Learning and memory in spin-glass-like immune networks

Abstract

In this paper we study the dynamics of a simple spin-glass-like network, which should model the behaviour of the idotypic network of antibodies. Contrary to what happens in neural networks models, the synaptic strengths here are fixed (i.e. time independent) and the learning which we investigate is induced by the action of an external field. We also study the persistency of this effect (i.e. memory) following the learning of new ‘antigens’ induced by other external fields.