10th International Brazilian Meeting on Cognitive Science
Keynote 3: Simulation and Adaptivity
Presenter: João José Neto – Polytechnic School of Engineering, University of São Paulo
December 7, 2015 – 2:00 PM
Simulation is among the most popular computing techniques, feaqturing a multitude of applications in almost all areas of knowledge. Simulators and their variations turn computers into versatile tools that help their users to mimic behaviors in selectable degrees of fidelity that range from crude approximations to detailed and faithful reproductions, even in real time. In our current state of technology, an ambicious challenging goal remains steady in focus, despite its complexity and low feasibility: replicating the behavior of natural, biological and especially neuronal and cognitive processes. Rule-driven devices are those whose behavior can be fully encoded in a finite set of if..then rules, and for their simplicity, the behavior they describe can be easily interpreted by a computer program such as a simulator. Adaptivity is the feature that enables devices to self-modify the set of rules defining their behavior, without external intervention, just in response to its history. One may turn any rule-driven computing device into a corresponding adaptive one by allownig them to perform self-modifications, and their ease of simulation is then further improved by allowing them to change their own behavior by modifying their set of if..then rules. In order to do so, adaptivity is easily achieved with the help of a couple of functions: one for inserting new rules and another for erasing existing ones. Even under the limitations of current technology, the association of adaptivity to simulation (and to other standard techniques) has shown to be helpful to the search of solutions in the field of complex problems. Standard divide-and-conquer strategies usually help us to separately investigate, understand and explore the multitude of aspects considered of interest, by adequately modeling learning and cognition processes, and by upgrading them with extra features from adaptive methods. Therefore, adaptivity is not proposed here as a “magic” for solving all problems in this area, but it has already proved to be highly promising as an effective tool (when used in conjunction with standard techniques and methods) to make successful incursions into unexplored and hard-to-access fields of knowledge, as it happens in cognitive sciences.