ArgLab • Reading Group

Andrés Soria Ruiz

Explanatory expressivism

The next session of the reading group on Explainable Artificial Intelligence (XAI) and Inference to the Best Explanation (IBE) will take place this Wednesday, March 9, 4 – 5:30 PM (Lisbon time). In this session there will be no reading and Andrés Soria Ruiz will give the following talk:

I defend that explanatory statements (sentences of the form ‘φ explains ψ’) are expressive, in virtue of the fact that they do not contribute strictly new information. Rather, explanations presuppose all the information that they carry, and their assertive contribution is to package available information in particular ways. More specifically, I propose that ‘φ explains ψ’ is true just in case the truth of φ makes the truth of ψ strictly clearer than the truth of the alternatives to ψ (where “clearer” can be glossed as less surprising, more normal, more expected…). ‘Explanatory expressivism’, as I call the view, has advantages over putative descriptivist semantics for explanation. In particular, I defend that this view offers a better account of the non-monotonicity and contrastive character of explanations.

To join the session on Zoom, click here.

The Argumentation Lab (ArgLab) organizes a reading group on Explainable Artificial Intelligence (XAI) and Inference to the Best Explanation (IBE). This reading group is part of the activities organized by the research group ANTIDOTE. The purpose of this reading group is to explore topics related, but not limited to, explanation in science and in language, methods of inquiry, and its application to artificial intelligence (explainability, interpretability, debugging, etc.). We will start reading classics in the literature on inference to the best explanation and move from there to more contemporary topics in AI and philosophy of science/language.

Each session will be led by a rapporteur, who will introduce the reading and spark discussion. Please write to if you are interested in taking part.