TLDR: A research paper by Donato and Barton proposes a new ontological definition of risk, categorizing it as a ‘role’ within the Basic Formal Ontology (BFO) framework, rather than a ‘disposition’. The authors argue that risk is externally grounded in an agent’s aversion to potential outcomes, meaning its existence depends on the context and an agent’s mental state, not just the physical properties of an object. This approach offers a more precise understanding of risk, distinguishing it from mere potential events and laying groundwork for improved risk management systems.
Risk is a concept we encounter daily, from financial decisions to health choices and environmental concerns. Despite its omnipresence, a clear, universally accepted definition of “risk” remains elusive, leading to confusion in various fields, from databases to policy-making. This lack of conceptual clarity hinders effective strategies for preventing or mitigating risks.
A recent research paper, “An ontological analysis of risk in Basic Formal Ontology,” by Federico Donato and Adrien Barton, delves into this fundamental problem. The authors propose a novel way to understand risk by categorizing it within the Basic Formal Ontology (BFO), a widely used framework for organizing knowledge. Their work aims to provide a precise, logically-based definition of risk, addressing long-standing questions about its nature: Is it a material entity, a belief, or a type of probability? Is it objective or subjective?
Understanding Risk: A New Perspective from Ontology
Ontologies are powerful tools that help us define terms, categorize entities, and enable logical reasoning across different information systems. The Basic Formal Ontology (BFO) is a foundational ontology that divides everything into two main types: continuants (things that endure over time, like an object) and occurrents (things that happen over time, like a process). Within continuants, BFO identifies “realizable entities” – properties that can be activated or realized in a process. These include dispositions and roles.
A disposition is a latent property that depends primarily on an entity’s physical makeup. For example, an orange has a disposition to dry out in the sun. If this disposition were to cease, the orange’s physical structure would change. In contrast, a role is a property that exists because an entity is in specific circumstances – physical, social, or institutional – and its existence doesn’t necessarily alter the bearer’s physical makeup if the role disappears. Think of an orange playing the role of a “competition winner” after being selected by a jury; if it loses this role, the orange itself doesn’t physically change.
The Debate: Is Risk a Disposition or a Role?
The paper examines two main views on how risk fits into the BFO framework, using a simple scenario: Paul is walking under an unstable ceiling. The first view, which the paper contrasts with its own, suggests that risks are dispositions. In this “dispositional” approach, the ceiling’s instability is a disposition. It becomes a “risk for Paul” only if Paul finds its potential collapse undesirable. The instability itself exists regardless of Paul’s feelings, but its status as a risk is dependent on Paul’s desires. If Paul leaves the room and no longer cares, the instability remains, but it’s no longer a risk for him. This implies that a disposition is a risk “accidentally” – it can gain or lose its risk status.
The authors, however, propose an alternative: risks are roles. In this “role” approach, the unstable ceiling bears a “risk role” (e.g., “hazard”) specifically when Paul is under it and wants to preserve his health. Unlike a disposition, this risk role exists because of the context, including Paul’s mental state. If Paul’s desires change, the “hazard” role simply ceases to exist; it cannot exist without being a risk. This view suggests that a risk is a risk “essentially” – it is defined by its context and the aversion of an agent.
A key difference also lies in how these views handle the “realization” of risk. In the dispositional view, the ceiling’s instability realizes in a physical collapsing process, which is then classified as undesirable. In the role view, the instability still leads to a physical collapse, but the risk role itself realizes in a separate process: causing damage to Paul. These are distinct, though often co-occurring, processes.
The Role of Human Aversion in Defining Risk
The paper emphasizes that for something to be a risk, it must involve an unwanted outcome for an agent. Crucially, the authors argue that an agent’s knowledge or beliefs about a situation are not relevant to whether a risk exists. Even if Paul is unaware of the ceiling’s instability, or is asleep, the risk is still present if the conditions for it are met. What matters is a “conative” mental state, specifically an “aversion.”
An aversion is defined as a mental disposition opposite to a desire. For instance, Paul’s aversion is directed towards the process of being bodily hurt. This aversion exists even if the damaging process never actually occurs. The specific target of this aversion – what the agent wants to avoid – determines what constitutes a risk. If Paul also cared about his clothes being ruined, a separate risk related to that outcome would emerge, grounded in a distinct aversion.
A Formal Definition of Risk
Based on their analysis, Donato and Barton propose the following sufficient conditions for something to be considered a risk:
If: (i) r is a role & (ii) r is externally grounded in an aversion that an agent or group of agents have towards the possible realizations of the role, then r is a risk.
This means that for something to be a risk, it must function as a “role” within a given context, and its existence must be dependent on an agent’s (or group’s) active dislike or avoidance of the potential outcomes associated with that role. The “external grounding” highlights that the risk isn’t inherent solely in the object itself, but arises from its interaction with an external agent’s mental state.
Also Read:
- Predicting Future Activities: How Knowledge Graphs and Markov Chains Offer Auditable Insights
- Connecting Bayesian Inference and Embodied AI for Real-World Adaptation
Looking Ahead: Implications and Future Research
This ontological analysis provides a clearer, more structured way to think about risk, moving beyond vague natural language descriptions. By categorizing risk as a BFO:Role, the authors offer a framework that can be used to build more precise risk ontologies and improve interoperability between diverse risk databases.
The paper acknowledges that this is an exploratory investigation and suggests several avenues for future work. These include determining if the proposed sufficient conditions are also necessary, clarifying the nature of “damaging processes,” and integrating quantitative aspects of risk, such as probabilities and severity degrees, into the ontological model. Further research will also explore how risks propagate through causal chains and how this framework can be evaluated with real-world data to structure information in risk registers more effectively.
