Bayesian networks and causal models provide frameworks for handling queries about external interventions and counterfactuals, enabling tasks that go beyond what probability distributions alone can address. While these formalisms are often informally described as capturing causal knowledge, there is a lack of a formal theory characterizing the type of knowledge required to predict the effects of external interventions. This work introduces the theoretical framework of causal systems to clarify Aristotle's distinction between knowledge that and knowledge why within artificial intelligence. By interpreting existing artificial intelligence technologies as causal systems, it investigates the corresponding types of knowledge. Furthermore, it argues that predicting the effects of external interventions is feasible only with knowledge why, providing a more precise understanding of the knowledge necessary for such tasks.

Pearl observes that causal knowledge enables predicting the effects of interventions, such as actions, whereas descriptive knowledge only permits drawing conclusions from observation. This paper extends Pearl's approach to causality and interventions to the setting of stratified abductive logic programs. It shows how stable models of such programs can be given a causal interpretation by building on philosophical foundations and recent work by Bochman and Eelink et al. In particular, it provides a translation of abductive logic programs into causal systems, thereby clarifying the informal causal reading of logic program rules and supporting principled reasoning about external actions. The main result establishes that the stable model semantics for stratified programs conforms to key philosophical principles of causation, such as causal sufficiency, natural necessity, and irrelevance of unobserved effects. This justifies the use of stratified abductive logic programs as a framework for causal modeling and for predicting the effects of interventions.

Persuasive dialogue plays a pivotal role in human communication, influencing various domains. Recent persuasive dialogue datasets often fail to align with real-world interpersonal interactions, leading to unfaithful representations. For instance, unrealistic scenarios may arise, such as when the persuadee explicitly instructs the persuader on which persuasion strategies to employ, with each of the persuadee's questions corresponding to a specific strategy for the persuader to follow. This issue can be attributed to a violation of the "Double Blind" condition, where critical information is fully shared between participants. In actual human interactions, however, key information such as the mental state of the persuadee and the persuasion strategies of the persuader is not directly accessible. The persuader must infer the persuadee's mental state using Theory of Mind capabilities and construct arguments that align with the persuadee's motivations. To address this gap, we introduce ToMMA, a novel multi-agent framework for dialogue generation that is guided by causal Theory of Mind. This framework ensures that information remains undisclosed between agents, preserving "double-blind" conditions, while causal ToM directs the persuader's reasoning, enhancing alignment with human-like persuasion dynamics. Consequently, we present CToMPersu, a multi-domain, multi-turn persuasive dialogue dataset that tackles both double-blind and logical coherence issues, demonstrating superior performance across multiple metrics and achieving better alignment with real human dialogues. Our dataset and prompts are available at https://github.com/DingyiZhang/ToMMA-CToMPersu .

We consider how human-centered causal theories and tools from the dynamical systems literature can be deployed to guide the representation of data when training neural networks for complex classification tasks. Specifically, we use simulated data to show that training a neural network with a data representation that makes explicit the invariant structural causal features of the data generating process of an epidemic system improves out-of-distribution (OOD) generalization performance on a classification task as compared to a more naive approach to data representation. We take these results to demonstrate that using human-generated causal knowledge to reduce the epistemic uncertainty of ML developers can lead to more well-specified ML pipelines. This, in turn, points to the utility of a dynamical systems approach to the broader effort aimed at improving the robustness and safety of machine learning systems via improved ML system development practices.

This paper continues the line of work on representing properties of actions in nonmonotonic formalisms that stresses the distinction between being "true" and being "caused", as in the system of causal logic introduced by McCain and Turner and in the action language C proposed by Giunchiglia and Lifschitz. The only fluents directly representable in language C+ are truth-valued fluents, which is often inconvenient. We show that both causal logic and language C can be extended to allow values from arbitrary nonempty sets. Our extension of language C, called C+, also makes it possible to describe actions in terms of their attributes, which is important from the perspective of elaboration tolerance. We describe an embedding of C+ in causal theories with multi-valued constants, relate C+ to Pednault's action language ADL, and show how multi-valued constants can be eliminated in favor of Boolean constants.

The verification and validation of automated driving systems at SAE levels 4 and 5 is a multi-faceted challenge for which classical statistical considerations become infeasible. For this, contemporary approaches suggest a decomposition into scenario classes combined with statistical analysis thereof regarding the emergence of criticality. Unfortunately, these associational approaches may yield spurious inferences, or worse, fail to recognize the causalities leading to critical scenarios, which are, in turn, prerequisite for the development and safeguarding of automated driving systems. As to incorporate causal knowledge within these processes, this work introduces a formalization of causal queries whose answers facilitate a causal understanding of safety-relevant influencing factors for automated driving. This formalized causal knowledge can be used to specify and implement abstract safety principles that provably reduce the criticality associated with these influencing factors. Based on Judea Pearl's causal theory, we define a causal relation as a causal structure together with a context, both related to a domain ontology, where the focus lies on modeling the effect of such influencing factors on criticality as measured by a suitable metric. As to assess modeling quality, we suggest various quantities and evaluate them on a small example. As availability and quality of data are imperative for validly estimating answers to the causal queries, we also discuss requirements on real-world and synthetic data acquisition. We thereby contribute to establishing causal considerations at the heart of the safety processes that are urgently needed as to ensure the safe operation of automated driving systems.

Recent research on algorithmic fairness has highlighted that the problem formulation phase of ML system development can be a key source of bias that has significant downstream impacts on ML system fairness outcomes. However, very little attention has been paid to methods for improving the fairness efficacy of this critical phase of ML system development. Current practice neither accounts for the dynamic complexity of high-stakes domains nor incorporates the perspectives of vulnerable stakeholders. In this paper we introduce community based system dynamics (CBSD) as an approach to enable the participation of typically excluded stakeholders in the problem formulation phase of the ML system development process and facilitate the deep problem understanding required to mitigate bias during this crucial stage. Problem formulation is a crucial first step in any machine learning (ML) based interventions that have the potential of impacting the real lives of people; a step that involves determining the strategic goals driving the interventions and translating those strategic goals into tractable machine learning problems (Barocas et al., 2017; Passi & Barocas, 2019).

We will explore the use of disjunctive causal rules for representing indeterminate causation. We provide first a logical formalization of such rules in the form of a disjunctive inference relation, and describe its logical semantics. Then we consider a nonmonotonic semantics for such rules, described in (Turner 1999). It will be shown, however, that, under this semantics, disjunctive causal rules admit a stronger logic in which these rules are reducible to ordinary, singular causal rules. This semantics also tends to give an exclusive interpretation of disjunctive causal effects, and so excludes some reasonable models in particular cases. To overcome these shortcomings, we will introduce an alternative nonmonotonic semantics for disjunctive causal rules, called a covering semantics, that permits an inclusive interpretation of indeterminate causal information. Still, it will be shown that even in this case there exists a systematic procedure, that we will call a normalization, that allows us to capture precisely the covering semantics using only singular causal rules. This normalization procedure can be viewed as a kind of nonmonotonic completion, and it generalizes established ways of representing indeterminate effects in current theories of action.

We consider a simple language for writing causal action theories, and postulate several properties for the state transition models of these theories. We then consider some possible embeddings of these causal action theories in some other action formalisms, and their implementations in logic programs with answer set semantics. In particular, we propose to consider what we call permissible translations from these causal action theories to logic programs. We identify two sets of properties, and prove that for each set, there is only one permissible translation, under strong equivalence, that can satisfy all properties in the set. As it turns out, for one set, the unique permissible translation is essentially the same as Balduccini and Gelfond's translation from Gelfond and Lifschitz's action language B to logic programs. For the other, it is essentially the same as Lifschitz and Turner's translation from the action language C to logic programs. This work provides a new perspective on understanding, evaluating and comparing action languages by using sets of properties instead of examples. It will be interesting to see if other action languages can be similarly characterized, and whether new action formalisms can be defined using different sets of properties.

We provide a logical representation of Pearl's structural causal models in the causal calculus of McCain and Turner (1997) and its first-order generalization by Lifschitz. It will be shown that, under this representation, the nonmonotonic semantics of the causal calculus describes precisely the solutions of the structural equations (the causal worlds of the causal model), while the causal logic from Bochman (2004) is adequate for describing the behavior of causal models under interventions (forming submodels).