Pellionisz, A. (1987) Modeling and Theory of Neurons and Network Functions, Proceedings of the AAAS 1987 Annual Meeting, Chicago 14-19 February
(Presentation in Section: Computers in Brain Research: Where are the Frontiers? Organized by John M. Gibson, Charleston, Univ. South Carolina)
Modeling and Theory of Neuron- and Network Functions. A. PELLIONISZ (Dept. of Physiol. & Biophys., New York Univ. Med.Ctr., NY, 10016)
Profound change can be observed today in the role of mathematical abstraction and its quantitative expression by computers in brain research. Neuronal Modeling and Brain Theory had long furnished Experimental Neuroscience with the quantitative power of computers and mathematics, necessary for an abstract understanding of complex neural systems. The auxiliary role of phenomenological modeling of neuron- and network function is, however, giving way to direct attempts at forging a mathematically exact Brain Theory. Similarly to other fields of sciences, conceptually and formally homogeneous theories could Iend experimentation a Ieading edge eg. by their organizing power and direction, and also provide with means for addressing not merely local phenomenological problems, but also global philosophical issues of understanding the brain. In Tensor Network Theory that transpired from decades of computer modeling of neurons and networks, brain function is conceived as geometrical representation within multidimensional brain-spaces that are spun by general coordinates intrinsic to the CNS. Sensorimotor research, as will be shown, is a most effective proving ground for theories emerg-ing from computer modeling. First, the physical character of both the sensoryand motor expressions of external objects can be quantitatively observed (thus theory can stay experimentally proven). Second, neurons and neuronal networks underlying simple reflexes, eg. in gaze or Iimb control, are the simplest and most readily available both for experimentation and computer modeling (and thus theory can stay relevant to real neurons and networks, known in neuroscience). Third, the evolution of robotic systems follows Nature, from purely motor "limbs", via limbs under visual & tactile sensory control, to systems controlled by means of intelligent representation (thus, given that robotics assumes the forrmalism of general coordinates, sensorimotor brain theory can co-evolve with brain-likecomputer-controlled robots; jointly benefitting from the impact).