Incremental Design Revision in Biologically Inspired Design

ion built in (Price’s [116] “[e]ncapsulate complex behavior within a component” or their use of function labels to simplify the output; Iwasaki et al.’s [78] model fragments), but their simulations and thus verifications produce results for the entire model in one chunk or otherwise, like with [78], are producing output for a specific aspect of themodel. Simulation enables individual verifications to stay focused on one function or one behavior at a time even if the functional decomposition is very large while not requiring that a user scope the verification themselves. This automatic focusing should make results to handle compared to those that produce output for the entire model. Additionally, DESC is a step towards the “[m]ulti-level modelling” mentioned by Price et al. [117] as necessary to achieve future targets for qualitative reasoning. Second, DESC leverages the same causal process representation (state diagrams, a.k.a., behaviors) in SBF* for both reasoning and representing simulation results, whereas other work uses representations for reasoning (e.g., component models or model fragments) that differ from the state-based representations in their simulation results. From the perspective of SBF* modeling, DESC thus does not require modelers to learn a new representation for reasoning since it leverages aspects of the models that are already part of the SBF* models being built–with the exceptions that modelers will need to learn equation syntax and simulation semantics (although the latter should be generally intuitive relative to modeling expectations). Additionally, that the verification results are couched in terms of the alreadymodeled elements (i.e., as issues in state conditions and function provides conditions) should lessen the cognitive burden relative to other systems because users do not need to learn a second representation to interpret the results as they would if one built a component model and then interpreted state-based representation results. DESC also differs from [78, 116] by incorporating functions into the simulation. They impact the simulation process by acting as pointers to subbehaviors rather than just being used in evaluation/output. This empowers the simulator to use existing structural aspects of the model (i.e., the functional decomposition) in its reasoning, which should also, in the