Pattern Recognition Based on Holonic Information Dynamics: Towards Synergetic Computers

One of the most striking features of biosystems, their ability to percept and generate information is in contrast to the properties of conventional machines which only transform information. Conventional pattern recognition machines identify and classify the input information only. Since their ability is limited to strongly restricted situations, they have only deductive but not inductive capacity. This is not the case for biosystems. Biosystems have a high ability of cognition of unknown events. Animals have survived in the long history of biological evolution under severe circumstances, aquiring the ability to see immediately whether an object is food, enemy or indifferent (neither food nor enemy). This is also true even if the object to be proved is new as is frequently seen in the wild world. WITTGENSTEIN pointed out that in cognitive processes input information is interpreted in terms of stored information, “Vorverstandniss” [1]. A similar mechanism has been proposed by MARR for pattern recognition in his book “Vision” [2]. However, no clear explanation has been given to the mechanism of the self-interpretation of input information.