Environmental determinism and archaeology. Understanding and evaluating determinism in research design

Abstract

With the emergence of modern techniques of environmental analysis and widespread availability of accessible tools and quantitative data, the question of environmental determinism is once again on the agenda. This paper is theoretical in character, attempting, for the benefit of drawing up research designs, to understand and evaluate the character of environmental determinism. We reach three main conclusions: (1) in a typical pattern of research design, studies seek to detect simultaneous shifts in the environmental and archaeological records, variously positing the former to have influenced, triggered or caused the latter; (2) the question of determinism involves uncertainty about the justification for the above research design in particular in what comes to biologism and the concept of environmental thresholds on the one hand and the externality of the drivers of transformation in human groups and societies on the other; (3) adapting the concepts of the social production of vulnerability and the social basis of hazards from anthropology may help to clarify the available research design choices at hand.

Introduction

Determinism, in particular environmental determinism, is a recurring topic of debate in modern archaeology and related disciplines (Meggers 1954; 2001; Wheatley 1993; Contreras 2016; Middleton 2017; Bicho and Cascalheira 2018). The issue of determinism has become more prominent in recent years, partly due to the rise and dissemination of new scientific techniques in archaeology, such as ancient DNA (aDNA) and isotope analysis that have been argued to ‘open up a new chapter in archaeological knowledge that demand similar changes in archaeological methods and theory’ (Kristiansen 2014, 12). These new techniques have fomented a desire to return to the questions and topics promoted in processual archaeology (e.g. Kintigh et al. 2014) motivated by the belief that the archaeological problems which could not be addressed thirty or forty years ago might now be addressable by new scientific methods. Most noticeable has been the strong resurgence of palaeo-environmental studies. In the background of academic debate, the worsening contemporary environmental situation adds its own colour to the discussion.

It seems fair to say that, at its core, determinism in archaeology and related disciplines has been a term of reproach. To speak of environmental determinism involves the contrast between understanding, explaining and describing prehistoric human action as a response or reaction to, and thus as it were determined by, environmental factors as opposed to studying the various cultural and social factors affecting the course of human (pre)history. Processual archaeology is sometimes taken to be advocating deterministic models of human behaviour, while postprocessual archaeology, by contrast, is taken to advocate the study of the contingency of historical events (Hodder 1985) or the agency of actors to produce non-deterministic behaviour (Dobres and Robb 2000). In the words of Ian Hodder,

It is argued by the processual school in archaeology that there are systems so basic in nature that culture and individuals are powerless to divert them. This is a trend towards determinism – theory building is seen as being concerned with discovering deterministic causal relationships. There is a close link here between discarding notions of cultural belief and of agency (Hodder and Hutson 2003, 7).

Although archaeology has in many ways moved on beyond the ideas espoused by processual and postprocessual archaeology, archaeological research is still commonly framed as either deterministic or non-deterministic. Broadly speaking, argued Elizabeth Arkush (2011, 200), ‘our differences lie in the extent to which we stress contingency versus process, and agency versus conditions, in the making of diverse human histories’.

In the background to the contrast looms the larger, antagonistic ‘two-cultures’ debate (Snow 1998). The perception is that, modelled on the natural sciences, hard and proper scientific sciences ought to be concerned with causal relationships whereas softer, interpretive sciences generally fall short of that or, alternatively, must be conceived as reaching for some wholly different kinds of conclusions altogether – comparative or exploratory, perhaps (Sørensen 2017). The focus on the study of the impact of natural systems on human behaviour is often a methodological choice springing, on the one hand, from disciplinary preferences relating to the dominant or emerging paradigms (Kuhn 1996; Lucas 2017) or controlling models (Clarke 1972). The practical disciplinary reality of archaeology often plays its own role: the availability of particular kinds of often sparse data – and the fact that, in certain contexts, environmental data is the easier, or the only, one to come by.

Not only does the most recent palaeo-environmental work proclaim itself to be aware of the charge of environmental determinism and explicitly aim for more balanced accounts of the past, but there is also growing awareness of the distinction between identifying causal as opposed to probable or correlative connections (see e.g. Contreras 2016; Faulseit 2015; Coombes and Barber 2005). In philosophy and social theory, there exists a long-standing debate regarding the best way to conceive the kinds of knowledge that the human sciences, as opposed to natural sciences, are trying to attain: whether the human sciences, too, ultimately deal with causal facts – albeit disguised beneath a layer of so-called emergent phenomena – or whether both, the natural and the human sciences, actually are best conceived of as not dealing with causal facts at all (Dilthey 2008; von Wright 1971; Elder-Vass 2010; Lohse 2017). The indications are, accordingly, that an exploration of the core of the determinism debate will turn out to concern at least as much the self-referential foundations of our own scientific conduct as the causal foundations of prehistoric transformations. Critical self-reflection about the way we construe our own scientific practices is essential for scientific progress and should not be taken to automatically yield a social-constructivist position (Shennan 1994, Introduction; Bhaskar 2008).

We write as an interdisciplinary team of researchers working in a joint research project, and we are interested in gaining clarity about what it might mean, if anything at all, to be a determinist. Indeed, as soon as we set off to discuss the topic, we began to suspect that ‘determinism’ might be one of those catchphrases that carry much emotional baggage and diverse associations, but of which the precise meaning and implications are far harder to articulate.

In what follows we shall begin with a philosophical discussion of the problem and the concept of determinism. This will lead to us making explicit a typical form of reasoning that is commonly found in palaeo-environmental research in archaeology – one that is typically oriented to detecting certain parallels between societal and environmental changes. We will also discuss blind spots and potential problems with this form of reasoning. We conclude with thoughts on how the social aspects of the human–environment relationship could be brought into sharper focus in the studies that build upon said parallels. In our view, the study of environmental or climate triggers, influences or even causes of societal processes is understood as a rigorous research practice focused on empirical questions and data. Environmental determinism is not an a priori faulty form of structuring research design. That said, such studies typically draw from paradigms or controlling models that have their own weaknesses and blind spots, as we observe with reference to selected literature below. Such weaknesses and blind spots might be said to reflect a deterministic orientation, yet the issue is ultimately empirical.

The philosophical problem of determinism

The classic philosophical problem of determinism arises as follows. It starts off from the conviction that the modern natural scientific world view is in its fundamental aspects correct, namely that physical reality is a totality of physical ‘elements’ in the broad sense as these are revealed to us by the theories of physics, biology and so forth. It also belongs to this picture that these ‘elements’ interact in ways that are regular and law-governed; that is, that there are such things as laws of nature that govern the behaviour of the ‘elements’ which are studied in the respective sciences. To introduce further terminology, these interactions can be called chains of causes and effects that make up the causal world with a certain causal history. Such a picture comprehensively and exhaustively describes ‘the real’, the reality, the world or the universe, past and present. In philosophical discourse, such a picture is often referred to as the materialist or the physicalist picture.

Now, from this physicalist conviction alone the philosophical problem of determinism already follows, for in physicalism, any worldly processes, including human decisions and actions, can only be further links in the chain of causes and effects and therefore determined by the state of the antecedent universe. This ‘can only be’ is a highly significant expression because in the physicalist picture, anything that falls outside the chain of causes and effects is either non-existent or irrelevant to the chain (as it does not affect the chain). It is unreal.

In a somewhat neglected paper, the archaeologist Travis W. Stanton (2004) provides an excellent and nuanced analysis of the situation, arguing that, understood as a causal position, determinism can basically only offer incomplete and selective accounts of the causal reality it purports to describe. Stanton takes it that environmental determinism makes causal claims about the world and, as such, Stanton’s perhaps surprising view is that determinism as we might find it, for example, in palaeoclimatology is not deterministic enough, or, better said, it is incompletely deterministic (ibid., 7 and passim).

Stanton’s view can be understood as setting off from the physicalist picture and the view that for each event an astonishing array of multiple causes are or were at work. Of this array, even our best deterministic theories can only ever hope to grasp, understand and formulate a small fraction. In practice, Stanton argues, most theories focus on very few causes; that is, they are monocausal, or they are what is only slightly better, multicausal (ibid., 36). In either case, our accounts of the events can only remain hopelessly incomplete in the face of the immense causal complexity of reality itself.

Thresholds and biologism

Most palaeo-environmental scientists would probably be sympathetic to Stanton’s argument and make an admission along the lines of the palaeo-ecologists Coombes and Barber, saying, ‘Of course, a model cannot hope to reproduce all the processes and interactions in a particular culture’. That said, many would perhaps also make the caveat that Coombes and Barber go on to make, that ‘a representation of the system’s critical components should be feasible’ (Coombes and Barber 2005, 305). That is,

A simple model cannot hope to replicate all the complexities of environment–culture relationships across a civilization, but one basic approach that can provide valuable insights is to treat human populations in ecological terms, with their ranges shifting in response to changing conditions (ibid.).

Coombes and Barber articulate here a basic, widely shared approach to human–environment relations: the environment sets the frame or thresholds within which human groups exist, with changes to critical parts of the frame inevitably affecting the humans living within it. As a consequence, palaeo-environmental studies are frequently structured as investigations into shifts in environmental thresholds that are then related to parallel, simultaneous, changes in archaeological proxies such as those detecting the extent of human occupation in a given area. That is to say, palaeo-environmental research designs are typically structured such as to observe shifts that occur, in tandem, in the archaeological record on the one hand and in some environmental and/or climate record on the other.

The basic logic behind such parallelism is certainly plausible. There is a baseline of biologism built into modern reasoning about human existence – the view that, whatever else, human life is fundamentally a biological phenomenon and humans are therefore subject to biological conditions and needs. The biological needs are satisfied in the frame of available environmental affordances and, by implication, changes in environmental thresholds will affect all biological creatures living within them.

It is of no use to employ such contested labels as ‘environmental determinism’ here, but it seems fair to say that the determinism debate is fuelled by a particular uncertainty about the idea of biologism and thresholds: on the one hand, the biologistic idea of the climate and environmental thresholds being fundamental to human existence is eminently plausible, yet, on the other hand, many also feel that the picture overtly and unduly externalizes the drivers of change in human societies. Great potential for conflict – and arguably the gist of the determinism debate – exists precisely where, along the continuum from external influence to internal societal dynamics, to locate the moving forces of prehistoric transformations. In what follows, we discuss the issue further with selected examples from palaeo-environmental archaeology.

The tightness of thresholds

Consider the conclusion to a paper discussing, on the basis of high-resolution records of stable carbon and oxygen isotopes in ostrich, faunal and shellfish remains from the Middle Stone Age of southern Africa, the climate-change-based explanations of ‘the emergence of innovative behaviours’. Roberts et al. (2016, 16) argue that ‘[i]t seems that although climatic and environmental change clearly occurred in this region, a diversity of potential resources allowed human populations to absorb these changes’. There is, therefore, the question of how tight particular environmental thresholds envelope human life such that changes in the thresholds will elicit a human response. Roberts et al. would presumably also accept biologism as a basic tenet of scientific thinking about human existence, yet the precise way the thresholds impinge upon human life is still up for debate. At its heart, the question of determinism thus remains an empirical question, not an a priori charge levelled against a particular study.

The tightness of the thresholds is again a topic in another study from a different prehistoric context. Jones et al. (1999, 138) provide extensive observations and arguments for the general view ‘that the linkages between the physical/biotic environment and human subsistence and settlement are sufficiently tight to warrant serious consideration of environmental change as a potentially important factor in explanations of cultural change’. Their archaeological case studies focus on North America and their more specific contention is that

the interval between A.D. 800 and 1350, known to climatologists alternatively as the Medieval Warm Period, the Secondary Climatic Optimum, the Little Optimum … or the Medieval Climatic Anomaly … was a time of increased aridity that coincided with a unique pattern of demographic stress and frequent economic crises across much of Western North America (ibid.).

As a result, Jones et al. (ibid.) continue, ‘Large populations of agriculturalists and hunter-gatherers were confronted with serious and abrupt declines in productivity caused by repeated and prolonged droughts’. Extensive case study materials are offered to support the general view that the climatic and environmental changes in question ‘can and did cause cultural changes in the prehistoric past, and attribution of cause to environment in archaeological models need not be deterministic’ (ibid.).

However, at the end of the paper, a range of specialists provide comments to the main article. Pertaining to our present point regarding thresholds, one commentator points out,

A key problem here [in Jones et al. 1999] is that in almost none of the cases mentioned do we have anything like a clear understanding of precisely what is actually limiting population. There is certainly a basic connection between population and food supply, but the relationship is seldom direct and cannot simply be assumed to be so (Robert L. Bettinger in Jones et al. 1999, 158).

The commentator goes on to suggest that it is ‘unlikely … that drought would have had a uniformly adverse effect on all food sources of interest’, as well as that ‘drought may well create as many exploitative opportunities as it destroys’ (Robert L. Bettinger in ibid., 159; see also Bettinger, Garvey and Tushingham 2015). We are aware that we are providing here just one selected criticism of Jones et al. Yet, again, as in the context of Roberts et al. (2016) above, the criticism levelled against Jones et al. (1999) indicates that the issue of thresholds is an empirical question that ought to be acknowledged to exist and addressed in an empirical manner in our research designs.

At the core of the research design in Jones et al. we find the aforementioned parallelism: parallel, simultaneous, shifts in the archaeological record (‘a unique pattern of demographic stress and frequent economic crises’) and the environmental thresholds (‘the medieval climatic anomaly’) are detected and used on the basis of an argument positing a relationship between them – in this case the verb ‘cause’ is used. Bettinger’s critique challenges whether sufficient empirical evidence is presented and/or is available to support the conclusion of Jones et al.

Once again in a parallel critique, Grattan (2006, 11) argued that, while ‘[w]ithin the Holocene there are many volcanic eruptions coincident with climatic and/or cultural change’, there apparently still is room for debate as, according to Grattan, ‘the precise mechanism by which one could have influenced the other is obscure and a clear tendency exists to simply assume that the eruption and subsequent climate change was [sic] large enough to have triggered the coincident environmental or cultural change’. The observation made by Grattan here is strikingly analogous to Bettinger’s note quoted above that the relationship between changes in climate and their effects upon people is ‘seldom direct and cannot simply be assumed to be so’.

Again, a strikingly similar observation was made in the context of Andean prehistory, and the case of the 300-year drought from A.D. 1150 onwards faced by the Tiwanaku state, by Clark L. Erickson (1999). Erickson critically reviewed what he referred to as ‘neo-environmentalist’ accounts (citing e.g. Binford et al. 1997 and papers in Kolata 1993) of how ‘cultural change could be explained by climatic shifts in rainfall and temperature’ (Erickson 1999, 634). We do not have to go into details of the case here, but Erickson’s general argument is that while the neo-environmentalists had viewed the drought as bringing the populations of the area across their environmental thresholds, Erickson argued that in fact these people were accustomed to dealing with such environmental phenomena: ‘Peoples inhabiting the lake region have developed a complex indigenous knowledge system that includes a sophisticated agricultural technology and elaborate social strategies to mitigate both short- and long-term climatic fluctuations’ (ibid., 641).

One of the examples given by Erickson here concerns the effects of prolonged drought on the water level of Lake Titicaca. Erickson explained that while a drought could have obvious adverse effects on agriculture, in fact the receding water exposed ‘deep, organic-rich soil that is highly prized by local farming communities’. While others inevitably suffered from droughts, Erickson continued, ‘the communities who control lakeshore territories managed to become “rich” during the droughts by selling the abundant surplus produced on newly exposed lake bed and renting those lands to those less fortunate’ (ibid., 637). Therefore, again, the relationship between environmental impacts and social changes is seldomly direct. What kind of disaster a drought would wreak upon the wider area would thus arguably depend on the social and political systems in place in terms of which the changes are, or indeed are not, dealt with – more on this below.

Based on the above cases, thinking about the human relationship with the environment in terms of shifting thresholds appears to bear the risk of the shifts being taken as more significant than they were. The brief survey of literature provided above is, of course, not exhaustive, and it is not systematic. Yet, given that the palaeo-environmental research designs are typically built to detect parallel changes in environmental and archaeological proxies, we can expect that such studies can be challenged on the ground of whether the alleged transgressions of the tresholds identified in a given study really constituted real transgressions. The point of our brief survey above is to illustrate that the form of research is in principle – and often in actual fact – vulnerable in that way.

A possible lesson with regard to devising research designs might thus be this: while the detection of parallel shifts in some environmental and cultural proxies might be taken to indicate a relationship, it seems that previous studies have often been challenged precisely on the question whether particular changes in the thresholds are representative enough to carry through the general argument. It is important to note that none of the commentators cited above are challenging the general idea of biologism and that there may be environmental thresholds. They are, rather, calling for greater care in the evaluation of the impacts of changing thresholds.

It seems possible to draw two morals from the story. One moral is that simply greater care is needed to evaluate the ability of shifts in particular thresholds to affect human life. That is a matter of greater care in empirical evaluation and argumentation – and certainly also of the quality and breadth of available data sets.

The other moral is that there may be a systematic, paradigmatic, bias in favour of studies focusing on the detection of parallel shifts in environmental and archaeological proxies such that such parallels appear as a priori plausible, and are therefore easily asserted, then affirmed, in studies. After all, the very form of the research design is oriented to detecting parallel shifts in environmental and archaeological proxies, and when such parallels are detected the argument from external influence is easily made. We want to leave it to the reader to pick the moral they find more plausible.

Environment and human agency: the social basis of hazards

As noted above, a second thread in the discussion about environmental determinism has been the question whether ‘environmentally deterministic’ accounts afford too minor a role for human agency; that is, whether there is not only an environmental, but also a social, basis to the human–environment relationship. To open up that discussion, let us briefly turn to another case study.

Hoggarth et al. (2016) studied ‘the impacts of climate on complex, rainfall-dependent, agricultural societies’ with ‘multi-decadal droughts’ possibly behind the so-called Classic Maya collapse. In a related publication, Hoggarth et al. (2017) describe the severity of the drought episodes by a comparison with documented evidence from the later colonial period from the first quarter of the 18th century onwards. The aforementioned parallelism of the research design is present here as well: shifts in climate proxies are related to parallel shifts in the archaeological record. In addition, Hoggarth et al. (2016) consider further historical, comparative anthropological data from a later period. In their conclusion they posit ‘a strong relationship between political disintegration and climatic stress in the Maya lowlands’ (ibid., 25), although the word ‘cause’ is not used. This may be a case in which the impact of the changing environmental thresholds is purely a matter external to the communities affected: the droughts were so severe that the humans could not but come to suffer the consequences.

That said, Hoggarth et al. (2017, 83) also state that droughts’ ‘co-occurrence with other natural and social phenomenon (i.e., locust infestations, epidemics, and warfare)’ may greatly amplify the effects of the drought itself. Speaking in terms of the form of the research design, it should be pointed out that in Hoggarth et al. (2017) the parallelistic design is supplemented by a certain autonomy afforded to the social sphere in terms of how the influence of external events plays out in the social world. This takes us on to discuss what we might call the social basis of hazards.

Picking up on the distinction just articulated, there exists a branch of anthropology called the anthropology of hazards and disasters. We bring up this branch of anthropology here not because we ourselves are champions of that perspective, but in order to provide a clearly contrasting picture of how to conceive the human–environment relationship. A recent review article observed, ‘Since the 1980s, researchers have focused on concepts of vulnerability, resilience, and adaptation as ways to understand the social bases of disasters’ (Peterson and Broad 2016, 74, our italics). In the 1990s, Anthony Oliver-Smith (1996, 304), a chief protagonist in this area of academic research, wrote, ‘a new perspective has emerged that views hazards as basic elements of environments and as constructed features of human systems rather than as extreme and unpredictable events’.

That is to say, research in this field has proceeded on the assumption that, alongside attention to the effects of changing climate and the environment, societies are affected by changes in their environment because the societies exist in a ‘socially and technologically produced condition of vulnerability’ (ibid., 303). In particular, the form of decision making prevalent in a society or group is likely to have a great impact on how they deal with an external effect. Peterson and Broad (2016, 76) report that ‘anthropologists have started to focus on the central role of institutional flexibility for successful adaptation’ and on ideas such as that ‘inasmuch as climate is a constraint on decision-making, it can also provide an opportunity for creative agency’.

A similar approach drives the work conducted by the United Nations (2016) on poverty and vulnerability to climate change. The UN Climate Change Resilience report observes that the adverse effects of climate change are distributed unevenly, tending to fall specifically upon particular, poorer, regions of the world. Poverty probably is not a useful category for prehistoric purposes. However, as the report notes, there are ‘structural inequalities that perpetuate poverty, marginalization and social exclusion’ (ibid., 4). The report also argues that ‘addressing the root causes of inequalities requires transformative policies that generate change in the fundamental attributes of systems, particularly the existing governance systems and norms that perpetuate inequalities’ (ibid., ix). Governance systems are the kinds of phenomena that archaeology is able to detect.

There is no reason why such ideas and approaches could not be explored in archaeology and some have already begun to do so (see e.g. Leroy 2006; De Keyzer 2016; Cooper and Sheets 2012). Archaeology might benefit from giving systematic, creative thought to the question of how particular societal forms (kinds of ‘governance systems’) might fare in the face of climate and environmental change. There exist opportunities here to review known prehistoric cases from a comparative perspective and perhaps to arrive at generalizable, testable and revisable insights as to how different social formations might, on the one hand, produce vulnerability and, on the other, be able to deal with hazards.

However, the worry is that such a framing of the human–environment relation as producing vulnerability does not come to us naturally. As noted earlier, the human–environmental relationship is typically conceived using the thresholds model and our studies typically focus on detecting parallel shifts in environmental and archaeological proxies. In such a model, human action is likely to appear automatically as a response to external shifts (adaptation). By contrast, we would like to raise the possibility of conceiving human action as co-constitutive of the development of the human–environment relation.

There are (at least) two ways of attributing agency to humans in the face of external environmental change. One is to see humans as exercising agency in their coping in the aftermath of climate and/or environmental change. Here we afford humans agency, but that agency is exercised in response to an external event. The second way is to see the human–environment relationship as always already sociocultural in the way humans are positioned with respect to their environmental thresholds. The difference lies in the moment in time human agency enters the picture. There appears to be a tendency in the literature to focus upon the first kind of agency that views coping as a response to an external change. It is important to repeat, the very form of the palaeo-environmental research design, as focused on detecting parallels in changing environmental and archaeological proxies, automatically places that temporal focus on human coping, adaptation and response. The latter conception, which views human sociocultural organization as constitutive of the relationship from the beginning, seems underdeveloped in archaeology and, importantly again, the parallelistic form of research design implicitly and automatically militates against such a view.

A brief recourse to the case of modern Haiti in the wake of the 2010 earthquake is illustrative of the research design differences between the anthropology-of-hazards literature on the one hand, and palaeo-environmental archaeology on the other. In his account of the Haitian case, Anthony Oliver-Smith argued that the analysis of multiple catastrophic environmental impacts witnessed in the region across decades and centuries ‘reveals how deeply embedded they were in the historical processes that resulted in the unequal distribution of risk and vulnerability at the national, regional, and local levels in Latin America and the Caribbean’. In this sense, Oliver-Smith (2012) wrote, in 2010 Haiti ‘in some respects experienced the culmination of its own more than 500-year earthquake’. What then follows in Oliver-Smith’s account is a long history of colonialism and capitalism leaving its mark on Haiti and socially producing a grave state of vulnerability:

A lack of building codes, together with informal settlements, widespread undernourishment and hunger, disease, poor access to clean water or electricity, inadequate educational and health facilities and services at the national and municipal levels, and crime and corruption led to the construction of extreme vulnerability. In addition, Haitians were largely unaware of the seismic risk on the island, although seismologists had been warning of the possibility of a strong earthquake.

If we take anthropology of hazards as our model, a striking characteristic is that the analysis is focused on the character of the social system prior to the external event which forms a basis for how that system channels the challenges and opportunities posed by the event. By contrast, archaeology typically takes primary interest in the social system after it has been impacted by the external event to see how the system has changed as a result of the event. The contrast here is real – paradigmatic even – and the different paradigms give rise to different narratives of the past: one is focused on human innovation under pressure from external events, while the other is more political and focused on how certain societal dynamics might be producing vulnerability here and now, prior to the external event. In so far as palaeoclimatology and archaeology offer models beyond archaeology in how to think about the human–environment relationship, the scientists bear some political responsibility for where they see the crux of the matter, whether in posterior adaptation to external events or in the prior social production of vulnerability.

Conclusion

The above discussion has sought to understand the character of the debate around environmental determinism in archaeology. We reached three main conclusions:

• Palaeo-environmental studies in archaeology typically look for parallel shifts in the archaeological as compared to environmental and climate data.

• The biologistic idea of the environment as setting the thresholds within which humans exist underlies the view that changes in the thresholds are seen to necessitate a human response. There is an uncertainty as to how we should feel about biologism and the concept of thresholds (as plausible as they might be) and whether human beings should be viewed as merely responding to such external changes. Furthermore, reccurring cases from the literature suggest that the effects of the thresholds tend to be assumed rather than approached as an empirical question, which critics subsequently have indeed pointed out.

• Drawing inspiration from anthropology of hazards, archaeology might do well to think systematically about the social basis of hazards; that is, how societies might socially produce vulnerability in all of that society or in selected parts of it. Such social production would involve, for example, decision-making systems that are unable or unwilling to effectively coordinate – or, conversely, particularly well suited to coordinating – joint action to deal with changing conditions. It would also be desirable to broaden the temporal scope during which the phenomena of interest are considered to take place to include also the period of the social construction of vulnerability during which the institutions and other social dynamics of social vulnerability were shaped, and which precedes the putative environmental or climate event.

Our conclusions highlight the role and importance – indispensability, even – of interdisciplinary research into human–environment interactions. It is clear that specialists from a wide range of disciplines will be needed to answer questions about the thresholds on the one hand, and about the social production of vulnerability on the other, each of which concerns transformations that occur in a variety of very different scales and dimensions. Prehistoric research is uniquely placed to look at longer timelines and can view past environmental and cultural developments from a long-term perspective. The interaction of disciplines and the development of mutual, cultural and environmental explanatory models – while certainly more easily said than done – can help to avoid monocausal and unilinear determinism.