Purkinje local circuits with delays: mathematical conditions of stability for learning and retrieval

Mathematical conditions of stability of learning and retrieval by a Purkinje local circuit, the Purkinje unit, are investigated in the case of delay between any two neurons. The model used takes into account anatomical and physiological constraints: (i) real connectivity; (ii) specific activatory and inhibitory synaptic properties; and (iii) anatomical hierarchical structure. The Purkinje unit is defined in terms of the topology and the geometry of the network, i.e. the anatomical connectivity and the distance between given granule cells and a Purkinje cell. The neurons are assumed to be linear. The network of Purkinje units is general with regard to the number of elements, cells and synapses. For this linear model of a Purkinje unit with delay, we have obtained the conditions of stability for learning and retrieval in two cases: (i) for a single Purkinje unit, the condition is an inequality between the synaptic efficacies which occur in the granule cells-Golgi cell loop; and (ii) for a two-unit system, i.e. two coupled Purkinje units, a strong condition independent of the delay is obtained. This condition includes the granule cell-Golgi cell loop of the two units. We show that the condition of stability for the two-unit system implies the stability of each of the units. This work allows us to define the Purkinje unit in terms of the stability of its function: the physiological process of learning and retrieving must be stable within the anatomical structure that supports this physiological process. It is thus shown that a parameter of a biological nature, the ratio r between the delays of propagation from one unit to another and inside the granule cell-Golgi cell loop, governs the behaviour of the two-unit system. Another result, obtained in the framework of our theory of the functional organization [Chauvet, G. A. (1993). Phil. Trans. R. Soc. Lond. B, 339, 425-444], shows that the association of two unstable units provides a stable unit for certain values of the delay. These results constitute a basis for an eventual interpretation of the coordination of movement by means of a network of non-linear Purkinje units.

[1]  Khashayar Pakdaman,et al.  Interneural delay modification synchronizes biologically plausible neural networks , 1994, Neural Networks.

[2]  Reza Shadmehr.  Learning Virtual Equilibrium Trajectories for Control of a Robot Arm , 1990, Neural Computation.

[3]  Richard E. Plant,et al.  A FitzHugh Differential-Difference Equation Modeling Recurrent Neural Feedback , 1981 .

[4]  Dean V. Buonomano,et al.  Neural Network Model of the Cerebellum: Temporal Discrimination and the Timing of Motor Responses , 1999, Neural Computation.

[5]  R. F. Thompson,et al.  Neural mechanisms of classical conditioning in mammals. , 1990, Philosophical transactions of the Royal Society of London. Series B, Biological sciences.

[6]  P. G. Ciarlet,et al.  Introduction a l'analyse numerique matricielle et a l'optimisation , 1984 .

[7]  V. Braitenberg.  The cerebellum revisited , 1983 .

[8]  V. Braitenberg.  The cerebellar network: attempt at a formalization of its structure , 1993 .

[9]  M. Kawato,et al.  The cerebellum and VOR/OKR learning models , 1992, Trends in Neurosciences.

[10]  D. DeAngelis,et al.  Positive Feedback in Natural Systems , 1986 .

[11]  G A Chauvet,et al.  On associative motor learning by the cerebellar cortex: from Purkinje unit to network with variational learning rules. , 1995, Mathematical biosciences.

[12]  Richard S. Sutton,et al.  Neural networks for control , 1990 .

[13]  A. L. Leiner,et al.  Cognitive and language functions of the human cerebellum , 1993, Trends in Neurosciences.

[14]  K. Gopalsamy,et al.  Stability in asymmetric Hopfield nets with transmission delays , 1994 .

[15]  Leon O. Chua,et al.  Cellular neural networks with non-linear and delay-type template elements and non-uniform grids , 1992, Int. J. Circuit Theory Appl..

[16]  W T Thach,et al.  The cerebellum and the adaptive coordination of movement. , 1992, Annual review of neuroscience.

[17]  Pierre Chauvet,et al.  Mathematical conditions for adaptive control in Marr's model of the sensorimotor system , 1995, Neural Networks.

[18]  M. Paulin.  The role of the cerebellum in motor control and perception. , 1993, Brain, behavior and evolution.

[19]  D. Marr.  A theory of cerebellar cortex , 1969, The Journal of physiology.

[20]  T Tyrrell,et al.  Cerebellar cortex: its simulation and the relevance of Marr's theory. , 1992, Philosophical transactions of the Royal Society of London. Series B, Biological sciences.

[21]  W. I. Card.  Lecture notes in biomathematics. Volume 11: Mathematical models in medicine , 1977 .

[22]  Carlos Lourenço,et al.  Control of Chaos in Networks with Delay: A Model for Synchronization of Cortical Tissue , 1994, Neural Computation.

[23]  V. Braitenberg,et al.  Morphological observations on the cerebellar cortex , 1958, The Journal of comparative neurology.

[24]  C. Darlot.  The cerebellum as a predictor of neural messages—I. The stable estimator hypothesis , 1993, Neuroscience.

[25]  Gisela Håkansson.  Some quantitative aspects of Teacher Talk , 1982 .

[26]  Richard S. Sutton,et al.  Some New Directions for Adaptive Control Theory in Robotics , 1995 .

[27]  Amir F. Atiya,et al.  How delays affect neural dynamics and learning , 1994, IEEE Trans. Neural Networks.

[28]  François Chapeau-Blondeau,et al.  Stable, oscillatory, and chaotic regimes in the dynamics of small neural networks with delay , 1992, Neural Networks.

[29]  C. Darlot,et al.  The cerebellum as a predictor of neural messages—II. Role in motor control and motion sickness , 1993, Neuroscience.

[30]  F. Robinson,et al.  Role of the cerebellum in movement control and adaptation , 1995, Current Opinion in Neurobiology.

[31]  Richard S. Sutton,et al.  An Adaptive Sensorimotor Network Inspired by the Anatomy and Physiology of the Cerebellum , 1995 .

[32]  G A Chauvet,et al.  Hierarchical functional organization of formal biological systems: a dynamical approach. I. The increase of complexity by self-association increases the domain of stability of a biological system. , 1993, Philosophical transactions of the Royal Society of London. Series B, Biological sciences.

[33]  F. Attneave,et al.  The Organization of Behavior: A Neuropsychological Theory , 1949 .

[34]  William G. Faris,et al.  Reliable evaluation of neural networks , 1991, Neural Networks.

[35]  R. Westervelt,et al.  Dynamics of iterated-map neural networks. , 1989, Physical review. A, General physics.

[36]  Pierre Chauvet,et al.  The purkinje unit of the cerebellum as a model of a stable neural network , 1993, ESANN.

[37]  B. Womack,et al.  Adaptive Control Using Neural Networks , 1991, 1991 American Control Conference.

[38]  Masao Ito.  The Cerebellum And Neural Control , 1984 .

[39]  Kumpati S. Narendra,et al.  Adaptive control using neural networks , 1990 .

[40]  J. Albus.  A Theory of Cerebellar Function , 1971 .

[41]  J. Bélair.  Stability in a model of a delayed neural network , 1993 .

[42]  A. M. Uttley.  A two-pathway informon theory of conditioning and adaptive pattern recognition , 1976, Brain Research.

[43]  V. Braitenberg,et al.  Some Quantitative Aspects of Cerebellar Anatomy as a Guide to Speculation on Cerebellar Functions , 1984 .