Version classiqueVersion mobile

Settling the World

Léna Sanders

Chapter 16: Ontological points of view on the transitions of settlement systems

Lena Sanders, CNRS, Thérèse Libourel, Pierre Livet et Denis Phan

Note de l’éditeur

With the participation of Hélène Mathian, CNRS, UMR EVS (Environnement, Ville, Société)

Texte intégral


1The settlement systems studied in the TransMonDyn project concern various periods, spaces and scales. The chapters of Part 2 are devoted to the description and modelling of the transitions which were identified. Each of these ten chapters was co-authored by thematicians and modellers. In an initial section, the thematicians describe, in narrative form, the transition which interests them, following the framework of the harmonised description developed by the TransMonDyn collective (cf. inset 1). Each case study has given rise to a conceptualisation in ‘settlement system’ terms, whereby the elements, objects and relations which make up the system are characterised. The point for the thematician is to distinguish periods of permanence, i.e. those in which the changes observed empirically can be interpreted as simple fluctuations around a central tendency, from periods of more profound change. The first have given rise to an interpretation in terms of ‘regimes’, while the second correspond to transitions between regimes that are qualitatively different (cf. chapter 2). In a second section, each of these chapters presents a model permitting exploration of different hypotheses concerning the mechanisms of change at work within the settlement system under study. These models constitute the material that we analyse in this chapter, adopting an ontological approach (cf. the next section). The purpose is to make explicit the ontologies underlying the conceptualisations developed by the thematicians and the modellers in order to facilitate comparison between the transitions.

  • 1 Walker Brian H., Gunderson Lance H., Kinzig Ann P., Folke Carl, Carpenter Steve R., Schultz Lisen, (...)
  • 2 Cumming Graeme S., Collier John, ‘Change and identity in complex systems’, Ecology and Society, 10( (...)

2The first task consists, for each case, in identifying what objects, relations and processes exert a driving force in the evolution of the settlement system, both in its permanent aspects and in its transformations, in order to determine how the system becomes ‘qualitatively different’ from what preceded it. For certain researchers, a regime change implies a change of functioning such that the system changes its identity1. Indeed, a simple change in the values of the variables describing the settlement system is insufficient to enable the conclusion that the system is ‘qualitatively different’ from what existed previously and that a transition has taken place. One may in fact be dealing with a simple change in state compatible with the functioning of the system. Faced with a settlement system in constant evolution, moreover, Graeme Cumming and John Collier2 put the question in the following terms: ‘how to maintain a constant definition of a changing system?’. The distinction between a state change and a system change is especially difficult to determine in the domain of the human and social sciences (HSS), where the observations and measurements suitable for characterising the system under study are often lacking. In this project, this step of identification and interpretation of changes in terms of regimes and transitions has engendered ‘negotiations’ between thematicians. It is this collective development, combining various points of view, that assures a certain robustness in the distinction between small changes linked to the ‘ordinary’ functioning of the system and changes reflecting a veritable transition to a new regime.

3Once the transitions were identified and described, the challenge consisted in producing formalised models, capable of easily being made operational by means of mathematics and/or computer science, which would then permit exploration of different hypotheses respecting the evolution of the settlement system being studied. Realising this step also depended on numerous exchanges of knowledge and ‘negotiations’, this time between modellers and thematicians (see chapters 14 and 17 for a more thorough discussion of this co-construction and of the passage from narrative to model).

Inset 1: Reminder of the three-stage strategy put in place to identify the transitions for each case study (cf. chapter 2):
1- identifying a regime (
R1) of the settlement system functioning at a given period, with a dynamic tending to reproduce this system while maintaining its structure and functioning;
2- identifying a regime (
R2) of the settlement system qualitatively different from the first, functioning at a later period;
3- defining the transition as the process leading from one regime to the other (
i.e. from R1 to R2).

4The objective of this chapter is to adopt an ontological reflection in order to shed light on the different stages, proceeding from the formulation of the problematic by the thematicians in terms of a settlement system transition to the operational model permitting exploration of the mechanisms underlying the transitions. To begin with, we propose a framework explicitly setting out the different steps in the co-construction of a model in an interdisciplinary context. We demonstrate and illustrate the interest of an ontological approach for coherently connecting these different steps and facilitating comparison of the models developed with respect to the different transitions. Secondly, we propose a systematic comparison of the different models developed in Part 2 (chapters 4 to 13) from three points of view (knowledge engineering, thematics and philosophy).

Definition of a research framework linking the empirical realm to a formal model

  • 3 Phan Denis, ‘Ontologies et modélisation par SMA en SHS’, dans Ontologies et modélisation par SMA en (...)

5To move from empirical enquiry to a formal operational model, it appears essential to distinguish the different realms of knowledge which are involved and which will be combined (figure 1). The analysis relates to a sub-group of observable phenomena from the empirical realm which will be termed the ‘referent’: that is, the object of our research, what it refers to. This term has been chosen by analogy to its usage in linguistics, where the notion of ‘referent’ indicates ‘what we are talking about’. In the same way, ‘un modèle est un système de signes qui renvoie à "quelque chose" de concret dans le monde réel que l’on désignera comme le “référent” du modèle’ (a model is a system of signs which points to ‘something’ concrete in the real world, which will be designated as the ‘referent’ of the model)3. In our case, the referent is the evolution of the spatial forms of human settlement. This referent is represented at the extreme right of figure 1.

Figure 1: From the empirical referent to the formal system

Figure 1: From the empirical referent to the formal system

(the stages of the progression unfold from right to left).

  • 4 Hanson Norwood R., Patterns of Discovery, Cambridge (MA), Cambridge University Press, 1958. Traduct (...)

6A modelling approach consists in passing from this referent, with which the thematic question is concerned, to a formal system by which different conjectures may be explored. An initial step consists in expressing the question within the existing disciplinary conceptual frameworks. In this project, we are dealing with disciplines which produce knowledge and research practices regarding settlement, its spatial organisation and its evolution – namely, geography, history, archaeology and paleo-linguistics. Their conceptual background has led the thematicians to consider units of settlement in interaction as forming a system. ‘Settlement system’ and ‘unit of settlement’ are, therefore, conceptual constructions which characterise what seems important to retain from the observations bearing on the referent in the empirical realm (villages, agglomerations, territories and the forms of social and spatial organisation associated with them). The relation between the empirical domain and the conceptional domain of the thematician is not purely a one-to-one relation, however, because the observations pre-suppose ‘background’ theories which may differ from one discipline to another. Our perceptions, like our experiences, are actually attached to conceptual frameworks which act as a ‘filter’ at the interpretative level: thus ‘facts’ are in part cognitive, theoretical or social constructs. For Norwood Hanson4, for example, the shaping of facts is a necessary moment of discovery, one necessary to their intelligibility. The existence of such theoretical constructs, however, does not suffice to reformulate the question posed in the language of a formal operative system (mathematical, statistical, computational) represented in the ‘model domain’ at the extreme left of figure 1.

7Several intermediate steps are in fact necessary between the referent and the formal system. These imply a composite ‘conceptual domain’ with three conceptual frameworks in interaction – those of the thematicians, the modellers and the philosophers. The remainder of this section is organised in three parts. We first introduce and discuss the notion of ontology and the way in which it is understood in different scientific fields. We then show how specification of the ontology of the thematic domain can facilitate comparison between the transitions. Finally, we concentrate on the dialogue between the conceptual frameworks of the thematicians and the modellers (square in the middle of figure 1) and the passage to the formal model (square on the left-hand side of figure 1), which will permit exploration of different hypotheses concerning the changes in the settlement system being studied.

Ontological approaches

8Every researcher has a conceptual point of view based on an ontology. The differences between these ontologies induce us to employ the notion of ‘regional ontologies’ to designate those which are associated with the conceptualisation of the thematicians and modellers, while the ontology of the philosophers (cf. inset 2) is more generic in nature.

  • 5 Dhondt Urbain, ‘Science suprême et ontologie chez Aristote’, Revue Philosophique de Louvain, 59(61) (...)
  • 6 Livet Pierre, Ogien Ruwen (dir.), L'enquête ontologique. Du mode d´existence des objets sociaux, Pa (...)

Inset 2: The notion of ontology, philosophical points of view
The notion of ontology is more than two millennia old. It corresponds to an extremely large question engaged by the Hellenic philosophers Parmenides, Heraclitus, Gorgias and Aristotle. Ontology is the study of “being as being” (the definition proposed by Aristotle
5), that is, the study of the general properties of all that exists. The term ontology currently extends to a great many definitions. It is useful for what follows in this chapter to explain the philosophers’ points of view on ontology.
The philosopher is primarily interested in the generic categories of fundamental ontology in order to furnish a concise system of description to account for what is supposed to exist in general, whether scientific or ordinary knowledge. From the point of view of the taxonomy of fundamental ontology (termed philosophical for simplicity’s sake), three types of entities are considered as the most basic: substrates (or substances), properties and relations. In this framework, objects are generally considered as substrates defined by properties and processes as temporal relations.
Pierre Livet and Ruwen Ogien
6 have proposed an ontology which seems to them more suitable for the social sciences than classical ontology (that of substances and properties). This is an ontology of processes, running from the ‘degré zéro des processus’ (degree zero of processes) – that is, from processes occasioning no change to the temporal scale considered (for instance, the process of erosion of a granite column on the scale of a human life) – to transformative processes (the same process but considered on large time scales), whether these result from endogenous factors (for example, a process of rotting) or from interactions (destruction effected by an agent, for example).

  • 7 Smith Barry, ‘Ontology’, in Blackwell Guide to the Philosophy of Computing and Information, ed. L. (...)
  • 8 Smith Barry, ‘An introduction to ontology’, dans The Ontology of Fields, ed. D. Peuquet, B. Smith a (...)
  • 9 Gruber Thomas R., ‘A translation approach to portable ontologies’, Knowledge Acquisition, 5(2), 199 (...)
  • 10 Uschold Mike, Gruninger Michael, ‘Ontologies : Principles, methods and applications’, The Knowledge (...)
  • 11 Gruber T., ‘A translation approach’ … op. cit.

9Barry Smith, philosopher, specialist in applied ontology and founder of the ‘National Center for Ontological Research’, defines ontology as ‘the science of what is, of the kinds and structures of objects, properties, events, processes and relations in every area of reality'…/… [the science] of what might exist)7. Notable is that conditional added at the end of the sentence, which thus refers to a given conceptualisation of the world such as it ‘might be’, while in computer science the objective of ontology is to describe the world in its ‘reality’8. Modellers most often bring to bear the definitions proposed by specialists in knowledge engineering. In this field, Thomas Gruber9 defines an ontology as ‘an explicit specification of a conceptualisation’. For him, it is a description of the concepts and relations which ‘exist for an agent or a community of agents’. In this sense, ontology is a shared understanding of a field of interest10. This vision, stemming from knowledge engineering, is coherent with that of philosophy to the extent that the ‘specification’ evoked in knowledge engineering consists in spelling out the properties and relations evoked in the philosopher’s definition. While the latter refers to the ‘things’ of the world, the definition of Gruber situates itself within the framework of knowledge-based systems. That which ‘is’, that which ‘exists’, more specifically ‘is exactly that which can be represented’11.

  • 12 Livet Pierre, Sanders Lena, ‘Le “test ontologique”: un outil de médiation pour la modélisation agen (...)
  • 13 Phan Denis, Müller Jean-Pierre, Sibertin Blanc Christophe, Ferber Jacques, Livet Pierre, ‘Introduct (...)
  • 14 Mathian Hélène, Sanders Lena, Objets géographiques et processus de changement. Approches spatio-tem (...)

10Ontology thus represents a framework of explicit description, aimed at eliminating ambiguities when characterising a phenomenon. The notion can help to render coherent the different blocks represented in the ‘conceptual domain’ of figure 1. In the following section, we examine how an ontological approach can be used to link the ‘referent’ framework to the ‘conceptual framework of the thematicians’. We illustrate this on the basis of the transition studies realised in this volume. In order to do this, we will adopt a definition which is situated at the intersection of the approaches of the philosopher and the modeller, and which is appropriate for this chapter: ontology consists in analysing a domain by identifying the pertinent entities (objects, properties, relations, events and processes) and the operations which can be applied to these entities12 1314. These operations make it possible to render operational the characterisation of the different entities and the functioning of the relations which connect them and which lead to change (such as, for example, the equations and rules governing the behaviour of the entities studied, whether individuals or spatial entities).

From the ‘referent’ to the conceptual framework of the thematicians: reflection on the ontological choices (explicit or implicit) made to model the different transitions

11This last definition, consisting in spelling out the pertinent entities and the operations characterising their functioning, can be applied to the ten cases of transitions which constitute the empirical corpus of this volume. The five types of entities figuring in the above definition can therefore be listed for the transitions of the TransMonDyn project:

  • 15 We should emphasise that there is no consensus in the literature on the difference between entity a (...)
  • 16 See table 1 of chapter 2 for a presentation of all the transitions studied.

12The objects15 introduced in the conceptualisation of the transitions belong to different scales and different types, social as well as spatial. The social objects are individuals, households (for instance, the peasant households of transition 816, the Pueblo households of transition 3), groups (the hunter-gatherers of transitions 1 and 2), ‘macro-actors’ (the Roman or Gallic elites for transition 6, the lords for transition 8, the Pueblo leaders for transition 3), and finally the central power (Roman, royal). The spatial objects are portions of space (transitions 1, 2 et 8), villages (transitions 3, 7 and 8), agglomerations (transitions 5, 6 and 7), cities (transitions 4, 9, 10 and 12), ‘mega-cities’ (transition 12), nodes of communication (transition 9), and finally, at a higher level of organisation, networks and systems of cities (transitions 4, 9 and 10), as well as regional systems (transitions 3, 7 and 8). These varied objects are characteristic of the settlement systems and the choice of them to describe a transition reflects the hypotheses of the work-groups concerning the key elements of change. Simply enumerating the objects introduced provides a glimpse, in fact, of the level at which the motive force of the system operates.

  1. The properties are characteristics of these objects. With regard to empirical practice, they refer to the ‘attributes’ which may stem from counting (number of households in a village, number of cities in a region) ou measurement (accessibility of a city), or they may be qualitative (a group of hunter-gatherers is Bantu or forest forager, a given elite is Roman or Gallic).

  2. The relations between these entities are of multiple kinds. Some serve to connect objects which may be of different types, for example, when ‘group’ objects take resources from ‘portions of space’ objects or a ‘peasant household’ object transfers a tax to a ‘lord’ object. Other relations concern objects of the same type, for example, the relations of cooperation between Roman and Gallic elites or the relations of proximity and exchanges (flows of migration, information, resources) between two cities or between two groups of hunter-gatherers.

  3. The events characterise the changes which take place in a time-frame interpreted as instantaneous. On the macro-geographic level, this is the case, for example, with the appearance, at a given moment, of a context of violence and of a religious obligation of exogenous origin (transition 8). Another case concerns the influence of exogenous communities with which exchanges start to develop from a given point in time (transitions 5 et 6). This is also the case with climatic perturbations which affect a territory. Such perturbations are either reconstructed on the basis of estimations of past climatic data (the case of transition 3) or simulated in a random manner (the case of transition 1). Other events, by contrast, are endogenous, produced by the model, such as, for instance, the emergence of a new lord (transition 8).

  4. The processes refer to the changes of the entities (objects, properties, relations) over time and thereby raise the question of the factors involved in the changes. Some transitions lead to the emergence of new objects – for instance, political communities and leaders (transition 3), cities, systems of cities and ‘mega-cities’ (transitions 4, 9 and 12). Others see new properties of the settlement system emerging, observed at a macro-geographic level, with, for example, a concentration of habitat (transitions 5 et 8) or a hierarchic organisation taking place (transition 4). Finally, other transitions are founded on processes of change affecting the relations between objects. Such is the case, for instance, with the relations of dependence between the Bantu and forest forager groups in transition 2, or between the settlements at different hierarchic levels in transition 5. A change in relation may also be expressed at a more generic level, for instance, with reference to the increase in the distance travelled in an hour or a day (transitions 12 and 9, respectively).

13The operations bearing on the totality of these entities specify the way in which they are connected and in which they evolve. Thus, the principal operations consist of the mathematical functions and/or rules which enable a change to be linked to an event or to generate a series of successive changes. This is the case, for example, with the equation that makes it possible to evaluate the lack of satisfaction of the peasant household in transition 8 or the preference of a Bantu agent-group when making mobility choices in transition 2. Such specifications belong to the formal model, but they must also have been already considered and anticipated during the ‘dialogue’ between the conceptual frameworks of the thematician and the modeller (figure 1).

14For each transition, it has been necessary to identify the elementary objects essential to accounting for the change. From a systemic perspective, at issue are the objects whose interactions supply the driving force for change. The choices depend at once on the characteristics of the settlement system under study, of the epistemological positions of the researchers, and of the levels at which their hypotheses concerning the driving force of the changes are played out. Thus, a posture of methodological individualism leads to a consideration of individuals as elementary objects (as in transitions 3, 5, 6, 8 and 12, for example), while a more holistic position will lead to considering objects at a meso- or macro-geographical level (the case notably of transitions 1, 2, 4, 7 and 9). In the face of the same empirical question, moreover, several ontological choices are possible, and the discussion between thematicians and modellers in this choice is stimulating. Some examples taken from the transitions modelled in this volume make it possible to illustrate this:

15- Example 1: Transition 2 deals with the influence exerted by the interactions between Bantu and forest foragers on the form and the rhythm of the formers’ expansion towards the south (chapter 5). The objects of the model represent groups of hunter-gatherers. Bantu and forest foragers had different capacities with respect to exploitation of the environmental resources and different forms of behaviour with respect to moves, and the authors had the choice between two possibilities of formalisation. One initial possibility would have consisted in introducing two social objects of different kinds (Bantu groups on the one hand, forest forager groups on the other), each being characterised by its distinct functioning with regard to exploitation of resources and mobility, and by a specific structure of interactions (possibility of diffusing a technical innovation in one case and not in the other, for example) (figure 2, left). A second possibility consisted in considering only a single type of object, made up of groups endowed with the same structure of functioning, the distinction between the two types of groups being figured only by way of their properties (figure 2, right). The authors chose the latter solution. It enabled them to adopt a higher level of generality and to align themselves with the treatment of agent-groups as they had been formalised for transition 1 (chapter 4). By contrast, the first solution would have made it possible to reduce the number of parameters brought to bear in the model. The problematic privileges, in fact, the Bantu point of view, as the objective is to model their expansion towards the south. The behaviour of the forest forager groups was therefore much simplified, the authors retaining only the elements capable of influencing the survival and the moves of the Bantu. As a result, it was necessary to ‘neutralise’ certain properties, behaviour and relations of the forest forager groups which did not concern them (technical level, accumulation, energy, adoption of an innovation) with the aid of the parameters associated with these properties and forms of behaviour. If the problematic had required more precise formalisation of the functioning of the forest forager groups and this functioning were supposed to be qualitatively different from that of the Bantu groups, an ontology founded on two different types of objects would have been more appropriate.

Figure 2: Schemas characterising two different ontological choices (UML formalism).

Figure 2: Schemas characterising two different ontological choices (UML formalism).

16- Example 2: transition 7 (chapter 10) deals with the evolution of the organisation of space under the impact of the co-existence of different powers. It characterises the passage from the regime of Antiquity to the medieval regime in Septimania (Languedoc-Roussillon in southern France). The aristocracy plays an essential role in this transition; it represents the elite within which power is installed, whether Germanic power or religious power. Two ontological choices were possible in formalising the role of the aristocrats. The first, privileging the actors as a point of entry, would have consisted in formalising individual aristocrats as objects and localising them in the urban and rural agglomerations. The second privileges places as an entry point, as objects having the property of ‘possessing’ aristocrats. Because historical knowledge does not permit characterising the properties of the aristocrats and the processes associated with their decisions, it was the second point of view that was applied. Indeed, it was possible for the thematician who specialised in the field to evaluate the number of aristocrats present on a spatial entity. Such a choice made it possible to conduct analysis on the scale of the objects ‘urban agglomerations’ and ‘rural entities’, and to refrain from defining the choices of localisation at the level of the actors themselves (i.e. the individual aristocrats). Since attaining a certain threshold was a condition sine qua non for power to become installed in a given place, it was sufficient subsequently to formalise how the number of aristocrats evolved in each spatial entity in the course of time.

  • 17 Mathian H., Objets géographiques…, op. cit.
  • 18 Garmy Pierre, Villes Réseaux et Systèmes de villes, Paris, Errance, 2012.

17- Example 3: for a settlement system including the totality of settlement entities, from the village to the city, two ontologies may be envisaged: 1) one single object type, the unit of settlement, the city/village distinction then being perceived as a ‘property’ of that object; 2) two object types ontologically different, the villages and the cities. The choice of one or the other of these ontologies depends on the interpretation of the differences between the objects ‘city’ and ‘village’ and the rules which one wishes to implement concerning the processes transforming one into the other17. If one considers that there is a semantic continuity between villages and cities, and that they may be characterised by the same collections of properties and relations, only with different values, then the first choice is preferable. Such is the case with the modelling of the emergence of cities from an initial situation in which there existed only agricultural villages with minimal differentiation (transition 4). This is also the case with transition 8, where the authors chose to use the term ‘aggregate’ in order to signify their neutrality with regard to the classic concepts of village or agglomeration and to stress only the intensity of the concentration of peasant households. If, on the other hand, the hypothesis is that the properties and relations associated with the village are qualitatively different from those associated with the city, the second choice is called for18. This was the case with transition 7, in which the rural entities were differentiated from the urban entities. The former participates in the dynamic of the territory and determine the implantation of churches but do not intervene in the localisation of civic powers. The latter do.

18These few examples illustrate the interest of explaining the different possible ontological choices and specifying the reasons leading to the choices made. These reasons related to the hypotheses which the thematicians wished to test concerning the dynamic of the settlement system being studied. To return to the terms of figure 1, this step has everything to gain from a conceptual co-construction of the model at the interface between the conceptual frameworks of the thematician and the modeller.

Strategies of co-construction of the models

19Between the conceptual model of the thematicians and the operational model, a number of steps remain to be traversed (figure 1). Specification of the conceptual modelling framework allows for elimination of the possible incompatibilities between the conceptual model developed by the thematicians and the formal system used to render the model operational. In fact, the ‘conceptual framework of the modellers’ makes use of concepts more generic than those of the disciplinary thematic frameworks, which are more equivocal. It makes it possible to remove the ambiguities which might introduce uncertainty into the construction of the formal system and make it ineffective. Indeed, the principal object of the conceptual framework of the modellers is to study the properties of generic modelling concepts as ‘systems’, agents, relations, etc., regardless of their applications.

  • 19 Phan D., ‘Ontologies et modélisation par SMA … ‘, op. cit., p.66

20It is useful, especially in an interdisciplinary context, to be precise about what is meant by ‘model’, so as to avoid any misunderstanding between thematicians and modellers. The notion is in fact polysemic. The initial significance of the word ‘model’ derives from the Italian modello which in the sixteenth century designated in the domain of fine art, something intended to be reproduced (see chapter 14). It is interesting to extend this idea of reproduction, in more abstract fashion, to those properties between which one can establish a ‘relation of similarity’. Such a relation forms a connection between what is represented (which was designated above as the ‘referent’) and the representation of that ‘referent’. Thus, the model is an abstraction constructed with reference to ‘something’ which exists outside it. Moreover, the nature of the model is linked to the language in which it is expressed. Since natural language is rich in ambiguities, for modelling a more precise formalism is necessary. This formalism may be expressed in a purely conceptual manner, as, for instance, by means of an arrow diagram. It may also be expressed by a ‘formal language’ with which syntactic rules are associated. Such an approach takes place in the ‘formal system’ (on the left of figure 1). In this system, the properties may be explored independent of semantics. On the basis of these elements, the definition of a model for scientific exploration can be formulated as follows: ‘Un modèle est une représentation exprimée dans un langage formel, qui porte sur un domaine de référence déterminé, relatif à des phénomènes observés, en relation avec un point de vue finalisé et contextualisé relativement à un cadre conceptuel qui lui donne son sens.’ 19(A model is a representation, expressed in a formal language, bearing on a determined field of reference connected with observed phenomena and reflecting a point of view oriented towards a goal and contextualised in relation to a conceptual framework which imparts its meaning.). This definition connects all of the elements represented in figure 1.

21Every modelling approach, then, is oriented towards a goal, in this case the bringing to light of ‘transitions’ between regimes of different settlement systems and the understanding of the processes underlying them. The models are therefore influenced by this objective, but also by other contextual elements, such as, for example, the theoretical and methodological frames of reference of the academic domains to which the thematicians and the modellers who co-construct the model belong. Some comparisons between pairs of transitions will illustrate the consequences of these different influences for the modelling choices made:

  • 20 There is, in fact, a hiatus, which is explained in chapter 17.

22The transitions 6 (process of Romanisation) and 7 (passage from the world of Antiquity to the medieval world) follow each other in time (almost20) and deal with the same space (corresponding to a part of present-day Languedoc-Roussillon). The different disciplinary bases of the modellers associated with the study of these transitions have nevertheless led to different formalisations. Game theory has been used in the case of transition 6 (chapter 9) to explore the effects of different strategies of cooperation or conflict between Rome and the Gallic elites to dominate the conquered Gallic territory. Transition 7 (chapter 10), for its part, has given rise to the proposition of a conceptual model of the agent type, in which the games of power are perceived on the basis of places.

  • 21 The objective of the model of transition 8 was to analyse the spatial organisation stemming from th (...)

23Transitions 7 and 8 are concerned with different spaces but also follow each other in time. Both are related to processes of polarisation and territorialisation and put in play the competition between different powers. These transitions were conceptualised according to different conceptual backgrounds corresponding to particular relations between thematicians and modellers. In transition 7, the existence of a certain conceptual parallel between the historian’s treatment of the hierarchy of powers and the geographer-modeller’s treatment of the hierarchy of places led this team to adopt place as a point of entry for modelling the transition. The team for transition 821, by contrast, adopted a conceptual framework centred on the individual choices of the lords and small farmers, and for this reason committed itself to defining behaviour at the individual level. In the one case, the choice was made to analyse places as vectors of power, the interactions between these places constituting a driving force in the dynamic of the system; in the other case, the places represent the sites of the individual actions which are considered as the driving forces of change.

  • The models of transitions 3 et 4, devoted respectively to understanding the emergence of villages and cities and, at a higher level of organisation, of communities of villages directed by the same leader and of systems of cities in interaction, are both formalised using multi-agent systems (MAS). On the other hand, there is a difference between the hypotheses regarding the level at which the interactions exert their motive force. In the case of transition 3 (chapter 6), the focus is on households which exchange and which are prepared to pay taxes to a leader to manage equality of access to resources. In transition 4 (chapter 7), by contrast, the decisions of the actors are not explicitly taken into account. The exchanges of information between cities are considered at a collective level, and their intensity depends directly on the hierarchic and spatial structure of the system of cities.

  • 22 This term designates an urban region made up of a collection of metropolises and cities geographica (...)

24The link between the emergence of new urban entities and the communication network is at the centre of transitions 9 and 12. The focus is on the networks of cities in the nineteenth century and of ‘mega-cities’22 in the transition underway in our time. Despite this thematic convergence, the methodological approaches differ fundamentally. In the first case (chapter 12), the approach is statistical and based on the data on the size of cities and the time needed for travelling between them by postal routes. Graphs were constructed and analysed to filter the data, a technique which permitted precise identification of the start of a network already functioning in the mid-eighteenth century. In the case of transition 12 (chapter 13), the transport infrastructures of an imaginary city were simulated, using a multi-agent system, by way of the decisions of mayor-agents. The objective was to explore the consequences of a number of scenarios relating to an increase in energy costs. These two approaches are perfectly complementary, the statistical approach making good use of data obtained by means of different sources, the simulation model exploring hypotheses regarding the functioning of a theoretical case. In the latter, the model is founded on expert knowledge rather than on data and is used to explore the possible effects of different transport policies.

25These examples illustrate to what extent the process of modelling is guided by the modelling tools privileged by the modeller for dealing with a problem. A permutation of the modellers intervening in the different transitions would thus have resulted in different models. The project includes cases where the methodologies for dealing with relatively similar problematics differ (examples are transitions 6 and 7, on the one hand, and 7 and 8, on the other, as presented above). On the other hand, there are cases of methodological continuity with different narratives. This is the case, for instance, with transitions 1 and 2, concerned with thematics separated by tens of thousands of years, but where the same collective of authors mobilized the same family of models.

26All the constitutive elements of figure 1 have now been expounded with reference to the studies effected within the TransMonDyn project. The ‘referent’ concerns the same object, the settlement system, which is placed in a variety of spatio-temporal frameworks. The ‘conceptual framework of the thematicians’ was negotiated between the thematicians involved, all basing themselves on a particular ‘regional ontology’ of reference involving knowledge relating to their fields of competence (archaeology, geography, history and paleo-linguistics). From this emerged harmonised criteria for describing the transitions (summarized in figure 4 of chapter 2) and a common point of departure for the diverse thematic cases. On their side, the modellers likewise mastered varied methodological fields – agent-based modelling, game theory or the theory of graphs, for example. From that point (moving from right to left on figure 1), a number of forms of co-constructions involving thematicians and modellers were developed, which issued in varied transition models.

Different ontological points of view for comparing the transition models of the TransMonDyn project

27The authors of this chapter, by virtue of their different conceptual ‘backgrounds’, have varied points of view and practices with respect to ontology. Accordingly, we propose in this final section successively to present three comparative views of the transition models.

  • 23 UML (Unified Modelling Language) is a modelling language with a standardised notation for construct (...)

28The first privileges the computer science point of view and introduces a UML23 formalism to represent, in a synthetic manner, the elements which are present in the structure of all the models and which constitute their skeleton (figure 3). It expresses the fact that a social object, whether individual, community or household, has a social functioning governed by a form of power. This social object resides in a habitat situated in a space constituted by zones serviced by networks of diverse kinds. All of these elements appear in the harmonised description of the transitions, as well as in their modelling.

Figure 3: UML representation of the structural elements common to the different transitions studied (landscape format)

Figure 3: UML representation of the structural elements common to the different transitions studied (landscape format)
  • 24 This term is introduced in the treatment of transition 3. It designates the association of a number (...)

29A second point of view is proposed by the thematicians. They chose to carry out the comparison by systematically setting out the objects, properties and relations deployed in the different models (table 1). Each transition is characterised by the emergence of a new system associated with the appearance of new entities (new objects such as ‘polities24’, networks of cities, ‘mega-cities’ or new centres of power). Two types of factors may trigger this emergence (cf. chapter 1) – exogenous perturbations (environmental or political) or interactions endogenous to the initial system. In order to take account of the diversity of cases in the comparison, we have synthesised the models of the transitions on the basis of five items (table 1):

  • The elementary objects of the model, i.e. those whose interactions are taken to be driving forces in the transition from one system to another (cf. inset 1);

  • The relations driving change: these are the relations which govern the dynamic of the system. According to the hypotheses, they come into play between households, between groups (the domination of the Bantu over the forest foragers allows the former, for example, to have access to more resources), between settlement units (for example, exchanges between cities foster the growth of some to the detriment of others, which decline), or indeed between entities of different kinds (a lord levying a tax on a peasant household, a mayor constructing a highway).

  • Triggering mechanisms: certain mechanisms are considered in the model as exogenous, whether environmental perturbations, the installation of a new power or the arrival of a migratory flow. Other mechanisms are endogenous (crossing of a critical threshold, amplification of a fluctuation).

  • The spatio-temporal dimension of the model outputs: most of the models produce results in the form of indicators summarising the characteristics of the settlement systems modelled, notably in terms of space and time. These indicators notably take account of the spatial and hierarchical organisation of the distribution of population.

  • Type of modelling: as has been expounded at length, the models of the different transitions are various. Most of the models were implemented and have given rise to experimentation. With regard to ontology, few have issued in a formal ontological formalisation (expressed in UML language), but most have passed through a relatively rigorous conceptualisation in terms of objects, properties and relations. The term ‘discursive ontology’ has been used to designate the latter approach.

Table 1: Comparative synthesis of the 10 transition models developed in Part 2.


Elementary objects

Relations driving change

Triggering mechanisms (exogenous and/or endogenous)

Spatio-temporal dimension of model outputs

Type of modelling

1- Sapiens peopling of the World

Groups of hunter-gatherers

-Search for resources;

-diffusion of innovations

-Exogenous environmental perturbation (exo)

-Arrival of groups from the exterior (exo)

Size and spatial extent of the colonisation

-Discursive ontology

- MAS model implemented

2- Bantu expansion

Groups of hunter-gatherers and ‘hunter-farmers’

-Search for resources

-Interactions between forest foragers and Bantu

-Arrival of groups from the outside  (exo);

-Relations of exploitation (end)

Extent of the migratory expansion within a constrained space (obstacle of the forest)

-Discursive ontology

- MAS model implemented

3-Emergence of Pueblo villages


-Access to resources and protection

-reciprocal exchanges between households

-competition between ‘villages’

-Environmental perturbation (exo)

-Conflict between ‘villages’ for resources (end)

-Emergence of ‘villages’ with or without leader;

-Emergence of ‘complex groups’ (several villages)

-MAS model implemented incorporating a rule from game theory

4-Emergence of cities

Units of settlement

Interactions between units of settlement according to their sizes

- Diffusion of innovations (end)

Emergence of a rank-size organisation of the settlement units

-MAS model implemented

5-Habitat concentra-tion in the Iron Age

Units of settlement


-Diversification of ‘occupations’

-Warrior hierarchy

-Arrival of exogenous flows

-Foundation of Massilia (exo)

-Concentration of habitat

-Model conceptualised in UML

- Netlogo implementa-tion of a MAS model

- Platform Metaciv


Collective actors

Strategies of cooperation / confrontation between Rome and Gallic elites

‘Game’ between two ‘competing’ collective actors (end)

A-spatial model (different configurations of power relations for administration of territories)

Game theory

7-Passage from Antiquity to the Middle Ages

Spatial entities (agglomerations)

Competition between powers

-Appearance of a new power (ecclesiastic)


-Hierarchisation of settlements

-Spatial imbrication of different powers

- Discursive ontology

- conceptual model

8- Emergence of concentrated and hierarchised settlement patterns

-Peasant households


-Parish churches

-‘Dissatisfaction ‘of peasant households

-Power relation between lords and peasant households (fees)

-Decline of the central authority (exo)

-New religious obligations (exo)

-Concentration of the population

-Emergence of networks of feudal obligations

-Coverage of the territory (parishes)

-Spatial tangle of different powers

-Discursive ontology

- MAS model  with 2 powers in a single space

9-Emergence of integrated urban systems

-Postal relays and cities

- ‘Journeys’ by road between postal relays

-Co-evolution of networks of communication and networks of cities

-distance between relays

-Development of the road network (service, speed)

-Growth of large-scale maritime trade

-fine scale paving of the territory

-Reinforcement of mid-size cities

-emergence of integrated urban systems

-Emergence of nested networks

-Indicators of network structure (theory of graphs)

-Confrontation with the sources

12-Emergence of polycentric ‘Mega-cities’

- Micro-economic agents

- Mesoscale agents (mayors)


-Dynamic balance between transport supply and demand

-Construction of transport infrastructures (end)

-increase in transportation costs (exo)

Emergence of centralised governance at the level of new ‘mega-city’ functional entities

SMA coupled with a

LUTI model (Land Use Transport Interaction)

NB : in column 4, ‘exo’ designates exogenous and ‘end’ endogenous.

30A third comparative point of view originates with the team’s philosopher, who also bases himself on an ontological analysis to compare the transitions according to the kinds of differences between the regimes preceding and following them. If the spatial dimension has been affirmed from the beginning of the project, certain other properties have emerged in the course of the comparative studies of the different transitions. Such is notably the case with the relations of power or control (power indicates domination over the activities of other persons, whereas the notion of control can apply not only to activities but also to material or symbolic arrangements). It is therefore interesting to parallel the spatial dimension of the transition (considered a priori) with the relations of control (emergent during the comparison). A ‘control relation’ implies that the properties and relations between the entities subject to control depend on processes operating at the level of the controlling entities. The entities present in transition 8 will serve to illustrate these statements: the lord represents a controlling entity and the peasant household one under control. The objective of the former is to obtain a tax from the latter, and to attain that objective, he imposes a levy. These relations of control may be related to the spatio-temporal characteristics of the regimes preceding and following each transition. To do this, we will introduce the concept of a spatio-temporal span, which defines the extent and the duration over which the influence of an entity (agent, institution, group, spatial entity) is exercised.

31For most of the transitions, it is possible to discern an evolution from one regime to the other of the spatio-temporal span which the entities can control. For instance, in transition 1, as in transition 2 (chapters 4 and 5), this span increases between regime 1 and regime 2, passing from the scale of Africa to that of the world in the first case, and from northern Cameroon to all of southern Africa in the second. The passage to agriculture (transition 2) also implies temporalities different to those of hunting and gathering, the delays between seeding and harvesting being longer than those associated with a hunt or with foraging. One may therefore pose the question of whether this relation of power and control can simply be reduced to its spatio-temporal projection: an agent has more power and control when it can control a more extensive spatio-temporal span. But some portrayals of transition show us that such a reduction is not systematic. Thus, in transition 5 (chapter 8) this span does not change, but the population is more concentrated, a situation which enables the leaders to manage and control a larger population. Finally, in other cases, there is a diminution of this span, as with transition 7 (chapter 10), where the appearance of local religious powers entails a spatial fragmentation. Although this is not necessarily incompatible with stronger local control, it appears that the relation between control and spatio-temporal span is not straightforward. This is still more evident in transition 8 (chapter 11). In the course of the transition, competitions emerge between different powers (feudal and ecclesiastic) corresponding to the varied divisions of the territory. In the second regime, one thus arrives at superpositions of different powers on a given territory, with relations of vassalage between the feudal ones and a heaping up of fees for the local populations. In these cases, therefore, the same span of territory sees a superposition of different controls and powers. Comparison of the spatial characteristics of the settlement systems associated with different regimes implies taking into account these superpositions of powers over the same territory. If one cannot reduce power to its spatio-temporal projection, one can utilise the latter as a good indicator of the specificity of the power under consideration. In other words, the differences in the spatio-temporal span associated with one power relative to that of another enable characterisation of its specificity.

32To take account of the changes from one regime to the other, a suitable ontological approach should therefore consist in the identification and specification of two systems: 1) the system of space-time-population relations (SSTP), a notion which differs in its definition of the settlement system, in that it specifies the temporal factors associated with the relations between entities; 2) a system of control (SC) which could be revealed for each transition. The point would be to identify the nodes and the links of these systems of relations and the processes which may modify them, resulting in different transformations of the SSTP being controlled, hence of the spatial distribution of the associated population.

33For example, the settlement system associated with transition 2 puts into play a confrontation between two SSTP systems in an interdependence of two populations: the system of the Bantu (more localised, sedentary, with a more extensive temporal span linked to the agricultural cycle and the storage of products), which succeeds in controlling that of the forest foragers (more dispersed, dependent on fruits and hunting, with very limited storage). Transition 3 shows the transformation of the SSTP on a single territory by a change in the mode of control (emergence of a leader). Transition 5 begins with a diversification of craft activity and exchanges, which implies an SSTP within a network and capacities for control in the nodes of the networks involved. Transition 4 implies the constitution as an SSTP of a network of cities, with an extension and hierarchisation of the capacities of control of each city. Transition 6 extends its SSTP at the same time as levels of control are added, entailing the emergence of a new settlement system. We have already stressed that transitions 7 and 8 each involve modifications (competitions, redistributions) of the SC. In transition 9 and its sub-transitions, one observes modifications in the inventions but also in the technical limitations and the systems of control associated with the domains of transportation. The dominant importance of transition 11’s port cities (cf. inset 2 of chapter 2) shows that it is the global modes of mass transportation that dominate: the element that permits a larger extension of the SSTP is also the factor to be controlled. The ‘mega-cities’ of transition 12 (chapter 13) are networks in themselves, combining the features identified for transition 4 but also controls linked to the specialisations of the components of these ‘mega-cities’.

34Passage from the first to the second regime often goes together with a complexification of these two systems of relations, SSTP and SC. Thus, Roman colonisation implies a significant increase in its spatial and temporal span which imposes one or several layers of control and supplementary power on that of the Gallic leaders and elites – which it recruits, moreover, in its organisation of control. Transition 7 corresponds to a still more complex design, and in transition 8 there is interference between the structures of control, with the combination of a number of SC which are imbricated and superposed within the space.


35The activity of modelling entails a number of steps, which have been represented in figure 1. Their linkage has served as the through-line of this chapter. If a researcher (thematician-modeller or modeller-thematician) were alone in the project and consequently took on all the roles, the progression from the ‘referent’ to the operational model would have every chance of being rapid. Imposing coherence on the different elements of the ‘conceptual domain’ would then be no challenge, but at the same time, there would be less chance that an innovative approach would emerge. The context of the TransMonDyn project corresponds, on the contrary, to a collaboration implicating numerous participants with diverse disciplinary areas of competence. The strong interdisciplinarity of the project has necessitated ‘negotiations’ at several levels: 1) between thematicians, first of all, in an initial phase of conceptualisation, which issued in a harmonised description of the transitions as a whole (cf. chapter 2); then 2) between thematicians and modellers, this time in work-groups which gave rise to the models presented in the chapters of Part 2.

36This change of scale in the collaborative enterprise, first implicating the collective as a whole around the concept of transition, then a specific work-group around each transition model, explains why the thematic conceptualisations are harmonised but the models are varied. Indeed, the decision was made not to proceed to a ‘negotiation’ between modellers so as to arrive at the development of a single model for the different transitions. On the contrary, it was the interaction between the thematicians and the modellers (most often a single one but sometimes two or three) of each sub-group which was the driving force in the construction of the model. The choice of method of each sub-group depended on the type of information available (access to data or to expert knowledge) and on the attraction to data processing and/or simulation of stylised facts. This approach gave a great deal of scope for exchanges within the ‘conceptual domain’ of figure 1. Starting from the same base, a diversity of ways of modelling was thus explored. The ontological approach, by virtue of the level of specification it implies, played a mediating role and facilitated the exchanges between thematicians and modellers.

37In terms of knowledge construction, this approach has made a double contribution: 1) the models developed make it possible to filter the data and to explore different hypotheses regarding the interactions at work within settlement systems; 2) the confrontation of points of view has enabled new light to be shed on phenomena which had already been much studied.

  • 25 Even the chapters which have been composed by more limited collectives (transitions 3, 4 and 12) ha (...)

38This diversity of approaches moving outward from a common base lends full meaning to comparison a posteriori, a question to which the chapters of this third part are devoted. An ontological framework, even one developed a posteriori, facilitates inter-group dialogue and comparison of the models of the different transitions. No single approach has emerged as more productive of knowledge than the others; rather, comparison has brought to light the complementary aspects of the different approaches. Most of the transitions have given rise to a veritable ‘encounter’ and to the establishment of new collaborations25.

39One further step might consist in changing the composition of the teams – something which would give rise to new negotiations, new models, and furnish a new basis for comparison…in order to pursue the reflexion on the driving-forces of the great changes in settlement systems.


1 Walker Brian H., Gunderson Lance H., Kinzig Ann P., Folke Carl, Carpenter Steve R., Schultz Lisen, ‘A handful of heuristics and some propositions for understanding resilience in social-ecological systems’, Ecology and Society, 11(1), 13, 2006, [Online] URL:

2 Cumming Graeme S., Collier John, ‘Change and identity in complex systems’, Ecology and Society, 10(1), 2005.

3 Phan Denis, ‘Ontologies et modélisation par SMA en SHS’, dans Ontologies et modélisation par SMA en SHS, éd. D. Phan, Paris, Hermes-Lavoisier, 2014, p. 55.

4 Hanson Norwood R., Patterns of Discovery, Cambridge (MA), Cambridge University Press, 1958. Traduction française, Les Modèles de la découverte, Dianoia, 2000, p. 3.

5 Dhondt Urbain, ‘Science suprême et ontologie chez Aristote’, Revue Philosophique de Louvain, 59(61), 1961, pp. 5-30.

6 Livet Pierre, Ogien Ruwen (dir.), L'enquête ontologique. Du mode d´existence des objets sociaux, Paris, Editions EHESS, coll. Raisons pratiques, 2000.

7 Smith Barry, ‘Ontology’, in Blackwell Guide to the Philosophy of Computing and Information, ed. L. Floridi, Oxford, Blackwell, 2003, pp. 155–166.

8 Smith Barry, ‘An introduction to ontology’, dans The Ontology of Fields, ed. D. Peuquet, B. Smith and B. Brogaard, Report of a specialist meeting held under the auspices of the Varenius Project, Panel on Computational Implementations of Geographic Concepts, NCGIA, Bar Harbour, Maine, 1998.

9 Gruber Thomas R., ‘A translation approach to portable ontologies’, Knowledge Acquisition, 5(2), 1993, pp. 199-220.

10 Uschold Mike, Gruninger Michael, ‘Ontologies : Principles, methods and applications’, The Knowledge Engineering Review, 11(2), 1996, pp. 93-136.

11 Gruber T., ‘A translation approach’ … op. cit.

12 Livet Pierre, Sanders Lena, ‘Le “test ontologique”: un outil de médiation pour la modélisation agent’, in Ontologies et modélisation par SMA en SHS, ed. D. Phan, Paris, Hermes-Lavoisier, 2014, pp. 95-110.

13 Phan Denis, Müller Jean-Pierre, Sibertin Blanc Christophe, Ferber Jacques, Livet Pierre, ‘Introduction à la modélisation par SMA en SHS: comment fait-on une ontologie?’ in Ontologies et modélisation par SMA en SHS, ed. D. Phan, Paris, Hermes-Lavoisier, 2014, pp. 21-52.

14 Mathian Hélène, Sanders Lena, Objets géographiques et processus de changement. Approches spatio-temporelles, Londres, ISTE-Editions, coll. Systèmes d'Information Géographique , 2014, 178 p.

15 We should emphasise that there is no consensus in the literature on the difference between entity and object, which are sometimes considered to be synonymous. In this chapter, ‘entity’ will be used in the broadest sense, corresponding to the definition of Livet expressed in the preceding paragraph, and ‘object’ will be reserved for the more restrictive sense of entity-object.

16 See table 1 of chapter 2 for a presentation of all the transitions studied.

17 Mathian H., Objets géographiques…, op. cit.

18 Garmy Pierre, Villes Réseaux et Systèmes de villes, Paris, Errance, 2012.

19 Phan D., ‘Ontologies et modélisation par SMA … ‘, op. cit., p.66

20 There is, in fact, a hiatus, which is explained in chapter 17.

21 The objective of the model of transition 8 was to analyse the spatial organisation stemming from the competition of powers resting on two different principles, one hierarchic and the other ecclesiastic, exercised from the bottom up.

22 This term designates an urban region made up of a collection of metropolises and cities geographically closely situated and functioning together. In the final version, the author of chapter 13 opted for the term ‘polycentric metropolitan region’.

23 UML (Unified Modelling Language) is a modelling language with a standardised notation for constructing and specifying systems of all kinds. Its main characteristic is its generic nature; it can be adapted to very different fields and, moreover, in computational terms, it can easily be translated into object-oriented coding.

24 This term is introduced in the treatment of transition 3. It designates the association of a number of villages under the same leader. In the model, the terminology deploys the terms ‘group’ to designate the association of a number of households and ‘complex group’ to designate the association of several groups. The thematics corresponding to these objects are, respectively, the village and the ‘polity’.

25 Even the chapters which have been composed by more limited collectives (transitions 3, 4 and 12) have benefited from the reactions of the collective as a whole, as well as from the inter-group exchanges concerning comparison.

Table des illustrations

Titre Figure 1: From the empirical referent to the formal system
Légende (the stages of the progression unfold from right to left).
Fichier image/png, 62k
Titre Figure 2: Schemas characterising two different ontological choices (UML formalism).
Fichier image/jpeg, 121k
Titre Figure 3: UML representation of the structural elements common to the different transitions studied (landscape format)
Fichier image/jpeg, 172k


Le texte et les autres éléments (illustrations, fichiers annexes importés) sont sous Licence OpenEdition Books, sauf mention contraire.

Rechercher dans OpenEdition Search

Vous allez être redirigé vers OpenEdition Search