Anticipations as Abductions in Human and Machine Cognition Deep Learning: Locked and Unlocked Capacities

Abstract

In my opinion, it is only in the framework of a research dealing with abductive cognition that we can analyze important cognitive aspects of human and machine capacities. From the point of view of human (and of some animal, mammals for example) capacities the phenomenological concept of anticipation (seen as a kind of abduction), which is related to the problem of the spontaneous generation of spatiality and its three-dimensionality, will be central. I will describe that anticipations can be seen as types of visual and manipulative abduction and also fruitful to illustrate, in the case of human and machine capacities, the respective role of two kinds of strategic reasoning: locked and unlocked abductive strategies, which characterize the basic cognitive pro- cedure of “reading ahead”. The specificity of these contrasting inferential strategies is also related to their potentiality in producing different kinds of hypothetical outcomes, which in turn represent dissimilar levels of knowledge creativity. This diversity is also fundamental to depict the special character, the kind of creativity (often amazing), and the limits of current computational deep learning AI systems, such as AlphaGo, which realize abductive cognitive processes.