This talk is based on on-going joint work with J. van Benthem and S. Smets.

The main claims of the talk are:

(1) ``Social informational dynamics" is the missing ingredient in the Platonic definition of knowledge (as ``true justified belief").

(2) Dynamic Epistemic Logic (DEL), in its recent ``Belief-Revision-friendly" incarnation, is a useful formal tool that incorporates this missing ingredient, and can thus offer a better account for belief change than more traditional logical approaches, and a better insight into the nature of the various forms of `knowledge' considered by modern epistemologists.

It is often recognized that, in order to understand `knowledge' as a form of belief, one has to first understand the dynamics of belief. It is less often admitted that this task must involve going beyond the traditional Belief Revision theory, in at least three different ways. First, one needs to take into account the ``social" features of justified belief: things like communication, persuasion, belief merge or information merge etc. Secondly, one has to look at higher-level beliefs: beliefs about other beliefs, group beliefs etc. Thirdly, one has to move from a purely propositional setting to a truly ``dynamical" one: the new information does not come only in propositional form, but also in dynamic form. Beliefs are never revised by ``propositions", but by the specific ``doxastic events" through which the learning of new propositions is achieved.

I argue that the above three issues are related. E.g. the ``social" aspect is essential even when dealing with a single agent's beliefs: one agent's dynamic belief revision can be seen as an instance of ``two-agent" belief merge (in the sense of ``preference merge" from Social Choice Theory) of her ``old" and her ``new" information/beliefs/preferences. Dually, a dynamic approach to belief merge (and preference merge) seems more realistic (and gives more insight) than the traditional abstract approach, by investigating how the merging of beliefs can be actually realized via specific forms of communication, persuasion and deliberation.

By incorporating the above-mentioned features, the ``Belief-Revision-friendly" version of DEL can formalize various concepts of knowledge proposed in the philosophical literature, and can make subtle distinctions between different versions of these concepts (e.g. by distinguishing between different variants of the Lehrer-Klein defeasibility theory: knowledge as ``stable" belief versus knowledge as a ``fixed point" of belief revision). I also briefly refer to other on-going developments, such as the extensions of DEL into ``counterfactual dynamics", and their use in understanding Nozick's and Dretske's conceptions on knowledge.