Exploring ideas of knowledge co-production

Introduction

As outlined in Chapter 1, the organisation and delivery of health and social care services require multiple forms of knowledge and expertise drawn from a wide range of sources.33,72 Service managers and health service researchers increasingly recognise the limitations of producing research separately from the sites where this knowledge will be applied.66,73,74 What was previously interpreted as a knowledge ‘pipeline’ problem,75 a knowledge ‘gap’76 or a knowledge ‘transfer’ issue77 is now increasingly reconceptualised as a need for collaborative knowledge creation in context,65,78,79 or ‘knowledge co-production’, in short.

Knowledge co-production comes laden with a variety of positive expectations.80,81 In the context of embedded research, there can be an easy assumption that co-production is integral or even ‘naturally’ occurring. Against this, we begin to see concerns voiced about a potential ‘dark side’.82 Some suggest that knowledge co-production, undertaken without caution, can smuggle in hidden interests and disguised power relations, potentially outweighing any proposed benefits with unacknowledged risks.38 Critiques also focus on the overly instrumental and normative nature of much co-production literature (and practice), both within and outside health care.83

Ideas of knowledge co-production are germane to any understanding of embedded research, but there remains a lack of clear conceptual underpinning or any precise formulation as to how these ideas may play out in practice. One persistent challenge is the differing assumptions, expectations and frames of reference that collaborators bring with them, and the ways in which these influence methods of working together.84 There is a need to unpack what is meant by knowledge co-production, to develop greater conceptual clarity and an appropriate language for discussing its components and varied manifestations. This narrative literature review, the first of two in workstream 1, set out to address this issue.

Methods

We conducted a narrative literature review, to ‘synthesise representative literature on a topic in an integrated way such that new frameworks and perspectives on the topic are generated’,85 carrying out a framework analysis of the academic literature on knowledge co-production. The scope of the review was literature exploring a direct collaboration of academic researchers and non-academic stakeholders aimed at generating new knowledge.

Assembling the published literature

To gather a wide array of perspectives, we collected literature that featured theoretical or conceptual discussion of knowledge co-production, including reports of co-productive research activities incorporating reflective and theoretical insights into knowledge co-production, and editorial/review articles in which the main focus was knowledge co-production.

We conducted systematic searches of the Web of Science™ Core Collection™ (Clarivate Analytics, Philadelphia, PA, USA) in March 2018. The search terms and strategy are set out in Table 1. Because we wanted to bring together insights from a wide range of health, science and social science fields, we did not limit this search by discipline or by year, although the earliest article that satisfied our inclusion criteria was from 2003.

Screening the search results, we removed sources in which knowledge co-production between academic researchers and non-academics was not the principal or main topic. We therefore excluded writings on value and service delivery co-production, public service co-production and other kinds of collaborative co-production not involving researchers working with non-academics, or an explicit concern to make knowledge.

Other relevant articles identified through the reading and networks of the project team were subject to the same process of formal inclusion/exclusion. The numbers of articles screened, removed and included are shown in Figure 1.

FIGURE 1 Flow chart of literature-gathering and screening. (Figure not reproduced here; see the original article.)

The articles included for analysis are listed in full in Report Supplementary Material 1. As well as articles from health and directly health-related disciplines (n = 24), we reviewed articles from management studies (n = 15), environmental science (n = 15), sociology and social policy (n = 9), sustainability studies (n = 8), ecology (n = 5), human geography (n = 3), communication studies (n = 2), and one each from humanitarian and disaster studies, engineering, psychology and genomics. Source articles dated from 2003 to 2018.

Analysing the data

The framework analysis technique is well established for analysing large bodies of literature subsequent to searching, screening and preliminary reading.86–88 This approach was appropriate because of its flexibility (allowing us to interrogate sources from diverse disciplines); its capacity for ongoing modification and revision in response to data reading, re-reading and re-interpretation; and its orientation towards descriptive understanding rather than prediction or normative goals.89 Framework analysis facilitates an iterative, team-based approach to analysis, giving increased confidence that the categories and themes discerned have stability and interpretive utility.

An initial interpretive reading conducted by one member of the team (BR) aimed to discern and label groups of concepts occurring within and across the source articles. For instance, writers might report concerns around institutional politics, or the difficult emotions of perturbed professional identities, without necessarily using the words politics, identity or emotions.

The review team (BR, VW, HD and TT), which had access to the full set of sources as well as the evolving analytic categories, then discussed this working set of conceptual labels, or domains, for their face validity, coherence, completeness and overlaps, and made modifications as agreed. A closer re-reading of source articles was then conducted to see if the conceptual categories emerging provided a sufficient framework for interpreting the data. This second reading allowed us to identify additional content that would not fit easily into the domains, and to refine or develop further domains and, more significantly, subthemes within domains.

In discussions across the whole project team, it became evident that differing perspectives on knowledge co-production created tensions within each domain. We recorded and labelled these tensions and grouped them together into connected accounts of meaning.

What emerged was a set of five conceptual domains overlaid with a continuum of perspectives from the conventional to the radical. A conventional view of knowledge-making would see the boundaries between different knowledges and different roles largely maintained, with the goal of worthwhile but incremental change. A more radical approach would see boundaries more thoroughly dissolved, in pursuit of a more substantial overhaul of priorities and possibilities.

Given our goal of creating a language of shared meaning that would make knowledge co-production easier to discuss, we sought categories that were defined and coherent enough to allow for shared understanding, broad enough to encompass the messiness of divergent practices, and flexible enough to allow ongoing debate and refinement. The structured reading we lay out is not the only one available, but it has survived repeated interrogation across the wider project team of knowledgeable, diversely experienced collaborators, and through presentations at various professional events.

Findings in outline

The key findings were that there are five key conceptual domains in knowledge co-production; that, in each domain, there is a continuous spectrum of differing perspectives; and that each domain comprises two subthemes, also characterised by a spectrum of perspectives.

Five domains of meaning

Meanings and perspectives in knowledge co-production can be located, understood and compared within five related domains of meaning: politics, knowledge, identity, space–time and aesthetics:

1. The politics domain brings to the surface those negotiated and meaning-laden processes (among large and small groups, and individuals) that in conventional knowledge production are obscured or unspoken.
2. The knowledge domain concerns implications around the tangible product that co-producing parties reach for – what is this knowledge, and what is it for?
3. The identity domain allows a close examination of people who co-produce knowledge – what makes them who they are, individually and collectively?
4. The space–time domain concerns the physical processes and happenings of knowledge co-production. More than just a practical question of ‘where and when’, it involves different configurations and enactments of collaboration.
5. The aesthetic domain makes visible aspects of knowledge co-production experience and expression that may seem intangible and subjective, but that are highly significant – human likes and dislikes, and emotions that escape easy expression but are fundamental to collaboration.

These five domains mark out a conceptual space in which to consider the complex interplay of diverse ideas and understandings involved in knowledge co-production.

Spectrums within domains

In each domain we found that, underlying any superficially isolated points of opposition and difference, there were subtle variations of thinking that merge and separate, creating a spectrum of interpretations and actions. These range from non-collaborative or reservedly collaborative approaches that leave existing boundaries relatively unchallenged (we term these ‘conventional’ research or ‘provisional’ co-production), to approaches that challenge or unravel such boundaries in pursuit of more disruptive and substantial change (what we term ‘committed’ or ‘radical’ knowledge co-production). This spectrum is set out in Figure 2.

FIGURE 2 Styles of knowledge co-production. (Figure not reproduced here; see the original article.)

The extremes of this spectrum (conventional and radical) serve as idealised anchor points for clear thinking, but also reflect knowledge-making processes actually (in some cases) performed. Rather than a barrier or problematic void between them, there is a rich space of variation among subtly different approaches. Real-life knowledge co-production mostly happens in this intermediate space. To guard against binary thinking, we included a middle category of moderate co-production, which models a roughly equal balance and interplay between ideals.

These labels proved useful to unpack, interrogate and lay out subtle contrasts within domains, but should not be thought of as overly rigid or proscriptive. Labelled marker points do not indicate static types of knowledge co-production, but dynamic approximations, signposts that we have settled on after repeated interrogations of the literature and discussions among the team.

Subthemes within domains

In further exploring and categorising the themes in the literature, we identified two subthemes within each domain:

1. politics includes ideological and transactional politics
2. knowledge includes the substantive product and output of collaboration, and is also something that exists and has nature in itself
3. identities includes both the roles fulfilled by individuals and the way in which those individuals are less or more socially intertwined
4. space–time incorporates time and space as distinct physical dimensions that can be configured in a variety of ways, according to how they are realised and occupied in collaborations
5. aesthetics includes the preferences and affinities of people and groups involved in knowledge co-production, and the valuation and influence of emotions in collaborative proceedings.

In the subsequent section, we summarise each domain and subtheme, giving examples from the literature. These discussions are encapsulated graphically in Figure 3, and also in Table 2).

Findings in detail

Domain 1: politics

The politics domain incorporates ideologies of research, ranging from the apolitical (conventional co-production: politics seen as separate from knowledge-making) to the emancipatory (radical co-production: knowledge-making as a tool for social justice and structural change). Intermediate points are the utilitarian (provisional co-production: knowledge for instrumental application), the engaged (moderate co-production: research and practice/policy substantially shape each other) and the transformative (committed co-production: sustained micro- and meso-level change is part of the programme).

This domain also incorporates the instrumental politics of transactions, the often mundane institutional politics of co-producing knowledge. Attitudes across that subthematic spectrum range from politics as pollution (conventional), through politics as necessary (provisional) or valuable (moderate), to politics as essence (committed), and politics as all-encompassing (radical).

Politics as ideology

Diverse authors have observed the contrast between politically provocative styles and more functional styles of knowledge co-production. Orr and Bennett31 (social policy context) distinguish between co-production ‘based on a critique of the status quo’ and the de-politicised ‘anodyne’ use of co-production terminology in service of established interests, ‘not necessarily advancing the broader interest or public welfare’. Similarly, Lövbrand (climate science context)90 identifies:

. . . tension between the critical/reflexive ambition built into the co-production idiom, and the more utilitarian interpretation of the term. Whereas the former sets out to expose and interrogate the ontological assumptions underpinning public policy, the latter seeks to be useful by responding to the knowledge needs of societal decision-makers.

Analytically, one could accentuate the mutual distinctiveness of these positions.91 In the narratives we reviewed, however, we found instability and equivocation between utilitarian and critical co-production styles. ‘Useful’ forms of knowledge co-production do not necessarily preclude more thoroughgoing change. Co-production with a stated transformative intent, meanwhile, often also carries slower-moving and locally specific change processes. Both understandings frequently pertain within projects, and between collaborators in how they see the value of a project developing; hence the value of exploring intermediate points along a spectrum. Dynamic exchange between intermediate points softens the edges of opposition between extremes.

Transactional politics

Alongside ideological politics, we have the prosaic politics of social interactions between institutions, organisations, policy bodies, academic disciplines, professions and service user groups. Negotiating the intricate mechanics of such transactions is necessary to secure and maintain a productive connection between stakeholders.92,93 Hewison et al.37 report an initiative that aspired to full engagement with non-academic collaborators,94 but whereby mutually incompatible policies and pressures on employees in the collaborating institutions made this impossible in practice. Subsequently, McCabe et al.23 developed a typology among ‘contextual factors of partnerships’ in which collaborative structures determine what kinds of knowledge gains can practically be made.

Although the transactional politics of collaboration are usually difficult and demanding, this is sometimes seen as productive and valuable in itself.95 Klenk and Meehan’s96 critique (environmental science context) of imperatives towards ‘integration’ of stakeholder contributions, which they argue neutralises the critical value of non-academic perspectives, advocates that disparate, incompatible voices should be protected. Such contributions imply a committed disposition in which transactional politics hold the value of co-production, contrasted with a more provisional disposition in which they are a barrier to be negotiated to arrive at expected knowledge products, amalgamation of perspectives, and effective knowledge use.

To be clear, the value of a politics lens on knowledge co-production is not in theoretically settling the political-or-otherwise status of research knowledge.97 What differentiates styles of knowledge co-production is where the different stakeholders draw the boundaries between knowledge production and politics. Do they encourage a restricted, instrumental political engagement towards a foreseeable output? Do they confidently pursue a political re-ordering of knowledge processes and hierarchies? Or is there some hybrid of these? Whichever they choose, the collaborative political process is made visible, conscious and deliberate, instead of being kept below the threshold of awareness.21,98 Politics becomes intentionally woven together, in contestable ways, with knowledge.

Domain 2: knowledge

The knowledge domain incorporates perspectives on the outputs of knowledge co-production, and implied positions on the nature of knowledge. Beyond being a pure commodity made for truth’s sake (conventional co-production), knowledge can have utility, transfer and ‘impact’ potential (provisional co-production). Or knowledge can be something through which people become inclusively engaged in social life (moderate co-production), or through which they gain new productive capacities (committed co-production). Knowledge may even enable actors to become empowered (radical co-production). Accordingly, knowledge has different kinds of existence: as an object in itself; as something exchanged or negotiated among those who hold it; as something held communally by ever-changing collections of people; or finally, as a constitutive property of a wider collective group, in a deeper, ongoing and embodied sense.

Output/product

A common framing in knowledge co-production literature is the agenda for research impact,92,95,99,100 reflecting a utilitarian impulse to show use-value and policy relevance.101 Antonacopoulou102 upheld this view of co-produced knowledge, inviting researchers to pursue delivery that had an impact, explicitly turning away from political preoccupations (a provisional approach). In contrast, Pain et al.103 (human geography context) critically explored the tensions between impact as an issue of economic accountability and as a matter of social responsibility – both inescapably political concerns. They advocated co-production as a way to ensure socially equitable and ‘radically transformative’ knowledge outputs.103

Capacity-building (committed) and empowerment (radical style) are also common framings in the literature,40,104–106 whereas others speak of engagement, engaged scholarship and inclusion (a moderate style).24,107,108 Each of these indicates a subtly different narrative of who and what knowledge is for.

Nature of knowledge

Associated with this array of output purposes are implicit understandings on the nature of knowledge itself, frequently presented through metaphor.

For Rycroft-Malone et al.,33 the first move of health-care knowledge co-production is to challenge ‘pipeline’ conceptualisations of knowledge, knowledge transfer, implementation, the gap, and the ‘two communities’ model of knowledge production and use.21 A pipeline conceptualisation imagines knowledge as a thing that must originate on the outside and be transported to the context where it is applied, or as Kitson and Harvey109 see it:

. . . the process by which knowledge moves from where it was first created and refined to where it has to get to in order to make an impact.

To sustain a ‘gap’ conceptualisation, these two places must be different, with knowledge travelling as an object package from one to the other. To discuss knowledge implementation techniques in terms of the success or failure of their transfer to the point of use109,110 is to retain a conventional sense of knowledge as a commodity that exists autonomously and is delivered post production to people and places without altering its nature.

Radical knowledge co-production, rejecting the object-and-transfer conceptualisation, imagines knowledge instead as a valued attribute of people located in time and place, a process performed and embodied, a presence that empowers and includes, a pervasive energy in movement and change, and a collective and communal good. In short, knowledge as a positive mode of human being.111 For Kothari and Wathen,112 this understanding:

. . . is not just about knowing but encompasses ways of being and relating . . . making space for an additional ‘value-added’ communal perspective.

Between these idealisations is a middle ground of ambivalent metaphors. The idea of knowledge exchange, for example, imagines collaborators each contributing a genre of object-like knowledge to a sharing process that simultaneously creates knowledge anew, collectively.18 Some emphasise tension in this dialogue, a ‘balancing act between imposing control’ (provisional) and ‘opening up a plurality of voices’ (committed).113

When knowledge is said to be negotiated in co-production,114 a knowledge object again precedes the interaction, which results in a more diffusely social knowledge (moderate style). However, when there is any seeking of shared wisdom and ‘integrated or transformational understanding’ (committed style – Schuttenberg and Guth115), there is also the possibility of troublesome conflict and re-assertion of power status (provisional),98,114 felt as a ‘need to exercise a stronger voice’.115

These tensions are political, not merely between related actors, but between perceiving knowledge as an object property held individually, unequal but shared and received, or an elusive and transient presence realised through communion. Terms such as impact, negotiation, exchange and transformation carry variable meanings in the literature, suggesting a range of tangled understandings that might be disaggregated, given the right tools for dialogue.

Domain 3: identity

Politics experienced at the individual level can be conceptualised through identity, both in terms of the roles that participants occupy, and by acknowledging individual feelings, affinities and sensibilities.

We found that roles in co-production might be separate and securely fixed (conventional style), or they might fit together as complementary and negotiable where they meet (provisional style, in which case conflict is more likely found to be counterproductive). Sometimes there is an expectation that roles naturally overlap (moderate style), interfere and contradict (committed style), or disintegrate and multiply (radical style), in which cases conflict may be naturalised and valued as significant.

The individuals involved might be thought by nature to be autonomous, or connected to each other in singular, transactional ways; they might be more interwoven and interdependent; or, in the extreme cases, less clearly separable from each other in terms of the multiple knowledge roles they embody.

In this domain, it is useful to hold in mind some typical characters in narratives of knowledge co-production. The basic players are academic researchers (whose accounts are usually of central interest for academic writers and audiences, even under committed and radical modes of co-production) and non-academic stakeholders, such as (health) professionals, policy-makers and service users (whose accounts rarely surface in enduring forms). Recurring figures include non-academics who gain skills and empowerment from involvement in research,116 researchers who skilfully ‘surf the boundary between ways of knowing’96 and ‘critical friends’.117,118 ‘Critical friends’ implies researchers who are aligned with non-academic institutions, but remain able to be critical, or non-academics who develop an expert voice and agency in knowledge production by virtue of their insider status.119

Role

This spectrum spans the ways in which people are understood to occupy their socially designated roles.

If those roles are thought of as naturally ordered, given and fixed to individuals, it indicates a conventional or provisional style in which one simply ‘is’ an academic or non-academic, and remains unproblematically so. If roles become partly negotiable, this typifies moderate co-production. In committed or radical styles, roles become performances, contingently (re)made by people and subject to ongoing change; possibilities arise of people disconnecting from rigidly defined roles, crossing role boundaries and enacting multiple performances, actively and creatively. In recognising an underlying common humanness at this end of the spectrum, researchers can, as Greenhalgh et al.95 note:

. . . view research as a creative endeavour, with strong links to design and the human imagination . . . [placing] individual experience at the heart of this creative design effort.

Klenk and Wyatt120 (environmental-economics context) similarly advocate ‘engagement with partners that is creative and transformative rather than mainly informative’, suggesting experimentation with role boundaries, and scepticism of pre-conceived roles and duties.

Fenge et al.121 (health and social care context) describe knowledge co-production ‘involving lay people, volunteers, service users and carers in scholarly writing and dissemination’, exploring how academia can embrace the expertise of these non-academics. Usually this suggests a single identity, such as a patient, professional or health-care policy-maker, that serves as a warrant for involvement, but becomes plural as collaborators contribute, not just information, but part of their personhood to knowledge-making.122

Almost always, original role markers remain meaningful. The truly radical possibility is to abandon them entirely and see all collaborators as people originally equal, communally connected through their shared interest.

Individuals

Collaborative work can place demands on individuals. We see this reflected in O’Hare et al.’s118 interest in ‘coping strategies’ for critical tensions, Orr and Bennett’s123 reflections on their ‘at times disturbing and debilitating’ experiences of co-production, and Vindrola-Padros et al.’s65 account of researcher-in-residence strategies for facing challenges to their professional identity. This implies an individual consciousness that is separable from the role a participant occupies and is able to reflect on competing demands.

For researchers, knowledge co-production brings perceived risks, such as dominance by particular stakeholders, eroded independence and credibility, interpersonal conflict, burnout and stress, and damaged careers.82 These threats relate to the individual status of the researcher, which, under the individualistic model of personal worth dominant in academic research, is vulnerable.

For non-academics, collaboration is largely expected to be an edifying and empowering process. By challenging the individualistic academic culture and directly confronting knowledge inequalities and cultural boundaries, it could be seen to ‘democratise knowledge’.124 But such a framing, although notionally moving non-academic individuals into positions of power, may also place considerable responsibilities and expectations on them.

Underlying such anxieties are equivocations over the nature of people as reflexive individuals. If people are thought to be comfortable as autonomous beings (conventional style) then provisional or moderate ways can be found to account for their connections, but more committed or radical approaches are problematic. If the natural state of people is thought to be as communal, interconnected and multiply oriented beings, then the challenges of more radical co-production may be considered worthwhile and intrinsically valuable for the way they decentre individualistic orthodoxies.

This ontology of individuals meshes with an ontology of roles as fixed and stable (if restrictive) at one extreme, or mutable and multiple (allowing for creativity, if sometimes disorienting) at the other. The space between has possibilities for dialogue and the ongoing reconstruction of roles, and of the reshaping of the identities of the individuals occupying those roles.

Domain 4: space–time

Time and space are highly visible points of interest in knowledge co-production literature. Conventionally, time is imagined in a linear way, with knowledge first produced, then moved into policy or practice settings; provisional arrangements may be made around a variable but linear process of production and dissemination. Moving along this spectrum, time can enfold different collaborative stages that repeat and overlap or become more consciously iterative. In the radical ideal, ongoing production and ongoing application are in a mutually constitutive continuous cycle of renewal.

Space may be strictly divided between separate research and practice locations, or attention may be given to in-between, liminal meeting points where institutions begin to overlap. A blurring of boundaries and conscious challenging of spatial designations characterise more committed and radical forms of co-production.

Time

Time is crucially important in knowledge co-production. With non-academic contributors, time is usually prioritised by their primary identities (as managers, practitioners, etc.); negotiation is often necessary to divert that time towards knowledge collaboration. Co-ordinating the schedules of diverse stakeholders for meetings is a basic difficulty.37 Supposing collaborators are brought together in a timely way, writers often emphasise the long time required to nurture collaboration: time invested in building trusting relationships and accumulating confidence among collaborators, and time taken to agree and implement knowledge-making plans, with the implication that the added value of outputs should correspond to added time inputs. Lehmann and Gilson125 speak on behalf of many co-producing researchers:

We spent an awful lot of time in meetings, formal and informal, big and small, in cycles of conversations to plan and implement strands of work, reflect on their outcomes and replan.

These cycles of conversations allude to a characteristic aspect of committed and radical styles: the iterative nature of knowledge co-production, in which different stages of the research process overlap more than in orthodox research – ‘planning, execution, dissemination and implementation are not separate and linear phases but interwoven’.95

This contrasts also with provisional co-production in which consultations at specific points punctuate the path of knowledge into practice. A moderate style of co-production signifies ambiguity and overlap between pre-made knowledge being moved into application, and an ongoing process of making and modifying knowledge at the same time as ‘using’ it.107

Space

In the conventional ideal, a ‘gap’8,78 envisages a one-way flow of knowledge in space from sites of production to adoption and routine use. In the radical ideal, knowledge attaches to communities in places, so is made just as it is inhabited and enacted, in a closed, reflexively aware loop. In between these extremes are a range of possibilities for co-production embedded in space and place.

Ideas about the in-between spaces that separate and connect institutions are ever present in knowledge co-production. These liminal interstices are valuable, for example, in creating ‘a shared space where multiple communities can come together’,23 and where collaborators are ‘able to challenge ideas and existing policies or entrenched beliefs’126 in ways otherwise not possible.

Rather than refer directly to specific physical sites important for knowledge co-production, writers sometimes evoked space through social-psychological metaphors. They wrote of space beyond hierarchies and boundaries; indeterminate social space; permeable space;114 exploratory space;79 nurturing, safe and inclusive space;127 space ‘both embedded in and insulated from research and practice’ and spaces ‘open, autonomous, unpredictable, dynamic, reflexive, and shared’;128 and spaces of ‘dialogic co-inquiry’ and the ‘cramped’ space of co-production.40

Across the intermediate spectrum of spatial designations, these metaphors range from those that look for liminal gaps between original single-purpose spaces (provisional), those that bring different purposes to coincide and overlap in one space (moderate), and those that challenge and blur the boundaries between spaces (committed).

All of these pragmatically repurpose and reinterpret the physicality of extant institutions, whereas, in radical knowledge co-production, there is an acknowledged political awareness in the redefining of spaces. Bremer and Funtowicz129 draw on the political geography of Doreen Massey to highlight this awareness of distinct narratives of space that ‘coexist, meet up, affect each other, come into conflict or cooperation’. For many collaborative purposes, however, a language of spanning and blurring boundaries, temporarily displacing orthodox routines and interactions, and flexibly redrawing the borders between institutions – a metaphorical language of space – is found sufficient.

Domain 5: aesthetics

Knowledge co-production involves an aesthetics of attitudes, dispositions and emotional attunement that resonates across domains. Preferences for simplicity, rigidity and predictability in more conventional forms merge into a tolerance and interest in social complexity in intermediate forms of knowledge co-production. Beyond this, there is a positive solicitation of emergent outcomes, and an attraction to collective knowledge states that cannot be foreseen (implied in more radical approaches).

Alongside this, we find different valuations and language around emotions. Emotions can be marginalised, or a manageable complication to knowledge mobilisation and application; can be a complementary and useful form of rationality; or can be of foundational value in themselves as part of knowledge-making processes. At the radical extreme, the emotion/rationality distinction disappears, as communal production becomes shared, embodied and affect infused.

Preferences

Preferences, affinities and aversions in co-production writing are easily detected in the literature we reviewed, although not always so openly acknowledged. Our interpretation brings these recurring themes into open conversation. For instance, appreciative considerations of multifaceted, evolving connections between co-producing individuals and institutions suggest an ethic of connectedness and mutual responsibility between academia and other social sectors.130 This leads away from a simple schema of entities with isolated purposes towards complexity of social patterns and trajectories.131 A complex system is more likely to appear untidy, or messy, and lead to outcomes that are unforeseeable or not predictable, and is often thought to be more valuable for that.

Borg et al.,132 writing on ‘valuing uncertainty’, tell of attempts to explore multifaceted realities through complex processes that ‘can be intense, unpredictable and at times rather chaotic’. Interest in complexity,21,133,134 in messiness105,135 and tolerance of messiness,33 in plurality and mutability136 and unpredictability95,128 is woven through narratives of knowledge co-production. These tropes create a graded succession, from knowledge transfer that is accountable and predictable (provisional), knowledge exchange that is attentive to complexity (moderate), and co-productions seeking creativity and social innovation (committed). Radical intentions are signalled by a challenge and subversion of orthodox constructions and a scepticism of norms conventionally made to appear simple, in which messy boundary work is moved backstage, out of view.137,138

Emotions

Human actors embedded in messy realities may not be well placed to grasp and govern their situations in the manner of singular purposive and rational minds holding unitary object knowledge. They must find ways to create collective wisdom, drawing on diverse sources and expressions of connectedness to the world,139 indicating the importance of the ‘more-than-rational’.139 Orientation to emotional knowledge and things known instinctively through experience,140 as well as attention to shared feelings and affect,141 are often active in knowledge co-production.

One example of emotionally rich language is the motif of ‘nurture’ common in this literature, whether used appreciatively102,112 or sceptically.38 ‘Trust’ is also usually cited as a foundational condition of successful partnership.142–144 The nature and conditions of this trust may suggest a strategic management and positive use of emotion, for example when Jagosh et al.116 carefully engaged community members ‘who were already known to be knowledgeable, sincere, compassionate, and understanding’, so that others would ‘feel safe participating’. In other cases, emotional currents are detectable as forces motivating co-production. For instance, when Klenk et al.145 envisage knowledge practices that ‘move and affect not only present concerns but also future solutions’, hope is a foundational emotion for energising the project.

Attunement to emotional knowledge does not imply positive emotional states. More often than being joyous, the emotional relations of knowledge co-production are fraught and difficult. What distinguishes styles of co-production is the extent to which emotions are given legitimacy in juxtaposition with rational object knowledge. In moderate styles of co-production, emotions and rationality are imagined to complement one another in working towards consensus or productive difference. In committed knowledge co-production, emotions may be foregrounded as sources of knowledge, emphasising the human and creative processes of project design. In radical knowledge production, mirroring the collapse of a knowledge–politics binary, the dichotomy of reason and emotion is more unstable. Rational and affective streams of consciousness are woven together into a thinking–feeling fabric of being.

Representing the conceptual framework

To visualise the five-domain framework, we created a diagram (Figure 3) in which co-production styles appear as concentric circles, and domains and subthemes as pairs of ellipses arranged around a central locus. An arrangement with conventional research on the outside and radical co-production at the centre evokes the impression of distancing and separateness in the conventional style, and of closeness and intensity in radical co-production. The ellipses representing each subtheme convey a dynamic sense of motion, we hope, as projects and collaborators may fluctuate between possibilities.

FIGURE 3 Domains and subthemes of knowledge co-production. (Figure not reproduced here; see the original article.)

Table 2 characterises each marker point along the subthematic spectrums with a word or two, creating a grid of possibilities for co-production projects that can be read horizontally (to see progression across spectrums) or vertically (to see the properties of each style). Gradations in shading suggest the idea of blending between styles. A perspective that is in some ways committed, for example, could be, in other ways, radical or moderate, but is less likely provisional or conventional, if our domains are a sound analysis of thinking around co-production. However, all approaches are legitimate; dissonance between aspects of a project may indicate tension and dysfunctionality, but may also reflect different goals and intentions co-existing within a project, or changes in a project’s character over time. To emphasise, we propose these labels of recurring patterns to be descriptive, not normative.

Concluding remarks

Through our narrative literature review and detailed framework analysis, we have offered some conceptual categories, language and metaphors for articulating and disaggregating knowledge co-production aims and processes. These were derived from academic literature on knowledge co-production from a wide range of fields and traditions. We have provided a theoretically rich, yet robust and pragmatic, way of understanding and talking about the different aspects of knowledge co-production.

Conceptually, the framework explores five domains of meaning seen repeatedly in the literature, each with two distinct subthemes. Representing these on a spectrum, from more conventional knowledge production (shared subsequent to being created) to more committed or radical co-production, has allowed us to see the variegated nature of co-production possibilities. Should this way of mapping the field take hold, there is the possibility of greater depth and nuance to conceptual conversations, and more precision in theoretically informed empirical explorations of co-production practices.

More practically, the framework provides a coherent and consistent language for surfacing assumptions, sharing perspectives and making sense of difference. Therefore, we hope that knowledge collaborators, such as those embarking on embedded research initiatives, will use the framework to articulate their views on shared projects, both at inception and as they evolve. Primarily, we hope they will use it to identify their own and each other’s positions, to better communicate around and organise projects, and hence to be more successful in delivering those projects to meet various goals and outputs.

Of course, such a process may be as revealing of dissonance and divergence as alignment. For example, any specific project may have aspects that suggest different points on different spectrums at the same time. Sometimes such differences could promote productive tensions; at other times they may be sources of conflict and misunderstanding. Moreover, dynamic change, not static positioning, is to be expected: projects may shift between points on the various spectrums over time. The language and labels we have presented may thus be useful for analysing coherence or incongruence, and for articulating and examining dynamic shifts.

Through this review work, we have revealed the considerable complexity, nuance and contestability of simple ideas of knowledge co-production. But we have done so in a way that provides a practical framework for interrogating the idea and translating its aims and methods into practice. The second part of workstream 1 (and the second literature review) moves, in turn, to unpack ideas of embedded research practice (see Chapter 3). In this way (through these two reviews combined), we can get a better conceptual grasp of the diverse models of embedded knowledge co-production that lie at the heart of this project. The nature and findings from that second literature review are presented next.