By Stephen Crowley and Graham Hubbs

2. Graham Hubbs (biography)
How can we improve our understanding of knowledge integration? What are the elements of integration?
Sometimes what gets integrated are products of science, such as data sets or scientific models. Sometimes it is not the products that are integrated but instead the methods, as can happen on interdisciplinary teams. On these teams, scientists work together, so sometimes it is the people themselves (scientists are people!) or their disciplinary cultures that get integrated.
These are only some of the possible elements of integration. There is just as wide a variety of processes and products of integration as there are elements. The process of integrating data sets might be a sort of analysis, and the result might be a table or graph that displays the results of research in a conspicuous manner. Integrating diverse scientists into an interdisciplinary team, by contrast, is a matter of people working together, and the result of the integration is not a table or a graph but the team itself. A successful interdisciplinary research project might require integrating data sets, and scientists, and methods, and more—what it means to integrate each of these seems different in every case.
Can anything general be said, then, about all of these sorts of integration?
We at the Toolbox Dialogue Initiative (TDI) have developed a model of integration—the IPO model—that we find informative for both general and case-specific accounts of integration. Each part of the model focuses attention on a collection of questions that can help illuminate a given integrative process, and the model as a whole allows for broader, more global questions about a specific integration.

The model starts by separating
- an initial stage (INPUTS), in which there are things to be combined, from
- an intermediate stage (PROCESS) in which combining occurs, from
- a final stage (OUTPUTS) in which there are new whole(s) to be identified and described.
These divisions, in turn, allow us to separate out the following elements and questions:
- INPUTS – Thinking about inputs begins with appreciating their diversity. How many are there? What are their natures? What is the nature of their differences?
- PROCESS – If you are thinking about the process itself you can begin by asking about features of the integrative relation – just what is being done to the inputs in order to combine them? Further, is the combining intentional? Algorithmic? How much change (number and kind) is occurring?
- OUTPUTS – Thinking about outputs, like thinking about inputs, involves a focus on their diversity. What kinds of thing(s) are they? How many are there? How do they differ among themselves and (where possible) how do they differ from the inputs from which they ‘descended’?
In addition to questions that arise out of a focus on particular parts of the integrative process, the model helps us identify more global issues. More specifically, it encourages us to home in on the following three parameters:
- Commensurability – How easy (or hard) is it to combine the inputs? What are the barriers to combining and what is involved in doing so?
- Scale – Is the integration operating on a more local (eg., combining data from different experiments within a single lab) or a more global (eg., combining major theories within a discipline) level?
- Comprehensiveness – Roughly, does the integration yield new perspectives that are broader (address phenomena not covered by the inputs) or deeper (fuller understanding of some focal phenomena) but perhaps with reduced breadth?
The aim of this model is not to answer questions but rather to systematize them. By highlighting these basic features of integration, the model helps researchers and practitioners to be clearer on the aspects of integration they are investigating, to compare different integrative efforts, and to work on new aspects of integration.
While there is clearly more to be said—and we hope you will in the comments!—we’d like to wrap up with a puzzle about the model that we find particularly engaging.
What, exactly, counts as an integrative relation that might occur during the PROCESS stage? There are some paradigm examples, as when a bunch of bits of disparate information comes together to reveal the perpetrator in a murder mystery. What, though, about resolving an apparent contradiction? That feels like some sort of move toward combination, but is it enough for integration? Or, consider someone sorting pieces of multiple jigsaw puzzles. Recognizing that two pieces belong to the same puzzle is real progress, but it’s not the same as actually connecting the pieces and solving the puzzle. Or, what do we say about a situation where we recognize one alternative is better than another and discard the weaker alternative? Does this sort of “winner take all” scenario count as integration? If so, why, and if not, why not? We have our own hunches about how to address these questions, but we don’t have well-developed answers, and we’d love to hear what you reckon.
Finally, we and others are just getting started applying the model to case studies. If anyone has material they think fits or undermines the model, we’d love to hear about that too!
To find out more:
O’Rourke, M., Crowley, S. and Gonnerman, C. (2016). On the nature of cross-disciplinary integration: A philosophical framework. Studies in History and Philosophy of Science Part C: Studies in History and Philosophy of Biological and Biomedical Sciences, 56: 62–70. (Online) (DOI): http://doi.org/10.1016/j.shpsc.2015.10.003
Toolbox Dialogue Initiative (Online): http://tdi.msu.edu/ with information about the research program on integration at https://tdi.msu.edu/research-overview/tdi-integration-research/
Biography: Stephen Crowley PhD is chair of the Philosophy Department at Boise State University, Idaho, USA. He is also a member of the Toolbox Dialogue Initiative. He helps facilitate team science projects (as part of the Toolbox Dialogue Initiative) in a variety of areas as well as working on models of such collaborations and how to support them.
Biography: Graham Hubbs PhD is Associate Professor and Chair of the Department of Politics and Philosophy at the University of Idaho in Moscow, USA. He is also a senior member of the Toolbox Dialogue Initiative. He contributes to cross-disciplinary integration through his work with the Initiative.
Hi, this is a very interesting framework, very easy to understand. It helped me to structure my thinking about knowledge integration. However, how can one practically apply this with methods, for instance, in a transdisciplinary research project with researchers from different disciplines and in addtion non-academic actors? Do you have any suggestions on this? Thank you so much for sharing your experiences!
David – thank you for the positive feedback and the excellent question. My reading of your question (I hope I have this right) is as follows – Suppose you had a collaboration with folk from across the academy and outside and you wanted to create an integrated method for the group to use what practical advice does the IPO model give you? My view (Graham may differ) is that the IPO model won’t tell you what to do but it can help you with working out what to worry about. If we use your story as an example, the IPO model won’t tell you how to combine the methods of the groups you mention but it will point you at a number of relevant questions such as, what are the ‘inputs’ to be considered (i.e. what views on methodology do these folk start with), how similar or different are they (see ‘commensurability’ above), how ‘aligned’ do you need them to be (i.e. what kind of integration do you seek), do you need this alignment to work broadly or just for this collaboration (see ‘scale’ above) etc. I think (hope) that knowing what to worry about can be a valuable stage in the collaborative process. One thing that it does is allow a group to ‘target’ whatever consensus building mechanisms they choose; focussing those consensus activities/discussions on just the worries identified rather than trying to address every aspect of the collaboration just in case that aspect is problematic.
Two further thoughts – 1) the IPO model will help identify some but not all the areas of a collaboration that need to be worried about. 2) The number and variety of tools for consensus building are growing all the time. I’m a fan of the Toolbox workshop approach (https://tdi.msu.edu/) but its possible I’m a little biased in that regard 🙂 For a classic discussion of many concrete mechanisms for building consensus I’d recommend – McDonald, Bammer and Deane – Research Integration Using Dialogue Methods (ANU e-press 2009; http://doi.org/10.22459/RIUDM.08.2009).
To sum up – I think what the IPO model gets you is some clarity about what challenges you face – what it doesn’t do is tell you how to address those challenges. I hope these thoughts are of some value mate – it would be great to know what you think about them. Take care, be well and thanks for both the feedback and the questions – they are really helping me clarify my thinking on this stuff – thank you once again!
Stephen is right–the model’s goal is to be descriptive, so it won’t tell you what to do. If you want help there, be sure to check out the rest of TDI’s resources, especially our workshops at: https://tdi.msu.edu/.
Dear Stephen and Graham, Thank you so much for answering my question. This is helpful. Also, thank you for the link to your toolbox, I will check it out. Best, David
Dear Stephen and Graham, thank you for your post. I remember reading a couple of papers about the ‘Toolbox project’ and the ‘dialogue-based’ seminar, so I am a bit familiar with the topic. However, I find the end of your post a bit confusing. Some of your examples surely counts as “integration”, but maybe not as “cross-” or “interdisciplinary integration”. For instance, you ask: “What do we say about a situation where we recognize one alternative is better than another and discard the weaker alternative?”. It seems to me that if in an interdisciplinary team everybody recognize that one alternative is better than another, then we say that that group is already very integrated. The trouble is that researchers from different backgrounds and with different disciplinary identities may actually disagree in their ‘ranking’ of the alternatives, with none of them being necessarily wrong or irrational. Sometimes, the supporters of one of the two (or more!) alternatives will not be able to convince the supporters of the other(s) alternative(s). And yet, a critical decision must be taken at the group level. In a situation like this, I doubt that we can talk of ‘knowledge integration’ (or, rather, the problem is not with the integration of knowledge). What we would need is a theory of group decision and group deliberation. Can your IPO model account for such a dynamics (or could at least hep clarifying what is at stake in the situations like the one I described)?
Thanks a lot,
Vincenzo Politi
Vincenzo – thanks heaps for these thoughts and questions mate! For me your thoughts fall into 2 sections – first a concern about whether some of our examples of integration really apply in a cross- or inter- disciplinary context. Second – what, if anything, does IPO model say about group level decision making. Let me have a shot at each of these in turn.
The ‘examples problem’ – my sense (Graham may have a different view) is that you’re right about our examples – they are focussed on ‘integration’ simpliciter rather than integration in a complex (multi, inter, trans etc) disciplinary context. That’s because we’re still trying to get clear on ‘integration’ in the simple cases first and as you can see we are not there yet. Take the example you mention – when a group all picks one idea as preferable to another. Since the group has gone from having a diversity of views on this issue to having a single view we might view this as ‘integration’ (from many to one) *but* at the same time the surviving idea has not been informed by the other idea and so we might see this as a case where no integration (informing of one thing by another) has taken place. In sum, our views about what counts as integration are unstable even before we add in the complication of trying to achieve this state (integration) in complex disciplinary contexts. I’m pretty sure we haven’t been as clear on this point – integration simpliciter vs integration in complex disciplinary contexts – as we should have. Thank you for pushing me to clarify my thinking on this.
The ‘group decision making problem’ – here I take the question to be – does the IPO help us with group decision making. I don’t think that it does. The goal of the model is to get clearer about what we might be saying when we call the intellectual products of a team ‘integrated’. This feels ‘orthogonal’ to group decision making. A group may make a decision about how to go forward with a project in a way that has no impact on the diversity of intellectual products the team is working with. For example team members may have distinct ontologies and make a decision about what to do next that ‘side steps’ the question of multiple ontologies. In such a case we have no integration and so not much for the IPO model to address although we clearly have group decision making going on. While such a scenario is certainly possible and maybe even plausible a lot of the work of integration will involve decision making. That in turn suggests that the IPO model would benefit by being linked to work on group decision making. The epistemology, ethics and methodology of group decision making will impact how a group manages the integration of their conceptual resources. Finding a way to capture that impact would be a good idea. It looks like I’ve got some work to do!
Once again – thanks for the really helpful thoughts – I hope what I’ve said here goes some way to addressing your concerns. It would be great to hear what you think. Be well mate and I hope we can continue to think together.
Hi Stephen! Thank you for the reply!
Thank you for this post Steve and Graham! It’s always illuminating to board the philosopher’s space shuttle to enable zooming out on a problem that practitioners are grappling with. On a recent INTEREACH webinar, one participant attested to the difficulty in helping researchers understand what knowledge integration (KI) is and what it can do for their research. I think this framework will not only be fruitful in clarifying integration for practitioners, but also in helping others wrap their heads around KI in order to be able to employ both it and the expert in doing the KI.
Kristine – thanks heaps for these thoughts. There’s stuff that’s not so obvious from the ‘space shuttle’ so feedback from practitioners is super valuable! In this case you mention that there is a challenge around ‘what KI is and what it can do for their research’. I think the IPO model does help think about what KI is or at least the sorts of things it can be. I’m not so sure that the model is as useful for answering the ‘what it can do for their research’ question. For me, this second question is one of value; roughly, ‘how does KI make my research more valuable?’. To answer that we’d need to know more about what the researcher in question values. But, in general, knowing more about what KI is should help with the task of connecting (or not) KI up with this person’s values. The more you know about something, the clearer you can be about what it can and can’t do and that in turn makes it easier to see how it connects to specific values.
Once again – thanks! You’ve got me thinking in new ways about the model – and that is a blessing! I’m intrigued to see what Graham thinks.