Fatigue in a Dynamic Neural Network

The addition of neural fatigue, due to adaptive membrane currents, can greatly expand the computational repertoire of recurrent (Hopfield-type) networks to include the rich, dynamical behavior of real biological neu-ronal networks such as the production of rythmic oscillations, central pattern generators, and the storage and retreival of temporal sequences of memories. Numerical simulations of a simple model of a recurrent network with neural fatigue, described by a lattice of coupled, 2-D nonlinear maps, shows that temporal sequences of memories can be reliably stored and retrieved using only fixed asymmetric weights and that memories with overlapping representations can be sequentially retrieved in a process resembling “free association”.

[1]  G. Shepherd.  The Synaptic Organization of the Brain , 1979 .

[2]  Tom Heskes,et al.  Retrieval of pattern sequences at variable speeds in a neural network with delays , 1992, Neural Networks.

[3]  G. W. Strong,et al.  A solution to the tag-assignment problem for neural networks , 1989, Behavioral and Brain Sciences.

[4]  Horn,et al.  Neural networks with dynamical thresholds. , 1989, Physical review. A, General physics.

[5]  L. Abbott.  A network of oscillators , 1990 .

[6]  R. Palmer,et al.  Introduction to the theory of neural computation , 1994, The advanced book program.

[7]  D Kleinfeld,et al.  Sequential state generation by model neural networks. , 1986, Proceedings of the National Academy of Sciences of the United States of America.

[8]  Kanter,et al.  Temporal association in asymmetric neural networks. , 1986, Physical review letters.

[9]  J J Hopfield,et al.  Neural networks and physical systems with emergent collective computational abilities. , 1982, Proceedings of the National Academy of Sciences of the United States of America.

[10]  Edward E. Smith,et al.  Categories and concepts , 1984 .

[11]  Eytan Domany,et al.  Models of Neural Networks I , 1991 .