Cognitive signal processing: An emerging technology for the prediction of behavior of complex human/machine systems

One of the themes of central interest, to the public and private sectors alike, is the prediction of the behavior of complex human/animal/machine systems. Typical examples of such systems are the national economy, education systems, transportation systems, etc., as well as individual humans and animals viewed anatomically and physiologically as complex systems. This lecture presented a new approach to the prediction of behavior of complex human/animal/machine systems based on cognitive signal processing. By a "cognitive signal" we mean a behavioral variable expressed as a waveform or a time series associated with a complex system which is partially or wholly driven by human and/or animal cognitive/perceptual skills. The new approach emphasizes two points: (1) from strictly abstract space considerations pioneered by the speaker, the engine that performs the cognitive signal must be neuromorphic. For background information on this viewpoint, please refer to R. J. P. de Figueiredo, "Beyond Volterra and Wiener: Optimal Modeling of Nonlinear Dynamical systems in a Neural Space for Applications in Computational Intelligence", in "Computational Intelligence: The Experts Speak", edited by Charles Robinson and David Fogel, volume commemorative of the 2002 World Congress on Computational Intelligence, IEEE and John Wiley & Sons, 2003. (2) In general, the point-by-point prediction to be made must be fast so that one may accurately insert in the signal behavior the features of the cognitive mechanism that controls its generation. In order to dramatically emphasize the above considerations for a non-technical reader, we point to the example of a cat pursuing a mouse. In the complex system consisting of cat, mouse, and environment, the trajectories of the cat and of the mouse are, according to our definition, cognitive signals. Note that they cannot be predicted by means of Newton's laws alone, (a) The cognitive processes going through the sensing and cortical systems of the two animals must be taken into account; (b) the prediction algorithm must be at least as fast as the mechanisms in the brains of the cat and mouse controlling their movements; (c) to achieve the desired prediction, it is not necessary, as some neuroscientists are attempting to do, to replicate, in hardware/software, the structures of those animals' brains, for the same reason that, in the design of an airplane, it does not make sense to replicate the structure of an eagle; (d) finally, the present formulation of the cognitive signal processing problem is not related to those currently being pursued by the cognitive radio community.