To support the trustworthiness of AI systems, it is essential to have precise methods to determine what or who is to account for the behaviour, or the outcome, of AI systems. The assignment of responsibility to an AI system is closely related to the identification of individuals or elements that have caused the outcome of the AI system. In this work, we present an overview of approaches that aim at modelling responsibility of AI systems, discuss their advantages and shortcomings to (...) deal with various aspects of the notion of responsibility, and present research gaps and ways forward. (shrink)

Problem solving by autonomous, interacting computersystems has attracted much attention in the ArtificialIntelligence community. These autonomous computersystems, called agents, provide a promisingperspective for the legal knowledge-based systemscommunity, as legal problem solving often involvesdistributed problem solving capabilities that gobeyond the capabilities of individual knowledge-basedsystems.We focus on the coordination of agents andcommunication between agents by proposing a model ofcommunication between various agents using modellingtechniques such as communication primitives and statetransition diagrams. Our representation concerns theDutch Algemene Wet Bestuursrecht (AWB; GeneralAct on Administrative Law). (...) A proposal for an agentarchitecture describes how these communication aspectscan be incorporated into an architecture. (shrink)

The aim of this study is to gain an understanding of why employees misuse information systems resources in the workplace. Rather than consider “intention,” as existing behavioral research commonly does, this study investigates actual behavior and employs IS resource misuse as the dependent variable. Data from a web-based survey are analyzed using the partial least squares approach. In light of the dual-process approach and the theory of planned behavior, the findings suggest that IS resource misuse may be both an intentional (...) type of behavior and an unreasoned action. Perceived behavioral control influences employees’ IS resource misuse actions via their desires or intentions, whereas attitude toward such misuse affects these actions via employees’ desires alone. Subjective norm is found not to affect employees’ IS resource misuse via either desires or intentions. In terms of its theoretical contribution, this study considers unethical behavior in information systems by incorporating a dual-process model and the theory of planned behavior. With regard to its managerial significance, the study’s results will help managers to better understand why employees commit IS resource misuse within organizations. (shrink)

In this article we strive to provide a detailed and principled analysis of the role of beliefs in goal processing—that is, the cognitive transition that leads from a mere desire to a proper intention. The resulting model of belief-based goal processing has also relevant consequences for the analysis of intentions, and constitutes the necessary core of a constructive theory of intentions, i.e. a framework that not only analyzes what an intention is, but also explains how it becomes what it is. (...) We discuss similarities and differences between our approach and other standard accounts of intention, in particular Bratman’s planning theory. The aim here is to question and refine the conceptual foundations of many theories of intentional action: as a consequence, although our analysis is not formal in itself, it is ultimately meant to have deep consequences for formal models of intentional agency. (shrink)

This paper presents a systematic analysis of the various steps of goal-processing and intention creation, as the final outcome of goal-driven action generation. Intention theory has to be founded on goal theory: intentions require means-end reasoning and planning, conflict resolution, coherence. The process of intention formation and intentional action execution is strictly based on specific sets of beliefs (predictions, evaluations, calculation of costs, responsibility beliefs, competence, etc.). The origin of an intention is not necessarily a “desire” (which is just a (...) kind of goal). Intention is a two-layered goal-structure: the intended action(s) to be executed, and the intended outcome motivating that action—with two distinct kinds of “failure”. This belief-goal perspective also allows to examine two stages/types of intention, and the relations and differences between intention “in agenda” (future directed) and intention under execution (intention in action). I will argue that the will is much more than the intention driving an intentional action. I will also claim that intentions are not there just for motivating and regulating intentional actions (from the motor level to more complex behaviors), but that they play also several other important roles. (shrink)

Realism for agents with unconditional beliefs, desires and intentions has been analyzed in modal logic. This paper provides a logical analysis of realism for agents with conditional beliefs and desires in a rule based approach analogous to Reiter's default logic. We distinguish two types of realism, which we call ‘a priori' and ‘a posteriori' realism. We analyze whether these two new properties are compatible with other properties discussed in the literature, such as existence of extensions. We show that Reiter's default (...) logic is too strong, in the sense that a weaker notion of maximality of extensions is needed to satisfy realism. Finally we show that several existing approaches do not satisfy the new realism properties, and we introduce a new construction that does satisfy them. (shrink)

This paper studies intentional action in stit logic. The formal logic study of intentional action appears to be new, since most logical studies of intention concern intention as a static mental state. In the formalization we distinguish three modes of acting: the objective level concerning the choices an agent objectively exercises, the subjective level concerning the choices an agent knows or believes to be exercising, and finally, the intentional level concerning the choices an agent intentionally exercises. Several axioms constraining the (...) relations between these different modes of acting will be considered and discussed. The side effect problem will be analyzed as an interaction between knowingly doing and intentionally doing. Non-successful action will be analyzed as a weakening of the epistemic attitude towards action. Finally, the notion of ‘attempt’ will be briefly considered as a further weakening in this direction. (shrink)

We introduce quantificational modal operators as dynamic modalities with (extensions of) Henkin quantifiers as indices. The adoption of matrices of indices (with action identifiers, variables and/or quantified variables as entries) gives an expressive formalism which is here motivated with examples from the area of multi-agent systems. We study the formal properties of the resulting logic which, formally speaking, does not satisfy the normality condition. However, the logic admits a semantics in terms of (an extension of) Kripke structures. As a consequence, (...) standard techniques for normal modal logic become available. We apply these to prove completeness and decidability, and to extend some standard frame results to this logic. (shrink)

Important decisions are often based on a distributed process of information processing, from a knowledge base that is itself distributed among agents. The simplest such situation is that where a decision-maker seeks the recommendations of experts. Because experts may have vested interests in the consequences of their recommendations, decision-makers usually seek the advice of experts they trust. Trust, however, is a commodity that is usually built through repeated face time and social interaction and thus cannot easily be built in a (...) global world where we have immediate internet access to a vast pool of experts. In this article, we integrate findings from experimental psychology and formal tools from Artificial Intelligence to offer a preliminary roadmap for solving the problem of trust in this computer-mediated environment. We conclude the article by considering a diverse array of extended applications of such a solution. (shrink)

Representing an epistemic situation involving several agents obviously depends on the modeling point of view one takes. We start by identifying the types of modeling points of view which are logically possible. We call the one traditionally followed by epistemic logic the perfect external approach, because there the modeler is assumed to be an omniscient and external observer of the epistemic situation. In the rest of the paper we focus on what we call the internal approach, where the modeler is (...) one of the agents involved in the situation. For this approach we propose and axiomatize a logical formalism based on epistemic logic. This leads us to formalize some intuitions about the internal approach and about its connections with the external ones. Finally, we show that our internal logic is decidable and PSPACE-complete. (shrink)

This paper investigates the influence that social ties can have on behavior. After defining the concept of social ties that we consider, we introduce an original model of social ties. The impact of such ties on social preferences is studied in a coordination game with outside option. We provide a detailed game theoretical analysis of this game while considering various types of players, i.e., self-interest maximizing, inequity averse, and fair agents. In addition to these approaches that require strategic reasoning in (...) order to reach some equilibrium, we also present an alternative hypothesis that relies on the concept of team reasoning. After having discussed the differences between the latter and our model of social ties, we show how an experiment can be designed so as to discriminate among the models presented in the paper. (shrink)

This article introduces definitions for direct, means-end, oblique (or indirect) and ulterior intent which can be used to test for intent in an algorithmic actor. These definitions of intent are informed by legal theory from common law jurisdictions. Certain crimes exist where the harm caused is dependent on the reason it was done so. Here the actus reus or performative element of the crime is dependent on the mental state or mens rea of the actor. The ability to prosecute these (...) crimes is dependent on the ability to identify and diagnose intentional states in the accused. A certain class of auto didactic algorithmic actor can be given broad objectives without being told how to meet them. Without a definition of intent, they cannot be told not to engage in certain law breaking behaviour nor can they ever be identified as having done it. This ambiguity is neither positive for the owner of the algorithm or for society. The problem exists over and above more familiar debates concerning the eligibility of algorithms for culpability judgements that mens rea is usually associated with. Aside from inchoate offences, many economic crimes with elements of fraud or deceit fall into this category of crime. Algorithms operate in areas where these crimes could be plausibly undertaken depending on whether the intent existed in the algorithm or not. (shrink)

In this paper I focus on a recently discussed phenomenon illustrated by sentences containing predicates of taste: the phenomenon of " perspectival plurality " , whereby sentences containing two or more predicates of taste have readings according to which each predicate pertains to a different perspective. This phenomenon has been shown to be problematic for (at least certain versions of) relativism. My main aim is to further the discussion by showing that the phenomenon extends to other perspectival expressions than predicates (...) of taste and by proposing a general solution to the problem raised by it on behalf of the relativist. The core claim of the solution (" multiple indexing ") is that utterances of sentences containing perspectival expressions should be evaluated with respect to (possibly infinite) sequences of perspective parameters. (shrink)

Our preferences determine how we act and think, but exactly what the mechanics are and how they work is a central cause of concern in many disciplines. This book uses techniques from modern logics of information flow and action to develop a unified new theory of what preference is and how it changes. The theory emphasizes reasons for preference, as well as its entanglement with our beliefs. Moreover, the book provides dynamic logical systems which describe the explicit triggers driving preference (...) change, including new information, suggestions, and commands. In sum, the book creates new bridges between many fields, from philosophy and computer science to economics, linguistics, and psychology. For the experienced scholar access to a large body of recent literature is provided and the novice gets a thorough introduction to the action and techniques of dynamic logic. (shrink)

In this article, I first elaborate and refine the Principle of Intention Agglomeration (PIA), which was introduced by Michael Bratman as “a natural constraint on intention”. According to the PIA, the intentions of a rational agent should be agglomerative. The proposed refinement of the PIA is not only in accordance with the spirit of Bratman’s planning theory of intention as well as consistency constraints for intentions rooted in the theory, but also reveals some deep rationales of practical rationality regarding resource-limited (...) agents. Then I defend the PIA against various objections and counterexamples, showing that the refined PIA survives attacks based on both conceptual analyses and psychological considerations. (shrink)

Although the change of beliefs in the face of new information has been widely studied with some success, the revision of other mental states has received little attention from the theoretical perspective. In particular, intentions are widely recognised as being a key attitude for rational agents, and while several formal theories of intention have been proposed in the literature, the logic of intention revision has been hardly considered. There are several reasons for this: perhaps most importantly, intentions are very closely (...) connected with other mental states—in particular, beliefs about the future and the abilities of the agent. So, we cannot study them in isolation. We must consider the interplay between intention revision and the revision of other mental states, which complicates the picture considerably. In this paper, we present some first steps towards a theory of intention revision. We develop a simple model of an agent’s mental states, and define intention revision operators. Using this model, we develop a logic of intention dynamics, and then investigate some of its properties. (shrink)

Rational agents are important objects of study in several research communities, including economics, philosophy, cognitive science, and most recently computer science and artificial intelligence. Crudely, a rational agent is an entity that is capable of acting on its environment, and which chooses to act in such a way as to further its own best interests. There has recently been much interest in the use of mathematical logic for developing formal theories of such agents. Such theories view agents as practical reasoning (...) systems, deciding moment by moment which action to perform next, given the beliefs they have about the world and their desires with respect to how they would like the world to be. In this article, we survey the state of the art in developing logical theories of rational agency. Following a discussion on the dimensions along which such theories can vary, we briefly survey the logical tools available in order to construct such theories. We then review and critically assess three of the best known theories of rational agency: Cohen and Levesque's intention logic, Rao and Georgeff's BDI logics, and the KARO framework of Meyer et al. We then discuss the various roles that such logics can play in helping us to engineer rational agents, and conclude with a discussion of open problems. (shrink)

In this paper a formal framework is proposed in which variousinformative actions are combined, corresponding to the different ways in whichrational agents can acquire information. In order to solve the variousconflicts that could possibly occur when acquiring information fromdifferent sources, we propose a classification of the informationthat an agent possesses according to credibility. Based on this classification, we formalize what itmeans for agents to have seen or heard something, or to believesomething by default. We present a formalization of observations,communication actions, (...) and the attempted jumps to conclusions thatconstitutes default reasoning. To implement these informative actionswe use a general belief revision action which satisfies theAGM postulates; dependent on the credibility of the incominginformation this revision action acts on one or more parts ofthe classified belief sets of the agents. The abilities of agents formalizeboth the limited capacities of agents to acquire information, and the preference of one kind of information acquisition to another. A very important feature of our approach is that it shows how to integratevarious aspects of agency, in particular the (informational) attitudesof dealing with information from observation, communication and defaultreasoning into one coherent framework, both model-theoretically andsyntactically. (shrink)

We demonstrate ways to incorporate nondeterminism in a system designed to formalize the reasoning of agents concerning their abilities and the results of the actions that they may perform. We distinguish between two kinds of nondeterministic choice operators: one that expresses an internal choice, in which the agent decides what action to take, and one that expresses an external choice, which cannot be influenced by the agent. The presence of abilities in our system is the reason why the usual approaches (...) towards nondeterminism cannot be used here. The semantics that we define for nondeterministic actions is based on the idea that composite actions are unravelled in the strings of atomic actions and tests that constitute them. The main notions used in defining this semantics are finite computation sequences and finite computation runs of actions. The results that we obtain meet our intuitions regarding events and abilities in the presence of nondeterminism. (shrink)

The Little Nell Problem was formulated by Drew McDermott in 1982. It reveals unexpected complexities in the interaction of the beliefs and intentions of a planning agent. This paper discusses the problem and proposes a solution.

In adversarial problem solving (APS), one must anticipate, understand and counteract the actions of an opponent. Military strategy, business, and game playing all require an agent to construct a model of an opponent that includes the opponent's model of the agent. The cognitive mechanisms required for such modeling include deduction, analogy, inductive generalization, and the formation and evaluation of explanatory hypotheses. Explanatory coherence theory captures part of what is involved in APS, particularly in cases involving deception.

I show how a conversational process that takes simple, intuitively meaningful steps may be understood as a sophisticated computation that derives the richly detailed, complex representations implicit in our knowledge of language. To develop the account, I argue that natural language is structured in a way that lets us formalize grammatical knowledge precisely in terms of rich primitives of interpretation. Primitives of interpretation can be correctly viewed intentionally, as explanations of our choices of linguistic actions; the model therefore fits our (...) intuitions about meaning in conversation. Nevertheless, interpretations for complex utterances can be built from these primitives by simple operations of grammatical derivation. In bridging analyses of meaning at semantic and symbol‐processing levels, this account underscores the fundamental place for computation in the cognitive science of language use. (shrink)

Team members co-regulate their activities and move together at the collective level of behavior while coordinating their actions toward shared goals. In parallel with team processes, team members need to resolve uncertainties arising from the changing task and environment. In this exploratory study we have measured the differential neurodynamics of seven two-person healthcare teams across time and brain regions during autonomous and collaborative segments of simulation training. The questions posed were: whether these abstract and mostly integrated constructs could be separated (...) neurodynamically; and, what could be learned about taskwork and teamwork by trying to do so? The taskwork and teamwork frameworks used were Neurodynamic Information, an electroencephalography derived measure shown to be a neurodynamic proxy for the pauses and hesitations associated with individual uncertainty, and inter-brain EEG coherence which is a required component of social interactions. No interdependency was observed between NI and IBC, and second-by-second dynamic comparisons suggested mutual exclusivity. These studies show that proxies for fundamental properties of teamwork and taskwork can be separated neurodynamically during team performances of ecologically valid tasks. The persistent expression of NI and IBC were not simultaneous suggesting that it may be difficult for team members to maintain inter-brain coherence while simultaneously reducing their individual uncertainties. Lastly, these separate dynamics occur over time frames of 15–30 s providing time for real-time detection and mitigation of individual and collaborative complications during training or live patient encounters. (shrink)

This paper formalizes part of a well-known psychological model of emotions. In particular, the logical structure underlying the conditions that trigger emotions are studied and then hierarchically organized. The insights gained therefrom are used to guide a formalization of emotion triggers, which proceeds in three stages. The first stage captures the conditions that trigger emotions in a semiformal way, i.e., without committing to an underlying formalism and semantics. The second stage captures the main psychological notions used in the emotion model (...) in dynamic doxastic logic. The third stage introduces a BDI-based framework (belief–desire–intention) with achievement goals, which is used to firmly ground the preceding stages. The result is a formalization of emotion triggers for BDI agents with achievement goals. The idea of proceeding in these stages is to provide different levels of commitment to formalisms, so that it remains relatively easy to extend or replace the used formalisms without having to start from scratch. Finally, we show that the formalization renders properties of emotions that are in line with the psychological model on which it is based. (shrink)

Since the early days of artificial intelligence (AI), many logics have been explored as tools for knowledge representation and reasoning. In the spirit of the Crossley Festscrift and recognizing John Crossley’s diverse interests and his legacy in both mathematical logic and computer science, I discuss examples from my own research that sit in the overlap of logic and AI, with a focus on supporting human–AI interactions.

We introduce a conceptual model for reaching group decisions. Our model extends a well-known, single-agent cognitive model, the recognition-primed decision (RPD) model. The RPD model includes a recognition phase and an evaluation phase. Group extensions of the RPD model, applicable to a group of RPD agents, have been considered in the literature, however the proposed models do not formalize how distributed and possibly inconsistent information can be combined in either phase. We show how such information can be utilized by aggregating (...) it using a specific social choice method, namely judgment aggregation. Our model is applicable to hierarchical groups of agents containing at least one RPD agent. (shrink)

Intentions are an important concept in Artificial Intelligence and Cognitive Science. We present a formal theory of intentions and beliefs based on Discourse Representation Theory that captures many of their important logical properties. Unlike possible worlds approaches, this theory does not assume that agents are perfect reasoners, and gives a realistic view of their internal architecture; unlike most representational approaches, it has an objective semantics, and does not rely on an ad hoc labeling of the internal states of agents. We (...) describe a minimal logic for intentions and beliefs that is sound and complete relative to our semantics. We discuss several additional axioms, and the constraints on the models that validate them. (shrink)

While logical theories of information attitudes, such as knowledge, certainty and belief, have flourished in the past two decades, formalization of other facets of rational behavior have lagged behind significantly. One intriguing line of research concerns the concept of intention. I will discuss one approach to tackling the notion within a logical framework, based on a database perspective.

Here is a prima facie plausible view: since the metaethical error theory says that all positive moral claims are false, it makes no sense for error theorists to engage in normative ethics. After all, normative ethics tries to identify what is right or wrong (and why), but the error theory implies that nothing is ever right or wrong. One way for error theorists to push back is to argue for “concept preservationism,” that is, the view that even though our ordinary (...) moral discourse is deeply flawed, we should nevertheless continue to engage in moral thought and talk. However, in this article, I pursue a different strategy. I argue that even if we completely abandon moral discourse, thus endorsing “concept abolitionism,” the discipline of normative ethics survives. While traditional normative ethics uses as its “starting points” moral claims and beliefs, instead, concept abolitionists can make use of alternative utterances and attitudes that share salient characteristics with moral claims and beliefs, allowing for a kind of theorizing that is practically oriented, impartial, involves the traditional subject matters and methods of normative ethics, and allows engagement with the arguments of traditional moral philosophers. (shrink)

The paper argues that there are two main kinds of joint action, direct joint bringing about (or performing) something (expressed in terms of a DO-operator) and jointly seeing to it that something is the case (expressed in terms of a Stit-operator). The former kind of joint action contains conjunctive, disjunctive and sequential action and its central subkinds. While joint seeing to it that something is the case is argued to be necessarily intentional, direct joint performance can also be nonintentional. Actions (...) performed by social groups are analyzed in terms of the notions of joint action (basically DO and Stit).A precise semantical analysis of the aforementioned kinds of joint action is given in terms of time-trees. With each participant a tree is connected, and the trees are joined defining joint possible worlds in terms of state-expressing nodes from the trees. Sentences containing DO and Stit are semantically evaluated with respect to such joint possible worlds. Intentional joint actions are characterized in terms of the notion of we-intention (joint intention), characterized formally by means of a special operator. (shrink)

In this paper I study two ways of transforming decision problems on the basis of previously adopted intentions, ruling out incompatible options and imposing a standard of relevance, with a particular focus on situations of strategic interaction. I show that in such situations problems arise which do not appear in the single-agent case, namely that transformation of decision problems can leave the agents with no option compatible with what they intend. I characterize conditions on the agents’ intentions which avoid such (...) problematic scenarios, in a way that requires each agent to take account of the intentions of others. (shrink)

In this paper I present a dynamic-epistemic hybrid logic for reasoning about information and intention changes in situations of strategic interaction. I provide a complete axiomatization for this logic, and then use it to study intentions-based transformations of decision problems.