reached via the exercise of the agential resources of the Australia that existed in the moment before. The exercise of Australia’s agential resources precipitated the transformation of those very same agential resources, where those later resources are connected to the earlier ones by being the transformation of them. There is thus a stream of ever-changing agential resources, which connects Australia-at-one-moment to Australia-at-the-next-moment to Australia-at-the-moment-after, and so on into the future. This process links Australia’s various manifestations through time. The process that connects Australia-in-1924 to Australia-in-1925 carries through to Australia-in-2024.

The intuitive idea, then, is not that Australia-in-2024 descends from Australia-in-1924 in a way that ‘leapfrogs’ the intervening years: we cannot look just at Australia-in-1924 and Australia-in-2024, to determine whether there is inheritance of culpability. These two temporally distant states must be viewed as joined by a stream of moment-to-moment descent relations that run through the intervening years, with each changing into the one that follows. But if this applies to a state over the course of a century, then why should it not also apply—perhaps to a lesser degree or for a more circumscribed set of actions—over a longer period? This is simply what the descent view proposes. Thus, the first argument is that the descent view captures our intuitions in ‘easy’ or ‘obvious’ cases of states’ culpability through time.

A second argument is that the view deals well with fission and fusion. If States B and C secede from State A after State A performs a culpable action, it can be that all three later states are culpable for that action. This is attractive: it prevents seceding states from unilaterally ‘cancelling’ their culpability, when their agential resources are descended from those of the state they seceded from. Any seceding state inherits culpability for a past action from whose underpinnings it is descended. (That said, secession won’t always imply descent—for example, it won’t if the secession is the result of an all-out civil war or revolution, so that the seceding state completely bypasses the agential resources of the earlier state. This is a plausible result that provides an argument for Sect. 3.1’s requirement that changes be ‘endogenous’ rather than ‘exogenous.’⁵²)

A third reason for the descent view is that it walks the tightrope between under-inclusivity and over-inclusivity. An underinclusive view precludes state culpability when it shouldn’t. For example, if a view required that the present state be substantively similar to the past culpable state (for example, requiring Parfitian ‘connection’), then a present state could cancel its culpability by becoming sufficiently dissimilar to its previous self. This might rule out contemporary Germany’s culpability for the Holocaust, for example.⁵³ Yet the descent view preserves such culpability, if the state’s self-changing actions were endogenously-created evolutions of the agential resources—including values, culture, and traditions—that underpinned the culpable act.

Conversely, an overinclusive view generates state culpability when it shouldn’t. Imagine a view that posits state culpability whenever there are two states who are merely ‘similar’ to one another, despite no history of interaction. The descent view blocks such culpability transfers. It requires that the agential resources of one produced the existence of the other, and the nature, style, or type of its central endorsed agential resources. Nor does it say that an invaded or annexed state is culpable for its own invasion or annexation, if the invaded or annexed state continues to exist with self-determined agential resources (for example, in an exiled government or subordinated national group). In such a case, the exiled or subordinated group is not a transformation of the agential resources of the invading state. Again, this plausible result derives from the characterisation of morphing provided in Sect. 3.1. The account thus avoids over-inclusivity, at least in these respects. To fully interrogate its over-inclusivity, though, it is worth turning to objections.

4.3 Objections

A first over-inclusivity objection concerns puppet governments: if one state invades another and installs a ‘puppet’ government, or otherwise uses its relevant agential resources to determine the nature of the invaded state’s central and endorsed agential resources, then the invaded state inherits culpability for the invasion, on the descent view. Isn’t this over-inclusive?

I suggested it is not: the puppet regime has central endorsed agential resources that transformed from the agency that was manifested in the invasion. The vast majority of the population will likely not be complicit in the new regime, nor share in its culpability: they may instead be the conduit for the agential stream of the invaded state.⁵⁴ But the new regime itself should be morally assessed in light of its descent from the invader. Here, as elsewhere, it matters that states’ agency and culpability is distinct from that of members, as explained in Sect. 1.

A second objection concerns copycat states. Consider a case from Janna Thompson:

Suppose that the members of a neo-fascist group commit an atrocity against a racial group that they despise. Afterwards, they put out a leaflet that calls on people who share their racist views to commit similar deeds. A group of young people are inspired to follow their example. They act in the name of the neo-fascist group and think of themselves as cooperating to achieve its racist objectives. However, neither their identification with the neofascist group nor their sharing of its objective makes them responsible for the wrongs that this group committed.⁵⁵

In Thompson’s case, there is no descent between the neo-fascist group and the group of young people: the later group’s agential resources are not an evolution of the agential resources of the earlier group. Again, this plausible result follows from the characterisation of morphing given in Sect. 3.1, so provides support for that characterisation. The descent view would not assert culpability in Thompson’s case. If the case differed in the respects just outlined, the descent view would indeed hold the group of young people culpable—and, I suggest, rightly so.

Another objection might come from advocates of Peter French’s view:

a corporation’s diachronic responsibility for a past act is dependent on the degree to which and the way in which the corporation’s internal decision structure’s policies and procedures that reflect its values, plans, desires, and intentions are causally connected to those that were the executive springs of the past untoward corporate action.⁵⁶

For French, there is a ‘connection’ between an earlier agent and a later agent just in case the later one both (1) was caused (to a high degree) by the earlier one and (2) is similar (to a high degree) to the earlier one. My proposal is a nuanced version of (1): mere ‘causation’ is not enough; rather, causation must occur such that the later state’s agential resources are an offshoot of the earlier state’s. My account rejects (2). If French’s view were true, it would provide strong objections to my view.

However, French’s account both under-generates and over-generates culpability. It under-generates by failing to assign culpability when there are few similarities between the earlier and later entities: consider Nazi Germany and contemporary Germany. My account correctly captures this example, because the latter morphed from the former, even if they are dissimilar. And French’s account over-generates, insofar as relations (1) and (2) can hold between two corporations who have parallel and interacting histories, yet who lack the shared cross-temporal agential stream characterised in Sect. 3.1.

French’s arguments reveal something important: my view applies to states, not humans. In defending his connection-based view, French defers to Andrew Khoury’s connection-based account of diachronic culpability for humans. My descent view does not readily apply to humans. If it did, it might imply that children are culpable for wrongdoings perpetrated by their parents: aren’t my agential resources in some sense an ‘off-shoot’ of my parents’ agential resources? It would be highly objectionable to render children culpable for their parents’ deeds. Yet human children’s agential resources are not part of a cross-temporal agential stream that includes the agential resources of their parents. This is because of how humans develop: they go through several years of not being agents at all, during which time various environmental influences help shape the agent they will become. There is no agential stream or flow between parents and children, because babies are not born as agents. States, by contrast, are. A colony’s agency thus morphs from the agency of the coloniser, in a way that children’s agency does not straightforwardly morph from the agency of their parents. The development of human agency is distinctive. My proposal, therefore, does not get a grip on humans. I believe French errs in copying-and-pasting from humans to organizations for diachronic culpability purposes.

5 Conclusion: the invasion of the Australian continent

Where does this leave the examples with which we began? Does the descent view attribute culpability to contemporary settler-colonial states for historical actions of invasion, dispossession, and genocide that were perpetrated before those states (seemingly) existed? This is a question for historians and political scientists. The descent relation must be assessed via a careful scrutinising of the nature, style, or type of central endorsed agential resources of the later state and the mechanisms by which these came about.

Yet a positive answer is plausible for some settler-colonies. Take Australia. This is a state whose structures, procedures, values, decisions, attitudes, and character have been heavily determined by a moment-by-moment flow in which Australia became an offshoot from Britain. The ‘first fleet’ of British colonists arrived at the Australian continent in 1788. For more than a century, the six British colonies in Australia were under power of the British Parliament, until the Australian Constitution was passed in the British parliament in 1900. Even after this, the UK’s Privy Council could be appealed to from Australia’s High Court until 1975. The Australia Act was passed only in 1986: this was a pair of bills, passed respectively in the Australian and UK parliaments, that eliminated the possibility for the UK to legislate with effect in Australia, be involved in Australian government, or receive appeals from Australian courts. Even today, Australia’s Governor-General is appointed by the British monarch and is commander-in-chief of the Australian Defence Force. This story is one of an outgrowth from Britain’s procedures and structures, where that outgrowth transformed into Australia’s, in ways that, at each point, were permitted by the agential resources of the relevant integrated rational whole at that point. At some points, the transformation was quite rapid—such as with the passing of the Australian Constitution in 1900—but even this moment of rapid transformation conformed to the integrated agential resources of Britain at the time.

This much concerns Australia’s procedures and structures. Australia’s values, decisions, attitudes, and character traits are also heavily indebted to Britain—that is, many of its central endorsed agential resources are in a narrative stream that includes British agential resources. This is especially true regarding the ways Australia and Australians are racialized—which is relevant when the historical wrong at issue is the invasion of the continent, and resultant dispossession and genocide of Indigenous Australians. As Aileen Moreton-Robinson has analysed, Australia’s conceptions of heroism, beauty, gender, ownership, and belonging all have strong roots in British culture, and more particularly in those agential resources of the British state that are bound up with racism and imperialism.⁵⁷ These contemporary agential resources of Australia plausibly descend from Britain’s historical wrongful act of imperialism, in the sense of ‘descent’ developed in Sect. 3.1.

Pending a deeper analysis of this history, we can tentatively conclude that Australia is culpable for Britain’s invasion of the Australian continent, under the descent view. Australia’s agential resources were arrived at from Britain’s underpinning resources, in a way that cohered with the ever-evolving rational unity of Britain. Now, it might still seem bizarre that a state could be culpable for actions performed before it existed. If so, recall that, on the descent view, the later state is not culpable for absolutely everything the earlier state did. The later state is culpable only for those actions (or, more generally, ‘exercises of agential resources’) that had underpinnings from which the later state’s existence and central endorsed agential resources descended.

For example, Australia bears culpability for Britain’s action of colonising Indigenous Australian territory, because Australia’s agential resources descended from the agential resources that underpinned that action. Australia also shares culpability for the exercise of Britain’s broad character trait of imperialism, because Australia grew out of this agential resource. But Australia isn’t culpable for Britain’s specific imperialistic actions in (say) Canada: Britain’s general character trait of imperialism, but not the more direct agential underpinnings of its actions in Canada, gave rise to Australia’s existence and central endorsed agential resources. The ‘more direct underpinnings’ here are the decisions and attitudes of colonial Britain that specifically related to the Canadian territory: for example, Britain’s decisions to incentivise Canadian settlement or send officials there. Call these ‘Britain’s Canada-specific agential resources.’ Assume Britain’s effects on Australia would have been roughly what they were, even if Britain had not had these Canada-specific agential resources. Britain’s Canada-specific agential resources therefore had no effect on Australia, so Australia avoids culpability for the actions that flowed from them. However, both Australia and Canada bear culpability for Britain’s imperialism, since both countries’ central endorsed agential resources flowed from this more general agential resource.⁵⁸

In this way, contemporary states’ culpability is determined by the extent to which those contemporary states exist as the agents they are (that is, exist and have the central endorsed agential resources they have) in a way that descends from historical agential resources. I have focused on the example of settler-colonial states, because they demonstrate that two states need not be ‘identical’ for one to inherit the culpability of the other. Settler-colonies can, provocatively, be culpable for actions that were performed before they even existed. But the proposal is applicable beyond these examples. The proposal is likely to expand state culpability well beyond what is currently politically recognised. When state officials are reluctant to acknowledge this culpability, it is up to citizens to hold them to account.