Neural Memories and Oscillatory Operation

This paper develops two themes. The first is to show that the task of ‘recognition’ can be realized by an artificial neural net, and that the structures thus generated can be in close analogy to those observed in nature. The second is to note that absolute signal levels are almost meaningless in biological contexts; for these, the current state of a system is revealed by dynamic rather than static behaviour. This was Walter Freeman's observation, which led him to the oscillatory models which he has developed so notably; see the reference list. Coupling of the two sets of ideas leads to some striking conclusions, developed in Whittle (Neural Nets and Chaotic Carriers (1998) and Neural Nets and Chaotic Carriers, 2nd edn. (2010). Imperial College Press, London).