Hermeneutics of complexity and the limits of learning pre-specification

Abstract

Contemporary work confronts researchers and educators with a range of challenges, including those posed by complex environments in which workers grapple with uncertainty in high-stakes situations. A different challenge lies in the assumptions brought to learning design and facilitation, assumptions which together constitute a ‘paradigm’ concerning how to approach these tasks. Without a paradigm, designers and practitioners may exercise an incoherent practice; but if the paradigm is in some way problematic, then the practice may be compromised. In this article, we focus on work in complex environments, problematizing a particular paradigm—the ‘pre-specification’ approach—that is often brought to bear on it. The key to this paradigm is that work can be analysed and represented such that it can be specified in advance as a reference point for educational endeavours. We problematize key epistemological and ontological assumptions of this paradigm. Hermeneutics draws attention to the limits of these assumptions regarding what we can understand and represent about work, while complexity theory allows us to recognize ontological features of work that make pre-specification difficult or untenable. We conclude that in complex work, representation-making needs to be a distributed activity rather than a specialty, and it occurs at multiple points on the learning continuum rather that solely at the beginning.

Introduction

Professional, continuing, and vocational forms of education present peculiar conceptual difficulties. It is perhaps only within the last two centuries that education for work, and what work involves, has attracted serious attention from educationalists and philosophers (such as Kerschensteiner and Dewey respectively). While significant inroads into the thicket of questions posed by these forms of education have been made (e.g. Winch 2017), a distinct set of approaches to them has been entrenched which, although it draws on meagre theoretical resources, retains a powerful grip on practice. We offer a rubric for this set of approaches and attempt to tease out some of its key weaknesses. What we call the ‘pre-specification paradigm’ is the idea that expertise can usefully be known and described in advance of learning and those situations in which this learning is activated. We develop our argument in relation to emergency healthcare—a kind of work to which the pre-specification paradigm has been applied for the purpose of developing professional education programmes. Educators in this area contend with a range of difficulties created by work in complex environments that involve uncertainty in high-stakes situations (such as those described in Bahl et al. 2013). How to prepare workers for this type of setting (which this article will refer to as ‘complex work’) is more than a matter of developing competence with respect to knowable elements of work situations. Rather, these situations contain much that is unforeseeable—elements which frustrate attempts by learning designers to prepare workers through typical educational interventions.

Contemporary work environments produce such situations in occupations like management, and in professional endeavours such as theatrical and musical performance (Hager and Beckett 2019). While in some of these occupations life itself may be at stake, in each the unknown is an element that impinges on the application of expertise, causing uncertainty and stress. Yet learning designers and educators may approach these situations with assumptions that screen off the difficulties produced by the inherent openness or indeterminacy of critical situations for which workers are being prepared. These assumptions serve instead to focus what is and can be known and represented in advance, and to downplay the remainder as something temporarily deviant which need not hinder the application of design and education measures. Designers and educators working in this mode are positioned, wittingly or not, in the Platonic tradition, which has for more than 2,000 years promoted the view that pure theory has inherent priority over its potential applications in the everyday world. The paradigm we problematize in our analysis is a contemporary reflection of Plato’s idealism.

The purpose of our analysis is two-fold. First, we attempt to draw out the assumptions indicated above, showing that they amount to a paradigm which systematically overstates what can be known in advance about work situations and expertise. We use critical hermeneutics and complexity theory to exhibit these assumptions. We then employ these theoretical resources so as to sketch an alternative approach that decentralizes design and education efforts in order to more fully engage with the types of work with which we are concerned. We conclude that in certain acute, complex work settings, ‘representation-making’, as a fundamental ingredient in competent engagement, should be a distributed activity as opposed to an exclusive specialty. That is, in these types of work, the representation of situations should be continually revised, and workers themselves must take a central role in forming and reforming them. As such, representation occurs at multiple points along the design and learning continuum, rather than solely at the beginning. We signal a paradigm shift with significant implications for the work of education professionals.

The pre-specification paradigm

Education and training endeavours are inherently oriented towards the future. They approach learners as capable of being prepared in some way for life or work, and evince confidence about our current understanding of the world such that we are already in a position to intentionally undertake that preparation. Cartesian dualism asserts itself here—the priority of mind over matter, thoughts preceding actions. Thus, there is a temporality intrinsic to education and training, without which the whole point of these endeavours would be lost. But there are different ways of theorizing and enacting educational temporality. For example, in Mezirow’s (2000) theory of transformative learning, educators can play a role in facilitating learning; but the ‘outcomes’, such as they are, cannot be predicted, and no programme can be built that would systematically elicit the benefits claimed for perspective transformation. What can come of learning cannot be pre-specified for given learners. Freire’s (1970) educational theory offers another example of learning that defies pre-specification. Here, it is the material–political status quo shaping the context of learning that changes in unforeseeable ways. We might say that a revolutionary temporality is indicated by these educational endeavours, one that seeks to overturn the order of the present. In the educational present, an understanding of the ‘now’ both shows us what we must transcend, and brings focus to activating powers in people to effect that movement. The pre-specification paradigm, as we call it, reflects a contrasting take on educational temporality. The future is here regarded as qualitatively continuous with the present—a place for the enactment of knowledge and skills already understood. There is no important difference between ‘now’ and ‘then’, only novel permutations of more or less well-known elements. For an example of the pre-specification approach, consider competency-based education and training (CBET). Investigation into the literature on this approach to learning reveals deep, unexamined assumptions about temporality (Hodge et al. 2020). It is assumed, for example, that the work learners will undertake can be exhaustively described and documented ahead of enactments of the work itself. Descriptions of tasks, performances, or behaviours are made as complete and precise as possible, and analysts and writers strive to ensure that little, if anything, is left out of the picture. A qualitative temporal dimension is conspicuously absent from this conception of learning. What was, is, and will be, are treated as part of a single static moment subject to analysis and description, with learners, educators, and future work situations all assimilated to an essentially ‘tenseless’ system. This framing, however, assumes much about temporality. Indeed, an a-temporal theory of education and training presupposes as much about temporality as does, for example, Freire’s.

A thorough attempt to articulate the pre-specification paradigm can be found in the work of researchers working for the US military in the 1950s and ’60s. Robert Miller (1962) describes a peculiar challenge in training design for operators of missile systems. In the period in question, that of the Cold War, the ‘arms race’ prompted very rapid development of military technology. One challenge was the need to prepare operators for sophisticated equipment that did not yet exist in physical form. There was no way to provide the trainees with opportunities for practice without compromising the project of military supremacy. Rather, the demand was that they be trained to effectively use the systems as soon as the equipment and environments were constructed. This principle was known as the doctrine of concurrency: ‘that the human components and the remainder of the system must be programmed to arrive at a man-machine assembly point at the same moment in time’ (Melton 1962: vii). Miller and fellow psychologists worked with engineers who designed these weapon systems to determine what future operators would need to know and do. At that stage in the development of education and training theory, according to Miller, learning design was only based on ‘heuristics’—but the goal was that their efforts would eventually find a scientific basis:

We must hope that fundamental research may eventually provide us with a descriptive system for behavior which can provide the factors and measurement operations for an equation of all input-output relationships of potential interest in man-machine systems. This hope may, however, contain a paradox. One of the major uses for the human in systems is as an improviser, to meet contingencies the nature of which can be anticipated only in part, and these imperfect anticipations of system activities make for imperfect specifications of human input-output requirements. Nevertheless, it is a properly pious hope, not only that science will be reduced to practice, but that on occasion that practice will be reduced to science. (Miller 1962: 188)

The pre-specification paradigm is expressed here, both in the heuristic method then underpinning the doctrine of concurrency, and in Miller’s ‘pious hope’ for a scientific basis of the doctrine; but the significance of this exemplar is perhaps the impetus to behavioural objectives, and performance- and competence-based education and training movements, generated by the conjunction of military need (and resources) and the discipline of psychology (Hodge et al. 2020). To be sure, the separation of expertise from the labour of its representation—a key assumption in Miller’s and his colleagues’ work—was conceptualized decades earlier in Frederick Taylor’s Principles of Scientific Management (1998 [1911]). Here, the argument was made that effective management in part required analysis of the work done on the factory shop floor, breaking it down into discrete tasks and later regulating individual workers’ labour on this basis. The problem, as Taylor saw it, was that workers sought to keep their expertise to themselves, and so long as their managers were ignorant about what was involved in their work, there could be no scientific management of production. But wresting the expertise from workers in order to implement Taylorism in factories was more than a moral and political struggle. How to represent the expertise itself was unclear. Colleagues of Taylor, Gilbreth, and Gilbreth (Kirwan and Ainsworth 1992) developed the ‘time-and-motion’ study methodology in an attempt to analyse skilled work, setting in train a whole new discipline of task analysis. When psychologists later became involved in this broad project, there were significant advances. Behavioural and then cognitive psychology each contributed to the effort which, in a sense, culminated with the intense scrutiny of the problem of how to represent work that was evident during the Cold War.

The pre-specification paradigm, operative in these different examples of education and training, theory, and practice, has endured even as those theories and practices themselves have come under attack by researchers (Hodge et al. 2020). Potentially, it is the power of the paradigm itself, rather than the efficacy of its instantiations, which accounts for the continued faith many have in systems based on the upfront definition of learning outcomes (cf. Taylor 1985). A focus on the paradigm reveals two kinds of assumption that warrant analysis, both for gaining critical traction and for imagining alternatives. The first is assumptions of an epistemological nature relating to the construction and use of representations, usually textual but sometimes graphic, fundamental to the implementation of the pre-specification paradigm. Here, we draw on ‘hermeneutics’, or the theory of interpretation, in order to pick out untenable presuppositions. Hermeneutic dimensions of education and training—consistently overlooked, yet ubiquitous in practice—become a source of real difficulty for the paradigm. The second kind of assumption is ontological, concerning work as such, which creates other challenges for the paradigm. We consider implications of complexity theory germane to understanding types of work that pose stubborn problems for educators and trainers (Doll 2012). We suggest that some kinds of work, when illuminated by ideas such as non-linearity and emergence, simply cannot be brought under the pre-specification paradigm. In the next two sections, we present analyses of each set of assumptions.

Epistemological assumptions

Over the course of about a century, job, task, and function analysts—from Gilbreth and Gilbreth (Kirwan and Ainsworth 1992) to Gagné (1985) to Clark et al. (2007)—have laboured to create methods for representing work for the purposes of management, control, selection, education, training, and assessment. Defaulting to a more or less positivist perspective on their endeavour, work is taken as appearing objectively in the time it is observed, or through the reflections of workers with the aim of construing the reasons for their behaviour. Perhaps natural phenomena can be ‘read off’ their empirical presence, even if that presence has formed through unknown expanses of time. Applied to human phenomena, this type of positivism is familiar as behaviourism. Human phenomena, however, are not necessarily manifest in their observable appearances in the same way, such that causes are really contained in or knowable as such through those manifestations (Taylor 1964). There is an inherent difficulty in interpreting human actions, which may be acute in the case of historical artefacts, fragmentary texts, psychopathic symptoms, or art, or potentially less so in relation to everyday utterances and simple tasks (Ricoeur 1972). Yet the difficulty remains that there is no guarantee that what we can observe or elicit (even with respect to obvious behaviour) really exhibits its causes.

In the 19th century, as debate arose about whether human and social phenomena could really be subsumed by the positivist programme, thinkers steeped in the ‘hermeneutic’ tradition such as Dilthey developed a position that undermined these aspirations (Schmidt 2016). With a rich interior life, formed socially and culturally with both conscious and unconscious material, and a propensity to (self-)interpret the dynamics of that inner world and outer behaviours, human and social phenomena possess a depth such that isomorphic relationships between causes and manifestations can never be assumed. From the start of the 20th century, hermeneutics theorized the difference between natural and human phenomena in this way—the lack of isomorphic relations in human phenomena demanded a different approach to research. In addition, Ricoeur (1972) introduced a distinction between hermeneutics that seeks to reconstruct the meaning of human and social phenomena, and a critical version concerned with distortions of meaning. For Ricoeur, thinkers like Marx, Nietzsche, and Freud were engaged in this latter kind of hermeneutics. Thus, theories of interpretation emphasize the need to actively interpret human phenomena, for they are not simple reflections of, and can even be misleading with respect to, their causes.

With its constitutive thrust running against the contention that work can be specified in a straightforward way, hermeneutic theory makes some evocative contributions to our investigation into the pre-specification paradigm. Specifically, bringing hermeneutics to bear problematizes the epistemological dimension of the paradigm. As Hodge et al. (2020) argue, this problematization occurs at two levels. First, there is analysis of work with a view to producing written specifications. Here, behaviours and/or cognitions effecting those behaviours become the focus. A hermeneutic view of this endeavour is that such manifestations are fundamental expressions of constellations of individual and social purposes, and are, in turn, individual and group interpretations of the requirements of the work. A behavioural analysis necessarily misses the dynamics of meaningful involvement in a significant occupation of which the behaviour is a momentary—and perhaps arbitrary or even misleading—glimpse. Picturing a behaviour using a document such as a skills inventory cannot furnish insight into why anyone would act in the way pictured and, crucially, does not provide a basis upon which to generate alternative behaviours that accomplish similar work in unforeseen conditions. Conversely, a cognitive analysis, which acknowledges and addresses the idea that behaviours are the product of something unmanifest, is fundamentally reliant on the veracity of the model of cognition that is brought to bear. Developments in the discipline of psychology are ongoing. For example, Gagné’s (1985) taxonomy gives way to more sophisticated analyses, such as that of Jonassen and Tessmer (1996/7). The composition of that picture and insights into behaviour from other disciplines undermine faith in the adequacy of psychological models, such as behaviourism and cognitivism, for representing determinants of human phenomena.

Hermeneutics problematizes the pre-specification paradigm in a second way. Specifications, however derived, must be read and deciphered to create learning programmes and assessment mechanisms with which they are consistent. Systems within the pre-specification paradigm such as CBET tend to pass over this matter in silence, perhaps supposing it unproblematic to leave to designers and educators the work of reconstituting the specified work for learners. This presupposition is not questioned in the literature surrounding CBET, although it is difficult to determine the warrant for the assumption. That there is such a presupposition is evident in the behavioural objectives theory, according to which:

a usefully stated objective is one that succeeds in communicating an intended instructional result to the reader. It is useful to the extent that it conveys to others a picture of what a successful learner will be able to do that is identical to the picture the objective writer had in mind. (Mager 1962: 19)

The effectiveness of such objectives to transmit in this direct way is a function of a systematic approach to formulating statements of intent, that is, of ensuring that a plain description of desired performance is accompanied by statements of criteria and conditions, as well as deliberate efforts to use language that ‘excludes the greatest number of possible meanings other than your intent’ (p. 20). Mager’s instructions here are based on a perspective on language and texts that is, as it turns out, idiosyncratic. Almost all disciplines that consider the nature of language, meaning, and communication arrive at positions that contradict Mager’s view. Translation theory, for instance, highlights the impossibility of suppressing diverse meanings for a single text. As Steiner (1998: 296) puts it, ‘We mean endlessly more than we say’—an expression of the central difficulty of all translation. Mager’s faith in our ability to suppress the proliferation of meanings seems naïve with respect to this standpoint. Hermeneutic theory, meanwhile, offers a full range of tools for comprehending the processes by which educators (and learners) respond to learning specifications.

Dilthey’s work, drawn on above, was taken up and transformed in the 20th century by key thinkers including Heidegger (1962), Ricoeur (1986), and Gadamer (1989). Through this development, the notion of the ‘hermeneutic circle’ was elaborated, a concept which presents a potent challenge to Mager’s (1962) ideal of transmission. Our encounter with any new text activates extensive and often tacit prior learning in an involved process of positioning the text within the structure of the known. Of course, an unfamiliar text also promises an influx of new knowledge, and it is the play and negotiation between known and unknown in our active reading that is encompassed in the idea of the hermeneutic circle. The positioning process occurs with respect to a range of reference points, which can be expressed in questions about the new text such as: what genre does it belong to? Who was the author and in what time and culture were they located? What was the author’s intent? What are the structural features of the text and how do they contribute meaning? What does the text mean to me (the reader) and the use(s) I may make of it? How does the text ‘sit’ with me—do I agree with it? Is it congruent with my experience? The positioning process as a dialectic between known and unknown thus involves continuous, iterative negotiation that never entirely settles but functionally reaches a point where the reader can say they ‘understand’ the text (or not). Ricoeur (1986) clarifies, additionally, that texts must be regarded as ‘autonomous’ of their author’s intentions. He explains that while in conversation our ability to question a speaker stands as a guarantee of some correspondence between authors’ and receivers’ understandings, once meaning is committed to a text, a gap opens between the author’s intentions and the reader’s comprehension. Over time, and as they are disseminated through multiple social and cultural contexts, texts take on a ‘life of their own’, weakening the link between authorial intentions and readings. This is an important dislocation that ultimately shifts the reader’s interpretative activity to resemble the potentially intricate process described in terms of the hermeneutic circle. Circularity thus replaces direct, linear transmission, and makes Mager’s ideal of transmission appear an unlikely scenario. Authors’ intentions fall away as texts enter into new relations with unknown people, through their practices.

Ontological assumptions

A particular difficulty of representing work for learning arises in situations where the parameters of the task or function to be analysed are uncertain. Aviation emergencies and acute healthcare situations are examples of such events, but a wide range of occupations confront workers with similar challenges, including air traffic control (Owen 2022), construction (Tichon et al. 2019), and the maritime industry (Bailey et al. 2023). These are situations that contain elements of the unknown and so are ‘open’ in an important sense, yet are, at the same time, high stakes—lives may be at risk, and a ‘solution’ of some kind is imperative. Educators and trainers in these fields generally use a combination of representations and simulated practice to address this challenge (Kearns et al. 2016). In these contexts, the pre-specification paradigm holds sway. The global aviation industry, for example, has adopted a competency-based approach and, in some settings, variants of this, such as ‘evidence-based training’ where assessment of performance must rely only on observable behaviour and data about incidents. However, the fields of healthcare and aviation include occupations that place practitioners in situations that exceed the scope of pre-specification approaches to inform effective professional preparation. It is a question as to why these approaches remain the default option. From current practices, it is evident that the issues highlighted here are not considered in terms of inherent limitations of prevailing approaches, but rather as failures to pre-specify in a comprehensive enough way. We suggest that complexity theory offers an alternative analysis, and illuminates ontological features of, for example, acute healthcare situations which suggest different ways of thinking about education and training.

Complexity theory is not the only perspective from which to critique the ontological assumptions of the pre-specification paradigm. Practice theory also embeds a critique of the pre-specification of learning (e.g. Lave and Wenger 1991). This approach sees all learning as fundamentally situated, with the context construed as a distinctive, multilayered material, cultural, and temporal setting for the pursuit of a socially significant endeavour in which knowledge, skills, relationships, identities, and power struggles entwine. In this view, ‘learning’ is a by-product of participation and commitment. For our purposes, however, the fact that a thoroughgoing practice theory approach offers little, if any, potential for meaningful preparation ahead of rare situations rules out its application to our challenge. Nevertheless, we accept the main ontological premise of the practice approach, that is, that the intricate social–cultural–material realities of work form a horizon against which our own analyses proceed. Another way to problematize the pre-specification paradigm is via so-called actor-network theory (ANT) (Pickering 2017). To be sure, ANT embeds complexity theory as an ontological orientation, viewing the kinds of situations we contend with as emerging via relationships among material and human entities. We stop short of a full utilization of ANT because, in its critical modality, the theory denies a constitutive role to representations of situations, preferring a ‘performative’ analysis. While, philosophically, the situations we are concerned with are performances in the sense conveyed by ANT, we need to revert to the concept of representation in our alternative analysis if we are to have anything practical to contribute to the problems of learning (albeit acknowledging the contingent nature of representations). So, while at one level, our analysis is consistent with ANT, for the sensibilities of that theory our proposals would give too great a role to representations. We do not believe, however, that at least tactical recourse to representations can be avoided given the imperatives of our chosen problem. Aware of the potential of practice theory and ANT as alternatives to the pre-specification paradigm, we opt instead for an analysis informed by complexity theory—a way of thinking about the unknown and its relationship(s) to a problem situation, and the expertise required for dealing with it optimally. It also gives us insight into the problems facing existing approaches to education and training directed at work in which the unknown is a salient factor.

The ‘science of complexity’ has its beginnings in the 1980s, when physicists and economists formed the conviction that ‘the kind of linear, reductionist thinking that has dominated science since the time of Newton … has now gone about as far as it can go in addressing the problems of our modern world’ (Waldrop 1993: 13). Complexity theory thus embeds a critique of traditional science. This critique is worked out in greater detail by Morin (2007), who argues that distinctive of traditional science is its disdain for complexity. Morin enumerates three principles of explanation basic to traditional science that conspired to suppress the real complexity of the world in the minds of earlier scientists. The first is the assumption of universal determinism, the principle that everything is determined and therefore knowable (in principle) through the past and into the future, and in a linear, systematic way. The second is the principle of reduction, or the idea that any whole can be explained by breaking it into parts or constituent elements. Central to this idea is that nothing essential is lost through such reduction. Finally, Morin suggests a third principle of ‘disjunction’ is in play, evident in the separation of disciplines, and notable for the ‘silo effect’ that plagues efforts to comprehend phenomena that transcend particular disciplines. Nevertheless, determinism, reductivism, and disjunction have contributed to real advances in science and technology; but for Morin, ‘In this scientific conception, the notion of “complexity” is absolutely rejected’ (p. 5). We suggest that the traditional determinist and reductive assumptions identified by Morin are also fundamental to the pre-specification paradigm. That is, the idea that the exercise of expertise can be broken down through analysis to simpler components that can be taken as determinable in the present and reconstituted for future effect by learners drilled in those components. However, this approach to learning and development comes up against a limit in emergency and acute situations, and it is complexity theory which may help us understand the difficulties faced by contemporary educators and trainers as they attempt to prepare workers for such conditions.

The phenomenon of complexity, then, poses a challenge to traditional ways of understanding the world (and learning). The generative part of complexity theory includes a set of concepts that serve to project thinking beyond the limits described by Morin (2007). Mennin defines complexity science as:

a collection of concepts and principles for the study of open systems that have nonlinear dynamical, self-organizing and emergent properties. Complexity science is the study of the dynamics of patterns and relationships rather than objects and substance in systems that are open and far from equilibrium. Complexity science focuses on processes and interactions of local agents that result in the emergence of new patterns as a whole. (Mennin 2010 : 839)

Some twenty such concepts and principles are differentiated by Mennin. An overarching feature of this thinking is its focus on systems of different kinds, reflecting a refusal to reduce complexity to constituent elements (entities, tasks, cognitions) for the purpose of explanation. The systems of particular interest to complexity theory are composed of ‘agents’. Agents have some degree of freedom, and their behaviours are not entirely predictable. In this context, agents are more than what a traditional perspective would identify as human individuals.

Systems can also be ‘non-linear’, a situation in which ‘small changes can result in large effects and large changes may result in no or in small effects’ (Mennin 2010: 840). Non-linearity poses a particular challenge to planning, and points to the intrusion of the unknown into our appreciation of acute situations. Complex systems also exhibit ‘self-organization’—but note that ‘[t]here is no “self” in self-organization and there is not a central hierarchical command and control centre’ (p. 840). That is, the system is the site of self-organization, not necessarily the agents that compose the system. Crucially, these systems are characterized by ‘emergence’, which refers to:

the arising of new, unexpected structures, patterns, properties or processes in a self-organizing system. Emergent phenomena exist on a higher level than the lower-level components that give rise to the emergents. For example, fertilization gives rise to a new emergent structure. The music that emerges from the coordinated interaction and exchanges among musicians in a jazz group and the learning that occurs as a result of the re-organization of structure leading to new patterns and possibilities. (Mennin 2010: 839)

There is consensus among complexity theorists that emergence is the unique phenomenon arising from complex systems and, as such, it moves to centre stage in the analysis developed here. This phenomenon is the locus of the unknown in the types of work with which we are concerned.

The pre-specification paradigm, already under question as a representative of the mentality described by Morin (2007), comes up against a fundamental challenge in the phenomenon of emergence. By its definition, emergence cannot be meaningfully represented in advance. But if emergence is productive of situations that become tasks for practitioners to negotiate and resolve, then we must look beyond the pre-specification paradigm for clues about how to prepare people for work characterized by emergence. We note that the strategies of classical science for dealing with complexity, such as ramifying determinism and analytic reduction, describe the quandary in which pre-specification approaches like CBET are caught when faced with the demands of complex work. Such a response is to add to more and more specifications to supposedly cover all varieties of possible situation, only to find that they have greatly exceeded the experience of even the most respected practitioners (Kearns et al. 2016). But just as pre-specification approaches a limit in emergence, what chance is there for any kind of preparation for complex work? How can people be prepared if their most important and demanding work situations cannot be specified in advance?

Combining the two analyses: a hermeneutics of complexity

Under the pre-specification paradigm, education and training for work (both routine and more complex) hinges on authoritative representation that precedes learning and the application of knowledge and skills. So far, we have canvassed the major hermeneutic assumptions of this kind of representation and elaborated an ontology of complexity in order to reveal a critical problem for traditional hermeneutics of work. The question that arises, then, is what alternative hermeneutics is possible for guiding learning for acute and critical situations? To approach this question, we note first that complexity theorists have themselves become embroiled in the problem of representation—a question that appears inevitable when human individuals and groups engage with complexity. While exhaustive representation of complex situations is no longer considered possible, it is recognized that in particular situations representation has practical efficacy. What theorists seem to have resolved at a general level is that the structure of complex systems can be understood and represented apart from actual complex situations (Cilliers 2001). The appeal to structure responds to the premise that relationships rather than entities are important for recognizing complexity, along with the critical position that the traditional reductive strategy has tended to remove relationships from representations.

However, as Hager and Beckett contend, while general structural features of complex systems can be known in advance, representation of particular systems is crucial—yet it can only ever be tentative. Representation (‘picturing’ or ‘modelling’) serves as a temporary means of coping by reducing the puzzlement produced by emergence. There is a place for representation of complexity, but if representation becomes fixed and generalized then it ceases to be relevant and effective. The quandary of CBET is succinctly captured here: its reliance on essentially static, widely disseminated written representations. On the other hand, the generative (if limited) role of reductive representation in the face of complexity must be acknowledged:

particular organisations, institutions, communities and the like engage in their own distinctive forms of complexity reduction shaped by their own particular aims and purposes. In these cases, the given entity develops its ethos and identity, as well as particular ways of doing things, all in the service of reducing complexity so as to facilitate timely achievement of the entity’s aims and purposes. Of course, in a changing world, the aims and purposes are very likely to evolve, so the nature of the complexity reduction should not be static. (Hager and Beckett 2019: 160)

The legitimate role of representation, then, is relative to the aims and purposes of groups in context, and must be dynamic with respect to changing conditions. In our context, dealing with complex work involves rapidly shifting situations with continually evolving aims, demanding a fluid approach to representation through temporary reductions. While Hager and Beckett have in mind the work of entities such as orchestras, formed over a period of time, their argument applies as well to the faster-paced situations of acute care and emergency response which are our concern.

The effective modes of representation indicated by Cilliers (2001) and Hager and Beckett (2019) respectively—that is, the possibility of representing general structures or relationships in advance and the tentative nature of representations of actual complex situations—are reflected in research applying complexity theory to problems of education and training for complex work environments. Arrow and Henry, for instance, consider the functioning of medical teams responding to acute situations. They find that effective teams are characterized by ‘internal complexity’, or different forms of expertise across the group, and exhibit integration of diverse knowledge, skills, and attitudes for addressing complex problems. The specific challenge and value of integration comes to the fore when there is a ‘spike’ in the problem situation:

the complexity of inputs spikes because of variation outside of the range the group normally handles, the group may either fail to attend appropriately to the change, or lose coordination and shift into a chaotic dynamic state, characterized by highly divergent energy flows and minimal integration. (Arrow and Henry 2010: 864)

The potential for ‘spikes’—that is, emergencies—characterizes the kind of work we focus on here. In such cases, the group ‘must communicate actively and continuously to provide a dynamically updated integrated picture’ (p. 864) of the unfolding situation. A shared representation must be maintained through the dynamics of the event, in effect a ‘model’ (Hager and Beckett 2019) of the complex situation and the group’s participation in it that is collectively constructed. In acute healthcare emergencies, for example, a team member will periodically describe to part or all of the team the overall situation, incorporating current diagnosis, care delivered, patient response, and likely next steps (Schmutz et al. 2018).

With respect to the non-specifiable aspect of complex work, representation (picture-making or modelling) is central to the resolution of acute situations; but this process becomes intrinsic to the work itself, not something that can be envisaged in advance. ‘Learning’ comes to the fore here as a process which, while it may be ‘formal’ (in the sense of the pre-specification paradigm), may be highly situated as well. As Mennin (2010), Doll (2012), and Fenwick and Dahlgren (2015) all emphasize, the learning potential of complex situations is significant. Such learning may be (temporarily) crystallized in the ‘pictures’ described by Arrow and Henry (2010) as a product of coping with complex situations. These pictures capture the learning, creating a platform for the next wave of activity and for the coordination of diverse expertise and efforts by a team. Learning is thus distributed throughout the group by means of temporary, localized representations. In their study of ‘complexity thinking’ in professional settings, Hager and Beckett (2019) point to modelling that takes place both at the individual and group levels. The creation of representations of emergent phenomena is thus integral to the competent functioning of such groups.

In a sense, within the complex work situation itself—away from environments designed and curated by educators—elements of the traditional hermeneutics of work are retained. Technical skills and knowledge represented in advance are essential, as they constitute the diverse forms of established expertise wielded across the team (Arrow and Henry 2010). However, they need to be set within a dynamic, rather than static, framework. The traditional division of educational labour by which analysts specify representations in advance, which are in turn translated by educators and trainers for learners, can be replaced with a dynamic schema of phases of representation-making, testing, and remaking performed by those dealing with the complex situation in the moment. As that situation is engaged by teams or individuals, it is empirically analysed in the present, and a provisional picture emerges. Learning is presupposed by the construction of a picture; but then that picture is adopted (learnt) by the whole team and, through additional learning about the emerging situation, the picture is revised.

A practical implication of this reconceptualization of the traditional division of educational labour is the potential for the dynamic notion of representation to form part of the repertoire of concepts that can be shared with and applied by those learning and practising complex work. Consistent with the analyses of Arrow and Henry (2010) and Hager and Beckett (2019), ‘pictures’ or ‘models’ of complex situations are produced by individuals and groups involved in them as a normal part of coping and responding to the pressures associated with non-linearity and emergence. Learning and communication about the evolving situation spring from the temporary efficacy of these interpretations, although as the non-linear processes unfold, revisions and new representations are demanded. Knowledge of this hermeneutic process of picturing and repicturing could be built into education, training, and assessment. Following Arrow and Henry’s suggestions relating to group processes, and Fenwick and Dahlgren’s (2015) suggestion that we explicitly teach complexity concepts in order to promote incisive analysis of emergent situations, can go towards constructing the conditions that enable the dynamic creation and revision of representations.

Conclusion

The hermeneutics of complexity offers tools for the critique of dominant approaches to contemporary education and training, and generative indications for those tasked with preparing others for complex work. In this article, we identified core assumptions of the ‘pre-specification’ paradigm. These assumptions include the convictions that we have sufficient access to the nature of competent work to be able to represent it for the purposes of learning, that our representation practices are effective, that educators and trainers can consistently decode and program learning on their basis, and that what we represent at one point in time is sufficient for use in future situations. The hermeneutics of complexity shows these epistemological assumptions to be limited, and they appear to fail when an ontology of complexity is introduced. However, in view of such criticisms, we sketched new possible directions for education and training practice. The picture- or model-making activity of individuals and teams responding to situations in which the known and unknown compete can potentially be enhanced; and learners can be allowed to own crucial aspects of educational representation as a means for sharing learning in the midst of situations marked by non-linearity and emergence. This localized, temporary representation-making invites analysis in terms of Gallagher’s (1992) ‘local hermeneutics’, which argues that the generic structures of hermeneutics need to be differentiated for particular educational settings and challenges. The theory of local hermeneutics could sensitize researchers to the possibility that distinct modalities of hermeneutic circularity occur, with different reference points, depending on the domain of work. For instance, qualitative differences between acute maternity care situations and aviation emergencies (for example, well-defined versus contingent team-member roles) may activate different kinds of representation-making, and analysis of such difference could in turn help nuance preparation of workers in these and other fields.

We envisaged a new division of educational labour—one that limits learning designers to generic, structural features of complex systems and the specificities of technical skills while promoting in learners knowledge of complexity, as well as the confidence and heuristics to make and use the representations which may be pivotal to resolving acute, high-stakes situations. Here, we indicated a different educational temporality. Instead of authoritative representations fixed in advance of situations as a province of task and function analysts, representational work is distributed, with the educational reduction of acute problems allocated to those coping in the moment. The hermeneutics of complexity suggests that Robert Miller’s (1962) ‘properly pious hope’ for a scientific and a-temporal reduction of practice must be reframed so that the human element he acknowledges becomes central to learning to work in complex environments. And it is not only emergency and acute situations that could be served by this hermeneutics. As Biesta (2010) explains, complexity reduction is a pervasive strategy in education policy and practice that affects entire education systems, including the preparation of teachers. The pre-specification paradigm is endemic, and ways to move beyond it—such as the hermeneutics presented above—deserve the attention of education philosophers and researchers.