PeC-HiCA: A Perception Centric Human-in-loop Cognitive Architecture

A layered model of the mammalian cortex is an accepted one. One of the major challenges of using this architecture in real-life use cases is the lack of reliable implementations of the symbolic and cognitive layers. On another parallel stream, the adoption of cognitive and artificial intelligence (AI) solutions in industries has been slow. One of the reasons for this is the fact that the current generation of AI architectures does not sufficiently take care of the perception levels of the human operator. This paper presents an architecture that solves both of the above challenges. We propose a cognitive architecture where a natural-human in loop replaces most of the symbolic and cognitive layers. At the same time, we propose the use of instrumentation to measure the perceptual level of the human in real-time. This enables us to bridge the symbolic non-symbolic gap to some extent as well. Lastly, the perception-centric block also adds a novel attribute to the existing AI architectures. The current generation of AI algorithms does not incorporate the users' perception as an integral part of the training and operation of the algorithm. To our knowledge, this is the first time user perception is an essential part of the algorithm. We call this architecture Perception-Centric Human-in-loop Cognitive Architecture (PeC-HiCA). PeC-HiCA is not only a more implementable cognitive architecture, it will also be more successful in industrial applications.