Roles remain such a nebulous concept notwithstanding its ubiquity in a wide variety of domains, including conceptual modeling, that no clear consensus exists over the nature of roles in the foundational ontology research. In this paper I argue that there are three closely intertwined, but conceptually separate role-related notions: a role specification, a role position, and a role performance. I further contend that different accounts of roles might depend on which of the three concepts takes priority over the other two. Additionally I propose that there be three possible interpretations of the ontological nature of roles, each of which requires careful cost-benefit analysis: a family resemblance concept, a functionally definable concept, and a practically unifiable concept.
The notion of role is present in a number of different domains, ranging from
knowledge representation [
Intimately connected to this topic is the extant issue of whether a single definition
of role is possible [
2For instance, Guarino [4, p. 14] says: “I have been always fascinated by the subtle aspects of this notion
[role], and by its ubiquitous relevance for practical applications. (...) It is not a surprise therefore to see roles
appearing in BFO [Basic Formal Ontology [
In this paper I endeavor to investigate the nature of role from the perspective of foundational ontology. To do so, I leverage the notion of grounding that has been recently developed in the field of philosophical ontology. As we will see below, grounding is supposed to specify how some phenomena (e.g., a table exists) hold in virtue of more fundamental phenomena (e.g., subatomic particles are ‘arranged table-wise’) and, quite importantly, to be closely linked to the notion of explanation (which may be called ‘ontological explanation’) in philosophical ontology.3
More specifically, I consider what grounds the concept of role-playing that is central to the general discussion on role. This work amounts to an attempt to seek an adequate explanation of role-playing, thus leading to a better understanding of role. It is found on close examination that there are three notions (which I call the ‘role triad’) that ground role-playing: a role specification, a role position, and a role performance. Defying easy analysis, each of the role triad is to be elucidated by analogy and with examples.
Then I hypothesize that different accounts of role might hinge on which of the role triad is ontologically prior to the other two concepts. To illustrate this, I explore three theories of roles and which element of the role triad each theory take to be primary. I further clarify that and how each theory’s ‘role choice’ is conceptually firmly glued to the (meta-ontological choices of) foundational ontology on which the theory is based.
I finally suggest that there be three possible interpretations of the nature of role: a family resemblance concept, a functionally definable concept, and a practically unifiable concept. As will be detailed below, each of them has both advantages and disadvantages. For instance, a family resemblance conception of role lacks practical virtue while it is arguably most theoretically tenable. All these findings will contribute to bridging the gulf between foundational investigation into the nature of role and and modeling processes associated with roles in various application domains.
The paper is structured as follows. Section 2 offers some preliminary knowledge on basic ontological assumptions and grounding. Section 3 presents and elucidates each of the role triad. Section 4 illustrates, with three selected accounts of roles, the relationship between meta-ontological choices in foundational ontologies and their ‘role choices’. Section 5 proposes three possible understandings of the nature of role. Section 6 concludes the paper with some brief remarks on future directions of research.
Since this paper is partly devoted to comparative analysis of foundational ontologies, I assume only the basic categories and relations that are relatively widespread in foundational ontologies. Concrete individuals (which exist in space and/or time) fall into two types: continuants (aka endurants) such as objects and occurrents (aka perdurants). Characteristically, continuants can persist, whereas occurrents extend through time. Continuants (e.g., a stone) can participate in occurrents (e.g., a fall of the stone).
I stipulate throughout the paper that roles are continuants, since roles are most
frequently take to be a special kind of properties (in the broad sense of the term) in the
relevant literature on role (e.g., [
As a conceptual tool for my investigation, I exploit the notion of grounding that is
supposed to provide ontological explanation.5 In particular, I employ its most standard
version: fact-grounding [
This theory is typically coupled with the claim that the notion of grounding is (a kind
of) ontological explanation (e.g., [
To be concrete, I will examine which fact grounds the fact that Mary is a student, given that a student is paradigmatic example of a role. For the sake of simplicity I introduce the notation ‘<>’ to refer to facts: e.g., <Mary is a student>. More specifically, I will consider which fact grounds <Mary plays a student role>, which is plausibly taken to ground <Mary is a student>. The gist of my argument is that there are three candidate facts which correspond to the role triad and which ground <Mary plays a student role> and, by the transitivity of grounding, <Mary is a student>.
What grounds <Mary plays a student role>? In order to be a student, Mary must gain admission to the university (say ‘ABC-U’) of her choice. To attain this goal, she needs to read and understand the admission policy and then make an effort to satisfy all its requirements. Mary’s role-playing has, in this respect, a deontic or normative dimension.
This observation may lead us to interpret role-playing as meeting the constraints or conditions that are ‘embedded’ in the role. In my terminology, role-playing in this sense is meeting a role specification, or the specification that is determined by the role. Therefore <Mary plays a student role> is grounded in <Mary meets a role specification>.
I appeal to the notion of specification to capture the above-mentioned normative
feature of role. The ontological nature of a specification remains obscure, but Turner [
5For an introduction to the general notion of grounding, see e.g., Schneider and Correia [
6The proponent of this view might argue that, just as some kind of causal explanation is given merely by
stating the causal relation (what causes what), so some kind of ontological explanation is given merely by
stating the grounding relation (what grounds what) [
A role specification is thus well understood with an analogy with artifacts (cf. [
Not surprisingly, the specification view of role-playing fits well with the notion of social role (e.g., president) because a specification is determined by our intentionality.8 To paraphrase in my framework, <Mary plays a student role> is grounded in <Mary meets a role specification>, which is in turn grounded (via a complex chain of grounding relations) in some relevant social facts: e.g., <ABC-U is nationally authorized>.
There is nevertheless another candidate for what grounds <Mary plays a student role>. As an ABC-U student, Mary can use various facilities and enjoy educational opportunities (e.g., taking classes). Seen from another perspective, she locates herself, in playing a student role, in a specific situation where she can do something role-related.
Following this intuition, one may think that Mary’s role-playing consist in her occupying the kind of special place (which I call a ‘role position’) that allows her to do something that is associated with the role. For this reason one may say that <Mary plays a student role> is grounded in <Mary occupies a student role position>.
A role position can be elucidated by an analogy with a relative place [
A role position is like a relative place. Role positions stand in a fixed conceptual
relation towards one or more entities (which I call ‘role reference entities’ and which
are sometimes called ‘context’ in the literature [
The analogy between role positions and relative places offers an interesting
interpretation of the alleged relational nature of roles [
8Turner [19, p. 147] says: “Our intentional stance determines what we take to be the specification: something is a specification if we give it normal force over the construction of an artefact.”
9I am using the term ‘conceptual’ in the very broad sense of the term. It may be argued that the partisan of role positions is responsible for clarifying the relationship between them and their role reference entities. earth, places with the ship as their reference object (e.g. the ship’s hold) move relative to places with the earth as their reference object.
Similarly, role positions may ‘conceptually move’ (e.g., changes their related functions) relative to one another when their role reference entities ‘conceptually move’ relative to one another, although the notion of conceptual movement is currently a placeholder and it is to be further clarified. For instance, when a human resource department changes its importance with respect to its company, personnel director role positions (whose role reference entity is the human resource department) change their relationship with executive role positions (whose role reference entity is the company).
The positional view of role-playing has two characteristics. First, it would lead to a
classification of roles according to what role reference entities are and what the
‘occupiers’ of role positions are (e.g., [
There is yet another possibility for what grounds <Mary plays a student role>. As said above, Mary can do many things (e.g., getting a student discount) because she is an ABC-U student. A clearly visible difference between Mary and non-ABC-U students is, for instance, that she is able to acquire a degree from ABC-U, whereas they are not.
This consideration may result in the idea that Mary’s role-playing resides essentially in her role-related performance, or rather her ‘power’ to do something role-related.11 I say that, generally speaking, role-playing in this sense is giving a role performance. Thus <Mary plays a student role> is grounded in <Mary gives a student role performance>.
It is important to note that a player of a role (performance) has only to possess the
role-related ‘power’ instead of actually demonstrating it, although I use the phrase ‘give a
role performance’ for the sake of simplicity. In playing a student role, Mary does not need
to use any facility or take any class; she only needs to be able to do them.12 In this sense,
the notion of a role performance may be intimately related to deontic powers [
On the one hand, the performance view of role-playing would be easier to
understand than the other two approaches discussed above, since it is explicable in terms of an
analogy with the intuitively less complicated notion of ‘power’ than a specification or a
relative place. On the other hand, the onus is on the proponent of this view to pin down
10See Marmodoro and Yates [
11Boella, Torre and Verhagen [3, p. 5] say: “(...) behavior should not be disregarded as a main feature of roles.”
12Strictly speaking, Mary would have to comply with the rules and regulations imposed by ABC-U in order to play a student role. It could be therefore argued that, in general, a player of a role needs to give some sort of role performance, however trivial it may be. precisely the interrelationship between a role performance and its intimately related concepts: e.g., dispositions, functions, and capabilities.13
Masolo et al. [
A concept and a description both fall into the DOLCE category of non-agentive social concept: “an endurant that: (i) is not directly located in space and, in general, has no direct spatial qualities; (ii) has no intentionality; (iii) depends on a community of intentional agents, e.g., a law, an economic system, a currency, an asset ...” [6, p. 272]. Furthermore, some basic features of descriptions are offered as follows [6, p. 271]: descriptions are created by (communities of) intentional agents at the time of their first encoding in an expression of a ‘public’ (formal or informal) language different expressions (possibly in different languages) can be associated to the same description, provided they have the same semantic content. I.e., descriptions have a unique semantic content descriptions must be encoded on (possibly multiple) physical supports [Original footnote: “Printed or recorded texts obviously count as physical support, but memory or other cognitive processes should probably be considered as well (think of orally transmitted tales, rules and contracts)”] descriptions are usually accepted (adopted) by (communities of) intentional agents, but a description can exist even if no one accepts it, as long as it remains encoded; acceptation can change in time descriptions cease to exist when their last physical support ceases to exist It is not hard to see that the approach by Masolo et al. to social roles pivots around a role specification in the role triad. My notion of role specification coincides with the DOLCE notion of description. Both of them are based on agents’ intentionality and aim to specify how continuants should be like by satisfying the constraints provided by the role specifications and the descriptions, respectively.
On my view, the choice by Masolo et al. of a role specification is not only because
they focus primarily on social roles but also because they take DOLCE as a general
13On dispositions: see e.g., Mumford [
Loebe [
Loebe also offers a classification of roles according to the kinds of the player, roleplaying relation, the role-of relation, and the context. This is, as said above, characteristic of the positional view of role-playing. Roles fall into social roles and abstract roles, the latter being in turn classified into relational roles and processual roles.15 Processual roles are, roughly, roles that are participated in (played) by objects and that bear the parthood (role-of) relation with occurrents (context).16 For instance: “When John moves his pen, he and the pen form participants of that process, and the processual role which John plays captures what John does in that participation” [7, p. 135].
The GFO theory of role would deal better with abstract roles than social roles.17 This can be seen as a consequence of GFO’s choice of a role position in the role triad. As was above pointed out, relational roles are well treated with a role position, whereas social roles with a role specification.18 Processual roles are sufficiently characterized in terms of a role position because they consist in having occurrents as role reference entities.
Finally, part of the reason why the GFO account of role is committed to a role position may lie in GFO’s meta-ontological choice of what it calls ‘integrative realism’.
14“Regarding the content of the ontology, the aim of DOLCE is to capture the intuitive and cognitive bias underlying common-sense while recognizing standard considerations and examples of linguistic nature. DOLCE does not commit to a strong referentialist metaphysics (it does not make claims on the intrinsic nature of the world) and does not take a scientific perspective (it is not an ontology of, say, physics or of social sciences). Rather, it looks at reality from the mesoscopic and conceptual level aiming at a formal description of a particular, yet fairly natural, conceptualization of the world” [37, pp. 279-280].
15Loebe [7, p. 137] explains abstract roles as follows: “Due to their similarity, relational and processual roles are subsumed by a role type called abstract roles which is contrasted with social roles. Abstract roles can be functionally characterized in a uniform manner, namely as a mechanism of viewing some entity - namely the player - in a defined context, i.e., in a more complex entity with interrelated other “notional components”. Put differently, players of abstract roles are looked at in an external manner in contrast to viewing them as self-contained entities focusing on their internals like their properties or parts.”
16I am using the word ‘roughly’ because, assuming the type level, a player of a processual role is the GFO
notion of persistant and its context is the GFO notion of process. See Herre [
17Loebe [7, p. 136] says: “Social roles appear to be the least understood role type in our model. For instance, switching to role-of, we must admit that contexts remain fairly obscure for social roles.”
18Loebe [7, pp. 137-138] says: “(...) encouraged by the diverging ontological categories of contexts, we believe that it will be hard to find further commonalities, especially between abstract and social roles. It may thus be difficult to add much more to a general theory of roles, at least as long as a similarly broad range of examples is to be covered.” Loebe [7, p. 154] also says: “In our opinion, it turns out that the aspects of abstract and social roles are intermingled in the literature, especially concerning relational and social roles.” I would submit that the same roles (e.g., student roles and professor roles) are interpretable in terms of a role specification as well as in terms of a role position.
Herre [38, pp. 303-304] elucidates this doctrine by comparing it with what he calls
‘Smithian realism’ [
Basic Formal Ontology (BFO) defines a role as follows: “A realizable entity that (1) exists because the bearer is in some special physical, social, or institutional set of circumstances in which the bearer does not have to be, and (2) is not such that, if this realizable entity ceases to exist, then the physical make-up of the bearer is thereby changed. A role is thus always optional” [5, p. 184].
A realizable entity is: “A specifically dependent continuant entity that has at least one independent continuant as its bearer, and whose instances can be realized (manifested, actualized, executed) in associated processes of specific correlated types in which the bearer participates” [5, p. 183]. A specifically dependent continuant is then: “A continuant entity that depends on precisely one independent continuant for its existence. The former is dependent on the latter in the sense that, if the latter ceases to exist, then the former will as a matter of necessity cease to exist also” [5, p. 185].
The BFO conception of role clearly focuses on a role performance in the role triad. An entity a plays a role because, for an external reason, a has a BFO-role which is unique to a and which can be realized in the kind of occurrents (typically a’s behaviors) in which a participates. For instance, Mary is a student in virtue of her student role that can be realized in, for instance, her behavior of taking classes.
BFO’s ‘role choice’ would be primarily motivated by its meta-ontological adoption of ontological realism: “The realist methodology is based on the idea that the most effective way to ensure mutual consistency of ontologies over time and to ensure that ontologies are maintained in such a way as to keep pace with advances in empirical research is to view ontologies as representations of the reality that is described by science. This is the fundamental principle of ontological realism” [40, p. 139]. In the role triad, a role performance is arguably most likely to be the object of empirically scientific inquiry.
I make two further comments on BFO-roles that would help us understand better the relationship between meta-ontological choices and their related ‘role choices’. First, BFO specifies the role-having relation, but not the role-playing relation: “An entity is sometimes said to play a role, as when a passenger plays the role of a pilot on a commercial plane in an emergency, or a pyramidal neuron plays the role occupied by a damaged 19Herre [38, p. 305] further elucidates integrative realism as follows: “the nodes in an ontology are labeled by terms that denote concepts. Some of these concepts, notably natural concepts, are related to invariants of material reality. Concepts are represented in individual minds and are founded in society. The same is true for individuals to which individual concepts correspond. The interrelations between universals, concepts, symbols and society are realized by various relations, including the relation of correspondence (between concepts and universals, and individual concepts and real individuals), the relation of representation (between concept and individual mind), the relation of foundedness (between concept and society), and the instantiation relation. We summarize that the restricted view of Smithian realism cannot be an ontological-philosophical foundation for the field of conceptual modeling and, in particular, for computer-science ontologies.” stellar neuron in the brain; but neither the person nor the pyramidal neuron have those roles. BFO 2.0 only specifies the has role relation” [12, p. 58].
Seen from the viewpoint of grounding, the former ‘pilot-role fact’ and the latter ‘neuron-role fact’ are better grounded in <A passenger meets the pilot role specification> and <A pyramidal neuron occupies the stellar neuron role position> than <A passenger gives the pilot role performance> and <A pyramidal neuron gives the stellar neuron role performance>, respectively.
BFO’s specification of the role-having (but not role-playing) relation implies that BFO-roles focus primarily on a role performance, but neither a role specification nor a role position. Using the Mary-student example, <Mary plays a student role> is grounded in <Mary gives a student role performance>, which is in turn grounded, within the BFO framework, in <Mary’s student role is (can be, more precisely) realized>.
Second, it is suggested that some usages of the term ‘role’ be covered in another way than BFO-roles: “The term “role” can, however, be used in a different sense in contexts such as Jane’s being the seventh person to fill the role of director of this institute, or Joe’s being the third person to play a particular role in a play. “Role” in this sense is being used to designate what BFO calls a generically dependent continuant” [5, pp. 100-101].
A generically dependent continuant is: “A continuant that is dependent on one or other independent continuants and can migrate from one bearer to another through a process of copying. We can think of generically dependent continuants as complex continuant patterns either of the sort created by authors or designers or (in the case of DNA sequences) brought into being through the processes of evolution” [5, p. 179].
It is not hard to see a close conceptual affinity between the BFO notion of generically dependent continuant and the DOLCE notion of description. It could be argued that a DOLCE-description is the sort of BFO-generically dependent continuant that is connected to agents’ intentionality. In this respect, BFO attempts to ground, e.g., <Jane fills the role of director of this institute> and <Joe play a particular role in a play> in <Jane meets the director role specification provided by this institute> and <Joe meets a role specification provided in a play>, respectively.
All the arguments given above would reveal the ontological nature of role: roles are family resemblance concepts [41, Section 3]. As shown above, there are equally plausible facts in which <Mary plays a student role> could be grounded. This means that there is no single privileged concept of role; the role notion is merely partly unified by the role triad. Moreover, different meta-ontological choices in foundational ontologies lead to different choices in the role triad. Granted that those meta-ontological choices are equally reasonable from a theoretical perspective, so are the associated ‘role choices’.
The family resemblance view of role is a direct consequence of the present work and it is arguably most convincing. It lacks practical virtue, however. As said, there is considerable need for formalizing this cross-disciplinary notion in order to conceptualize the world coherently. However theoretically tenable it may be, it would be of little help for actual modeling processes to say just that roles are family resemblance concepts.
I said that <Mary plays a student role> is grounded in either <Mary meets a student
role specification>, <Mary occupies a student role position>, or <Mary gives a student
role performance>. Instead of taking a family resemblance view of role, one may then
attempt to offer a functional definition [
The core part of my argument over role can be simplified as follows: For any an (individual) entity x, if x plays a role, then either x meets a specification determined by that role; x occupies a position determined by that role; or x gives a performance determined by that role.20 Replace the term ‘role’ with a variable R, and then existentially quantify it, as follows: 9R (if x plays R, then either x meets a specification determined by R; x occupies a position determined by R; or x gives a performance determined by R).
x plays a role =def. 9R [(if x plays R, then either x meets a specification determined by R; x occupies a position determined by R; or x gives a performance determined by R) and x plays R].
The view of role as a functionally definable concept may be a practical advance from
its family resemblance conception. Even if there may some room for improvement in
the formalization of the term ‘specification’ (e.g., [
One may still wish to provide a unified concept or definition of role even at the price of theoretical rigor. This task requires considering carefully which ‘role choice’ is practically appropriate. For instance, one may choose a role specification, based on the intuition that <Mary meets a student role specification> grounds <Mary gives a student role performance>. For another example, one may take a role position because it fits well with an apparently widely acceptable, moderate stance on ontology. This line of investigation is to be pursued together with careful cost-benefit analysis.
I have argued for three claims. First, there are three role-related concepts (the role triad): a role specification, a role position, and a role performance. Second, different accounts of role might depend which notion in the role triad is ontologically prior to the other two, as illustrated with close examination of three theories of role which are based on (the meta-ontological choices of) DOLCE, GFO, and BFO, respectively. Third, there are 20To simplify matters, I am setting aside the issue of whether logical entailment fully captures the notion of grounding that I have exploited so far. three possible understandings of the ontological nature of role: a family resemblance concept, a functionally definable concept, and a practically unifiable concept.
This work on role will bring us to many future directions of research and, in
particular, there are many thorny questions to be answered regarding various associated
modeling processes. For instance, the topic of qua-individuals [
22I thank Adrien Barton and the reviewers for some helpful comments of the paper.