Robots are being designed nowadays in various shapes and forms, in varied intelligence, capabilities and abilities, yet, with just one goal in mind- to closely mimic human beings as far as possible in order to deem it fit to be called as the "true peers" of the future human race (Chatterjee, 2012c). The structural and functional design aspects have altogether become exceedingly important with the aim of making such artificial creatures- or, rather entities, look-alike human beings as well, match in functional capabilities in order to nurture a close bond with its human counterpart. It is impertinent to say that we are not imitating a machine to mimic human behavior, indeed we are; since, human being is a machine in biological form decoded by genes and programmed by neural codes. But there is something peculiar about this natural machine; it is a self-organizing cognitive system having the capacity to think autonomously and have subjective states of feeling, the latter however, is unique to human beings and other life forms. So, it has a "life". In fact, the true endeavor is to create such "life- like" convincing interactive partners capable of responding to external social contexts. The basis for such evolution of imitative behavior in machines stems from the technical design aspects as interfaces that should have both embodied mental attributes as well; they should be able to abstract mentality from embodiment. In as much as these sentient artificial agents should be indistinguishable from a human being. Such a responsive artificial agent is deemed to be "conscious".
So, which is the best cognitive architecture for machine consciousness? Looking around the other way, what should be the nature of an ideal technological framework for artificial cognition and consciousness? It is beyond ambiguity that human ability is unique in this world. The very same architecture of awareness which makes it possible for a human being to achieve marvels also makes it possible for us to attempt to build systems which can operate autonomously and think in a way we do. Simply put, a 'humanoid robot' would be an ideal emulator of human activities, in thought and action. To achieve such levels of functional capabilities, it is essential to understand the nature of human cognitive architecture by decomposition of the concept of consciousness into its cognitive parts.
Cognitive Architecture: General Historical Framework
By cognition, we mean according to the definition given by Premack (1986), "The construction of mental representations on which one makes computations." The computational theories of the mind as described by Fodor's (1987) classic work "Psychosemantics" touches some key aspects on the nature of mental representations and the hierarchy of control processes comparing the human nervous system with a computer. Such a proposal for cognitive architecture of a humanoid robot (Peter and Helm (2012)) as outlined previously by Jerry Fodor (1981, 1987), Fodor and Pylyshyn (1998) discusses about Fodor's ideas having a combinatorial syntax and compositional semantics which compares functional architecture of a computer with reference to human brain.
However, prior to Fodor, Von Neumann (1958) mentioned about such a cognitive architecture in his work "The computer and the brain." The primary point is, it is better to recognize first what is required and then model something after. To achieve such design complexity similar to our brain for virtual machine functionalism, we have a very few choice at hand (Gelder, 1998), except that, to know more about the symbolic aspect of mental representation and the functional architecture of a cognitive computer, to which, we may append the unraveling of our own neural architecture to provide us with enough flexibility to implement similar cognitive architecture within 'this' framework. By "this" framework, we might refer to a hybrid brain-computer cognitive architecture based on layered, hierarchical behavior-specific modules (C. Burghart et. al., 2005). Such an approach to machine (artificial) cognition should be based on three core aspects; representational (Fodor, 1981), connectionist and dynamical systems approach (Fodor and Pylyshyn, 1988). The evolution of a functional artificial mind from a biologically inspired model (our own brain) is hence, an admirable choice, since, the human brain employ special forms of processing those which are still difficult to simulate virtually. To increase the capacity and speed of artificial perceptual processes, functionalists though employ ever-demanding novel tools to bring in efficiency in machine perception; such a self-organizing cognitive architecture similar to humans has always been a fascination for the emulators.
Limitations: Perception & Experience
However, it shall be born in mind that similar to any other technology, cognitive architectures generally have certain limitations, and by sustained innovative processes, the functional capacity of the architecture can be enhanced substantially, which is important for machine perception and experiential capacity. Experiential capacity depends much on the architecture and design of the network for cognition and consciousness. The more a system experiences, some involuntary errors of experiences are expected to be encountered. It is also pertinent to note that the processes of perception require minimal architectural complexity when compared to experience or information processing, hence, organizational complexity does not define the processes of perception, but sensation and experience do require organizational complexity. Design processes define experiential capacity and any deficiency or faults in design greatly increase the amount of error that a functional system produces. Fortunately, our brain although being the most complex system ever known, is remarkably well organized which filters out patterns from randomness by the processes of information processing from perception and by experience. Our brain hence has become the ultimate model of choice for designers to emulate something similar in robots. Yet indeed, cognitive architectures play vital role in casting the blueprint for intelligent agents (Duch, Oentaryo, and Pasquier, 2007, and Chatterjee, 2011a), systems and conscious robots.