By Joern Fischer
Just a few days ago, I spoke with a colleague about integration of research findings across scales of enquiry, and across disciplinary domains. We discussed this in the context of complex research projects that involve multiple components. It struck me there seemed to be two ways of knowledge integration that have different pros and cons.
The first is “linear” integration. You can picture this one as a flowchart, where one bit of knowledge flows to another, and where several boxes merge in one place. Such flowchart, or linear, integration is common in research projects. I see two main strengths for this. The first is intellectual rigour and transparency. Because you have mapped out an integration path from the beginning, you now collect the pieces of the puzzle, align them – and then you’re done. Anyone can follow this process because it’s transparent. The second advantage is that you can plan from the outset to have your data be in formats that fit together nicely. So, for example, your economic component can feed into the integration box, as can your ecological component – and if it’s planned well, you might find out something about the cost-effectiveness of alternative conservation methods. So, linear integration lends itself to integration via formal, quantitative models.
The second way of integration is “nonlinear”. This one may be pictured as a cloud of bits of knowledge about a system. Initially, you target the system of interest from a range of different perspectives. These could be disciplinary perspectives, or different scales, or a mixture of both. These perspectives may focus on the system in very different ways, including entirely different sets of methods. When it’s all said and done, all bits of knowledge thus generated can be visualized as points within a cloud. The system, in this analogy, is the whole cloud, and somehow, you never know the whole cloud. But if you have scattered your points nicely throughout it, when you step back and squint at your system, you start to see it nevertheless. This type of integration is non-linear, and difficult to plan. An advantage is that you might find things you could otherwise overlook, because you have no pre-conceived idea of how things might fit together. An obvious downside is that integration here is much more an art than a science (we’re squinting at a cloud, for crying out loud!).
Which is better? It probably depends on the purpose. My own research tends to favour the second approach, for better or worse.