Relating traditional and academic ecological knowledge: mechanistic and holistic epistemologies across cultures

Introduction

Indigenous, local, and traditional ecological knowledge (IEK, LEK, TEK) are increasingly recognized in the biological and environmental sciences but also raise fundamental methodological questions about the limits of knowledge integration. On the one hand, Indigenous scholars have emphasized the importance of TEK for understanding and managing local environments (Cajete 2000; Pierotti and Wildcat 2000; Wilson 2008) and have contributed to growing debates about “co-creation”, “co-management”, “participation”, and “transdisciplinarity” (Berkes 2018; Chapman et al. 2016; Hadorn et al. 2006; Parsons et al. 2016) in areas such as agriculture and conservation. By attacking what Agrawal (1995) has called the “divide between Indigenous and scientific knowledge”, these accounts typically combine epistemic and political concerns (Whyte 2018). Holders of TEK are epistemically recognized as experts (Byskov 2017) about local environments and this expertise is mobilized to challenge their political marginalization (Scott 1998; Shepherd 2006; White 2014) in negotiations of practice and policy.

On the other hand, attempts to integrate TEK into institutionalized science often create tensions with its definitional feature of being part of the heritage of specific communities. While there is a well-established academic consensus about the relevance of TEK in the biological and environmental sciences, the line between integration and assimilation is often far from clear. For example, El-Hani and de Ferreira Bandeira (2008: 7) warn that emphasis on the scientific character of TEK can lead to a loss of “the distinctiveness of other ways of knowing, and, also, their epistemic value in terms of their own validation criteria.” Related concerns are articulated in Nadasdy’s (2003) influential critique of integration projects that run the risk of extracting “scientifically useful” information from TEK while marginalizing (e.g. ethical and spiritual) aspects that do not fit the agendas (Castleden et al. 2017), methods (Marlor 2010), and ontologies (Ludwig 2018) of scientists.

The result seems to be a dilemma of division and assimilation in which either the epistemic potential or unique character of TEK becomes misrepresented through inadequate framing. For example, consider projects that aim to integrate TEK of Indigenous hunters with academic ecological knowledge (AEK)1 of conservation scientists. Indigenous hunters will often possess a lot of relevant expertise and may be best equipped to monitor vulnerable populations, identify causes of population decline, detect changes in ecological dynamics, identify requirements of sustainable hunting, and so on (Ens et al. 2015; Weiss et al. 2013). However, much of this expertise will also often be entangled with spiritual values and worldviews that diverge substantially from academic approaches to conservation management (Barbour and Schlesinger 2012). Treating holders of TEK as “collaborators” who can contribute evidence to AEK “just like scientists” therefore runs the risk of obscuring or completely missing core aspects of TEK.

The aim of this article is to develop an account that relates the epistemic resources of TEK and AEK while avoiding both horns of the dilemma of assimilation and division. Debates about the relations between TEK and AEK commonly start by characterizing the former as “holistic” and the latter as “reductionistic” or “mechanistic” (Aikenhead and Ogawa 2007; Dods 2004; Wood 1999). While such characterizations can convey important insights, this article argues that a simple holism–mechanism divide misrepresents the epistemic resources of both TEK and AEK. The following section emphasizes that holders of TEK are perfectly capable of identifying mechanisms that underlie ecological phenomena. Using the literature on mechanistic explanations in philosophy of science, we reinterpret Lansing’s (1991) classical case study of rice farming in Bali and argue that the neglect of TEK about ecological mechanisms turned out to be both epistemically inadequate and politically problematic. While holders of TEK can identify ecological mechanisms, the second half of the article emphasizes that TEK also often employs strategies that are holistic in two distinct meanings of the term. First, TEK commonly involves a methodological holism that avoids decompositional analysis and rather addresses dynamics of ecosystems as “wholes”. Second, Indigenous scholars often characterize TEK as holistic in a broader sense that emphasizes metaphysical, spiritual, and ethical dimensions of interrelatedness and reciprocity. The conclusion returns to the issue of differences between TEK or AEK by developing both a critical diagnosis and a constructive proposal. We argue that the prevalence of mechanistic resources in TEK and of holistic strategies in AEK undermines a simple holism–mechanism divide. Instead, we propose a more fine-grained analysis of heuristics in TEK and AEK that provides resources for engaging with both similarities and differences in a more substantial way that can inform negotiations of ecological knowledge between diverse stakeholders.

Identifying and managing mechanisms with TEK

While it has become widely agreed that there is no simple demarcation criterion that distinguishes TEK and AEK (Agrawal 2014; Drouin-Gagné 2014; Gorelick 2014; Whyte 2013), acknowledgment of substantial differences between knowledge systems is often a requirement of adequate and respectful engagement with TEK. A common strategy of describing differences is to contrast the “holistic” nature of TEK with the “mechanistic” or “reductionistic” character of AEK or Western science more generally. Furthermore, this contrast often motivates a wider framing according to which “traditional knowledge [constitutes] a challenge to the positivist-reductionist paradigm in Western science” (Berkes 2018: 20) as it generates understanding of local environments and sustainable practices in a holistic way.

This article focuses on mechanisms rather than reductionism, as a generic characterization of Western science as “reductionistic” is clearly insufficient for a nuanced analysis of epistemic strategies in TEK and AEK. While reductions can be important epistemic strategies in scientific practice (Dizadji-Bahmani et al. 2010), it has been commonly noted in philosophical debates (Brigandt 2010; Kellert et al. 2006; Ruphy 2016) that science is often not reductionistic in practice. Furthermore, limitations of reductionist perspectives have been especially prominent in the life sciences (Brigandt and Love 2017; Kaiser 2015) and debates about different levels of organization have even been characterized in terms of an “anti-reductionist consensus” in philosophy of biology (see Waters 1990; Griffiths and Stotz 2013).

While a general characterization of AEK as reductionistic would be based on an simplistic picture of “Western science”, a focus on mechanisms constitutes a more promising starting point. Mechanistic approaches have been extraordinarily influential in recent philosophy of science, in part because they are framed as an alternative to traditional reductionist programs (Andersen 2014; Craver 2007; Fehr 2004). While many “New Mechanists” (Levy 2013; Franklin-Hall 2016) reject traditional reductionism and stress the importance of components across multiple levels of organization, they still focus on explanatory strategies that can be plausibly contrasted with holistic perspectives in TEK. Most importantly, mechanistic approaches involve decompositional analyses of “parts whose activities and interactions are responsible for the phenomena” (Illari and Glennan 2017: 1). Even if mechanistic philosophers do not assume that a target phenomenon can be completely reduced to a lower level of composing parts, the mechanistic focus on interacting sub-components suggests an attractive contrast with holistic perspectives on ecosystems “as wholes” that are commonly emphasized in TEK.

For example, an academically trained ecologist may analyze a target phenomenon such as the spread of an agricultural pest in terms of a mechanism of interacting components (e.g. intensification and monocropping that create a favorable habitat for a pest, loss of biodiversity that reduces prevalence of parasites and predators of the pest, development of pesticide resistances). In contrast, holders of TEK may employ pest management strategies (e.g. ethical and spiritual requirements of engagement with a forest, agricultural norms regarding polyculture or cropping schedules) in their agroecological systems that are effective even if they do not aim for an explanation of how different components interact in creating pest vulnerability. This way, both TEK and AEK can guide effective strategies of pest management even if they employ very different epistemic strategies.

While this contrast between holistic and mechanistic approaches has some prima facie plausibility, it provides an overly simple picture of the epistemic resources of both TEK and AEK. This section develops the argument that a limitation of mechanistic analysis to AEK obscures the capacities of TEK to identify mechanisms of complex ecological phenomena. Holders of TEK are not only capable of identifying mechanisms of complex ecological dynamics but also of intervening in them through equally complex management strategies. Taking the epistemic resources of holders of TEK seriously therefore also requires to recognize that the identification of mechanisms is not exclusive to AEK or Western science more generally.

Developing a more adequate account of the scope of mechanistic perspectives in TEK and AEK requires some specifications of what is meant with “mechanism”. Mechanism talk has become ubiquitous in recent philosophy of science and is used to develop heterogenous frameworks and with very different agendas (Levy 2013; Nicholson 2012). In its most ambitious interpretation, mechanistic perspectives are associated with what Dijksterhuis (1961) famously described as the “The Mechanization of the World Picture” in early modern science in Europe. Given such an appeal to what Nicholson (2012) calls “mechanicism”, it is not only obvious that TEK is non-mechanistic but also questionable whether current trends in AEK can be adequately described as mechanistic (Berkes 2018). However, substantial parts of mechanistic philosophy of science focus on specific epistemic strategies rather than these more general debates about “mechanistic worldviews”. For example, Machamer et al.’s (2000) landmark article develops a mechanistic framework with examples from molecular biology and neurobiology. Their guiding questions are what it means for molecular biologists to refer to the “mechanism of DNA replication” or for neurobiologists to refer to the “mechanism of chemical neurotransmission”. In both cases, Machamer et al. argue that scientists’ appeals to mechanisms reflect a distinct epistemic strategy that involves “entities and activities, organized such that they are productive of regular changes” (2000: 1). Applied to the relation between AEK and TEK, the question is therefore not whether holders of TEK subscribe to a “mechanistic worldview” (they don’t) but rather whether they also employ this distinctive epistemic strategy that has become commonly emphasized in scientific practice.

Illari and Williamson’s (2012) proposal of a general account of mechanisms helps to further specify these issues. While there has been little agreement on a general definition of “mechanism”, Illari and Williamson aim to provide a minimal account of core components by stating that a “mechanism for a phenomenon consists of entities and activities organized in such a way that they are responsible for the phenomenon” (2012: 120). This definition includes three components that can be illustrated with Machamer et al.’s examples of DNA replication and chemical neurotransmission: First, mechanisms are composed of entities and activities. For example, the mechanism of DNA replication is composed of entities such as the double helix and base pairs as well as activities such as the unwinding of former and bonding of the latter. The mechanism of chemical neurotransmission is composed of entities such as neurotransmitter molecules and post-synaptic neurons as well as activities such as the release of the former and the depolarization of the latter. Second, mechanisms require the organization of these entities and activities into a mechanism. Both DNA replication and chemical neurotransmission are clearly not produced by an unordered set of entities and activities but rather by their intricate organization. Third, entities and activities have to be organized so that they are responsible for the target phenomenon. For example, organized entities such as the double helix and base pairs and activities such as their unwinding and bonding constitute a mechanism that is responsible for the target phenomenon of DNA replication. Target phenomena and their mechanisms stand in clear relations of dependency in the sense that the former are responsible for the latter.

Illari and Williamson’s account of mechanisms is intended to be broadly applicable in scientific practice for case studies from social psychology to astrophysics. It expresses the assumption that researchers in very different disciplines share a broad epistemic strategy that analyzes the organization of entities and activities that are responsible for target phenomena. Given this broad characterization for application in different disciplines, it becomes possible to see how not only AEK but also TEK can involve mechanistic epistemic strategies. Holders of TEK are often not only aware of many of the entities and activities that are involved in ecological dynamics but also of how the specifics of their organization are responsible for a target phenomenon. Furthermore, it is often precisely this intimate knowledge of entities, activities, and their organization that puts holders of TEK in the position to manage local environments by intervening in local mechanisms. The following section develops these claims in the context of Lansing’s (1991) classical case study of TEK about rice farming in Bali and of the marginalization of this knowledge in the wake of the Green Revolution.

Limitations of mechanistic approaches in TEK

In a collaborative project that integrates ecology and philosophy of science, Poliseli (2018) developed a detailed analysis of the potential of mechanistic approaches in AEK. Analyzing an autochthonous bee community in an agricultural pollination service in Brazil, Poliseli et al. emphasize the guiding role of heuristics in the construction of their mechanistic model. Understanding heuristics broadly as epistemic tools that guide engagement with target phenomena, the project developed a diverse toolbox of heuristics that was employed to generate an account of the pollination service in terms of underlying mechanisms. For example, Poliseli et al.’s heuristics include phenomenon characterizations that specify explanans and explanandum, mechanism sketches that visualize relations within the pollination service, hierarchical structures that locate components of the explanans at different levels of organization, and enabling conditions that identify core variables of the mechanisms in the pollination service.

Given the arguments from the last section, it will not surprise that some of Poliseli et al.’s heuristics are not restricted to AEK but can also be applied in contexts of TEK. For example, the complex Balinese system of rice farming requires identification of core enabling conditions such as irrigation and cropping schedules and tools of visualization have been widely documented in TEK and Indigenous knowledge even if they are not formally structured as mechanism sketches (Joyce 2012; Begossi and Caires 2015). That being said, Poliseli et al. developed their heuristics to guide the development of a specific mechanistic model for a specific ecological phenomenon and does not suggest that their heuristics will be applicable in all contexts of AEK or even TEK.

While some TEK heuristics can contribute to a better understanding of mechanisms, this section argues that many TEK heuristics are valuable precisely because they circumvent questions about the organization of entities and activities that is responsible for a target phenomenon. Mackinson’s (2000) influential study of local knowledge of fishery in British Columbia provides a helpful starting point for developing the argument that avoiding mechanistic perspectives is a productive feature of many epistemic strategies in TEK. Based on interviews with First Nations, non-indigenous fishers, fishery managers, and fishery scientists, Mackinson collected a large set of IF … THEN clauses about the behavior of herring shoals and integrated them in the expert system CLUPEX. Mackinson refers to these IF … THEN clauses as heuristics and emphasizes the expertise of local informants in identifying them. Examples of locally communicated heuristics include: “If there is a bright moon the fish spread out fast and it is difficult to catch them” and “If weather is bad close to spawning then schools get disrupted”. Heuristics in Mackinson’s case study share the feature of relating specific attributes such as bird predator type, current strength, moonlight, water depth, or weather with equally specific descriptors of herring shoals such as size, packing density, depth, shape, cohesion, and movement of the shoal. If such relations are consistently and robustly communicated across several informants, they are translated into more formalized IF … THEN clauses such as “IF time of day is dusk OR night AND state of the moon is full and bright, THEN shoal depth mid-range AND packing density very low AND ease of capture very low” (Mackinson 2000: 540).

Mackinson’s and Poliseli et al.’s heuristics differ as the former aim at dynamics of ecosystems at the macro-level without providing an account of underlying mechanisms. Poliseli et al.’s heuristics address pollination by developing a detailed account of the organization of entities and activities that are responsible for the target phenomenon. Their case study therefore illustrates how AEK can be mechanistic in the sense of Illaris and Williamson’s (2012) criteria from the last section. In contrast, Mackison’s heuristics do not satisfy Illari and Williamson’s core criterion of responsibility as they identify descriptors that can be used as indicators for the target phenomenon of herring behavior but do not include assumptions about how (and whether) the former is responsible for the latter. Of course, some informants may combine descriptors such as moon phase or weather conditions with accounts of their causal contribution to the target phenomenon. However, the successful application of the heuristics does not require such accounts and can also be embedded in belief systems that develop very different explanations of regularities. For example, a heuristic that relates brightness of the moon and packing density of herring shoals could be effective while being embedded in very different causal stories in TEK that strongly differ from AEK by being framed in terms of intentional actions of the moon.

Mackinson’s case illustrates the importance of TEK that is not mechanistic in character because it aims for patterns and regularities at the macro-level without attempting to explain how a target phenomenon is produced through the organization of sub-components. The importance of this knowledge about dynamics of ecosystems at the macro-level is acknowledged by Mackinson and constitutes his main motivation for integrating local knowledge in AEK about herring shoals. Given “our incomplete understanding of biological and ecological mechanisms underpinning behavioral responses of fish” (Mackinson 2000: 533), it is often not feasible to build predictive models exclusively on the basis of mechanistic analyses. At the same time, predictive modeling is needed for purposes of conservation and fishery management. The expert system CLUPEX aims to solve this problem of insufficiently understood ecological complexity by circumventing questions about mechanisms and by building a predictive model on the basis of regularities and patterns at the macro-level.

Both the problem and the solution are well-known to holders of TEK whose livelihoods depend on interactions with complex ecosystems and who often have to make rapid decisions on how to interact with their environments during fishing and other daily activities. Intricate models of the organization of productive sub-components in the sense of Illari and Williamson can be feasible in some situations such as Balinese rice farming where TEK institutions invest considerable resources to monitor and reorganize relevant entities and activities such as dams and cropping schedules. However, this does not mean that mechanistic perspectives will always provide the most successful strategies for dealing with complex ecological systems in TEK. Heuristics that circumvent issues of underlying mechanisms will often be preferable in applied contexts such as conservation management and are widely employed in TEK.

Conclusion: relating knowledge systems beyond the mechanism–holism divide

This article started with the common criticism of a “divide between Indigenous and scientific knowledge” (Agrawal 1995) as creating artificial boundaries and thereby contributing to the marginalization of the former. To some degree, our discussion can be framed within this wider critique as it suggests two shortcomings of a simple mechanism–holism divide between TEK and AEK. First, generic characterizations of TEK as holistic obscure the diversity of its epistemic resources. Holders of TEK are perfectly capable of identifying and intervening in mechanisms of complex ecological phenomena such as the agroecological system of Balinese rice farming. Second, holistic elements are not exclusive to TEK and a characterization of AEK as entirely mechanistic also runs the risk of obscuring the diversity of its epistemic resources. At least holism in the more restricted methodological sense often plays an important role for successful engagement with complexity and uncertainty in AEK. Ecological phenomena are commonly shaped by a large number of insufficiently understood variables and feedback loops that limit the prospects of predictive modelling through mechanistic analysis. This is explicitly acknowledged both in Mackinson’s (2000) and Berkes’ (2018) incorporation of TEK into AEK and converges with more general debates about the limitations of mechanistic perspectives in philosophy of biology (Braillard and Malaterre 2015; Dupré 2013; Halina 2017; Woodward 2013). In other words: Mechanistic approaches are relevant in TEK and AEK without exhausting the epistemic resources of either of them.

Recognizing the inadequacy of a simple mechanism–holism divide matters for entangled epistemological and political reasons. Epistemologically, a simple contrast between holistic TEK and mechanistic AEK obscures the complexity of both knowledge systems in practice. In contrast with such a simple divide, this article has aimed to develop a more adequate account that acknowledges mechanistic and non-mechanistic features of epistemic practices beyond generic labeling of TEK and AEK as either holistic or mechanistic. While our discussion has mostly focused on these epistemological issues, it is important to acknowledge their entanglement with applied and political questions about the status of TEK. First, bringing the resources of mechanistic philosophy of science into debates about TEK is not only theoretically intriguing but helps to clarify epistemic resources that have often been neglected and led to the marginalization of holders of TEK in practice. Lansing’s Balinese case study provides a powerful illustration of this practical significance as the neglect of Balinese TEK in the wake of the Green Revolution led to pest outbreaks and water shortages with catastrophic consequences for local farmers. Second, it is equally important to avoid a reduction of TEK to its mechanistic resources as substantial parts of TEK cannot be isolated from the wider worldview and way of life local communities. Taking TEK seriously therefore also requires to look beyond its potential contribution to mechanistic analysis and to recognize its entanglement with holistic reasoning in TEK.

While this article has criticized a simple mechanism–holism divide, our introduction contrasted the problem of division with a problem of assimilation that downplays the unique (cultural, ethical, metaphysical, methodological, social, spiritual…) aspects of TEK for the sake of integration with AEK (El-Hani and de Ferreira Bandeira 2008; Ludwig 2016; Nadasdy 2003). And indeed, the analysis of this article is not limited to a rejection of the mechanism–holism divide but can also be used for a more fine-grained description of differences between epistemic resources in TEK and AEK. Rather than using generic characterizations of knowledge systems as “holistic” or “mechanistic”, we have argued for a heterogenous toolbox of heuristics with diverse mechanistic and holistic features. Although this general analysis applies to both TEK and AEK, a more detailed analysis of the application of heuristics in practice will also uncover substantial differences.

Even if both AEK and TEK can involve the identification of mechanisms, heuristics in AEK are commonly shaped by methodological requirements and standards of scientific research that strongly differ from those employed in TEK. For example, Poliseli (2018) heuristic of Evidence Frequency aims to identify causal relations between components on the basis of probabilistic and mechanistic evidence from various sources. While it may be possible to locate this general epistemic strategy in some contexts of TEK, Poliseli (2018) also specify this heuristic in terms of strategies such as analytic and synthetic analysis, forward and backward chaining, graphs and indexes that are largely specific to academically institutionalized research. Speaking more generally, heuristics in AEK are shaped by methods and standards (e.g. of experimentation, modelling, and quantification) that clearly differ from methods and standards that shape practices in TEK (see also Marlor 2010). One does not need a general mechanism–holism divide to acknowledge these differences in a more fine-grained account of the relations between epistemic strategies in AEK and TEK.

Furthermore, not only heuristics in AEK but also in TEK often involve unique aspects that need to be acknowledged in a more fine-grained comparison of epistemic strategies. For example, AEK commonly employs epistemic strategies that are holistic in the sense of addressing dynamics of ecosystems at the macro-level while avoiding mechanistic decomposition in terms of productively organized entities and activities. However, we argued in the context of Berkes’ case study of Caribou that many heuristics in TEK need to be understood as holistic in much wider sense of being entangled with the ethics, worldview, and way of life of Indigenous and other local communities. Of course, AEK is also entangled with values and worldviews of Western researchers and wider issues of interrelatedness are not exclusive to TEK. At the same time, Berkes discussion makes it very clear that many Cree heuristics are very different from heuristics that are employed by AEK as they are shaped by unique metaphysical beliefs (e.g. animals deciding to be caught) and values (e.g. how to respect a caribou) of Chisasibi Cree worldview. Even if AEK is not exclusively mechanistic, holistic heuristics in TEK will often be clearly distinguished from AEK heuristics in how they integrate metaphysical, spiritual, and ethical commitments of traditional communities.

The rejection of a simple mechanism–holism divide does therefore not collapse differences between AEK and TEK but provides opportunities to locate differences in more specific practices. Holders of AEK and TEK both employ diverse epistemic toolboxes that include mechanistic and holistic strategies but that does not imply that they are using identical strategies. Locating both similarities and differences in specific practices beyond a simple mechanism–holism divide is not only an interesting theoretical task for philosophy of science but can contribute to epistemologically and politically more adequate practices of co-creation. Understanding similarities (both through mechanistic resources of TEK and holistic aspects of AEK) is important for finding common ground in co-creation and avoiding an artificial divide between knowledge systems. Understanding differences (both in quantitative and qualitative characteristics of heuristics) is important for taking the unique characteristics of TEK seriously and avoiding its inadequate assimilation into academic research.