Scope & Description
Knowledge representation is a lively and well-established field of AI, where knowledge and belief is represented declaratively and suitable for machine processing. It is often claimed that this declarative nature makes knowledge representation cognitively more adequate than e.g. sub-symbolic approaches, such as machine learning. This cognitive adequacy has important ramifications for the explainability of approaches in knowledge representation, which on its turn is essential for the trustworthiness of these approaches. However, exactly how cognitive adequacy is ensured has been often left implicit, and connections with cognitive science and psychology have only recently been taken up.
The goal of this workshop is to bring together experts from fields including artificial intelligence, psychology, cognitive science and philosophy to discuss important questions related to cognitive aspects of knowledge representation, such as:
- How can we study the cognitive adequacy of approaches in AI?
- Are declarative approaches cognitively more adequate than other approaches in AI?
- What is the connection between cognitive adequacy and explanatory potential?
- How to develop benchmarks for studying cognitive aspects of AI?
- Which results from psychology are relevant for AI?
- What is the role of the normative-descriptive distinction in current developments in AI?
IJCAI 22 WorkshopThis workshop is part of the IJCAI 22 workshop programme.
May 13, 2022: Paper Due Date June 3, 2022: Notification of Paper Acceptance June 17, 2022: Camera-ready papers due
- July 23, 2022: Workshop
Ruth Byrne (Professor of Cognitive Science, Trinity College Dublin) University of Dublin will deliver the following keynote talk at the CAKR workshop:
Knowledge representation: Evidence from the cognitive science of counterfactual reasoning [slides]. Abstract: I discuss several discoveries about the cognitive processes that underlie human knowledge representation that are relevant to Artificial Intelligence models. I rely on cognitive science research on counterfactual reasoning to illustrate several key points. First, I outline evidence that people construct iconic mental simulations of possibilities, they compare multiple such representations, and in doing so, they must track the epistemic status of representations, a requirement currently neglected in many models. Next, I describe evidence that the representations people construct are iconic but nonetheless can include symbolic information when necessary. Moreover, people combine knowledge of different sorts in their mental representations, and experimental evidence indicates that they prioritize knowledge of background conditions above knowledge of presuppositions. Finally, I consider the implications of the findings that people simulate counterfactual possibilities to verify facts, and represent impossibilities as if they were possible. I suggest that these recent discoveries in understanding knowledge representation in human counterfactual reasoning provide a rich source for future developments in Artificial Intelligence.
Schedule (July 23)
|09:30-10:30||Keynote: Ruth M.J. Byrne: Knowledge representation: Evidence from the cognitive science of counterfactual reasoning.|
|11:00-11:30||Greta Warren, Mark T. Keane and Ruth M.J. Byrne Features of Explainability: How users understand counterfactual and causal explanations for categorical and continuous features in XAI.|
|11:30-12:00||Claudia Schon Associative Reasoning for Commonsense Knowledge.|
|12:00-12:30||Meghna Bhadra and Steffen Hölldobler Identification of Noise Variables in Singular Decisions using Counterfactual Reasoning.|
|14:00-14:30||Stefania Costantini, Andrea Formisano and Valentina Pitoni Cognitive Aspects in Epistemic Logic L-DINF (remote talk).|
|14:30-15:00||Mohan Sridharan Cognitive Adequacy: Insights from Developing Robot Architectures (remote talk).|
|15:30-16:00||Animesh Nighojkar, Anna Khlyzova and John Licato Cognitive Modeling of Semantic Fluency Using Transformers.|
|16:00-16:30||Jorge Fernandez Davila, Dominique Longin, Emiliano Lorini and Frédéric Maris A Simple Framework for Cognitive Planning|
|16:30-17:00||Umberto Grandi, Emiliano Lorini, Timothy Parker and Rachid Alami Logic-Based Ethical Planning|
|17:00-17:20||Michael Giancola, Selmer Bringsjord and Naveen Sundar Govindarajulu Novel Intensional Defeasible Reasoning for AI: Is it Cognitively Adequate? (remote talk)|
- Clayton Baker, University of Cape Town and CAIR, South-Africa
- Giovanni Casini, ISTI - CNR , Italy
- Federico Cerutti, University of Brescia, Italy
- Marcos Cramer, Technische Universität Dresden, Germany
- Steffen Hölldobler, Technische Universität Dresden, Germany
- Beishui Liao, Zhejiang University, China
- Michael Maher, Reasoning Research Institute, Australia
- Mohan Sridharan, University of Birmingham, UK
- Frieder Stolzenburg, Harz University of Applied Sciences, Germany
- Ivan Varzinczak, University of Artois and CNRS, France
- Jesse Heyninck Open Universiteit, the Netherlands
- Gabriele Kern-Isberner Technische Universität Dortmund, Germany
- Tommie Meyer University of Cape Town and CAIR, South-Africa
- Marco Ragni Technische Universität Chemnitz, Germany
- Matthias Thimm FernUniversität Hagen, Germany