Abstract: The talk will be comprised of two parts.

In part I, a notion of DEL-based state machines will be introduced and situated. The key notion is that of “transition rules”. A set of transition rules picks a next transition (update) as a function of the formulas satisfied at a given state, hereby specifying the subsequent state(s). Hence defining a DEL-based state machine is not done by directly providing a state transition function from states to states, but rather by specifying a set of rules triggered by local features. It will be shown how DEL-based transition systems may mimic traditional types of state machines (finite/infinite, deterministic/indeterministic), and their “tree generation power” is compared to the protocol-based approach known from the DEL literature.

In part II, it will be shown how transition rules as “knowledge-based programs” may be used to define agent types. In this part of the talk, we will move to a specific DEL setting using epistemic plausibility models and action plausibility models with postconditions. Using this setting, it will be shown how DEL-based state machines may be used to define completely specified models of agent reasoning and choice. This is exemplified by a DEL-based state machine that captures the reasoning and choices occuring in an informational cascade-prone environment. As the state machine is formally fully specified, the construction allows for rigid proofs pertaining to informational cascade dynamics.