Image Processing Substrate to Assist Cognitive Models Interact with Dynamic Environments

Abstract

Cognitive models have typically dealt with artificial environments or real environments that are simple. This is because the cognitive models either use indirect approaches to interact with environments, or in cases where they adopt direct approaches to interact, the image processing substrate is incapable of dealing with complex interfaces. However, it is imperative for cognitive models to interact directly with complex environments in order to ascertain the reliability of the underlying cognition theory. The image processing substrate proposed in this thesis overcomes the above-mentioned limitations and enables cognitive models to interact directly with complex environments. This is due to the functionality provided by the substrate that facilitates representation and identification of complex visual patterns. As part of the research work for this thesis, the substrate has been customized to process two interfaces and a cognitive model has also been built on the ACT-R cognitive architecture that uses the proposed substrate to control a driving game environment.