Toward systemic neural network modeling

Summary form only given, as follows. In recent years, the literature on neural network theory has evolved around the development of mathematical models of cortical activities. Almost all these models assume that a biological brain behaves as a layered network comprising a massive number of processing units connected to one another with modifiable strength. While the value of such efforts in neural network research cannot be denied, it is wise to question the systematic directions of the assumption and its evolving dogma. Does the ability of a brain to recognize patterns or images arise from the massive complexity of the cerebral cortex alone? Could one actually design a brain-like computing device based on the currently prevailing assumptions? The answer to both questions is ´highly unlikely´; the justification is elaborated. The deficiencies of the layered neural network models have given rise to the notion of ´systemic neural network modeling´ proposed in this study. This motion is discussed in the context of auditory systems, although it applies to all sensory systems.<>