A Preliminary Study On Dynamic Testimonial Logic

We propose a dynamic testimonial logic (DTL) to model communication and belief change among agents with different dispositions to trust each other as information sources. DTL is an extension of the dynamic epistemic logic approach to belief revision (of van Benthem2007), with the addition of sources and trust. It is also in the spirit of the modal logic approach to trust (of Liau2003), with the addition of dynamics for belief change. In the multi-agent framework of DTL, we can represent how communication by an information source leads other agents to revise their beliefs about the world, about the source's beliefs, and about the beliefs of other agents in the source's audience. We can also represent how an agent's uncertainty about whether another agent trusts a source can produce, after communication by the source, uncertainty about what the other agent believes, and how an agent can learn whom a source trusts from the source' communication. To capture these phenomena, we introduce a new class of testimonial models and model transformations, for which we give a complete axiomatization. Finally, we describe an application of DTL in modeling a special case of the phenomenon of information cascade discussed in the economics literature.