Tuesday, May 8, 2012

Conference report: Integration in Biology and Biomedicine, Sydney, May 3-4 2012

A personal view of the meeting from Paul Griffiths

Major new research initiatives at the University of Sydney emphasise the ‘integrative’ nature of their work. This conference focused on what ‘integration’ is and how it can be facilitated. Participants were leading Sydney scientists, philosophers of science whose recent work has focused on integration, and social scientists studying integration and developing practical interventions to promote integration. The conference was jointly supported by the Sydney Centre for the Foundations of Science, the Charles Perkins Centre, and the Institute for Sustainable Solutions. The conference program is available here.

A key concept at the meeting was ‘translational integration’, a term introduced by Sabina Leonelli. This is a very different conception of integration from that found in from traditional accounts of the reduction of one science to another, or of the ‘unity of science’, and also differs from more recent work on the emergence of new fields at the intersection of different disciplines. These differences can be brought out using some apparatus introduced at the meeting by Todd Grantham: to see what is meant by ‘integration’ in any given context it is necessary to identity the units that are being integrated, the nature of the connections made between them, and the purpose for which they are being connected. Traditional discussion of reduction and the unity of science focused on scientific theories or models of broad application – units that constitute the major achievements of scientific disciplines. The connections between these units were on the same scale, with the ‘reduction’ of one theory or model to another being particularly prominent. The aim of integration was to clarify the overall structure of scientific knowledge. This is very different from translational integration.

The units of translational integration are methods, data, and specific results (or even hypotheses). The ways they are connected are often temporary and only locally valid, and the aim of integration is to design interventions. Integration of the results of research might seem an odd idea in a discussion which has often focused on integrating diverse elements in the process of research, but it does seem to be what is intended in some talk of 'integrative research. For example, in research on obesity, we have findings from, to choose just a few fields, molecular biology, physiology, social science findings about the (in)effectiveness of education programs, information on the price elasticities of different foods, epidemiological  data on obesity in domestic pets paralleling that in humans, etc, etc.  Treated in isolation, these may support very different recommendations about what to do to improve health. One reason to describe research as ‘integrative’ is if it tries to articulate these diverse findings to make a case for trialling particular interventions  This seems to capture what is intended by some references to ‘integrative research’ in the research initiatives which sponsored the conference. It is also something along these lines that the social scientists at the meeting have been seeking to facilitate.

Gabrielle Bammer’s presentation of ‘integration and implementation sciences’ described a systematic approach to delivering integrative research. She assumed that such research will be problem-oriented, and emphasised that a major issue in translational integration is that knowledge is incomplete, so that part of translational integration is the management of unknowns. The aim of translational integration is not to achieve a more complete vision of nature, and so incompleteness of knowledge is a practical issue to be managed, not an insuperable barrier to integration. Whereas philosopher Sandra Mitchell’s presentation argued for the necessary incompleteness of any single scientific representation of the world, and a consequent need to revise our ideal of scientific enquiry, Bammer’s suggestion that the management of unknowns is central to integrative research would remain valid even on the traditional view that the aim of science is to produce a complete and consistent model of the natural world.

Bammer discussed two classes of tools for delivering integrative research: dialogic methods and modelling techniques. Dialogic methods were exemplified in a presentation by Michael O’Rourke and Stephen Crowley. These researchers have drawn on ideas from the philosophy of science to design facilitated conversations between members of research teams which draw out underlying presuppositions about the aims and standards of science which team members bring from their home disciplines. Bringing these into the open pre-empts misunderstandings and allows the management of disagreement.  

The use of models in translational integration ranges from exploring potential connections between different fields, as exemplified in the demonstration of systems diagrams by Robert Dyball, through the detailed causal models of complex systems seen in presentations by David James and John Crawford, to the deliberately abstractive mathematical models presented by Olaf Wolkenhauer. Much recent work in philosophy of science has focused on modelling as scientific practice, and models as scientific products. Whilst both these themes were well-represented in these presentations, another perspective on models present at the conference was their role as a tool for achieving translational integration.  The construction of a model to serve as the basis for action can be the activity through which diverse data, methods, results, and existing models from the contributing disciplines, are connected up and rendered commensurable.

What agendas for future research emerged from the meeting? The social scientists involved in research on integration all emphasised that it is early days in the process of developing a systematic approach to delivering integrative research. Those who work with research teams to facilitate integration are simultaneously engaged in research into the effectiveness of these interventions. However, given the ubiquity of interdisciplinary team science in today’s biology and biomedicine, the idea of learning from theory and experience, and embodying those lessons in a more systematic approach to integration is surely worth pursuing.


  1. PART 1/2

    Before I forget it: An updated version of the ‘Reductionism in biology’ entry (originally published in 2008) of the Stanford Encyclopedia of Philosophy has been posted 3 weeks ago. We have added many new references, and in particular the discussion on integration in Section 5 has been enlarged: http://plato.stanford.edu/entries/reduction-biology/#5

    Some of the issues that Paul mention align with the position I previously expressed in writing, and which I mention here as they were not covered in my workshop talk: “Beyond reduction and pluralism: toward an epistemology of explanatory integration in biology,” Erkenntnis 73:295-311 (2010). In this paper I develop further the problem-based perspective on integration previously developed by my colleague Alan Love (Philosophy of Science 75:874-886, 2008). The idea is that a certain complex problem (dubbed ‘problem agenda’) not only motivates integration (as no approach in isolation can solve the problem), but coordinates integration in that a problem agenda has an internal structure (components question standing in systematic relations) plus associated standards of explanatory adequacy, which entail what kinds of contributions (from what fields) are required and in what way they are to be integrated.

    The issue that aligns with the Paul’s post (in particular his discussion of ‘translational integration’) is my position that integration is not a regulative ideal or aim in itself, but that integration may only be required for the purpose of a particular problem. Some problems do not require any integration, and the degree and kind of integration depends on the problem at hand. So integration is more transient, in that items of knowledge are brought together relative to an issue addressed in a local research context. So there is not one set of intellectual relations across fields (but many sets of relations, each for a different problem), in contrast to the traditional philosophical vision of the unification of different fields.

    In his talk, Todd made the useful distinction between different units that are integrated: (1) fields or (2) individual items of knowledge (explanations/models/data/methods). My above problem-based model is about the latter and this is what most appear to have in mind (at least Paul and some of the views expressed at the workshop): it is individual ideas (from different fields) that are integrated. To the extent to which some talk about the ‘integration of fields,’ this may just be a façon de parler, which does not mean the stable unification of different fields. For instance, Bill Bechtel has discussed how cell biology led to the integration of disciplines, but he immediately notes that such integration at the same time results in some “disintegration,” due to new institutional specialization (Biology and Philosophy 8:277-299, 1993). I agree with Todd that it is useful to distinguish between the integration of ideas (which can be vary local and transient) and the integration of fields in the sense of more systematic connections across fields (which happens, but is more rarer and the result of lengthy processes). [Bechtel’s discussions have paid substantial attention to the institutional aspects of integration, and we need more of this beyond the epistemological considerations that philosophers focus on. Likewise, integration should be conceived of as a process and a practice, as opposed to focusing on the intellectual results, e.g., an integrative explanation.]

  2. PART 2/2

    In a forthcoming paper, Alan Love and I discuss the situation of evo-devo, which is often hailed as an emerging “synthesis” of evolutionary biology and developmental biology (and possibly other disciplines). However, the label synthesis is clearly overstated; and in fact the future shape of evo-devo is not settled yet and biologists disagree about what evo-devo is or should be (including its relation to overall evolutionary biology), with some preferring the view that evo-devo is an autonomous discipline or just an approach at the intersection of many others. Evo-devo is clearly integrative research (by using items of knowledge from different disciplines), but its institutional status is less clear. (Brigandt, I. & A. C. Love “Conceptualizing evolutionary novelty: moving beyond definition debates,” Journal of Experimental Zoology Part B: Molecular and Developmental Evolution 318(6), to appear in 2012. For a shorter, published account see Brigandt, I. & A. C. Love “Evolutionary novelty and the evo-devo synthesis: field notes,” Evolutionary Biology 37:93-99, 2010).

    I have some misgivings about Paul using the label ‘translational integration’ for what he has in mind, and would just talk about ‘integration’ or ‘integrative research’. (Paul took over the label ‘translation’ from Sabina Leonelli, who needed a special name to distinguish it from other kinds of integration, e.g., cross-species integration, but it strikes me that these other types are just very narrow special cases of translation integration, if they are types of substantial integration at all.) One drawback of the label ‘translation integration’ (as opposed to ‘integration’) is that (a) it suggest that integration consists in / is only motivated creating a technological product, and (b) it may erroneously suggest that that the direction of interaction / knowledge flow is unidirectional. But (a) integrative research is done for all sorts of reasons (my above talk about a ‘problem’ motivating and coordinating integration is meant to capture different options); and (b) in integration, it is not just that existing knowledge is travelling from one place to another without being transformed (e.g., one field merely using the experimental instruments or data from another field), but different items of knowledge from different approaches come into contact so as to create something that did not previously exist.

    One word that Paul’s account focused on integration and integrative research hardly made use of was ‘interdisciplinary’. I guess that the two cross-cut each other: on the one hand, one can integrate ideas that come from one discipline only (though one may wonder how substantial such an integration is), on the other hand, one can have very loose types of interdisciplinary interactions that are not genuine types of integration.

  3. I am happy to see Ingo's development of the idea that integration is required for specific problem solutions.

    For my early defenses of this position you might want to look at my "On Pluralism and Integration in Evolutionary Explanations", American Zoologist , vol. 32, 1992, pp. 135-144, "The Why's and How's of Interdisciplinarity", Sandra D. Mitchell, Lorraine Daston, Gerd Gigerenzer, Nevin Sesardic and Peter Sloep, in Human by Nature: Between Biology and the Social Sciences , Erlbaum Press, 1997, pp. 103-150, and “Integrative Pluralism” Biology and Philosophy, 17, January 2002, pp. 55-70 and “Integration without Unification: An Argument for Pluralism in the Biological Sciences” co-authored with Michael R. Dietrich, American Naturalist. 2006. Vol. 168, pp. S73–S79. The 1992 and 2002 papers are reprinted in Biological Complexity and Integrative Pluralism , Cambridge University Press, 2003.

    More recent discussions are in Unsimple Truths: Science, Complexity and Policy, University of Chicago Press, 2009.