ABSTRACT

Robots that can genuinely feel, perceive, readily associate both sensory and symbolic information with items from memory, and thus experience the world subjectively will be the crowning glory of embodied cognition. Self awareness is easier to observe and has often been the first goal in the quest of conscious agency among researchers. The robots will have to elicit this consciousness in higher behaviours which are richer than navigation or pick-and-place operations. Robots are then confined to be nothing more than smart assistants who lack autonomy, helping in a human initiated project rather than being a genuine critic in human actions. The semi-sentient paradigm which relies on humanoid appearance and voice based communications reduces the robots to neo-Asimovian smart slaves. Since traditional symbolic AI is insufficient to lead to agency similar to human behaviour, connectionist approach and layering of behaviour as in subsumption architecture may be a good starting point. Mirror cognition has been a cornerstone in developing self-awareness in robots.