Title:
Epistemic Contextualism and Cognitive Modelling: an Interdisciplinary Inquiry

Speaker:
Yingjin Xu
School of philosophy, Fudan University

Abstract:
Cognitive modelling is branch of Artificial Intelligence which is devoted to build the computable model of certain aspects of human cognitive architecture. A sub-branch of this inquiry is “context modelling”, i.e., an umbrella concept covering varieties of approaches to build a computable system which is expected to behave flexibly in different problem-solving contexts. However, by relying on classical approaches to context modelling, such as John McCarthy and Ramanathan V. Guha’s approach to objectify contexts as some high-order parameters in some form of axiomatic systems, theorists are required to provide a sufficient knowledge about all of the contexts that the system is expected to be adaptive to, and such requirement is both philosophically and computationally demanding. However, AI theorists who are interested in philosophy may expect that some brain-storming epistemological studies of “context” may be illuminating in the case of context modelling. And this is the reason why I want to introduce “contrastivism” into the picture. In a typical contrastivist picture, the notion of “context” is no longer explicated to explain why a knowledge-attributor can behave in a context-sensitive manner, e.g., to raise the “epistemic bar” in some contexts and to lower it in others. Rather, what is made explicit is the ternary relationship among the belief that the agent is considering, the contrastive cue that leads its consideration onto certain tracks, as well as the relevant evidence of the belief. I will argue that all of the three variables and their interactions could be mimicked by the Non-axiomatic Reasoning System (Wang 2006, 2014), and the resulted model is not only practically useful but also philosophically interesting in the sense that it could be used to explain why certain types of skeptic problems could arise in certain contexts.