A Biological Plausible Maturation of an ART Network

We present a numerical bifurcation analysis of a shunting neural network (SNN). Fold bifurcations appear to occur in the plane of parameters that are subject to post-natal maturation: the range and strength of lateral connections. The SNN is implemented as a content addressable memory (CAM) in Exact ART, a complete implementation of Adaptive Resonance Theory. The stability and functionality of Exact ART with the CAM in different dynamic regimes is maintained. Moreover, through a bifurcation, the learning behavior of Exact ART changes from forming local representations to forming distributed representations. Presently, we extend Exact ART by adding biologically plausible evolution equations for activation dependent maturation of lateral connections in a SNN [14]. The resulting model is an epigenetic model of both morphological development and cognitive behavioral development.

[1]  J. Elman.  Learning and development in neural networks: the importance of starting small , 1993, Cognition.

[2]  Gail A. Carpenter,et al.  Distributed Learning, Recognition, and Prediction by ART and ARTMAP Neural Networks , 1997, Neural Networks.

[3]  M. Corner,et al.  Implications of activity dependent neurite outgrowth for neuronal morphology and network development. , 1995, Journal of theoretical biology.

[4]  Stephen Grossberg,et al.  Art 2: Self-Organization Of Stable Category Recognition Codes For Analog Input Patterns , 1988, Other Conferences.

[5]  Peter C. M. Molenaar,et al.  A third source of developmental differences , 1993, Behavior genetics.

[6]  S. Grossberg.  How does a brain build a cognitive code , 1980 .

[7]  Maartje E. J. Raijmakers,et al.  Is the learning paradox resolved? , 1997, Behavioral and Brain Sciences.

[8]  Peter C. M. Molenaar,et al.  Numerical bifurcation analysis of distance-dependent on-center off-surround shunting neural networks , 1996, Biological Cybernetics.

[9]  Peter C. M. Molenaar,et al.  Exact ART: A Complete Implementation of an ART Network , 1997, Neural Networks.

[10]  Peter C. M. Molenaar,et al.  On the impossibility of acquiring more powerful structures: A neglected alternative. , 1986 .

[11]  S. Quartz.  Neural networks, nativism, and the plausibility of constructivism , 1993, Cognition.

[12]  Stephen Grossberg,et al.  Contour Enhancement, Short Term Memory, and Constancies in Reverberating Neural Networks , 1973 .