Catastrophic Phase Transitions in Exact ART Networks
To study the occurrence of sudden transitions in code development, bifurcation analyses of ART networks were carried out In the interest of biological plausibility, we attempted to implement each ART network completely, including all regulatory and logical functions, as a system of differential equations capable of stand-alone running in real time. In particular, transient network behaviour thus remains intact because no asymptotic approximations were used. The most important functions of Exact ART are emergent properties of the network. Preliminary results of bifurcation analyses are presented. In closing, alternative connectionistic analyses of phase transitions are criticized and it is concluded that these analyses fall short on several accounts compared to Exact ART networks.
KeywordsManifold Mellon Subsys
Unable to display preview. Download preview PDF.
- 5.Thorn, R. (1975). Structural stability and morphogenesis. Reading, MA: Benjamin.Google Scholar
- 11.Grossberg, S. (1973). Contour Enhancement, Short Term Memory, and Constancies in reverberating neural networks. Studies in Applied Mathematics, vol LII, no. 3.Google Scholar
- 12.McClelland, J. L., Jenkins, E. (1991). Nature, Nurture, and Connections: Implications of Connectionist Models for Cognitive Development. In: Architectures for Intelligence: The Twenty-second Carnegie Mellon Symposion on Cognition, (pp. 41–73). Lawrence Erlbaum Ass., Inc., Hillsdale, New York.Google Scholar
- 13.Raijmakers, M.E.J., Molenaar, P.C.M. (1993): A continuous alternative for ART2. Internal Report, Department of Developmental Psychology, University of Amsterdam.Google Scholar