Navigation – Plan du site

AccueilTransversalitésDossiers200112ème Colloque Européen de Géogra...Opening Plenary : Communicating a...

12ème Colloque Européen de Géographie Théorique et Quantitative, St-Valéry-en-Caux, France, 7-11 septembre 2001

Opening Plenary : Communicating about Theoretical Geography

Conférence d'ouverture : Parler "théorie" en géographie
Denise Pumain

Texte intégral

1First of all, I wish to thank Yves Guermond for drawing a superb image as a symbol of our colloquium. This superimposition of bookshelves on the Etretat cliffs reminds us of the splendid metaphor developed by Abraham Moles, the specialist in the psychology of space, about the wall of knowledge, which is continuously under construction, and to which each new piece of research adds a new brick. But might we not wonder whether there is not a risk that, as in the case of the Normandy cliffs last summer, the wall of knowledge could not collapse, and a whole row of bricks disappear into the sea ?

Euroquant or Theoquant ?

2In his brilliant review of the activities of our group, presented in Durham two years ago, David Unwin (1999) insisted on their contribution to geography: "There has been a strong concentration on central concepts within geography" (p.1). He added : "Our postmodern colleagues would find this summary of over twenty years of work surprisingly "geographical" in its methods and objectives" (p.6).

3Where does this strange feeling that quantitative research could be excluded from the geographical stage come from ? It is paradoxical, since, as it will be shown during the next three days, when papers will be presented in three parallel sessions, quantitative geography is "very much alive and well !", as David says. However, it is doubtful that the feeling of exclusion originates solely in the contemptuous attitude towards quantification held by some post-modernist social scientists. A more common trend, and perhaps more worrying, is a relative absence of geographical theories and concepts from public and academic debates where they should or could be present. Not only most people in the street, but sometimes also our distinguished colleagues, ignore or pretend to ignore the present state of geographical research. There is a huge gap between the success and effectiveness of geographical methods and tools—which are now in use in so many versions of geographical information systems for an incredible variety of fields of application—and the general lack of awareness about the integrated corpus of knowledge that should be included under the name of geography. I would suggest that this may result, at least in part, from a much too discreet contribution of geographers, and especially the so-called quantitative geographers, to the general discussion about their geographical concepts and theories.

4According to Sylvie Rimbert (1999), "the success of commercial geocomputational production has depended upon twenty years [of] research in theoretical and quantitative geography".

5So the question is : what are we doing now which will become useful after twenty years ?

Theoretical bits and bobs or integrated theories ?

6In our individualistic societies, many questions are related to this interaction between micro-level and macroscopic organisations. To what extent are you merely a user of already existing spatial structures, or an actor who is able to modify or even create them ? Who is responsible for urban congestion or positive externalities in the agglomeration ? The preferred approach among social scientists is increasingly to break down these structures, from two opposite positions : either the very formal economic theory based on a micro-economic theory of firm individual behaviour, or the much less formalist thinking of those who adhere to biographic studies or family monographs. There is a widespread belief that in the social sciences, any concept or theory must explicitly refer to the (conscious or unconscious) motivations of individuals.

7This so-called methodological individualism, however, is not obvious. Some social scientists refuse to develop concepts that will not reflect the intentional dimension of individual behaviour. For others, the question remains open : whereas previously widely accepted notions concerning social groups, class, ideologies and other concepts of corporate life have been generally contested—if not rejected—in many of the social sciences, geographers should feel free to explore the collective properties of entities like territories or urban systems, which may in many respects be revealed to be non-intentional.

8Among the major conceptual inventions of geography in the second half of the twentieth century, two levels of formalisation can be recognised. Concepts related to the identification of meso-scale or macro-geographical spatial structures have been quite clearly defined and abundantly documented : the dual structure centre-periphery, at various levels of analysis, central place networks, different types of communication networks, for example. Measurement, comparisons, adjustments to models are easily made. But a second set of concepts has remained at the level of the intuitive approach. Tentative explanations have not yet been confirmed by demonstrations. There are mainly concepts related to individual spatial behaviour and cognition and their effects on the development of spatial structures. Very few systematic experiments have been made in relation to theoretical thinking, except for housing market searches, or investigations into shopping behaviour (with reference to discrete spatial choice theory), or migration moves (referring to spatial interaction models) and spatial diffusion of the innovations.

9The result is a mosaic of concepts and theoretical elements that are well developed at different levels of analysis but generally unconnected. At the same time, there is a demand for knowledge to be better integrated, and—for example—a desire for geographers to make an explicit contribution to contemporary debates about globalisation. Here I would like to mention Peter Haggett’s decision to change the title of his book : in the most recent version, "Geography, a Modern Synthesis" has become "Geography : a Global Synthesis". I would suspect that this change is not a sign that Peter Haggett is renouncing modernism, but rather, an indication that he recognizes what is expected.

A new deal for geographical theory : from borrowings to invention

10Borrowing concepts from other sciences is often fruitful, but it is now time to dare to produce our own theoretical thinking. What kind of theory are we developing ? What we call theory is not just a product of some logical or philosophical thinking. It is a construction made through a repeated process that includes observation, theory-building, modelling, and procedures of validation, leading to the formulation of a general and consistent explanation, then returning to the reformulation of a theory, new observations, and so on. This retroductive way of building theory is neither purely deductive nor inductive, but closer to the everyday practice of research. It also means that there is a limited commitment to a cumulative process as a means of developing knowledge.

11If geography shares with other social sciences an interest in individual behaviour in space, and may use the categories elaborated by psychology, sociology or political science in formalising this approach, only geographers are really interested in the emergence of structures and systems in space on a broader scale, and in ways of differentiating between them. Such geographical objects as territories, settlements, networks, borders... must then remain the focus of a theoretical project for geographers.

12Connecting geographical structures and individual behaviour through new theories of change has become possible. The paradigm of systemogenesis makes a distinction between two ways in which geographical change takes place : incremental adjustment to general trends (for instance, diffusion of innovations, simulations through cellular automata) including many local fluctuations; and amplification of some fluctuations (lags or advances in the general process), leading sometimes, during period of instability, to shifts in direction (changes in the relative geographical situation of geographical objects, or the emergence of new ones). We can try to develop such a theory.

13Self-organization theory, borrowed from physics, appeared fifteen years ago as a useful set of concepts to be imported (or better, transferred) to geography, because at one and the same time it provided an interesting theoretical framework for connecting natural and social sciences, and dynamic models whose behaviour was compatible with fundamental concepts of social systems (irreversibility of evolution, uniqueness of individual trajectories, uncertainty about the future state of the systems). Informative developments in geography were elaborated by A. Wilson or P. Allen or W. Weidlich and G. Haag and their research teams : for the first time, consistent spatial structures were emerging from a set of hypotheses about spatial interactions, and the shift between possible alternatives was linked to internal fluctuations or external perturbations that had taken place at critical periods in the history of the system.

14The dissatisfaction that ensued resulted not so much from the (real) difficulties of putting the dynamic models into concrete applications, as from the incapacity of this type of modelling to solve conceptual problems : during the applications, it is almost impossible to identify the effects of parameters that are embedded in a complex dynamics, and to measure their actual effects. The theoretical advantage of the theory, then—whose purpose is to connect the elementary behaviour of particles at the microlevel with the cumulative behaviour and evolution of structures at the macrolevel—is hardly put into practice in the social sciences, because the descriptors of individual action cannot be clearly identified and separated. Another difficulty is that the emergence of a true innovation cannot be simulated through such methods.

15We could be further tempted to borrow concepts and elements of theoretical thinking from theories related to living systems, and develop evolutionary theories of geographical change (as proposed here, for example, by G. Rabino). But again, discrepancies arise between models of living species and geographical objects, because essential mechanisms like cognition and reflexivity are not taken into account in the process of evolution.

16So, we believe it is now time to enter the "the age of maturity", and to develop a self-consistent theory for geographical systems, by bringing together the considerable amount of useful experimental results that have been established over the last fifteen years of research into systems of dynamics. Such a theory would rely on the state of our knowledge about spatial processes and the evolution of spatial structures. Again, a book title can provide an image for this : Alan Wilson’s recently published textbook makes "Complex spatial systems" understandable to a wider audience.

17Much has been said about the shift in geographical theory from morphology to processes, from single scale to multilevel analysis. But I would suggest that as a result of this, there has been a related shift in our way of thinking, which to my knowledge has not been sufficiently emphasized. The shift that has taken place between a static paradigm and a dynamic paradigm modifies our way of describing geographical systems. Through it, we are actually entering a new age for the theoretical representation of geographical objects. The notion of local combination was the basis for linking features in landscapes and the spatial organisation of society that had been observed elsewhere or in other cultures. This concept was still present, as a theoretical reference, in the form of a borrowing of mathematical combinational processing, for instance in the spatial interaction model of entropy maximisation. Now, instead of explaining any particular geographical object by looking for a specific combination of general features, we relate its peculiarities to a specific trajectory in the dynamics of a system. This is completely different from an explanation through history, as in a biography, because it implies that we can imagine a model for the development of a type of dynamics and can consider that only the peculiar series of shifts that have taken place over time in a given location are responsible for its appearance of uniqueness. Besides developing a way of understanding the articulation between the usual features of geography in general systematic geography and local peculiarities—the uniqueness of each case—this theory makes it worthwhile exploring why the same processes can give rise to a variety of configurations, but not to all possible combinations.

Theory and modelling

18Fortunately, in the last ten years, there has been marvellous progress in the quality of tools that can be used to help us in developing such a project. The exploration of complexity in geographical systems is not only necessary for developing geographical theory, but has also become possible, for at least two reasons.

19First of all, there is the increasing availability of individual databases elaborated by public agencies or trading and service programmes that are interested in geographical analysis; this is also true of land use, which is better and better documented, with an increasing precision in space and time, either from remote sensing devices, or from local surveys and measurements. This is the positive aspect of the growing involvement of geographers in the spatial engineering of society. For the first time in our academic life, we can hear research students complaining : "I have too much data"!

20In a parallel development, impressive progress has been made by many research teams in the processing of spatial data. Another reason for the new possibilities for developing geographical theory is the incredible development of integrated methods for simulating spatial processes. This is explained not only by the Moore law, but also by the fact that the essential work we were all clamouring to see only ten years ago has now been accomplished : many methods of spatial analysis are integrated into the GIS. Thanks to the joint efforts of dedicated specialists in computer science, smart geographers and some networks like the GISDATA that are supported by the European Science Foundation—and also a few large software firms interested in rewarding (profitable) applications like geomarketing or vehicle guidance.

21Some of these highly effective methods that allow for complex reasoning, like micro-simulation, cellular automata, multi-agent systems or animated mapping, and so on, are now fully integrated into everyday research procedures in our geographical laboratories, for instance, in Mike Batty’s CASA in London.

"Natural" hazards in modelling

22Of course we have to be aware of a few risks that are unavoidable in modelling, which is why one is tempted to call them "natural" risks. The new modelling tools associated with GIS allow for great flexibility in the use of models. Besides general effects, local factors can be included in order to adapt a simulation to very different conditions. We have gained more flexibility, and especially in the application of these simulations to local situations. We can modify locally the parameters of regression curves, and we can change the parameter values of our spatial operators. If this is done, the risk for geographical research is that researchers might merely produce what could be called ad hoc models, which would be significant only as applied and adapted to these local situations. Geographers would then better respond to the demand and needs of local authorities, agencies or firms, but they would hardly contribute to the accumulation and renewal of general knowledge. There is a temptation to accomplish technical feats, as in medicine, but these should be integrated into more general observation protocols (and then include theoretical objectives in their design) if each modelling case is to serve in the development of geographical theory.

23Formalisation is necessary for the diffusion of theory. Let us not forget that a hundred years before Christaller, in 1841, a full theory of central places was enunciated by Jean Reynaud (which Marie-Claire Robic rediscovered in 1982). But without tables of numbers or graphical representation of the already-mentioned hexagons, the theory continued to be ignored. There may be other reasons for this, but part of the success of Christaller central place theory probably lay in the possibility of experimentation and control given by centrality ratios and geometrical models.

24But formalisation is not formalism : in the literature, not so much in geography as at some fringes of the discipline, for example, the "new economic geography", many models, sometimes including a very long list of equations and a full Greek alphabet of parameters, have been written and never applied. To be useful, models must be formulated in such a way as to make experimentation through simulation possible, and to enable models to be applied to different experimental situations. If the definitions and hypotheses are too simplistic, if the parameters cannot be translated into indicators and measurements, the model remains pure speculation and cannot be submitted to validation and confrontation.

25Validation is a necessary step before we can decide that a model is making a new contribution to geographical explanation. But validation is a long and painful process. It takes time. It is easy to understand why a "quantitative geography", or, better, a theoretical geography built according to a "retroductive" method, cannot devote as much time to communicating and establishing the value of its results as can more discursive currents of the discipline. Perhaps quantitative geographers spend too much time in front of their screens and devote too little time for discussion among themselves, and for "selling" what they produce to the outside world.

26So, the isolation of modellers comes partly from outside and partly from within. One question is : how can we improve our internal and external communication about theories and models ?

Communication networks : interaction or fragmentation ?

27Communication networks in geography, particularly in Europe, remain too fragmented. Fortunately, a number of highly effective organisations are doing some important work, especially towards the improvement of all technical aspects of geographical research. Here it is worth mentioning the success of AGILE, following GISDATA. I have heard that after a period of hesitation, as many as eleven French research laboratories are joining the organisation, thanks to the co-ordination provided by Didier Josselin. In this group we have also developed a few opportunities for shared bilateral research. In addition, we have experimented with a number of significant modelling activities, some of them very successful, such as the meeting in Bristol about the most recent applications of GIS to the geography of change or the meeting in London College about cellular automata organised by M. Batty and P. Longley (thanks to Lidia Diappi, the next meeting on this theme will take place in Milan). In France, one could also mention the round of meetings concerned with presenting and comparing geographical models organised over the last two years by the GDR Libergeo, under the direction of Céline Rozenblat and Cécile Tannier.

28But there is still a great contrast on the Net between the very talkative lists devoted to solving technical problems in geomatics, for instance, and the faint echoes coming from time to time in discussion lists dedicated to more general problems of geography. We could also mention the failure of the attempt to build a European network in urban and regional modelling in a framework financed by the European Commission.

29We must react to this situation : my closing points, therefore, will be some concrete suggestions for developing a more regular system for interaction.

Suggestions for a European "research connection"

30A first suggestion is that we could make a better and more proper use of the tool in which we have heavily invested for more than five years now, following our decision taken in Spa, where we created an electronic European journal of geography for communicating about the European contribution to the international state of knowledge in geography. This journal is successful in the sense that it is still alive after five years of existence, and its editorial board is truly international. Its methods for selecting papers and controlling the quality of information that is put on line meet the standards of international publications. But Cybergeo is not yet included in the Social Science Quotation index. The only official objection to its inclusion is that papers published in Cybergeo are not quoted in other international journals. Of course this situation cannot improve as long as people who want to publish in Cybergeo cannot be certain that their paper will be considered an international publication. Such a vicious circle prevents many other European journals among them the most famous, from being quoted in the index. A first attempt to solve this problem could be to take collective action, perhaps with the help of the European Science foundation, to request that Cybergeo be included in the index.

31A second suggestion for developing communication about research within Cybergeo is that each researcher could write one or two pages about his or her line of research and particular focus, attempting to explain the role of this research in the development of a geographical theory, what its objectives are, and the reasons behind the choice of a given methodological approach. Such "research portraits" would summarize the essence of the theoretical thinking of the researcher behind a research project in the long term. Who thinks what in geography? Such a brief description of the researcher’s present—and modifiable—thinking about the research project, its objectives, the chosen method of exploration, the reasons for this choice and the expected results could help people who visit Cybergeo to learn more about theoretical thinking in geography and to understand the fundamental ideas of people who are actually developing such thinking and contributing to this geographical knowledge. As early as December 2001, this "Research connection" will appear as a new headline in Cybergeo.

32A third suggestion, from Sylvie Rimbert, is that we could also use Cybergeo as a basis for preparing European initiatives, in order to achieve a better reception for the projects we present in Brussels.

33Other initiatives may be imagined, as for instance the presentation and discussion of models and or research protocols with reference to theoretical orientations : cognitive geography, multilevel analysis, models for animated maps... all examples of possible themes for which debates are necessary. A discussion list could be joined to these interactive workshops.

34If necessary, we could adopt a name in order to make our informal association more visible. A still virtual but labelled association could help our external communication : I would like to suggest that we call ourselves the European Geographical Organisation (EGO). This would come as an echo to the name of the recently-organised ECU (European Cartographic Union). Of course the idea is to become better connected while remaining informal.


35The purpose of my contribution has been first to reassess the necessity and usefulness of theoretical thinking, and second, to develop it and communicate about it among geographers and towards the outside world. Let us work together to better coordinate our action in research, to develop more synergy among ourselves, so that our knowledge may become truly cumulative and not be destroyed by the erosion of time, so that our metaphoric tower of knowledge does not become a tower of Babel.


Haut de page


Batty M. 1995, New Ways of Looking at Cities. Nature, 377, 574.

Haggett P. 2001, Geography, a global synthesis. Prentice Hall, Pearson Education.

Rimbert S. 1999, A propos de la "key to the door". Cybergeo, discussion to 114.

Unwin D. 1999, Euroquant at 213 : "coming of age"? Cybergeo, 114, 16 p.

Wilson A. 2000, Complex spatial systems. Prentice Hall, Pearson Education.

Haut de page

Pour citer cet article

Référence électronique

Denise Pumain, « Opening Plenary : Communicating about Theoretical Geography », Cybergeo: European Journal of Geography [En ligne], Dossiers, document 200, mis en ligne le 04 décembre 2001, consulté le 23 avril 2024. URL : ; DOI :

Haut de page


Denise Pumain

Professor, University Paris I,

Articles du même auteur

Haut de page

Droits d’auteur


Le texte seul est utilisable sous licence CC BY 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont « Tous droits réservés », sauf mention contraire.

Haut de page
Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search