Toward a Realizable-Centered Approach to Artifacts in Applied Ontology

1.1. Background and Motivation

The world is brimming with artifacts. To name just a few, there are technical artifacts (e.g., screwdrivers), artworks (e.g., paintings), social artifacts (e.g., the European Commission) and abstract artifacts (e.g., software and Shakespeare's work Hamlet). These types of entities are usually contrasted with so-called “natural objects” such as molecules, stones, people and planets. They are intimately connected with many kinds of entities such as intentions, agents, actions (physical and mental), capabilities, capacities, functions, roles, and affordances. Artifacts are notoriously difficult to analyze ontologically because no clear consensus exists yet over how to understand systematically a number of different ontological characteristics of artifacts.

In this paper we will develop a new approach to the formal ontology of artifacts. Let us begin by explaining the general motivation behind our work. First, artifacts have been traditionally investigated in terms of functions, as is illustrated by the example of screwdrivers and the function to turn screws. There may be nonetheless some problems with the function-based accounts of artifacts. One theoretical problem is about their generality. While technical artifacts may be generally explicable in terms of functions to some degree, it is controversial whether so are all the kinds of artifacts as well (Franda & Souman, 2025; Houkes & Vermaas, 2009b; Juvshik, 2021b; Koslicki & Massin, 2025, in press; Olivero, 2019; Raven, 2025). For instance, what would be, if any, the function of Leonardo da Vinci's Mona Lisa? This question may be difficult to answer straightforwardly, given the variety of theories of functions as well as desiderata for a satisfactory theory of functions in the literature (Artiga, 2011, 2016; Röhl & Jansen, 2014; Wouters, 2005).

One related, practical problem with the function-based accounts of artifacts is that they could yield conceptual and linguistic confusion in the development of domain ontologies. This is because the term “function” has been used differently in different domains and upper ontologies, or even within specific domains such as engineering (Carrara et al., 2011; Yargan & Jansen, 2024). To take one simple example, the definition of the term “material artifact” as “a material entity that has some function” may be differently interpreted and formalized, according to different views of functions. This may, in turn, exacerbate the problem of the lack of semantic interoperability between information systems (Gardner, 2005), as it is arguably the case that: “neither in philosophy nor in formal ontology is there a generally accepted theory of functions. In particular, there is no consensus how to include functions into a top-level ontology or whether to include them at all” (Röhl & Jansen, 2014, p. 1).

Second, it is widely acknowledged that artifacts have an intentional dimension. However, there may be room for further exploration of the intentional dimension of artifacts. Suppose that I shaped and dried an amount of clay with the intention that it will be capable of containing liquid. What is, ontologically speaking, my intention to shape and dry this amount of clay in such a way as to be able to contain liquid? What kind of ontological relation holds between my intention and my intentional act of shaping and drying the clay to be able to contain liquid? What are the relationships among my intention, my intentional act and the resulting clay pot? These questions must be addressed by a complete analysis of the intentional dimension of artifacts. For instance, Borgo and Vieu (2009, p. 291) say: “although we need to model intentional aspects [of artifacts], we will avoid as much as possible the direct reference to theories of mental attitudes, which are not per se the subject of this paper and are by no means consensual.” Although there may be no solid consensus as to an ontology of mental attitudes such as intentions, it will constitute an important step toward a comprehensive ontology of artifacts to analyze the intentional dimension of artifacts in terms of some specific ontological account of mental reality.

To summarize, we are motivated to develop an alternative theory of artifacts in the field of formal ontology, partly because of the theoretical and practical limitations of function-based accounts of artifacts, partly because of the needs for further study of the intentional dimension of artifacts.

1.2. Purpose

The objective of this paper is to provide a theoretical foundation for a formal ontology of artifacts on the basis of the ontology of realizable entities, extending our previous works (Toyoshima et al., 2024a, 2024b). ¹ A realizable entity is a property that can be realized in associated processes of a specific correlated type in which the bearer participates. For example, the fragility of a particular glass can be analyzed as a realizable entity whose bearer is this glass and which can be realized in a process of the glass breaking when the glass is pressed with sufficient force. One paradigmatic subtype of realizable entities is the concept of dispositions, such as the fragility of a glass and the solubility of salt. Among upper ontologies, Basic Formal Ontology (BFO) features the category of realizable entity, which include dispositions (Arp et al., 2015; Otte et al., 2022) and the Unified Foundational Ontology (UFO) features the category of disposition (Guizzardi, 2005; Guizzardi et al., 2022).

We will develop a “realizable-centered approach to artifacts” in accordance with the aforementioned motivations for an alternative ontology of artifacts. Concerning the motivation from function-based accounts of artifacts, the notion of a realizable entity can be expected to help to develop a more general theory of artifacts (notably, beyond the scope of technical artifacts) than that of function. For one thing, as we will discuss in detail below, various kinds of functions may be generally characterized as subtypes of realizable entities such as dispositions (Roberts, 2024; Röhl & Jansen, 2014; Spear et al., 2016; Toyoshima et al., 2022b; Toyoshima & Jansen, 2026) and there can be correspondingly a group of artifacts that lack functions but have a special kind of realizable entities. This line of approach to artifacts can be found, for example, in Kassel's (2010) theory of artifacts, according to which artifacts have capacities and technical artifacts have functions, which are a subtype of capacities.

Concerning the motivation derived from the intentional dimension of artifacts, we argue that this dimension can be meticulously analyzed through the lens of intention as a realizable entity—specifically, as a disposition. Broadly speaking, an intention, as assumed here, is a disposition to act that comes into being as a result of the interactions between beliefs and desires (Toyoshima et al., 2020). Assuming the dispositional view of intentions, we can provide an ontological analysis (to be elaborated below) of the clay/pot example introduced above: my intention to shape and dry this amount of clay in such a way to be able to contain liquid was realized in my intentional act of shaping and drying the clay to be able to contain liquid.

1.3. Methodology and Scope

To achieve the purpose of developing a realizable-centered approach to artifacts, we will not adopt the traditional methodology of addressing the question of what is the correct definition of the term “artifact” as well as what are the correct definitions of domain-neutral artifact-related terms such as “material artifact” and “technical artifact.” This is because, like the term “function,” the term “artifact” has been employed in very different ways across various domains. As a result, it is doubtful whether seeking a single, universally applicable definition of the term “artifact” across all domains of ontology application would be theoretically and practically valuable (cf. Güngör, 2023, 2026; Preston, 2022b; Sperber, 2007).

We will instead adopt the alternative methodology of introducing two new terms whose ontological analysis can help to understand and formalize different usages of the term “artifact” in different domains: “canonical artifact” and “usefact.” A canonical artifact is an entity that is intentionally produced for a specific purpose. A usefact, by contrast, is an entity intended to be used for some accidental purpose, that is, for a purpose that is different from the one (if any) for which the entity was intentionally produced. Here, we use the term “accidental” without contrasting it with “essential”: its usage remains neutral as to the complex relationship between artifacts and essence (Irmak, 2024b; Passinsky, 2024). In particular, a natural object is a usefact when it is intended to be used for any purpose, since it was not intentionally produced for any purpose (or at least, this is the assumption that we make about natural objects throughout this article).

For illustration, when I intentionally shaped and dried this amount of clay in such a way as to be able to contain liquid, the resulting clay pot is a canonical artifact because it is intentionally produced for the purpose of being able to contain liquid. In contrast, when I intend to use a pebble to keep papers in place (cf. Borgo & Vieu, 2009), this pebble is a usefact because it is intended to be used for the accidental purpose of keeping papers in place, or more particularly, simply because it is a natural object that is intended to be used for some purpose.

Regarding the term “canonical artifact,” we introduce it with the understanding that the term “artifact” has traditionally been defined as an entity intentionally produced for a specific purpose (e.g., Hilpinen, 1992). Consequently, any well-developed theory of artifacts typically counts it as an artifact. Thus, an entity that is intentionally produced for a specific purpose can be considered as a canonical form of artifact, or simply as a canonical artifact.

Regarding the neologism “usefact,” we acknowledge that it may sound slightly etymologically problematic, as the suffix “-fact” derives from the Latin word “factum,” meaning “something made.” However, in introducing the term “usefact,” we draw inspiration from existing neologisms such as “naturefact” in anthropology (Oswalt, 1973, 1976) and “ecofact” in archeology (e.g., Kipfer, 2021, p. 422). To borrow Preston's (2022a) explanations, naturefacts are “naturally occurring objects used intentionally, but without modification, for some purpose,” such as a pebble intentionally used to keep papers in place; and ecofacts are “organic or inorganic remains of archaeological significance that have undergone no, or minimal, modification by humans, such as animal bones, stored grain, pollen, charcoal, and the like.” They are sometimes discussed in the literature on artifacts and functions to consider the difficulty of demarcating natural objects from artifacts (e.g., Artiga, 2016; Preston, 2022a; Spear et al., 2016). We introduce the notion of a usefact partly because, in some theories of artifacts (e.g., Borgo & Vieu, 2009), the term “artifact” refers to usefacts as well as to canonical artifacts, and partly because it will help to analyze artifact-related terms such as “naturefact” and “ecofact.”

More specifically, we will focus on the terms “material canonical artifact” and “material usefact” whose ontological analysis will help us better understand various usages of the term “material artifact.” We will define these two terms in terms of the ontology of realizable entities. Here we make some clarificatory remarks on the scope of this paper. First, technical artifacts are a typical subtype of material artifacts and they have been explored in formal ontology (Borgo et al., 2014a). By contrast, our focus lies on material artifacts in general, encompassing both technical and non-technical artifacts, including material artworks such as paintings.

At the same time, we will leave out of scope a detailed analysis of specific kinds of material artifacts. For instance, an in-depth study of material artworks would require an ontological analysis of art (Abell, 2012; Currie, 1989; Dickie, 1969; Dipert, 1993; Hick, 2019; Terrone, 2024; Thomasson, 1999, 2004, 2006; Xhignesse, 2020). Moreover, we will leave for future work the examination of so-called “abstract artifacts” (Friedel, 2021; Irmak, 2021; Reicher, 2022; Terrone, 2023), partly because their analysis will require an ontological analysis of abstract entities (Falguera & Martínez-Vidal, 2020; Friedel, 2021; Moltmann, 2013; Toyoshima & Niki, 2024). Abstract artifacts include repeatable artworks (Davies, 2024; Walter, 2013), musical works (Dodd, 2000; Mikalonytė, 2023), fictional characters (Thomasson, 1999, ch. 3), words (Irmak, 2019), languages (Evnine, 2025), and software (Irmak, 2012).

1.4. Organization of the Paper

The paper is structured as follows. Section 2 offers some preliminary background material. Section 3 develops a realizable-centered approach to artifacts. More concretely, we delineate and illustrate the terms “canonical artifact” and “usefact,” and define the terms “material canonical artifact” and “material usefact” from the viewpoint of the ontology of realizable entities. Section 4 examines the relationship between functions and artifacts from a realizable-centered perspective. Section 5 illustrates the flexibility and applicability of our framework through its adaptation to BFO. Section 6 discusses various issues arising from our account of artifacts. Section 7 examines preceding works on artifacts, especially those in the field of formal ontology, vis-à-vis our realizable-centered approach to artifacts. Finally, Section 8 concludes the paper and outlines directions for future research.

2.1. General Ontological Framework

We introduce a general ontological framework for our investigation that is largely (if not completely; see Section 2.3) realist in spirit (Smith & Ceusters, 2010) in contradistinction with “epistemic ontologies,” especially regarding artifacts (Kassel, 2023). Figure 1 provides a taxonomy of classes used in this article. The taxonomy consists of terms denoting classes that are already well-established in existing upper ontologies and in previous formal-ontological works, including ours. In particular, asterisked classes are attributed to Toyoshima et al. (2024a). For notational clarity, we will occasionally use italics for class or type-level entity names (e.g., “Material entity”) and bold for particulars or instance-level entity names (e.g., “clay₁”). We will also occasionally use subscripts for entity names (e.g., “BFO:Function”). In this taxonomy, a necessary and sufficient condition for a class A to be a subclass of a class B (as indicated by indentation) is that all instances of A are also instances of B.

OPEN IN VIEWER

Figure 1.

A taxonomy of classes used in this article.

We will now clarify the terms “material entity,” “natural object,” “process,” and “realizable entity” in the taxonomy. First, the notion of a material entity is widely accepted in upper ontologies, as illustrated by the BFO class Material entity and the class Physical endurant in the upper ontology “Descriptive Ontology for Linguistic and Cognitive Engineering” (DOLCE; Borgo & Masolo, 2010; Borgo et al., 2022; Masolo et al., 2003; Porello et al., 2025). Examples include an amount of clay, people and screwdrivers. We introduce the class Natural object as a subclass of the class Material entity, assuming that the term “natural object” can be understood intuitively (see Sections 1.1 and 1.3).

Second, processes and events remain widely discussed in formal ontology. They are very differently characterized by different upper ontologies (Rodrigues & Abel, 2019). In this paper we use the term “process” while remaining neutral regarding the detailed classification of occurrents (perdurants), such as the ontological distinctions between processes, events, and states. Our notion of a process aligns approximately with the BFO class Process and the DOLCE class Perdurant. Instances include the process of my shaping and drying this amount of clay, as well as the process of my using this pebble to keep papers in place.

Third and most importantly, the notion of a realizable entity has been explored in formal ontology, especially in connection with the BFO category of realizable entity (Rabenberg et al., 2024; Röhl & Jansen, 2014). In this paper we will draw upon Toyoshima et al.'s (2022c) framework for realizable entities that is theoretically underpinned by McKitrick's (2018) dispositional pluralism. The basic idea is that realizable entities can be generally understood as dispositions in McKitrick's broad sense of the term.

In particular, we will deploy the notions of an intrinsic disposition and an extrinsic disposition (McKitrick, 2003; McKitrick, 2018, ch. 8; Vetter, 2015, chs. 4.4–4.5) to specify important subclasses of the class Realizable entity. An intrinsic disposition is a realizable entity that is borne purely in virtue of the way the bearer is, for example, in virtue of the internal (physical) structure of the bearer. An extrinsic disposition is a realizable entity that is borne, at least partially, in virtue of the way the world that is external to the bearer is.

Classical examples include the intrinsic disposition of fragility and the extrinsic disposition d_ext of a particular key (key_a) to open a particular lock (lock_b). If lock_b is annihilated and ceases to exist, d_ext also ceases to exist, even without any physical changes of key_a, because key_a bears d_ext in virtue of the existence of lock_b, which is external to key_a. Our notion of an intrinsic disposition corresponds to the BFO class Disposition and the UFO class Disposition, as the former class is described by BFO as an “internally grounded realizable entity” (Toyoshima et al., 2022c; see also Section 5.1.1) and the latter is a subclass of the UFO class Intrinsic moment.

We will outline three philosophical views of diachronic identity as they are crucial for understanding the distinction between production and use, and thus between canonical artifacts and usefacts: the continuity view, the discontinuity view and the constitution view (the latter two grouped under the non-continuity view for expository purposes). We will illustrate them with two examples related to production and use, referred to as “Example A” and “Example B,” respectively (see Section 3 for a detailed discussion about the notions of production and use).

Before presenting Examples A and B, we offer a preliminary remark on the usages of the terms “pot” and “paperweight” as they appear in these examples. We assume that the pot in Example A is a canonical artifact (to wit, a material entity intentionally produced for the purpose of being able to contain liquid) and that the paperweight in Example B is a usefact (to wit, a material entity that is intended be used for the accidental purpose of keeping papers in place). Throughout the paper, however, we employ domain-specific artifact-related terms such as “pot” and “paperweight” primarily for illustrative purposes, highlighting our realizable-centered approach to artifacts. A meticulous analysis of these terms would require careful consideration of the difference between “artifact kind(s)” and “(specific) kinds of artifacts” (Irmak, 2024b; Jansen, 2013; Juvshik, 2025; Koslicki & Massin, 2025). As a matter of fact, the term “paperweight” may sometimes refer to a canonical artifact (to wit, a material entity that is intentionally produced for the purpose of keeping papers in place), as is shown by decorated paperweights found in souvenir shops. For this reason, we will provide fiat definitions of domain-specific artifact-related terms, whenever appropriate.

We will present Example A (diachronic identity and production) and Example B (diachronic identity and use), in which the terms “pot” and “paperweight” are employed as just explained, respectively:

We will articulate two different approaches to production and use from the perspective of diachronic identity: the “monolithic approach” and “hybrid approach.” The monolithic approach adopts the same philosophical view of diachronic identity for both production and use. Its strength lies in providing a philosophically coherent and unified framework for understanding both processes. However, it conflicts with intuitions about production and use—specifically, our tendency to assume the non-continuity view of production and the continuity view of use. For instance, Borgo and Vieu (2009) embrace the monolithic approach by assuming the constitution view of both production and use (see also Vieu et al., 2008). Their position has been criticized for its commitment to the constitution view of use: “it appears to us to be counter-intuitive to consider that a mental act gives birth to a new physical object” (Kassel, 2010, p. 242).

By contrast, the hybrid approach adopts different philosophical views of diachronic identity for production and use. A typical version of this approach combines the non-continuity view of production and the continuity view of use. Its strength lies in aligning well with intuitions about production and use. However, it faces the challenge of developing a philosophically well-founded and unified framework for understanding both processes. The hybrid approach may be implicitly assumed by many existing theories of artifacts, where the term “artifact” refers exclusively to canonical artifacts, and no explicit stance on diachronic identity (with respect to production, use, or both) is taken.

In this paper we adopt the monolithic approach to production and use, ensuring that our realizable-centered approach to artifacts serves as a solid theoretical basis for analyzing production, use, and their ontological distinctions. Specifically, we assume the continuity view of both production and use, acknowledging that this may involve a somewhat counterintuitive usage of the term “production.” Later, we will consider how our realizable-centered approach to artifacts can be reformulated under the non-continuity view of production and/or use; more concretely, within Borgo and Vieu's (2009) theory of artifacts espousing the constitution view of both production and use (see Section 7.1). This will show that our approach can be adapted to both the monolithic and the hybrid approaches to production and use, rendering it “foundation-agnostic” (Grüninger & Katsumi, 2019). In other words, it can be integrated into different foundational frameworks, accommodating different philosophical views of diachronic identity with respect to production and use.

2.3. An Ontology of Belief, Desire, and Intention

To analyze carefully the intentional dimension of artifacts, we will introduce an ontology of belief, desire, and intention, drawing on existing formal-ontological studies thereof (Barton et al., 2018a; Biccheri et al., 2020; Hobbs & Gordon, 2010; Toyoshima et al., 2020). First, our ontology of belief, desire, and intention will be limited in scope. For instance, we will not closely examine the representational dimension (intentionality, aboutness) of these mental entities (Biccheri et al., 2020; Mizoguchi & Borgo, 2022, 2025; Sanfilippo, 2021; Toyoshima & Barton, 2024a; Williams, 2019a; Yablo, 2014). Additionally, we will consider intentional acts only insofar as they relate to the development of a realizable-centered approach to artifacts. A comprehensive ontological analysis of actions falls outside the scope of this article (Piñeros Glasscock & Tenenbaum, 2023; Trypuz, 2007).

As for belief and desire, we import the notions of an occurrent belief and an occurrent desire from Toyoshima et al.'s (2020) dual account of belief and desire. An occurrent belief is a mental process of taking something to be the case (Barton et al., 2018a). An occurrent desire is a mental process of wanting something to be the case. Examples include my occurrent belief (belief₁) that clay₁ will be capable of containing liquid when it is suitably shaped and dried; and my occurrent desire (desire₁) for clay₁ to be capable of containing liquid. Our notions of an occurrent belief and an occurrent desire would correspond, respectively, to the notions of a “belief state” and a “desire state” as found in existing ontologies of mind (e.g., Ferrario & Oltramari, 2004), as we employ “process” as a generic term for, for example, DOLCE:perdurants in this article (see Section 2.1).

As for intention, we will leverage Toyoshima et al.'s (2020) dispositional theory of intention. Interpreted within our ontological framework, an intention is an intrinsic disposition to act that comes into being as a result of the interactions between occurrent beliefs and occurrent desires.

To further develop the view of intention as an intrinsic “disposition to act,” we will enrich it with the ontological analyses of goals by Hobbs & Gordon (2010) and Biccheri et al. (2020). Hobbs & Gordon (2010, p. 62) say: “The key concept in modeling intentional behavior is that of an agent a having some eventuality type e as a goal.” Their term “eventuality” is synonymous with our term “process.” Biccheri et al. (2020) formally characterize a goal as an entity that is desired or needed by an agent. By synthesizing these ontological accounts of goals, we will assume that a goal is a process type that is the object of an agent's desire, or more specifically, that comes into being through some occurrent desire. Examples include a process type (Goal₁) which comes into being through desire₁ and whose achievement results in a process of clay₁ being capable of containing liquid.

This ontological analysis of goals can be further elucidated as follows. First, because of our broad notion of a process (see Section 2.1), our process-based view of goals can be compatible with the existing state-based view thereof (e.g., Borgo et al., 2016; Mizoguchi et al., 2016). Second, Goal₁ is a “particular type,” that is, a type that is ontologically dependent on some particular entity such as clay₁ (Brodaric & Grüninger, 2018). Third, a formal representation of Goal₁ may warrant consideration from a realist perspective because Goal₁ can never be instantiated: a desire-directed goal ceases to exist when it is achieved, and the associated desire is satisfied (see Section 2.1; and also Section 5.2).

We introduce the class Act of trying and its subclass Successful act under the category of process, corresponding to Hobbs and Gordon's (2010) notions of “trying” and “success,” respectively. An act of trying is a process in which an agent attempts to achieve a goal, that is, a process in which the agent strives to bring into existence an instance of that goal (for in-depth discussions on trying, see Grünbaum, 2008; Hornsby, 1980; McCann, 1975; O'Shaughnessy, 1980; Stuchlik, 2013). A successful act, in turn, is an act of trying that results in goal achievement, that is, an act of trying in which some agent does bring into existence an instance of the relevant goal. To clarify the dispositional account of intention, we propose that an intention is an intrinsic disposition to perform an act of trying that arising from the interactions between occurrent beliefs and occurrent desires. For simplicity, we will continue to use expressions of the form “intention to V” (where “V” stands for a verb such as “use”) rather than the more explicit form “intention to try to V.”

To provide a general illustration of our ontology of belief, desire, and intention, suppose that I intended to shape and dry clay₁ to be able to contain liquid. At that time, intention₁ comes into being, where intention₁ is my intention to shape and dry clay₁ in such a way as to be able to contain liquid. We can analyze intention₁ as my intrinsic disposition to perform an act of trying to shape and dry clay₁ that comes into being as a result of the interactions between occurrent beliefs and occurrent desires such as belief₁ and desire₁. Suppose further that I intentionally shaped and dried clay₁ so that it is now capable of containing liquid. At that time, intention₁ is realized in my successful act (act₁) of shaping and drying clay₁ and, with the achievement of Goal₁, act₁ has as part a particular process of clay₁ being capable of containing liquid.

3.1. Canonical Artifacts and Usefacts

We introduced the terms “canonical artifact” and “usefact” in Section 1.3. A canonical artifact is an entity that is intentionally produced for a specific purpose. A usefact is an entity that is intended to be used for some accidental purpose, i.e., for some purpose that is different from the one (if any) for which the entity was intentionally produced (cf. Evnine's (2022) “counter-use”). We will further elucidate the notions of a canonical artifact and a usefact by illustrating them with four examples involving material entities.

In this example, clay₁ at time t₁ is a canonical artifact because it is intentionally shaped and dried for the purpose of being able to contain liquid. Assume that the term “pot” is defined as: “A material entity that is intentionally produced for the purpose of being able to contain liquid.” Under the continuity view of production (which may involve some counterintuitiveness; see Section 2.2), the production of a canonical artifact at time t₁ is explicable by clay₁ instantiating the class Pot at time t₁.

We now address the choice of Example 1 as a driving example of canonical artifacts. Some may view the clay example as a simplistic or toy example of canonical artifacts and prefer to consider practical examples of complex canonical artifacts, such as screwdrivers and cars, as found in the industry domain; cf. Houkes and Vermaas's (2009a) critique of the “artisan model” of artifacts. Nevertheless, we chose to illustrate canonical artifacts with this simple example. Indeed, complex canonical artifacts involve the aggregation of a multiplicity of diverse parts and their ontological analysis would require an in-depth study of their structure (Grüninger et al., 2020; Koslicki, 2008), in particular their features and components (Sanfilippo et al., 2016). This may call for a non-standard mereology, whose study extends beyond the scope of this paper (see Masolo et al., 2020 for a formal analysis of composition and Section 8 for a pointer to future inquiry).

In this example, pebble₂ at time t₂ is a usefact because it is intended to be used for the accidental purpose of keeping papers in place, or more simply, because it is a natural object that is intended to be used for some purpose (see Section 1.3). Assume that the term “paperweight” is defined as follows: “A material entity that is intended to be used for the accidental purpose of keeping papers in place.” Under the continuity view of use, pebble₂ being intended to be used in this way at time t₂ can be analyzed as pebble₂ instantiating the class Paperweight at time t₂.

A few clarificatory remarks concerning Example 2 are in order. First, the expression “be intended to be used to do” is employed throughout the paper independently of whether it is actually used for that purpose. The distinction between use intention (even without associated actual use) and actual use has been highlighted by Borgo and Vieu (2009) and Toyoshima et al. (2022b).

We characterize the notion of a usefact in terms of use intention rather than actual use. This offers two key advantages. Firstly, it helps to understand a wide range of theories of artifacts, including Borgo and Vieu's (2009) theory, according to which pebble₂ at time t₂ is an artifact (see Section 7.1). Secondly, it enables a clear, realizable-based distinction between canonical artifacts and usefacts (see Sections 3.3 and 3.4). Consequently, our notion of a usefact is broader than the anthropological notion of a naturefact, as the latter is associated with actual use (see Sections 1.1 and 3.5.1).

Second, there are at least two different, broad and narrow, senses of the term “use” in relation with production. According to the broad interpretation, production is a kind of use, as is illustrated by Houkes and Vermaas's (2010, 2014) statement that a “make plan” is a kind of “use plan” (see Section 7.2.2 for details). According to the narrow interpretation, use is distinct from production. In this article, we adopt the narrow interpretation of “use” to maintain a sharp distinction between canonical artifacts and usefacts. However, we will also investigate a type of use closely linked to production—aligning roughly with the broad interpretation—under the term “use for production” in Section 6.3.

In this example, clay₁ at time t₁’ remains a canonical artifact after having become so at time t₁. Simultaneously, it is also a usefact because it is intended to be used for the accidental purpose of holding a door open, which is different from the purpose for which the canonical artifact clay₁ was intentionally produced (in the sense of the continuity view) at time t₁, i.e., the purpose of being able to contain liquid. Assume that the term “doorstop” is defined as follows: “A material entity that is intended to be used for the accidental purpose of holding a door open.” Under the continuity view of use, the usefact clay₁ being intended to be used in this way at time t₁’ is explicable by clay₁ instantiating the class Doorstop at time t₁’.

In this example, clay₁ at time t₁” continues to be a canonical artifact after having become so at time t₁. Contrary to clay₁ at time t₁’, it is not a usefact because it is intended to be used for the purpose of containing liquid and this purpose is not accidental, as it is the same as the purpose for which the canonical artifact clay₁ was intentionally produced (in the sense of the continuity view) at time t₁, i.e., the purpose of being able to contain liquid. Informally speaking, the canonical artifact clay₁ is properly used at time t₁” but it is not a usefact anymore.

3.2. Intentional Realizable Entity

We will analyze the notions of a canonical artifact and a usefact in terms of the ontology of realizable entities. More specifically, we will provide a realizable-based definition of the terms “material canonical artifact” and “material usefact.” Let us first consider the canonical artifact clay₁ at time t₁. We will focus on the intrinsic disposition d₁ that clay₁ comes to bear at time t₁ purely in virtue of its specific physical (e.g., container-like) structure (see Section 6.1 for the fundamental issue of the individuation of realizable entities; and we will not repeat this point in introducing realizable entities below).

We propose to analyze the canonical artifact clay₁ at time t₁ in terms of the intrinsic disposition d₁; in particular, its intentional dimension can be understood in terms of two distinctive features of d₁. To this end, we will reintroduce the following particular entities, based on the discussion in Section 2.3:

belief₁: my occurrent belief that clay₁ will be capable of containing liquid when it is suitably shaped and dried.

First, d₁ comes into being through an intentional act, as it comes into being through act₁ and act₁ is a realization of intention₁. Second, a realization of d₁ is determined by an associated goal (i.e., Goal₁) in the sense that the former is a process of clay₁ containing liquid and an instance of the latter is a process of clay₁ being capable of containing liquid. In light of these observations, we define the term “intentional realizable entity” as follows ² :

By this definition, d₁ (borne by clay₁ at time t₁) is an intentional realizable entity.

Definition 1 can be further clarified as follows. First, this definition, especially item (ii), implies a close connection between intentional realizable entities and goals. However, this should not be conflated with the statement that any realizable entity is goal-directed in the sense that it is directed toward some associated type of processes (Tugby, 2024). We employ the term “goal” in the intentional sense of the term in this paper (see Section 2.3). For instance, the intrinsic disposition of uranium to decay is goal-directed in the broad, non-intentional sense of the term; but it is not goal-directed in the narrow, intentional sense of the term, since uranium can decay naturally without human intervention.

Relatedly, we will use the term “goal” rather than “purpose” when characterizing intentional realizable entities to encompass a wider range of canonical artifacts, extending beyond technical artifacts. For instance, McKitrick (2024) defines “artificial disposition” as “the disposition[s] that human beings deliberately cause something to have, in order to serve some purpose,” in alignment with her dispositional pluralism, which underpins our ontology of realizable entities (see Section 2.1). While our notion of an intentional realizable entity can be linked to McKitrick's notion of an artificial disposition, it is intended to be more general (for an illustration, see our analysis of Koslicki & Massin's (2025) “placebo capacity” in terms of intentional realizable entities in Section 3.5.2).

Second, one may argue that the term “intentional act” in Definition 1 is generally susceptible to a wide interpretation (cf. Borgo et al.'s (2014a) taxonomy of actions) and must be here interpreted in a constrained way in order to characterize the notion of a canonical artifact. For example, one may claim that intentional acts should involve physical modifications (e.g., shaping and drying).

On the contrary, we suggest that the term “intentional act” in the definition should be construed broadly enough to accommodate “mental selection” (cf. Borgo & Vieu, 2009). This broad understanding is motivated by the idea that, while physical modification has been a standard criterion for the production of artifacts (cf. Preston, 2022a; see also Section 7.2), the production of some kind of canonical artifacts may be better explained in terms of the notion of “appropriation” figuring in philosophy of artifacts (Hilpinen, 1992; Juvshik, 2021a; Preston, 2013, pp. 96–103; Thomasson, 2014). Examples of such canonical artifacts include ready-made artworks (e.g., Marcel Duchamp's Fountain; see also Evnine, 2013), where “appropriation” refers roughly to an intentional act of making an artwork from some pre-existing entity (or entities) with little or no physical modification (e.g., signing a urinal). Insofar as material artifacts are concerned, the philosophical notion of appropriation can be construed within our framework as an intentional act in which a material entity becomes a canonical artifact through little or no physical modification. We will delve into this point in Section 3.3.

Third, one might consider simplifying Definition 1 by omitting item (ii): “intentional realizable entity = _def. A realizable entity that comes into being through an intentional act.” While this simpler definition might seem sufficient to capture the intentional dimension of canonical artifacts, it would fail to distinguish canonical artifacts from so-called “by-products”: roughly, unintended or purposeless outcomes of production processes. Although a full-fledged ontological analysis of products falls outside the scope of this paper (Vignolo, 2010a, 2010b), the notion of a by-product is used in domain ontologies, as is witnessed by class terms such as Poultry by-product (http://purl.obolibrary.org/obo/FOODON_00004144) in FoodOn, a BFO-compliant ontology of food and food production (Dooley et al., 2018).

To illustrate, consider a classical example of a by-product (e.g., Preston, 2022a): when sawdust is unintentionally produced during a woodworking operation and, by hypothesis, lacks any intended purpose, its combustibility is not an intentional realizable entity according to Definition 1. However, under some interpretation of the simpler definition, it could implausibly count as an intentional realizable entity, as the woodworking operation is an intentional act.

Fourth and finally, one may argue that the terms “is determined by” and “an associated goal” in Definition 1 should be construed in a suitably restrictive manner. While the terms “determined” and “associated” may be difficult to characterize generally, they can be elucidated with examples. ³ We propose that these terms should be understood widely enough to encompass a wide range of canonical artifacts, extending beyond technical artifacts. In particular, regarding the term “associated,” consider the analysis of Example 1: Goal₁ is directly associated with act₁, as desire₁ (through which Goal₁ comes into being) and intention₁ (realized in act₁) both involve the idea of clay₁ becoming capable of containing liquid. However, the association between intentional acts and goals may be more indirect when it comes to non-technical artifacts. We will illustrate this point with Koslicki and Massin's (2025) notion of a “faith-based artifact” in Section 3.5.2.

3.3. Material Canonical Artifacts and Novel Intentional Realizable Entities

We introduced in Section 3.2 the term “intentional realizable entity” to analyze canonical artifacts, particularly their intentional dimension. One may attempt to define a “material canonical artifact” as “a material entity that bears an intentional realizable entity.” However, in our view, while bearing an intentional realizable entity is a necessary condition for a material entity to be a canonical artifact, it is not sufficient. This is because material usefacts can also bear intentional realizable entities.

To illustrate, consider Example 2: at time t₂, pebble₂ is intended to be used to keep papers in place. We will focus on two realizable entities involved in this example. One is the intention (intention₂) to use pebble₂ to keep papers in place that comes into being at time t₂, where intention₂ is an intrinsic disposition. The other is the extrinsic disposition d₂ of pebble₂ to keep papers in place that comes into being at time t₂, where d₂ is an extrinsic disposition, as pebble₂ bears d₂ in virtue of the existence of intention₂, external to pebble₂. Such an extrinsic disposition can be understood within existing foundational frameworks, such as a role in BFO (see Section 5). Furthermore, we can justify the existence of d₂ using a Borgo and Vieu’s (2009) analysis of Example 1: at time t₂, pebble₂ is “mentally selected” to keep papers in place, and simultaneously, the “attributed capacity” to keep papers comes to inhere in a new material entity paperweight₂ which is constituted by pebble₂ (see Section 7.1 for details on Borgo & Vieu's theory of artifacts).

According to Definition 1, d₂ is an intentional realizable entity for two reasons. First, d₂ comes into being through the intentional act of mental selection. Second, a realization of d₂ is a process of pebble₂ keeping papers in place and it is determined by a goal associated with the intentional act of mental selection (e.g., a process type whose achievement results in a process of pebble₂ keeping papers in place). Therefore, both the intrinsic disposition d₁ (borne by clay₁ at time t₁) and the extrinsic disposition d₂ (borne by pebble₂ at time t₂) are intentional realizable entities. Nonetheless, clay₁ at time t₁ is a canonical artifact, whereas pebble₂ at time t₂ is a usefact. To summarize, defining “material canonical artifact” as “a material entity that bears an intentional realizable entity” is implausible, as material usefacts serve as counterexamples.

More generally, the mere fact that a material entity bears an intentional realizable entity fails to distinguish canonical artifacts from usefacts. Arguably, this is because it fails to capture the ontological difference between production and use: a canonical artifact is an entity intentionally produced for a specific purpose, whereas a usefact is an entity intended to be used for some accidental purpose.

In our view, the notion of novelty of a realizable entity can serve to specify the distinction between production and use, as well as between canonical artifacts and usefacts. To elaborate, we will focus on the intrinsic disposition d₂’ of pebble₂ to keep papers in place (to be explained in more detail) in contradistinction with the extrinsic disposition d₂ of pebble₂ to keep papers in place, which comes into being at time t₂ (as previously introduced).

On the one hand, d₂’ is an intrinsic disposition, as pebble₂ bears d₂’ purely in virtue of its specific physical structure (e.g., its solidity). Furthermore, d₂’ can exist even in the absence of any associated use intentions, such as intention₂. In particular, d₂’ exists before time t₂, when intention₂ comes into being. Therefore, d₂’ is not an intentional realizable entity because it does not come into being through an intentional act.

On the other hand, d₂ is an extrinsic disposition, as pebble₂ bears d₂ in virtue of the existence of intention₂, external to pebble₂. Specifically, d₂ comes into being at time t₂ and ceases to exist when intention₂ ceases to exist, even without any physical changes of pebble₂. As previously established, d₂ is an intentional realizable entity.

Despite these ontological differences, d₂ and d₂’ are intimately connected: whenever d₂ is realized in a particular process, d₂’ is also realized in the same process. To put it differently, d₂ is not a “novel” realizable entity for its bearer pebble₂ in the sense that any realization of d₂ is also a realization of d₂’, which exists before d₂ comes into being. At time t₂, pebble₂ comes to bear d₂ and d₂ is an intentional realizable entity. However, pebble₂ does not become a canonical artifact at time t₂; rather, it becomes a usefact, due to the “non-novelty” of d₂ for its bearer pebble₂.

Let us now compare the intrinsic disposition d₁ of clay₁ to contain liquid, which comes into being at time t₁, with the extrinsic disposition d₂ of pebble₂ to keep papers in place, which comes into being at time t₂. Although both d₁ and d₂ are intentional realizable entities, clay₁ at time t₁ (unlike pebble₂ at time t₂) does not bear any realizable entities that existed before d₁ came into being and that would be realized in the same processes in which d₁ is realized. Unlike d₂, d₁ is an intentional realizable entity whose realization is of a “new” kind for its bearer clay₁. At time t₁, clay₁ is a canonical artifact in virtue of bearing d₁, because d₁ is an intentional realizable entity that is “novel” for its bearer clay₁.

To spell out the idea of the “novelty” of intentional realizable entities involved in canonical artifacts, we introduce the terms “novel realizable entity” and “novel intentional realizable entity” as follows ⁴ :

By these definitions, d₁ is a novel intentional realizable entity, whereas d₂ is not a novel realizable entity owing to d₂’ and consequently is not a novel intentional realizable entity. We can now define the term “material canonical artifact” in terms of the notion of a novel intentional realizable entity as follows:

By this definition, clay₁ at time t₁ (as well as at times t₁’ and t₁”) is a material canonical artifact in virtue of bearing the novel intentional realizable entity d₁. In contrast, pebble₂ at time t₁’ is not a material canonical artifact, despite bearing the intentional realizable entity d₂, as d₂ is not a novel realizable entity.

We offer some clarificatory remarks on the argumentation developed so far, including Definitions 2–4. First, our characterization of production in terms of novel realizable entities is based on Definition 4: the production of the canonical artifact clay₁ at time t₁ corresponds to the novel (and intentional) realizable entity d₁ coming into being at that time. However, our realizable-based approach to production rests on comparing the realizations of two different realizable entities, or more specifically, on the identity of the relevant realizations, which are processes, as is explicitly indicated by item (ii) of Definition 2. We will discuss the foundational topic of the identity of processes in Section 6.2.

Second, our analysis of Example 2 (usefact and natural object) can be applied, mutatis mutandis, to Example 3 (usefact and canonical artifact). In the latter example, consider the extrinsic disposition d₃ of clay₁ to hold a door open, which comes into being at time t₁’ and the intrinsic disposition d₃’ of clay₁ to hold a door open, which comes into being at and continues to exist after time t₁. On the one hand, d₃ is an extrinsic disposition, as clay₁ bears d₃ in virtue of the existence of the intention to use clay₁ to hold a door open, which is external to clay₁. On the other hand, d₃’ is an intrinsic disposition, as clay₁ bears d₃’ purely in virtue of its specific physical structure (e.g., its solidity). According to Definitions 2–4, clay₁ does not become a material canonical artifact at time t₁’ because d₃ is not a novel realizable entity due to the existence of d₃’ and thus is not a novel intentional realizable entity. Instead, clay₁ continues to be a material canonical artifact after time t₁ because it continues to bear the novel intentional realizable entity d₁ after that time.

Third, we propose to characterize canonical artifacts in terms of novel intentional realizable entities, rather than novel intentional intrinsic dispositions, even though d₁—featured in our driving example of canonical artifacts (Example 1)—is a novel intentional intrinsic disposition. This broader approach is motivated by the need to accommodate a wider range of artifacts beyond technical artifacts, even though many kinds of technical artifacts may indeed be characterized in terms of novel intentional intrinsic dispositions.

In the context of material artifacts, for instance, the philosophical notion of appropriation (see Section 3.2) can be understood within our framework as an intentional act in which a material entity becomes a canonical artifact in virtue of bearing a novel intentional extrinsic disposition. This is because acquiring an extrinsic disposition does not necessarily involve physical changes of the material entity that bears it. We will further illustrate novel intentional extrinsic dispositions with Koslicki and Massin's (2025) notion of a placebo capacity in Section 3.5.2.

Fourth and finally, it is important to examine whether our realizable-centered analysis of canonical artifacts—developed under the continuity view of both production and use—can be reconstructed under the non-continuity view of production and/or use (see Section 2.2). To address this, we need to consider how Definition 1 (intentional realizable entity) and Definition 2 (novel realizable entity) can hold under the non-continuity view of production and/or use.

Regarding Definition 1, it is straightforward to adapt this definition to both the non-continuity view of production and the non-continuity view of use. In particular, with respect to item (i) in the definition, if a realizable entity comes into being through an intentional act under the continuity view of production (or use), this condition is trivially satisfied under the non-continuity view of production (or use).

Regarding Definition 2, we will defer the question of how it can hold under the non-continuity view of production and/or use, as the answer depends on the specific details of the endorsed non-continuity view of production (or use), including whether it is the discontinuity view or the constitution view of production (or use). We will address this issue in Section 7.1 in discussing Borgo and Vieu's (2009) theory of artifacts which is committed to the constitution view of both production and use. ⁵

3.4. Material Usefacts and a Special Kind of Non-Novel Intentional Realizable Entities

We now turn to a close examination of the notion of a usefact. We will define the term “material usefact” based on the realizable-based definition of “material canonical artifact” (Definition 4) provided in Section 3.3. A naïve attempt might define this term as “a material entity that bears an intentional realizable entity that is not a novel realizable entity,” or more simply as “a material entity that bears a non-novel intentional realizable entity,” where the term “non-novel realizable entity” is defined as “a realizable entity that is not a novel realizable entity.” This naïve definition would correctly classify pebble₂ at time t₁’ as a material usefact, given that it bears d₂, a non-novel intentional realizable entity.

However, this naïve definition may be too broad. To illustrate, consider Example 4: at time t₁” (later than time t₁), clay₁ is intended to be used to contain liquid. Let us focus on the extrinsic disposition d₄ of clay₁ to contain liquid, which comes into being at time t₁”. We can justify the existence of d₄ in the same way we justified the existence of the extrinsic disposition d₂ of pebble₂ to keep papers in place in Example 2, drawing on Borgo and Vieu's (2009) theory of artifacts (see Section 3.3). Like d₂, d₄ is a non-novel intentional realizable entity. In particular, d₄ is non-novel because if d₄ is realized in a process, then the intrinsic disposition d₁ of clay₁ to contain liquid (which comes into being at time t₁) is realized in the same process. According to the naïve definition, clay₁ at time t₁” would be a material usefact in virtue of bearing the non-novel intentional realizable entity d₄. However, this would contradict our notion of a usefact as an entity that is intended to be used for an accidental purpose, that is, a purpose different from the one (if any) for which the entity was intentionally produced.

Therefore, usefacts should be considered as material entities that bear a specific kind of non-novel intentional realizable entities. To this end, we introduce the relation of “being non-novel because of” between two realizable entities, based on Definition 2 (novel realizable entity), as follows: ⁶

This definition enables us to classify non-novel intentional realizable entities (such as d₂, d₃ and d₄) into two types:

Non-novel because of some novel intentional realizable entity

The extrinsic disposition d₄ of clay₁ to contain liquid is non-novel because of the intrinsic disposition d₁ of clay₁ to contain liquid, where d₁ is both novel and intentional (Example 4; at time t₁”).

Non-novel because of some realizable entity that is not a novel intentional realizable entity

The extrinsic disposition d₂ of pebble₂ to keep papers in place is non-novel because of the intrinsic disposition d₂’ of pebble₂ to keep papers in place, where d₂’ is not intentional and thus not a novel intentional realizable entity (Example 2; at time t₂).

The extrinsic disposition d₃ of clay₁ to hold a door open is non-novel because of the intrinsic disposition d₃’ of clay₁ to hold a door open, where d₃’ is not intentional and thus not a novel intentional realizable entity (Example 3; at time t₁’).

Given that material usefacts can be characterized in terms of bearing the second type of non-novel intentional realizable entities, we define the term “material usefact” as follows:

Definition 6 material usefact

material usefact = _def. A material entity that bears an intentional realizable entity which is non-novel because of some realizable entity that is not a novel intentional realizable entity.

By this definition, pebble₂ at time t₂ is a material usefact in virtue of bearing d₂ and clay₁ at time t₁’ is a material usefact in virtue of bearing d₃. However, clay₁ at time t₁” is not a material usefact, despite bearing d₄.

We remark that Definition 6 could be simplified as follows. An alternative approach would be to first define the term “usefactual realizable entity” as follows: ⁷

By this definition, d₂ (borne by pebble₂ at time t₂) and d₃ (borne by clay₁ at time t₁’) are usefactual realizable entities. We note that usefactual realizable entities are a subtype of extrinsic dispositions because any usefactual realizable entity is borne by its bearer in virtue of the existence of some intention to use the bearer for some purpose, and this use intention is external to the bearer. This allows us to provide a simpler definition of the term “material usefact” in terms of the notion of a usefactual realizable entity as follows:

By this definition (as well as by Definition 6), pebble₂ at time t₂ and clay₁ at time t₁’ are material usefacts.

3.5. Theoretical Application

To illustrate the theoretical applications of our realizable-centered approach to artifacts, we will analyze two kinds of entities discussed in the literature: naturefacts in anthropology (Section 3.5.1) and faith-based artifacts in Koslicki & Massin's (2025) account of artifacts (Section 3.5.2).

4.1. Three Theories of Functions: Designer-Based, Etiological and Systemic

First, according to the designer-based theory, a function is determined by the intention(s) of the designer(s) of the function bearer (Baker, 2007). Here, “designer” is understood broadly to include authors, producers, inventors, and generally any agents responsible for creating the artifact or artifact kind. For instance, the function of a screwdriver to turn screws is determined by the designer's intention, as the designer intentionally produced (or designed) the screwdriver to perform this function. We provide the following realizable-based definition of the term “design function”:

We note that the phrase “is determined by” in this definition carries the same technical meaning as in Definition 1 (see Section 3.2 on intentional realizable entities).

Second, according to the etiological theory (aka causal-historical theory), a function of a bearer of a certain kind is determined by the successful production of an effect in past members of that kind, which explains why members of the kind continue to be selected or reproduced (Milikan, 1984, 1989; Neander, 1991; Wright, 1973). For instance, the function of this particular screwdriver (an instance of the type Screwdriver) to turn screws is determined by past successful instances of Screwdriver producing this effect, explaining why instances of Screwdriver continue to be used and reproduced. As a proper function is predicated upon a type of function bearer, we provide the following realizable-based definition of “proper function with respect to type T”:

Third, according to the systemic theory (aka the causal contribution theory) of functions—which we formulate here using the notions of a capacity and its manifestation—a function is determined by the manifestation of a capacity of the bearer that causally contributes to the manifestation of a capacity by the system of which the bearer is a component (Borgo et al., 2016; Burek et al., 2016; Cummins, 1975; Mizoguchi et al., 2016). For example, the function of a screwdriver to turn screws can be analyzed in terms of the capacity of this screwdriver whose manifestation causally contributes to the manifestation of a capacity of the “screw-turning system,” which has as component the screwdriver, a screw and an agent using the screwdriver.

The notion of a system function is based on the twin conceptual pillars: the notions of a system and causal contribution. Regarding systems, we assume that they can be intuitively understood as entities with two or more components related in a specific way, although their ontological analysis remains challenging (Miettinen, 2018; Mizoguchi & Borgo, 2021; Perez, 2020; Rodrigues, 2023). For the related notion of a mechanism, see Machamer et al. (2000) and Röhl (2012).

The notion of a system connects to the ontology of realizable entities through Vetter's (2015, ch. 4.5) systemic account of extrinsic dispositions (“extrinsic potentialities” in her terms). The core idea is that bearing an extrinsic disposition is both necessary and sufficient for the bearing of a joint disposition (“joint potentiality” in her terms) by a system having as component the bearer and other objects. Here, a joint disposition (in the Vetter-inspired sense) is an intrinsic disposition, as outlined in Toyoshima et al.'s (2022c) formalization of this systemic account of extrinsic dispositions. System functions can be considered as a kind of extrinsic dispositions, as they exist within the system that includes the bearer.

Concerning causal contribution, we interpret it in terms of realizations of realizable entities, since intrinsic dispositions, and more generally realizable entities, can be seen as causal properties in philosophy (Mumford & Anjum, 2011; N. E. Williams, 2019b; Ingthorsson, 2021; cf. Nolan's (2015) dissenting voice) and in formal ontology (Röhl & Jansen, 2011; Toyoshima, 2020; Toyoshima et al., 2022c). By synthesizing these analyses of systems and causal contribution, we can provide the following realizable-definition of “system function”:

4.2. Discussion: Beyond the Function-Based Accounts of Artifacts

For expository clarity, we examine the systemic, etiological and designer-based theories of functions in sequence, within the context of function-based accounts of artifacts. Our methodology involves scrutinizing, from our realizable-based perspective on material artifacts, how the notion of a material artifact is characterized in terms of bearing a [system/proper/design] function.

First, the systemic theory of functions struggles to account for many kinds of artifacts (as discussed in detail by Koslicki & Massin, in press). Consider, for example, material entities bearing novel intentional intrinsic dispositions: these are material canonical artifacts, yet they fall outside the scope of the systemic theory since any system function is an extrinsic disposition, and no intrinsic disposition can be a system function (see also Artiga, 2016; Röhl & Jansen, 2014).

It is worth noting that usefactual realized entities form a subtype of system functions—consider d₂ borne by pebble₂ at time t₂’ (Example 5; see Section 3.5.1). In the ontology of functions, the term “use function” refers roughly to a function ascribed to an entity in virtue of an agent's actual usage for a purpose intended by the agent. More precisely, it aligns with Bahr's (2019) “sporadic user-intended function” or Toyoshima et al.'s (2022b) “narrow use function”; cf. Artiga's (2016) “functioning as” in contrast with “having a function.” Examples include the use function of keeping papers in place of the naturefact pebble₂ at time t₂’. From a realizable-based perspective, use functions can be considered as usefactual realized entities (Toyoshima et al., 2024b). Consequently, the system theory of functions is sometimes motivated to accommodate the notion of a use function (e.g., Mizoguchi et al., 2016).

Second, the etiological theory of functions encounters difficulties with “innovative or atypical artefacts” (Vermaas & Houkes, 2003, p. 266) or prototypes, defined as “a first exemplar of a newly created artifact kind” (Koslicki & Massin, in press). This difficulty arises because proper functions of instances of type T requires a causal history involving the successful production of an effect that explains the continued use and reproduction of T instances—something that prototypes lack by definition.

A related issue with the etiological theory is its relativity of proper functions to types of bearers, which makes types of bearers ontologically prior to types of proper functions. This makes the etiological theory less suitable for developing a general ontology of artifacts intended to remain neutral regarding definitions of domain-specific artifact-related terms such as “pot”, “paperweight,” “doorstop,” and “amulet” (see Section 2.2; and Section 5.4 for a related issue concerning the BFO etiological account of functions).

Third, the designer-based theory of functions has limitations compared to our realizable-centered approach. Assuming that the notion of design primarily characterizes canonical artifacts, we now compare design functions with novel intentional realizable entities, or more generally with intentional realizable entities. First of all, design functions come into being through an intentional act of design (item (i) of Definition 11), whereas intentional realizable entities come into being through an intentional act simpliciter (item (i) of Definition 1). Clarifying their comparison requires disambiguating the term “design.”

There are at least two interpretations of “design”: a broad one, where an intentional act of design need not involve physical modification and may include mental selection, and a narrow one, where an intentional act of design must involve physical modification and excludes mere mental selection.

This ambiguity creates a dilemma. If one accepts the broad interpretation, defining “material artifact” as “a material entity that bears a design function” may fail to sharply distinguish canonical artifacts (e.g., clay₁ at time t₁ in Example 1) from usefacts (e.g., pebble₂ at time t₂ in Example 2). If one accepts the narrow interpretation, the same definition may exclude canonical artifacts involving appropriation, such as ready-made artworks (see Sections 3.2 and 3.3). Either way, the notion of design struggles to underpin a plausible conception of canonical artifacts.

Moreover, realizations of design functions are determined by the designer's intention (or designers’ intentions) (item (ii) of Definition 11), whereas realizations of intentional realizable entities are determined by an associated goal (item (ii) of Definition 1). To clarify the difference, consider (material) faith-based artifacts, which we analyzed as a kind of (material) canonical artifacts (see Section 3.5.2). The designer-based theory could characterize pebble₂ at time t₃ in Example 6 as a material artifact bearing the design function of dispelling evil spirits. However, this analysis commits to the controversial claim that some realizable entities have realizations that are nomologically impossible (though see Jenkins & Nolan, 2012). In contrast, our analysis of Example 6 in terms of placebo capacities avoids this problematic view of realizable entities and functions—cf. Preston's (2013, p. 177) “phantom function” as a function that a kind of artifact is “constitutionally incapable of performing”; see also Holm's (2017) critical examination.

Finally, a “pluralist” function-based account of artifacts combining two or more theories of functions would face similar issues (Koslicki & Massin, in press). For instance, one might attempt to characterize canonical artifacts and usefacts using design functions and use functions, respectively, on the assumption that both are reducible to systemic functions, and consequently, no design functions are intrinsic dispositions (cf. Mizoguchi et al., 2016). This systemic and designer-based theory (enriched with use function) would nonetheless inherit the challenges of each individual theory. To illustrate with Example 6, it remains unclear whether the agent at time t₃ is a “designer” or “user” with respect to pebble₂, making it uncertain whether pebble₂ at time t₃ is a canonical artifact or a usefact, unless “designer intentions” and “user intentions” are clearly specified. This uncertainty stems from the limitations of the intention-based approach to artifacts, particularly with respect to the designer-based theory of functions.

In sum, we examined traditional, function-based accounts of artifacts from a realizable-centered perspective, by considering three prominent theories of functions (viz., designer-based, etiological and systemic) or more concretely, by scrutinizing realizable-based characterizations of “design function,” “proper function,” and “system function.” This leads us to conclude that, despite their intuitive appeal (especially for design functions), these function-based accounts of artifacts (whether monist or pluralist) lack the generality of the realizable-centered approach to artifacts we develop here.

5.1. Preliminaries

5.1.1. A General Overview of BFO

Figure 2 provides a taxonomy of classes, including terms from BFO and our new terms to be introduced in Section 5.2. This taxonomy maps BFO class terms to class terms (in square brackets) from the taxonomy introduced in Section 2.1 (see Figure 1). Strictly speaking, BFO does not include the category of externally grounded realizable entity (parenthesized here), as the term “externally grounded realizable entity” is used to explain the BFO category of role; however, we include it to show the correspondence between this BFO term and our term “extrinsic disposition” (Toyoshima et al., 2022c).

OPEN IN VIEWER

Figure 2.

A BFO-based taxonomy of classes used in Section 5.

BFO includes the top-level distinction between continuants and occurrents. On the continuant side, independent continuants include material entities such as organisms and heaps of stones. They also include immaterial entities, with sites as one subtype. A BFO:site is defined as: “An immaterial entity in which objects such as molecules of air or organisms can be contained” (Arp et al., 2015, p. 184; see also Donnelly, 2005). Examples include my nasal cavity and the hold of a ship.

A specifically dependent continuant is a continuant that depends (existentially) on at least one independent continuant. A realizable entity is a specifically dependent continuant that can be realized in associated processes of specific correlated types in which the bearer participates. BFO currently includes three subtypes of realizable entities: dispositions, functions, and roles.

A disposition in BFO is a realizable entity that exists because of certain features of the physical makeup of its bearer—the independent continuant that bears it. It is an “internally grounded realizable entity”: if a BFO:disposition ceases to exist, then the physical makeup of the bearer is changed. Examples include the fragility of a glass and the flammability of a match.

A function in BFO is a BFO:disposition of a bearer with a specific kind of historical development. It is a BFO:disposition that its bearer possesses in virtue of its having a certain physical makeup because of how it came into being, either through evolution (for natural biological entities) or intentional design (for artifacts). Spear et al. (2016, p. 126) explain: “ ‘Came into being’ here strongly suggests that a functional disposition is one whose existence helps to causally explain the existence of the entity, or at least of the physical structure, that it is a disposition of.” Examples include the function of the heart to pump blood through the body or the function of a screwdriver to turn screws.

A role in BFO is 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 (optionality), and (2) is not such that, if this realizable entity ceases to exist, then the physical makeup of the bearer is changed (external grounding). Briefly, a BFO:role is an externally grounded and optional realizable entity. Examples include the role of being a student and the role of a stone marking a boundary. For further discussion on optionality here, see Röhl & Jansen, 2014; Toyoshima et al., 2022c; see also Toyoshima's (2021) foundational analysis of roles in BFO.

On the occurrent side, a process is an occurrent that exists in time by occurring—that is, by having temporal parts—and depends on at least one independent continuant as a participant. Examples of processes include cell division or a person's walk. A temporal region is an occurrent that is part of (the whole of) time.

5.2. Defining the Terms “Artifactual Capacity” and “Independent Continuant Artifact”

Based on the discussion in Section 5.1.2, we adapt the notions of a canonical artifact and a novel intentional realizable entity to the BFO ontological framework. Recognizing the importance of “terminological choice” (and ontological analysis) in ontology development (Neuhaus & Hastings, 2022), we introduce more accessible terms instead of using our technical terms “material canonical artifact” and “novel intentional realizable entity.”

For an alternative to “material canonical artifact,” we propose “independent continuant artifact.” First, we retain the term “artifact,” as existing BFO-compliant definitions of the term “(material) artifact” imply that it should refer to canonical artifacts (see Section 5.1.2). Second, we introduce “independent continuant artifact” instead of “material artifact” because some canonical artifacts are arguably independent continuants that are not material entities. For instance, trenches created during the First World War may be classified as BFO:sites. Given that BFO:sites can bear specifically dependent continuants (though see Neuhaus, 2026), they qualify as canonical artifacts by bearing novel intentional realizable entities.

For an alternative to “novel intentional realizable entity,” we propose “artifactual capacity.” We use here the term “capacity” for several reasons. First, capacity-based approaches to artifacts have been developed in formal ontology and in philosophy (Borgo & Vieu, 2009; Kassel, 2010; Koslicki & Massin, 2025; see also Section 7).

Second, while “capability” and “capacity” are sometimes considered as synonymous, a BFO-based analysis treats capabilities are a subtype of BFO:dispositions (Beverley et al., 2024). In contrast, we argue that canonical artifacts should be characterized by novel intentional realizable entities, rather than novel intentional intrinsic dispositions, where our term “intrinsic disposition” aligns with BFO:Disposition (see Section 2.1).

Third, the notion of a capacity can accommodate both intrinsic and extrinsic dispositions, including novel intentional extrinsic dispositions. For instance, Koslicki and Massin's (2025) “placebo capacities” can be analyzed as novel intentional extrinsic dispositions (see Section 3.5.2). Capacities are also characterized as “relational qualities,” synonymous with our term “extrinsic property,” according to a DOLCE-compliant analysis of functions, capabilities and capacities in the engineering domain (Borgo et al., 2021; Compagno, 2025; Compagno & Borgo, 2024b; Compagno et al., 2023).

We define “artifactual capacity” and “independent continuant artifact” as follows, acknowledging the needs for a BFO-friendly reinterpretation of the term “goal” in the definition of the former term:

artifactual capacity = _def. A realizable entity r (1) that comes into being through an intentional act; (2) whose realization is determined by an associated goal; and (3) such that the bearer has no realizable entity r’ such that (i) r’ exists before r comes into being and (ii) if r is realized in a process, then r’ is realized in the same process.

Second, we may need to reinterpret the term “goal” in a BFO-conformant way, as viewing goals as process types coming into being through some occurrent desire could conflict with BFO's realist methodology (see Section 2.3). The notion of a goal remains understudied in BFO-based ontology research. The BFO-compliant Information Artifact Ontology (IAO; Ceusters & Smith, 2015) defines “object specification” as “a directive information entity that describes an intended process endpoint” and lists “goal specification” as a synonym (http://purl.obolibrary.org/obo/IAO_0000005); for more details, see Duncan et al.'s (2024) analysis of the related IAO term “plan specification.” Based on this, we may loosely characterize a goal as a realizable entity that comes into being through some occurrent desire, such that it can be realized in “an intended process endpoint,” though this raises questions about which entity or entities bear the goal. We leave for future work a BFO-based rigorous characterization of goals.

5.3. Formal Characterization

We formalize “artifactual capacity” and “independent continuant artifact” in classical first-order logic, as this formalization aligns with existing ontologies (notably BFO) and avoids potential conceptual and technical issues with modal logical representations of realizable entities (including dispositions) in BFO (Toyoshima & Niki, 2021). Our first-order language includes a countably infinite supply of variables (for particulars only), standard logical symbols (including “¬,” “ ∧ ,” “,” “∀,” and “∃”) and predicates from Table 1 (imported from BFO unless specified otherwise). Free variables are universally quantified when ambiguity is absent and parentheses are used conventionally.

First of all, an artifactual capacity is a realizable entity:

A 1 AC ( x ) → RE ( x )

For items (1) and (2) in the definition of the term “artifactual capacity,” formal representation is challenging due to the complexity of notions such as intentional act, determination (“is determined by”) and goal (see Section 5.2). However, one possible way to formalize item (1) is to introduce the primitive predicate IA(x) (“x is an intentional act”), adopt the “results in formation of” relation (http://purl.obolibrary.org/obo/RO_0002297) from the BFO-compliant relation ontology (RO; Smith et al., 2005), reinterpret it more broadly as a relation (RF(x,y)) between a process and a continuant (not restricted to an anatomical entity) and state the following axiom: AC(x) → ∃y(IA(y) ∧ RF(y,x)). ⁸ Informally, this reads as: “any artifactual capacity is formed as a result of some intentional act.”

For item (3) in the definition, we formalize it using existing BFO relations:

A 2 AC ( x ) → ∃ y , t x [ INH ( x , y ) ∧ EXT ( x , t x ) ∧ ¬ ∃ r , t r [ ( RE ( r ) ∧ INH ( r , y ) ∧ EXT ( r , t r ) ∧ PREC ( t r , t x ) ∧ ∀ p ( REA ( x , p ) → REA ( r , p ) ) ] ]

(A2) can be informally read as: “Any artifactual capacity x inheres in an independent continuant y such that there is no other realizable entity r in y that exists before x comes into being and such that any of realizations of x is a realization of r.”

We formally define “independent continuant artifact” as follows:

D 1 ICA ( x ) ≜ INC ( x ) ∧ ∃ y ( AC ( y ) ∧ INH ( y , x ) )

(D1) can be informally read as: “An independent continuant artifact x is, by definition, an independent continuant in which some artifactual capacity inheres.”

We saw in Section 5.1.2 that existing BFO-compliant definitions of the terms “artifact” and “material artifact” tend to rely on the BFO category of realizable entity, particularly function. In that respect, our realizable-centered approach to artifacts can clarify realizable entities in BFO.

Initially, let us compare artifactual capacity with BFO:function. There are several key ontological differences between them. First, BFO:functions are a subtype of BFO:dispositions (or internally grounded realizable entities), whereas artifactual capacities can be externally grounded realizable entities. A case in point is Koslicki and Massin's (2025) “placebo capacities” (e.g., d₆ borne by pebble₂ at time t₃ in Example 6). Placebo capacities are artifactual capacities but are BFO:roles. In particular, they are optional realizable entities, as d₆ exists because pebble₂ is in contingent circumstances in which it does not have to be, to wit, the existence of intention₃.

Second, BFO's theory of functions seems to be committed to the non-continuity view of production, as the BFO notion of a function is closely tied to how the bearer “came into being” and helps to causally explain the bearer's existence. As the BFO theory of functions also seems to assume the continuity view of use, it may embody a typical hybrid approach to production and use (see Section 2.2). In contrast, the notion of an artifactual capacity is intended to be neutral with respect to whether the hybrid or monolithic approach is adopted; see Section 7.1 for discussion on adapting our approach to the non-continuity view, especially the constitution view, of both production and use.

Third, artifactual BFO:function is etiological and designer-based, as BFO explicitly endorses an etiological account of functions (Spear et al., 2016) and, for artifacts (to wit, canonical artifacts), intentional design is causally responsible for the physical makeup of the bearer of a BFO:function. We note, however, that while BFO claims functions to be etiological, artifactual BFO:functions would not qualify as proper functions in the sense of Definition 12 (see Section 4.1), as the intentional design involved in a given BFO:function pertains to the causal history of the particular bearer: “In the case of artifacts, the current [BFO:]function of an artifact is determined by its history of intentional design or selection by some agent” (Spear et al., 2016, p. 109, our italics added).

As a result, the BFO account of artifactual functions may inherit the issues of the etiological and designer-based theories of functions in providing a general ontology of artifacts (see Section 4.2). For instance, we raised concerns about types of bearers of proper functions being ontologically prior to types of these proper functions. Similar concerns might be formulated in BFO: “For BFO, […] if a function of a given kind exists, then its bearer must instantiate some independent continuant biological or artifactual universal. […] we take reference to kinds to be unavoidable in almost all talk of functions and take the question of whether or not a given thing has a function to be, at least in part, a question of the kind of thing it is” (Spear et al., 2016, p. 123).

Another issue is the ambiguity of the term “design” (e.g., whether it should be broadly construed to include mental selection or narrowly to require physical modification), as noted when we examined the designer-based theory of functions. This ambiguity is evident in BFO's explanations of functions, which sometimes refer to “selection” alongside intentional design: “this bearer [of a BFO:function] came into existence as a result of […] intentional design or selection for artifacts” (Spear et al., 2016, p. 107, our italics added) and “In the case of artifacts, the current [BFO:]function of an artifact is determined by its history of intentional design or selection by some agent” (Spear et al., 2016, p. 109, our italics added).

Following this, we compare BFO:role with artifactual capacity and usefactual realizable entity (see Section 3.4, especially Definition 7). First, we have already argued that some artifactual capacities (e.g., placebo capacities such as d₆) are BFO:roles and they fall outside the scope of BFO:function.

Second, usefactual realizable entities are BFO:roles, as illustrated by d₂ (borne by pebble₂ at time t₂ in Example 2) and d₃ (borne by clay₁ at time t₁’ in Example 3). Specifically, they are optional realizable entities, as d₂ exists because pebble₂ is in contingent circumstances in which it does not have to be, to wit, the existence of intention₂.

Furthermore, previous BFO-based analyses of functions have noted the existence of BFO:roles closely related to but distinct from BFO:functions. For instance, Spear et al. (2016) state: “when sticks from the woods are merely intentionally selected to be used as chopsticks, then they do not take on or have a function at all in the BFO sense, but rather a [BFO:]role” (Spear et al., 2016, p. 126) and “naturefacts and […] (merely intentionally) repurposed artifacts generally […] do not have functions in the BFO sense; but rather [BFO:]roles” (Spear et al., 2016, p. 127). As another example, Röhl and Jansen (2014, p. 14) argue: “the so-called ‘use functions’ […] are to be classified as roles […] in agreement with the BFO conception of roles.”

When BFO:roles come into being when their bearers (e.g., wooden sticks) “are merely intentionally selected to be used” for some accidental purpose (e.g., to serve as chopsticks), these BFO:roles can be analyzed as usefactual realizable entities. We can further add that naturefacts can be defined in terms of usefactual realized entities—a subtype of usefactual realizable entities (see Section 3.5.1)—and that, in Example 3, clay₁ at time t₁’ is a “(merely intentionally) repurposed artifact” and the usefactual realizable entity d₃ (borne by clay₁ at time t₁’) is a BFO:role. Recall also that use functions can be considered as usefactual realized entities (see Section 4.2).

In sum, our realizable-centered approach to artifacts clarifies functions and roles in BFO. On the one hand, artifactual BFO:functions can be understood as a specific kind of artifactual capacities. On the other hand, some BFO:roles are artifactual capacities while others are usefactual realizable entities. Although such use-related BFO:roles have been previously mentioned in the literature, we analyzed them systematically from a realizable-centered perspective on usefacts.

6.1. The Individuation of Realizable Entities

Since our realizable-centered approach to artifacts hinges on the ontology of realizable entities, the individuation and categorization of these entities are crucial—that is, determining which realizable entities exist and how a realizable entity is to be characterized. These determine which entities (particularly material entities) qualify as canonical artifacts, usefacts, or neither—and whether an entity can qualify as both.

To illustrate, suppose that at time t₄, pebble₂ is intentionally shaped into a cube (recall Example 2). On the one hand, pebble₂ at time t₄ would be a canonical artifact if the intrinsic disposition d₂’ of pebble₂ to keep papers in place ceases to exist due to intentional shaping, and a novel intentional intrinsic disposition d₂* of pebble₂ (say, to keep papers in place tightly) comes into being at time t₄. On the other hand, pebble₂ at time t₄ would be a usefact if d₂ “survives” the intentional shaping and continues to exist after time t₄ and d₂* is non-novel because of d₂’. Generally, the individuation of realizable entities (as exemplified by the issue of the existence of d₂’ at time t₄) is open to a variety of interpretations, sometimes leading to indeterminacy between classifying an entity as a canonical artifact or as a usefact.

Several approaches can tackle the thorny issue of individuating realizable entities. One pragmatic view holds that their individuation primarily requires domain-specific knowledge, with different domain ontologies adopting distinct principles established through “consensus creation” (Neuhaus & Hastings, 2022) among ontology developers and domain experts. For example, Arp et al. (2015, p. 102) acknowledge “the problem as to how [BFO:]dispositions are to be individuated” and state: “BFO itself does not provide a taxonomy of dispositions; it does not itself legislate concerning which types of dispositions exist, or how they are to be individuated. Rather, it leaves this task to the specific sciences” (Arp et al., 2015).

However, a domain-neutral principle for individuating realizable entities would be desirable. To explore this, consider Barton et al.'s (2018c) work on the (synchronic) identity of intrinsic dispositions. As a preliminary, we introduce the notions of a trigger and a categorical basis of an intrinsic disposition. A trigger of an intrinsic disposition is a process whose occurrence activates the realization of the intrinsic disposition; cf. McKitrick's (2018, ch. 6) event-based view of triggers within her dispositional pluralism. A categorical basis of an intrinsic disposition is a categorical (i.e., non-realizable) property (e.g., physical structure) of the bearer in virtue of which the intrinsic disposition exists (Prior et al., 1982; Röhl & Jansen, 2011). For example, the fragility of this glass is an intrinsic disposition that can be triggered by a process of pressing the glass with sufficient force and whose categorical basis is a specific molecular structure of the glass.

According to one criterion for the identity of intrinsic dispositions (“parthood criterion” in Barton et al.'s (2018c) terms), two intrinsic dispositions are identical if and only if they have the same categorical basis, the same class of minimal triggers and the same class of maximal realizations. The class of minimal triggers of a disposition d is the class of triggers of d for which no proper part is a trigger of d. The class of maximal realizations of d is the class of realizations of d which are not proper parts of another realization of d. To put it epistemically, according to this criterion, the identification of an intrinsic disposition amounts to the identification of the associated categorical property (or properties), the “smallest causal factor” which exceeds the threshold value for causation and the resulting whole causal chain of processes. Barton et al. argue that this criterion avoids “disposition multiplicativism”: the arbitrary proliferation of intrinsic dispositions that could undermine their causal and explanatory roles.

It would be valuable to extend the parthood criterion for the identity of intrinsic dispositions to the entire category of realizable entities, thereby developing a domain-neutral principle for individuating realizable entities. On the one hand, the notion of a categorical basis of an intrinsic disposition can be generalized to the broader notion of a categorical basis of a realizable entity (cf. Ferreira Ruiz & Hundertmark, 2024), as Toyoshima et al. (2022c) elaborate this notion with a focus on the causal import of realizable entities.

On the other hand, extending the notion of a trigger of an intrinsic disposition to the more general notion of a trigger of a realizable entity is more intricate. This difficulty arises partly because even the former notion remains largely unexplored in formal ontology and in philosophy. In biomedical ontologies, for instance, Röhl and Jansen (2011, p. 5) note: “It is true that in many cases the trigger of [an intrinsic] disposition might not be known, and in other cases we may even have good reasons to assume that no trigger is involved.” The latter claim aligns with McKitrick's (2018, ch. 6) philosophical argument for “trigger-less dispositions” (e.g., the intrinsic disposition of uranium to decay) within her dispositional pluralism. This presents a theoretical challenge to attempts to define a trigger notion characterizing all realizable entities. Moreover, Röhl and Jansen's former claim raises practical concerns about the applicability of a trigger-based characterization (if any) of realizable entities to ontology development. These theoretical and practical issues regarding triggers of realizable entities—encompassing both intrinsic and extrinsic dispositions—remain as key pointers for addressing the problem of individuation of realizable entities.

6.2. The Identity of Processes

Besides the individuation of realizable entities, the identity of processes is also a foundational topic relevant to our realizable-centered approach to artifacts, as Definition 2 (of “novel realizable entity”) relies on process identity—item (ii) being: “if r is realized in a process, then r’ is realized in the same process.” The same remark applies to the definition (i.e., Definition 5) of the relation “is non-novel because of” as well. The identity of processes significantly influences whether a realizable entity is novel or not, and consequently, which (material) entity is a canonical artifact or a usefact.

While the general notion of an identity criterion goes beyond the scope of this paper (Garbacz, 2022, 2023), we present two major views of processes in formal ontology: the manifestation view and the quality-change view; for details, see Rodrigues's (2023) discussion on these views. We briefly consider two corresponding identity criteria for processes (cf. Toyoshima & Barton, 2024b) with respect to our goal of elucidating the notion of a novel realizable entity.

First, the manifestation view is typically summarized as “events are manifestation of dispositions.” Paraphrasing in our terms, processes are realizations of realizable entities, particularly intrinsic dispositions. For instance, a process of this glass breaking can be analyzed as a realization of the fragility of the glass. This view underpins the UFO class Perdurant and has been practically applied in ontology-driven conceptual modeling (Almeida et al., 2019; Guizzardi et al., 2013; Guzzardi et al., 2016).

According to a simple version of the manifestation-based criterion for process identity, two processes are identical if and only if they realize the same intrinsic disposition(s) at the same time (Toyoshima & Barton, 2024b). As intrinsic dispositions are causal properties (see Section 4.1), this criterion may be seen as a (partial) dispositional interpretation of a causal criterion for the identity of processes: processes are identical if and only if they have the same causes and effects (Davidson, 1969).

Second, the quality-change view has been developed by Guarino et al. (2022), who use the term “event” rather than “process.” Central to their theory of events is the notion of a qualitative change, individuated by the triple <o, q, t> where o is an object (roughly, a material entity in our terms), q is an “(individual) quality” (as in DOLCE and UFO; see Section 7.1) and t is a time interval during which q exists. They also describe o as a “changing object” or the “object of change” (e.g., this flower) and q as a “changing quality” or the “subject of change” (e.g., the color of this flower), respectively.

Moreover, a qualitative change <o, q, t> is a direct qualitative change if q inheres in o and it is an indirect qualitative change if q inheres in one proper part of o. Then: “Direct and indirect qualitative changes are collectively called simple events” (Guarino et al., 2022, p. 261). To borrow their illustrative example, when a man gesticulates by moving his hand, we can identify at least two simple events: the direct qualitative change < the hand, the hand position, t’> (“hand-moving change”) and the indirect qualitative change < the man, the hand position, t’> (“gesticulation change”). A straightforward criterion for the identity of simple events is that two simple events <o₁, q₁, t₁> and <o₂, q₂, t₂> are identical if and only if (i) o₁ is identical with o₂, (ii) q₁ is identical with q₂, and (iii) t₁ is identical with t₂.

In our view, both the manifestation view and the quality-change view of processes offer potential criteria for the identity of processes in general. Further investigation is needed to adapt these views to elucidate the notion of a novel realizable entity. For instance, the development of the manifestation view will necessitate a robust ontological account of the identity of realizable entities (see Section 6.1). Both views of processes will need closer examination to be adapted to upper ontologies such as BFO, DOLCE and UFO; for related studies of processes in BFO, see Toyoshima and Barton (2024b) and Ceusters and Ruttenberg (2025).

6.3. Use for Production

As mentioned in Section 3.1, our realizable-centered approach to artifacts postulates the narrow interpretation of the term “use” (where use is distinct from production) in contrast to the broad interpretation (where production is a kind of use). To explore the idea underlying the broad interpretation, we investigate a kind of use (within the narrow interpretation) intimately linked with production, under the name of “use for production.” Let us introduce the following example:

Informally, we can think of clay₁ at time t₀ as a usefact because it is intended to be used for the accidental purpose of being able to contain liquid, or more simply, because it is a natural object that is intended to be used for some purpose. We focus on intention₀, the intention to use clay₁ to be able to contain liquid and the extrinsic disposition d₀ of clay₁ to be able to contain liquid that comes into being at time t₀. And d₀ is an extrinsic disposition, as clay₁ bears d₀ in virtue of the existence of intention₀, external to clay₁.

Recall now Example 1: at time t₁, clay₁ is intentionally shaped and dried to be able to contain liquid. We analyzed this as clay₁ at time t₁ being a canonical artifact in virtue of bearing the novel intentional realizable entity d₁, with d₁ coming into being through act₁—the successful act of intentionally shaping and drying clay₁. Given the continuity between Examples 7 and 1, d₀ is realized in act₁, even if it could have been, in principle, realized in alternative processes in which clay₁ comes to be able to contain liquid—such as casting it in a mold—rather than a process (e.g., act₁) of intentionally shaping and drying clay₁. Since act₁ is a production process (through which d₁ comes into being), we can regard d₀ as a realizable entity embodying the notion of “use for production”: clay₁ at time t₀ would be a usefact in virtue of bearing d₀, and d₀ is realized in a process (namely act₁) in which clay₁ becomes a canonical artifact at time t₁.

To justify the idea of use for production, we must show that clay₁ at time t₀ is a usefact in virtue of bearing d₀: specifically, in virtue of which realizable entity is d₀ non-novel? To address it, we introduce the notion of “pre-realizable entity”: a realizable entity that can be realized in a process in which another realizable entity of the bearer comes into being. In formal ontology, the notion of a “predisposition”—defined as “a disposition to acquire a (further) disposition”—has been developed and applied in the biomedical domain (Barton et al., 2018b; Scheuermann et al., 2009). From our perspective, a predisposition can be analyzed as an intrinsic disposition that can be realized in a process in which another intrinsic disposition of the bearer comes into being, as the term “disposition” traditionally refers to an intrinsic disposition. For that matter, Vetter (2015) elaborates the philosophical notion of an “iterated potentiality,” which is more general than the notions of a predisposition and a pre-realizable entity.

Drawing on the notion of a pre-realizable entity, we can analyze the extrinsic disposition d₀ of clay₁ to be able to contain liquid as an extrinsic disposition to acquire an intrinsic disposition of clay₁ to contain liquid. We now turn to the intrinsic disposition d₀’ of clay₁ to be able to contain liquid, which exists before time t₀. We can analyze d₀’ similarly in terms of the notion of a pre-realizable entity, or particularly of a predisposition: d₀’ is an intrinsic disposition to acquire an intrinsic disposition to contain liquid. From this interpretation of d₀ and d₀’, it follows that d₀’ exists before d₀ comes into being and, if d₀ is realized in a process of clay₁ coming to bear an intrinsic disposition to contain liquid, then d₀’ is realized in the same process. Therefore, d₀ is non-novel because of d₀’. As a matter of fact, both d₀ and d₀’ are realized in act₁ and, through this process, clay₁ comes to bear the intrinsic disposition d₁ to contain liquid.

Our notion of use for production rests on the hypothesis that there are pre-realizable entities such as the intrinsic disposition d₀’ of clay₁ at time t₀ to acquire an intrinsic disposition to contain liquid. Confirming this hypothesis is challenging due to its close ties to the complex issue of individuating realizable entities (see Section 6.1). However, for instance, there is a general agreement in the biomedical domain that an organism with a moderate genetic factor may possess an intrinsic disposition—a predisposition—to acquire a disease, where “disease” itself is treated as an intrinsic disposition in the BFO-compliant Ontology for General Medical Science (OGMS; Scheuermann et al., 2009). By analogy, we can argue that, due to its specific physical structure, clay₁ at time t₀ bears the intrinsic disposition d₀’ (which is a predisposition) to acquire an intrinsic disposition to contain liquid.

7.1. Borgo and Vieu's (2009) Theory of Artifacts: Mental Selection, Constitution, and Capacity Attribution

Borgo and Vieu (2009) develop a formal ontology of artifacts compliant with the DOLCE upper ontology that rests on three key notions: mental selection, constitution, and capacity attribution. For readability, we cite their work by page number.

As a preliminary, we outline DOLCE's theory of qualities (Borgo & Masolo, 2010), especially the terms “(individual) quality,” “quality space,” and “quale,” as they are important for understanding capacity attribution. A quality is a property particular of a particular entity (e.g., the color of this particular rose). Quality types partition qualities: the color of this rose, but not its size, instantiates the color quality type. A quality space (e.g., the “color space”) classifies qualities of the same quality type: it is the mereological sum of all the DOLCE:quality regions that are related to a certain quality type. A quale (e.g., a specific redness) is a “value” that a quality has in virtue of its position within a quality space.

First: mental selection (“intentional selection” in Borgo and Vieu's terms). Contrary to popular belief, they think that mental selection alone—even without physical modification—can specify the intentional dimension of artifacts: “we do not see the action of physical modification as an essential aspect of the creation of an artefact. […] Artefacts are, in nuce, created entities in which ‘created’ refers to a mental event, not to a physical modification” (p. 292). Their notion of mental selection also distinguishes between artifacts and by-products: “residues, such as sawdust, are intentionally processed but not intentionally selected for having certain properties and use” (Borgo & Masolo, 2010).

Second: constitution (see also Vieu et al., 2008). Borgo and Vieu maintain that constitution (or the “constitution view of production,” in our terms) explains the alleged “creation” of an artifact. They critique alternative views that treat “artefactuality as a property that physical endurants may or may not have” (p. 293), which aligns with our “continuity view of production.” They argue: “by not granting artefacts an ontological status, it would not do justice to the notion of creation […]. We are thus confronted with two co-located entities, the artefact and the underlying physical object” (Vieu et al., 2008). Their theory finds support in DOLCE's “multiplicative approach: it assumes that different entities can be co-located in the same space-time” (p. 285).

Third: capacity attribution. Before the creation of an artifact, the physical endurant bears “a single individual quality named capacity that characterizes all the capacities the physical endurant has. […] This quality maps into a quale that is a region […] in the capacity space” (p. 294). Borgo and Vieu illustrate this quality-based view of capacities with an example: “the capacity of this pen now has the quale of writing finely in black when drawn over paper, fitting in one's hand when grasped, and making a certain noise when it contacts the table” (ibid.). They acknowledge that the structure of the capacity space is complex and its detailed study is left for future work.

Borgo and Vieu characterize artifacts in terms of capacity attribution: “In addition to the capacity possessed by any physical endurant [footnote: We assume that an artefact's capacity is inherited from its constituting entity.], artefacts also have an attributed capacity, another quality associated with qualia in the same [capacity] space” (p. 294). They also explain the relationship between (unattributed) capacities and attributed capacities: “Although capacity and attributed capacity map into the same space of qualia, the former remains a physical quality whereas the latter is an intentional quality as it is dependent on the intentions of the creator at the time of the creation event” (p. 295).

We can illustrate Borgo and Vieu's theory with Example 2: at time t₂, pebble₂ is intended to be used to keep papers in place. Before time t₂, pebble₂ bears the (unattributed) capacity (capacity_pbl2), a physical quality of pebble₂ mapping into a quale (e.g., of blocking water flow), a subregion of the capacity space. At time t₂ (when mentally selected), pebble₂ constitutes a new material entity paperweight₂, which bears the attributed capacity capacity_pw2, an intentional quality of paperweight₂ that maps into the quale of keeping papers in place—a subregion of the same capacity space. For Borgo & Vieu, paperweight₂ is an artifact. This aligns with their summary: “The paperweight is the result of some agent intentionally selecting a pebble and attributing to it certain capacities. The artefact itself is the new entity whose physical realization is the selected object and which has attributed capacities” (p. 293).

Borgo and Vieu's theory and ours share key ideas: the intentional dimension of artifacts can be specified by mental selection only, without physical modification, and being an artifact is characterized by a new property of the bearer—an attributed capacity in their theory and an intentional realizable entity in ours.

However, their theory differs greatly from ours in other respects. First, their notion of an artifact encompasses both canonical artifacts and usefacts: paperweight₂ in Example 2 (usefact and natural object) is an artifact constituted by pebble₂ and bearing capacity_pbl2. Second, they endorse the constitution view of production, taking seriously the idea that artifacts are by nature “created.” While our realizable-centered approach to artifacts is developed under the continuity view of both production and use—which may involve a somewhat counterintuitive construal of “production”—Borgo and Vieu's constitution view yields similarly counterintuitive results, such as the “creation” of paperweight₂ (criticized by Kassel, 2010; see Section 2.2).

Despite these differences, our realizable-centered approach—particularly the notions of a canonical artifact and a usefact—can be reconstructed within Borgo & Vieu's framework, which assumes the constitution view of both production and use. To demonstrate this, we translate our novelty notion for realizable entities into their capacity-based theory.

The notion of a realizable entity approximates the notions of capacity and the quale into which it maps. Thus, an intentional realizable entity (e.g., d₂ in our analysis of Example 2) approximates an attributed capacity and its quale (e.g., capacity_pw2 and the quale of keeping papers in place).

To translate the notion of a novel realizable entity into Borgo & Vieu's constitution view of both production and use, we focus on their discussion of the inheritance of capacities through constitution: “The capacity of physical endurants is inherited through constitution, in the sense that the quale of the capacity of the constituted entity includes that of the constituting entity […]. Note that the two qualia need not be identical. The capacity of a physical object may include shape-based dispositions, while the capacity of the amount of matter constituting it cannot” (p. 299).

By extending this account of capacity inheritance via constitution, we can characterize canonical artifacts and usefacts in terms of capacities. For canonical artifacts, the quale of the capacity of the constituted canonical artifact is a proper super-region of the quale of the capacity of the constituting entity. Consider Example 1, where at time t₁, the quale of the capacity of the constituted pot₁—as introduced in Example A of Section 2.2—includes the quale of containing liquid, whereas the quale of the capacity of the constituting clay₁ does not. For usefacts, by contrast, the quale of the capacity of the constituted usefact is a subregion of the quale of the capacity of the constituting entity. In Example 2, paperweight₂ is a usefact because the quale (including keeping papers in place) of capacity_pw2 of paperweight₂ is a subregion of the quale of capacity_pbl2 of pebble₂.

Based on this discussion, we analyze the novelty of a capacity in terms of the inclusion relation between qualia of capacities: we define the terms “novel capacity,” “novel attributed capacity,” and “physical canonical artifact” (aligning with the terms “novel realizable entity,” “novel intentional realizable entity,” and “material canonical artifact,” respectively) as follows:

novel capacity = _def. A capacity c such that the quale of c is a proper super-region of the quale of the capacity of the material entity that constitutes the bearer of c.

We define the relation “is non-novel because of” between capacities and the term “physical usefact” (aligning with the relation “is non-novel because of” between realizable entities and the term “material usefact,” respectively) as follows:

A capacity c is non-novel because of a capacity c’

7.2. Other Theories of Artifacts in Philosophy and in Formal Ontology