Abstract
Interdisciplinarity and integration are topics that have drawn increasing interest from philosophers of science over the last decades. One feature that mark some such interactions is the sharing and transferring of problems between disciplines; what we call problem feeding. With a few notable exceptions, such as the work of Lindley Darden and Nancy Maull, it is an aspect of interdisciplinarity that remains underemphasised. The aim of this paper is to remedy this oversight by providing outlines of an account of interdisciplinarity as problem feeding; its prerequisites and the specific practical and epistemological challenges it involves. We proceed by first providing a philosophical framework and then applying this to two cases, both revolving around interactions between natural and social science disciplines.
Introduction–Interdisciplinary Choices
The following passage can be found in the introductory chapter to Paul Robbins (2012) Critical Political Ecology. In it he discusses the Gilbert Fowler White's doctoral dissertation Human Adjustment to Floods: a geographical approach to the flood problem in the United States (White 1945).
As [Gilbert Fowler] White so profoundly discovered in the 1940s, improved planning by state agencies and individual farmers who lived in the floodplains was retarded by the continued reinvestment in subsidies and massive investment in engineered structural solutions like dams and levees. But the urgent question raised by these results – why do structural solutions prevail in the face of better alternatives? – could not be answered within the hazards approach, which focused almost exclusively on individual choice, free markets, and rational regulation. Rather, the issue can only be addressed fully by examining the political economy of floodplain investment and the role of capital in agricultural development, and the control of legislative processes through normative ideologies, vested interests, and campaign finance. Similarly the risk of floods is not uniformly distributed through populations. Are poor and marginalized groups more vulnerable to such events? What is the role of power in the environmental system and its relationship to people? \cite{Robbins:2012uj} (35)
For Robbins White's realisation that floods, as writes "are 'acts of God,' but flood losses are largely acts of man" (White 1945, 2) marks an important even in the development of what would eventually become political ecology. White understood that although floods are often caused by natural events, the impacts they have on human communities, have to do with how those communities live. This in turn is a question of social organisation and economics. The problem, as it were, points outwards, towards other disciplines.
If we allow ourselves to idealise slightly, realisations of the sort that befell Gilbert Fowler White presents researchers with a decision to make. The tools afforded by their home discipline does not suffice to pursue the problem they are interested in. Do I import the tools I need, or export the problem? In this essay we shall concern ourselves with the latter choice. There are several reasons why this is interesting. First, a common conception of why interdisciplinarity is important is that interdisciplinary research—in contrast to its disciplinary counterpart—focuses on, and is driven by, problems. a common conception of why interdisciplinarity is important is that interdisciplinary research—in contrast to its disciplinary counterpart—focuses on or is driven by problems (Klein 1990, Jantsch 1972). The resources of different disciplines are required towards some common problem. But sharing problems necessarily involves transferring problems. Second, transferring problems between disciplines is a form of interdisciplinarity of considerable potency.
In this paper we will give an explicit account of problem feeding. Our aim is to answer a number of questions. What is problem feeding? When is problem feeding advisable? What needs to be in place? What are the main challenges in practical and epistemological terms?
We will structure the paper as follows...
What is Problem Feeding?
Before we get into the details, it is useful to start out with a schematic representation of what we have in mind. Problem-feeding (henceforth PF) concerns situations that can be conformed to the following schema.
1. Problem P(a) is formulated in discipline D(1)
2. Problem P(a) is transferred to discipline D(2)
3. Discipline D(2) provides a solution S(a) to problem P(a)
4. Solution S(a) is transferred back to discipline D(1)
This schema depicts what we shall call bilateral PF. This, we take it, is the most involved or substantive form. Unilateral PF, which we shall take a lesser interest in, does not involve step (4), and sometimes not strictly even step even(1).(1). stepForFor (1).instance,instance, there are disciplines that depend on other disciplines for some, or all, of their problems but
For instance, there are disciplines that depend on other disciplines for some, or all, of their problems where it may be the case that the solution to those problems—or even their very formulation—is never of interest to discipline that supplies the problem. For example, the philosophy of physics relies on physics for its problems, these problems may or may not be relevant or interesting to physicists. Todd Grantham has classed this kind of dependence on other disciplines for problems heuristic dependence (Grantham 2005).
We have already said something about the phenomenon under investigation; the transfer of problems between scientific disciplines or fields. So, an important immediate issue, is what more precisely a problem is, and what disciplines or fields are. Let us begin with the latter. Although the transfer of problems has not been widely discussed directly, neither within the philosophy of science, nor the broader discussion on interdisciplinarity, there are predecessors to our notion of problem feeding. The most important is to be found in Darden’s and Maull’s (1977) seminal paper on interfield theories. Interfield theories are essentially hypotheses about the ‘ontological’ connection between fields. An example is when one field studies the function of a structure studied in another field. The interfield theory, then, is precisely the supposition that the two fields are connected in that way. The chromosome theory of the gene is an example of interfield theory, as it places the gene—an entity stipulated in Mendelian genetics—on the chromosome, an entity observed by cytologists. Darden and Maull connect the establishment of interfield theories with the transfer of problems. They write:
In brief, an interfield theory is likely to be generated when background knowledge indicates that relations already exist between the fields, when the fields share an interest in explaining different aspects of the same phenomenon, and when questions arise about that phenomenon within a field which cannot be answered with the techniques and concepts of that field. (Darden and Maull 1977, 50)
The idea was further elaborated in a paper from the same year by Nancy Maull (1977). She approached the issue somewhat more explicitly under the label of problem shifts.
It is possible for problems to arise within a field even though they cannot be solved within that field. Their solutions may well require the concepts and techniques of another field. In this case, we say that the problem “shifts”. (Maull 1977, 156, emphasis original)
Maull argues that what she calls proper terms can connect fields (in her case on different levels of description). A term is shared between fields when it is a proper term of both fields. A proper term for a particular field, is a term which is part of the special vocabulary of that field. Proper terms are added to fields either by, quite simply, members of the field developing their terminology as they see fit by coming up with new terms, or by appropriating terms from elsewhere. Sometimes, when fields appropriate terms, however, the "knowledge claims previously associated with its [the terms] use are retained" (Maull 1977, 150). Inquiry connected to such a shared proper term, then, becomes the prerogative of all fields that indeed have that proper term in their special vocabulary. Thus the knowledge claims associated with the term in question can be modified and revised from several fields. In this way a problem—such as accounting for the concrete nature of mutations—can be ‘shifted’ from one field to another.
In both of these papers more specific issues are discussed than we are interested in here. In Darden and Maull (1977) the focus is on the relationships obtaining between fields, whilst for Maull (1977) the aim is to spell out relationships between fields that are on different levels of description. We will therefore refrain from using, for instance, Maull’s notion of a problem shift, as it is so strongly associated with her framework. We will instead deploy the label ‘PF’, since it is meant to be broader, and to include both fields and disciplines (to the extent these are in fact distinct) and also a range of other types of context.
PF can occur in many ways, but one important distinction that can be drawn is between unilateral and bilateral PF. The former concerns cases where one discipline depends on another for problems, but there is no reciprocity in the exchange. Todd Grantham (2004) mentions a form of practical integration he calls heuristic dependence of which this notion of unilateral PF is reminiscent. For instance, philosophers of physics may draw on physics to find interesting problems. However, these problems are not always problems that physicists themselves think of as important. That is to say, presumptive solutions are unlikely to be “fed back.” In bilateral problem feeding, on the other hand, there is a component of division of labour and reciprocity. A discipline encounters a problem that is perceived as important but resists solution within the discipline. The problem is thus outsourced to an appropriate alternative discipline or field, and when it is eventually solved the solution is fed back into the discipline of origin. In Darden’s and Maull’s examples the type of problem feeding that is occurring is bilateral. Again, the chromosome theory of the gene is a good example. The gene (or factor) had been stipulated in Mendelian genetics, but its physical nature was not known. As the gene was found to be located on, or in, the chromosome, Mendelian geneticists could use this to explain why assortment is not perfectly random. Genes close to one another tend to be inherited together, and this skews the ratios slightly (see Darden 1991, Thorén and Persson 2013).
Prima facie the PF model is of immediate relevance to sustainability science in particular, as that field can itself be said to be founded on an attempted problem transfer. The recognition that, for example, climate change is an issue of concern to both natural and social sciences was originally made by ecologists and climate scientists.
The model is also of more general relevance, as PF is fundamental to all kinds of problem sharing. Here is a general argument for this point. Let us assume a minimal case. Two disciplines, D1 and D2, are to be involved in solving problem P. Here either P is recognized, or taken note of, by both disciplines, or it is not. In the latter situation the case is trivial—the problem has to be transferred. In the former it appears that in order for D1 and D2 to recognize that they should both be involved in solving P there needs to be a mutual transfer of both versions of the problem so that the comparison can be made. Thus the transfer of problems between disciplines is fundamental to all types of joint problem solving.
Before we conclude this section it is important to say something about how problems can decompose into sub-problems. In situations of joint problem solving it will often be the case that an overarching problem is shared as a problem that falls apart into smaller sub-problems that can be solved individually, in the interdisciplinary case, by different disciplines. With such problems the solution to the overarching problem is the aggregate of the solutions to the sub-problems. Hence the transfers concern both the overarching problem and, many times, the various sub-problems. Alan Love (2008; see also Brigandt 2010) has used the notion of problem agendas to describe this kind of situation. Love (2008) distinguishes problem agendas from individual problems, the latter of which can be either empirical or conceptual (see Laudan 1977). He writes:
A problem agenda, by contrast [to an individual problem], is a “list” of interrelated questions (both empirical and conceptual) that are united by some connection to natural phenomena. For example, how do questions concerning greenhouse gas contributions from plant respiration, along with many other questions about emission-related phenomena (anthropogenic or otherwise), including their interaction with systematic cycling and atmospheric dynamics, get answered with respect to global warming phenomena? Problem agendas are usually indicative of long-term investigative programs and routinely require contributions from more than one disciplinary approach. Cross-disciplinary interactions of this kind rarely occur spontaneously and are often driven by a commitment to similar questions. (Love 2008, 877)
Although Love does not focus on transfers of problems directly, some transferring needs to be occurring in order for a problem agenda to be established in the first place.
There are plenty of examples of what appears to be PF and there is a good rational basis for PF as a practice, but issues arise over exactly how problems are to be transferred—at least, if we are to take the Kuhnian concerns raised in the previous chapter seriously.
PF can be construed as a cognitive or epistemic process; the transfer of problems between bodies of knowledge, conceptual schemes, distinct methodological frameworks, or perhaps theories. Or, we can think of it as a social process, the transfer of problems between disciplines conceived of as epistemic communities. On the former conception there is really no difference between importing tools and exporting problems, other than one of emphasis.
Two Phases of Problem Solving
Where in the problem solving process does PF occur? Let us begin by taking a step back and discussing problems and problem solving more generally. Thomas Nickles (1981) suggests we think of a problem as follows:
A problem is a set of constraints (better, a constraint structure) plus a demand that the object (or an object, etc., depending on the selection properties of the demand) delimited or ‘described’ by the constraints be obtained. (Nickles 1981, 31)
The constraints in question concern admissible solutions to the problem; they tell us what the solution should look like. We will use a definition based on Nickles’ account:
Definition: A problem is a pair D> where C is the set of all constraints on the solution(s) to the problem, and D is a demand that the problem be solved.
First, a general remark. This definition differs from other suggestions as to the nature of problems in that it does not construe problems in terms of the admissible answers themselves (see e.g. Belnap and Steel 1976). The main advantage of this is that it offers a way of understanding how features of a solution may be known to a problem solver without that problem solver actually having the solution. The idea that we can maintain a distinction between having knowledge of a problem and having the solution to the problem is highly intuitive, not least because not all problems have solutions despite appearing to be well understood. For example, the problem of providing an analytical solution to the n-body problem is challenging not primarily because we do not know what such a solution would look like, but because it has no solution.
Second, Nickles’ conception of a problem is very abstract, and the notion of a constraint is quite broad. For one thing, there are clearly different kinds of constraint, and one might find it useful to differentiate between them on occasion. Some constraints are open, as Reitman (1964) famously remarked. Some are explicit, and some are implicit. Some are necessary, others redundant or peripheral, and so on. Moreover, there may be relevant differences between different types of problem (e.g. producing a formal proof, explaining some hitherto unexplained phenomenon, predicting an event, the concrete operation of shifting a system from one state into another, and so on). These potential differences become muted when the above definition is accepted.
What precisely are these constraints, then? Nickles’ paper is curiously void of examples, but hints can be found in Love’s paper (2008). Love uses a somewhat different terminology. Rather than discussing constraints, he associates problems with criteria of explanatory adequacy which are necessary in order to assess whether a solution is acceptable or not. He mentions examples of such criteria—e.g. logical consistency for a conceptual problem, or the need to include a causal factor when addressing an empirical problem” (Love 2008, 877). Precisely what constraints are associated with specific problems is, as we shall see, often contentious, but consider the following. Suppose the resilience theorists are right in their thinking about things such as sustainability and sustainability transitions. If they are indeed so, then the problem of making a social-ecological system sustainable is really about making it as resilient as possible in the face of certain types of disturbance. This is in itself a substantive constraint on admissible solutions: if problems of sustainability are really about resilience, their solutions will have to be put in terms of certain types of mechanisms and interactions. For example, relevant events (such as collapses and regime shifts) are to be explained in terms of structural features, like interactions between driving variables and parameters in the system. Such constraints exclude large swathes of possible solutions.
Some constraints, clearly, appear to be more closely related to specific disciplines. For example, the expectations we have about what it would mean for a physicist to have solved a problem may be quite different from the expectations with which judge a literature scholar. There are, within each of these disciplines, restrictions regarding what solutions in general are allowed to look like. These will sometimes be trivial in the sense that they are not particularly exclusive—a matter of form only, perhaps. At other times they may be highly exclusive. Think only of the propensity of economists to solve problems within a formal framework. The precision offered by the method comes at the cost of often having to deal in obtrusive idealization.
Now, to return to the process of solving problems. Roughly speaking, we can, on this definition, discern two phases in the problem solving process. In the first, the exploratory phase, the problem itself is the immediate object of enquiry. The aim here is to acquire a sufficient understanding of the problem—that is to say, to reveal the set of constraints C that is constitutive of the problem we are trying to solve. In the second phase, what might be called the derivational phase, the aim is to obtain a solution given a full, or sufficiently articulated, set of constraints C. Let us abbreviate these phases as phase-1 and phase-2, respectively.