Dynamically stable associative learning neural system

An architecture of a neural network (200) comprising:

• a plurality of primary signal inputs (215) and a plurality of primary signal outputs (245);
• a plurality of unconditioned-signal means (110) for receiving input signals from unconditioned stimuli and a plurality of conditioned-signal means (100) for receiving input signals from conditioned stimuli, said means (100) and (110) interconnecting said primary signal inputs and primary signal outputs, said conditioned-signal means also including means for varying the weight of the input signals from the conditioned stimuli propagated therethrough, an unconditioned-signal means receiving an unconditioned signal from one and only one other unconditioned-signal means, the means for varying the weight of the signals from the conditioned stimuli increasing the weight along a conditioned-signal pathway which extends from a primary signal input to a primary signal output if, and only if a signal-carrying connection pathway carries a signal during a present time period for both the conditioned and the unconditioned signals and a preceding time period, and wherein means for varying the weight of the input signals from the conditioned stimuli retains the weight along a conditioned-signal pathway only if a conditioned-signal pathway carries a signal during a present time period and a preceding time period and the unconditioned-signal pathway has not carded a signal.